初创公司尽调
尽调报告 Industrial AI / Industrial Data Platform Late-stage private growth 2026-05-20

Cognite

Cognite:工业 AI 平台,蓝筹能源客户集中,私有估值不透明

Cognite 的工业 AI 落地和头部客户证据都真实存在;但私募市场定价不透明、财务披露不完整,因此更适合观察,而不是立即买入。

封面要素

最近确认估值 01
$1.6B [CV001]
FY2025 收入 02
>$170M [CO020]
已知融资额 03
>=$450M [CV003]
成立时间 04
2016/2017 (conflicting public sources) [CO004]

公司概况

Cognite 是一家工业 AI 和数据平台公司,创立于挪威奥斯陆,目前总部位于亚利桑那州 Tempe。核心平台 Cognite Data Fusion 为资产密集型行业的 OT 与 IT 数据赋予上下文;Cognite Atlas AI 和 Cognite Flows 则把这层基础延伸到工业智能体和一线工作流。Cognite 服务 Saudi Aramco、bp、TotalEnergies、Aker BP 等大型运营商,FY2025 报告收入超过 $170M,背后股东包括 Aker ASA、TCV、Accel 和 Saudi Aramco。

官网
www.cognite.com
成立时间
2016-01-01
创始人
John Markus Lervik, Geir Engdahl
创立地点
Oslo, Norway
总部
Tempe, Arizona, USA
产品
Cognite Data Fusion 工业 DataOps 平台、Cognite Atlas AI 工业智能体工作台,以及面向一线工业工作流的 Cognite Flows 行动层。
客户
石油与天然气、能源、化工、流程制造、制药及相邻工业领域的资产密集型企业。
商业模式
订阅制软件平台,在企业工业数据和 AI 部署之上叠加实施与服务。
阶段
Late-stage private growth
融资情况
公开确认的融资包括 2021 年 TCV 领投、估值 $1.6B 的 $150M 轮次,以及 2024 年 TCV 领投、估值未披露的约 $300M 轮次;已知累计融资至少 $450M。
[CO001, CO005, CO006, CO015, CO016, CO017, CO019, CO020]

执行摘要

主要优势

  • Industrial Knowledge Graph 和 Cognite Data Fusion 在重资产场景里嵌入工作流,形成真实切换成本。
  • Saudi Aramco、bp、TotalEnergies、Aker BP 等蓝筹工业客户验证了企业级相关性。
  • FY2025 收入超过 $170M,ARR 订单增长 36%;到 2026 年 Q1,ARR 同比增速提升至 57%。
  • Atlas AI 和 Flows 把 Cognite 从数据底座推向智能体和动作层工作流。

主要风险

  • 2024 年融资金额有公开引用,但投后估值未披露,当前私募市场估值不透明。
  • 尽管在多元化,油气集中度看起来仍然较高,使 Cognite 暴露于能源资本开支周期。
  • Aker ASA 控股和 Saudi Aramco 战略持股让治理和退出路径更复杂。
  • 毛利率、绝对 ARR、全客群 NRR、客户数和合并现金流可见度仍未披露。
  • AVEVA、AspenTech、Palantir 以及云厂商等既有玩家和超大规模云厂商竞争,可能压低倍数、削弱差异化。

未决问题

  • 2024 年投后估值、投资人优先权和股权结构所有权仍未披露。
  • 绝对 ARR、合并毛利率和全客群 NRR 没有公开数据。
  • 客户数、头部客户集中度和合同期限数据仍不透明。
  • 公开证据无法解决 2016 年与 2017 年创立年份不一致的问题。
  • 未发现公开诉讼或执法历史,但私营公司的法律风险仍是证据缺口。

目录

Chapter 01

01公司概况

1.1 身份与商业模式

Cognite 是一家 B2B 工业 AI 和数据平台公司,服务资产密集型行业,主要包括石油与天然气、 能源和流程制造。它帮助这些企业释放孤岛化运营数据、赋予上下文,并在企业规模部署 AI 驱动的解决方案。 公司平台由两款核心产品打底:Cognite Data Fusion(CDF)是工业 DataOps 平台,负责在异构 OT 与 IT 系统之间创建并维护工业知识图谱;Cognite Atlas AI 是 2024 年 6 月推出的低代码工业智能体工作台, 让非开发者也能在知识图谱之上构建和部署 AI 智能体。2026 年 5 月,公司又加入 Cognite Flows,作为平台的 “行动层”,面向一线员工提供统一的 AI 驱动工作流体验。 Cognite 的商业模式是软件即服务(SaaS):客户为平台访问支付订阅费,之上再叠加专业服务和实施支持。 公司瞄准企业级工业运营商,这类客户通常有复杂的多站点环境。Cognite 将 CDF 定位为数据底座,先把数十年的 运营数据统一起来,AI 智能体和应用才有机会释放价值。截至 FY2025,约 40% 收入来自石油与天然气以外的客户, 说明公司正有意义地分散到制造、化工、制药和可再生能源。公司总部位于 40 E Rio Salado Pkwy, Suite 900, Tempe, AZ 85281;近期已将全球总部从挪威奥斯陆迁至此处,以贴近亚利桑那州的高科技和半导体制造生态。 Cognite 还在 Austin TX、Houston TX、挪威 Lysaker(Oslo)、日本 Tokyo 和印度 Bengaluru 设有区域办公室。 [CO001, CO002, CO004, CO010, CO015, CO016]

Cognite 概览 KPI 表(FY2025 / 2026 年初)
指标数值 / 状态截至日期置信度缺口 / 注意事项
年收入>$170MFY2025(截至 Dec 2025)初步、未经审计;仅为公司披露
ARR 增长(YoY)36%FY2025ARR 预订增长;绝对 ARR 基数未披露
NRR(较新客户)150%FY2025队列定义为「较新客户」;全客户基数 NRR 未说明
累计融资(已知)≥$300M20242021 $150M TCV 已确认;2024 ~$300M TCV 领投轮来自 TCV 新闻页
最近已知估值$1.6BMay 20212024 轮后未披露估值;已过时约 5 年
员工数>800Jan 2026初步数据;2025 增长 +21%;确切人数未说明
MAU 增长26%FY2025平台整体;绝对 MAU 未披露
Atlas AI 客户占比(按 ARR)59%FY2025公司披露;分母和方法未披露
非 O&G 收入占比~40%FY2025公司披露;各行业未逐项拆分
知识图谱数据点>80T 时间序列2025公司披露;48% YoY 增长

收入和 ARR 数据为 Cognite January 2026 新闻稿所称的初步、未经审计数据。估值来自 May 2021 TCV 轮;之后没有融资轮披露估值。融资总额只代表公开确认轮次。

[CO017, CO018, CO019, CO020, CO021, CO022]
FO002: Cognite 商业模式与利益相关方地图

Cognite 的股权结构、平台层和关键客户 / 合作伙伴关系如何互相连接。

[CO001, CO012, CO015, CO024, CO041]

1.2 创立、领导层与治理

Cognite 的起源与挪威企业集团 Aker 深度交织。Cognite 官方 about-us 页面称,公司“自 2016 年以来”一直推进这项使命; 但其 2021 年 5 月新闻稿明确写道“Cognite 成立于 2017 年”,并提到成立以来“四年”。CNBC 2026 年 5 月 Disruptor 50 档案则称,Cognite “2016 年在奥斯陆成立,作为欧洲石油公司 Aker BP 的战略伙伴”。这个 2016 年与 2017 年的时间冲突,在公开记录中重要且尚未解决。创立命题是借工业数据释放来改造重资产行业, Aker BP 同时是孵化方和锚定客户。挪威连续创业者 John Markus Lervik 与前 Google 软件工程师 Geir Engdahl 共同创办公司;后者担任 CTO AI 和联合创始人。两人仍在公司活跃。 2022 年,John Markus Lervik 退居首席战略与发展官后,Girish Rishi 被任命为 CEO。Rishi 也担任董事会主席, 带来美国企业软件经验:他曾是 Blue Yonder(后被 Panasonic 收购的供应链软件公司)CEO,此前也在 Motorola 和 Tyco 担任高管。这次 CEO 更替是一个值得注意的治理事件:创始人交棒职业 CEO,往往意味着公司转向商业化和美国市场扩张; 随后总部迁往亚利桑那也与此一致。Cognite 现任高管包括 Jeff Coulter(CFO)、Chirayu Shah(首席产品官)、 Liat Berger(首席人力资源官)、Paul Grenet(全球现场运营总裁)、Sandy Joung(首席营销官)和 Håkon Bjerke(首席战略官)。区域总裁覆盖美洲(Bill Hendricks)、EMEA(Petteri Vainikka)、日本 (Ryoichi Egawa)、北亚和东南亚(TS Park)以及印度(Guru Ananthanarayanan)。Cognite 员工来自 60 多个国家。董事会监督包括 Aker ASA 关联董事,以及 Jake Reynolds(TCV 普通合伙人,2021 年加入董事会)。 鉴于 Girish Rishi 和 Geir Engdahl 分别承担战略与技术领导职责,公司对两位关键人物的依赖评估为高。 [CO002, CO003, CO004, CO005, CO006, CO007]

领导层与创始人表
人员职务创始人身份背景关键人物依赖
Girish RishiCEO 兼董事会主席否(2022 加入)前 Blue Yonder CEO;曾任 Motorola 和 Tyco 高管高 – 战略与对外负责人
Dr. John Markus Lervik创始人;首席战略与发展官是(联合创始人)挪威连续创业者;CDF 架构师中 – 有战略影响力,不担任运营 CEO
Geir EngdahlCTO AI 兼联合创始人是(联合创始人)前 Google 软件工程师;Atlas AI 技术负责人高 – 负责核心平台与 AI 产品
Jeff Coulter首席财务官未公开披露中 – 财务管理
Chirayu Shah首席产品官未公开披露中 – 产品路线图
Paul Grenet全球现场运营总裁未公开披露中 – 企业销售执行
Sandy Joung首席营销官未公开披露中低 – 品牌与需求生成
Håkon Bjerke(管理层)首席战略官未公开披露中低 – 战略执行
Bill HendricksCognite Americas 总裁未公开披露中 – 美国收入增长
Christine Tønsberg Nilsen(管理层)总法律顾问未公开披露中 – 法务与合规

枚举基于 Cognite 截至 May 2026 的官方高级领导层页面。非创始人高管团队的背景细节未公开披露。关键人物依赖为作者评估。

[CO005, CO006, CO007, CO008, CO009]

1.3 融资、资本与投资者

Cognite 最近一次公开确认的融资里程碑,是 2021 年 5 月来自成长股权机构 TCV 的 $150M 轮次,公司估值 $1.6B,并获得独角兽身份。TCV 的 Jake Reynolds 作为这笔交易的一部分加入董事会。Cognite 2021 年新闻稿称, 当时公司增长迅速,拥有“超过 500 名员工”,并将该轮描述为“欧洲 SaaS 公司最大融资轮之一”。在 2021 年 TCV 轮之前,Accel 已做过一笔更早期的成长股权投资(公开资料未披露金额);2021 年新闻稿称,TCV 补充了“全球风险投资机构 Accel 早前提供的资金”。 TCV 随后又领投了追加投资轮。TCV 新闻稿页面标题为“TCV Leads $300M Investment in Cognite”,而一个 2024 年 6 月 TechCrunch URL(现已返回 404)提到“cognite-raises-300m”。CNBC 2026 年 5 月 Disruptor 50 榜单报告了“$300 million”融资,这可能单指 2024 年轮次,也可能指已披露资本总额。由于 Cognite 是私营公司,2024 年轮后准确投后估值和完整资本历史均未公开。2021 年 5 月的 $1.6B 估值是最后一个公开确认数字。 挪威亿万富豪 Kjell Inge Roekke 的投资载体 Aker ASA 是 Cognite 的大股东,也扮演公司母公司角色。 双方关系由独立运营的关联公司结构管理。2022 年 2 月,Saudi Aramco 从 Aker BP(不同于 Aker ASA 的实体)购买了 Cognite 7.4% 股份,使一个具有战略意义的工业客户成为股权持有人。Saudi Aramco 也是 Cognite 最大客户之一, 这让财务利益与潜在产品路线图影响、客户准入影响发生对齐,值得持续尽调。董事会代表包括 Aker ASA 关联董事、 TCV 的 Jake Reynolds,以及可能存在的 Aramco 代表(公开资料未确认)。 [CO011, CO012, CO013, CO014, CO017, CO018]

利益相关方或投资者图谱
利益相关方类型角色 / 关系已知持股尽调问题
Aker ASA企业母公司 / PE多数股东;提供资本、工业领域资源和 Aker 生态访问多数股权(比例未披露)确认治理独立性;审查董事会构成和否决权
Aker BP原始战略合作伙伴 / 早期投资者锚定客户;孵化 Cognite;2022 将 7.4% 出售给 Aramco剩余持股未知确认当前经济权益及任何优先条款
TCV成长股权投资(领投方)领投 2021 $150M 轮和 2024 ~$300M 轮;董事会席位(Jake Reynolds)重要少数股权确认 2024 轮条款、估值和治理权利
Accel风险投资早期投资者;2021 年以前轮次少数股权(未披露)确认当前持股和董事会代表
Saudi Aramco战略企业投资者February 2022 从 Aker BP 购买 7.4%;同时也是大客户7.4%评估其对产品路线图、排他条款和客户数据访问的影响
Microsoft技术合作伙伴(Azure)连续 3 次 Energy & Resources Partner of Year(2022–2024);Azure 为首选云基础设施None确认 Azure 承诺条款及任何 Microsoft Fabric 数据边界协议

持股比例来自公开公告;这家私人公司的当前完整股权结构未披露。Aker ASA 比例未公开说明。Aker BP 按 Reuters(Feb 2022)报道,将其 7.4% 股权出售给 Saudi Aramco。

[CO011, CO012, CO013, CO014, CO017, CO019]

1.4 规模、指标与地域足迹

Cognite 在 2026 年 1 月新闻稿中将 FY2025 描述为“破纪录的一年”。年收入首次超过 $170M;同期年经常性收入(ARR) 预订额同比增长 36%。新客户净留存率(NRR)达到 150%,显示现有客户扩张强劲。到 2026 年初,公司全球员工数超过 800 人,FY2025 期间增长 21%。这些指标仍是初步且未经审计的数据,Cognite 自己在 2026 年 1 月公告中也明确说明。 Atlas AI 在 FY2025 显著拉动 Cognite 提速:Atlas AI 客户数同比增长近 700%;Atlas AI 被超过半数客户群使用 (按 ARR 计为 59%);2025 年新预订中超过 70% 包含 Atlas AI。平台月活跃用户(MAU)在 2025 年增长 26%。 支撑所有 Cognite 产品的工业知识图谱,到 2025 年底已在上下文中覆盖超过 80 万亿个时间序列数据点,增长 48%; 数据建模实例数翻倍。AI token 使用量同比增长近 500%。 FY2025 约 40% 收入来自石油与天然气以外客户,说明公司正从历史上的 O&G 集中度中实现有意义的分散。具名关键客户包括 bp、Saudi Aramco、TotalEnergies、Aker BP、Cosmo Energy(全部三家日本炼厂)、NOVA Chemicals、B. Braun (医疗技术)和 Idemitsu Kosan。Forrester 2025 年分析师博客记录了多项企业级部署证据:NOVA Chemicals 在一座工厂部署 CDF,并计划扩展到 11 座;TotalEnergies 三年内覆盖 39 个资产;Cosmo Energy 在三座炼厂存储 2500 亿个数据点。 毛利率、烧钱速度和 EBITDA 作为私营公司指标仍未披露,绝对客户数量也未公布。 [CO020, CO021, CO022, CO023, CO024, CO025]

FO003: Cognite FY2025 业绩 KPI

Cognite 在 2026 年 1 月披露的 FY2025 创纪录业绩关键指标(初步、未经审计)。

所有数据均由公司披露,并在 2026 年 1 月 moonshot 新闻稿中标注为初步、未经审计。绝对 ARR 基数未披露。

[CO020, CO021, CO022, CO023, CO025, CO027]

1.5 里程碑与负面事件

Cognite 近十年轨迹可分为三个阶段:创立 / 孵化阶段(2016/2017–2020),公司在 Aker 体系内打造平台,以 Aker BP 为锚定客户;规模化 / 独角兽阶段(2021–2023),TCV 投资、Microsoft 合作和国际客户扩张推动快速增长;美国优先的 AI 阶段(2024–2026),特征是 Atlas AI 发布、CEO 主导美国化转向,以及 Flows 平台扩展。 迄今最重要的治理事件,是 2022 年联合创始人 John Markus Lervik 退位,由 Girish Rishi 接任 CEO。公开传播中虽以正面方式表述, 但创始人向职业 CEO 交接本身带来执行风险和潜在文化震荡。本轮研究审阅的公开资料中,未发现负面事件、监管行动或重大诉讼记录。 UpGuard 针对 Cognite 的供应商风险报告截至 2026 年 5 月仍处于活跃状态,监控网站、电子邮件、网络和品牌等维度的安全态势; 详细评级位于付费墙后,但报告存在以及“开始免费试用以获得完整访问权限”的门槛,说明有独立第三方安全审查,企业客户和投资者应复核。 Cognite 为能源公司运行关键基础设施软件,安全态势因此是重大尽调事项。审阅资料中未发现公开事故、数据泄露或重大宕机;但没有证据不等于不存在, StatusGator 仍维护 Cognite 服务可用性的活跃监控页面。CNBC Disruptor 50 榜单错误地将 Girish Rishi 称为“创始人”—— 他是在 2022 年加入担任 CEO;这一事实错误出现在一个本报告依赖的来源中,需作为尽调卫生问题记录。 [CO002, CO003, CO004, CO005, CO007, CO030]

里程碑表
日期事件类型金额 / 估值 / 状态参与方含义
2016(据 about-us、CNBC)/ 2017(据 2021 新闻稿)公司在奥斯陆成立,是 Aker 拆分项目;Aker BP 是锚定客户和战略合作伙伴创立John Markus Lervik、Geir Engdahl、Aker BP 等创始相关方创立年份冲突未解决;起步时依赖单一客户
2019–2020(约)平台从 Aker BP 向外扩展;bp、Saudi Aramco、Mitsubishi 被列为早期客户规模化bp、Saudi Aramco、Mitsubishi 等伙伴TAM 验证;多客户模式建立
Nov 2020估值达到 ~$500M(暗含:May 2021 新闻稿称三倍增长)融资~$500M 估值(暗含)Aker, AccelTCV 前估值快速增长;融资前动能强
2021-05从 TCV 融资 $150M;以 $1.6B 估值成为独角兽融资$150M / $1.6BTCV(Jake Reynolds 加入董事会), Aker独角兽里程碑;顶级成长股权机构验证;当时欧洲最大 SaaS 融资轮之一
2022-02Saudi Aramco 从 Aker BP 购买 7.4% 股权融资金额未披露Saudi Aramco, Aker BP战略客户成为股东;潜在利益一致 / 冲突
2022Girish Rishi 被任命为 CEO;John Markus Lervik 转任首席战略与发展官治理Girish Rishi、John Markus Lervik 两名高管创始人退居一步;引入美国企业软件领导层;增长重心转向美国
2022–2024Microsoft 连续三年(2022、2023、2024)将 Cognite 评为 Energy and Resources Partner of Year合作Microsoft, Cognite平台生态验证;Azure 为首选云;与 Azure AI 深度集成
2024-06Cognite Atlas AI 发布;低代码工业智能体工作台产品Cognite, Microsoft Azure OpenAI重大产品转向 AI 智能体;随后 18 个月客户采用加速
2024-06TCV 领投 Cognite 约 $300M 融资轮融资~$300M / 估值未披露TCV最大已披露融资轮;重要规模化资本;未发布估值
2025全球总部从挪威奥斯陆迁至亚利桑那州坦佩治理Cognite美国优先的公司身份;贴近半导体和高科技制造
2025-09TotalEnergies 扩大 3 年期企业合作,在全球所有上游资产部署 CDF合作3 年期交易TotalEnergies, Cognite一线能源公司承诺在 39+ 资产上企业级部署
2025Verdantix 在 Industrial AI Analytics 和 Industrial Data Management(两份独立报告)中将 Cognite 评为领导者监管Verdantix2025 第二项独立分析机构领导者认定
2026-01FY2025 创纪录业绩公布:>$170M 收入、36% ARR 增长、>800 员工数、NRR 150%规模化>$170M 收入Cognite首次公开披露收入超过 $170M;确认持续高速增长
2026-03在 IDC MarketScape Worldwide Industrial DataOps Platforms 2026 中被评为领导者监管IDCDataOps 顶级分析机构定位;与 Verdantix 一起形成第三方验证
2026-05Cognite Flows 发布——工业 AI 平台的行动层产品Cognite、Idemitsu、B. Braun、Radix 等参与方平台完整性里程碑;发布时已有 30%+ 客户基础启用
2026-05CNBC Disruptor 50 榜单排名 #15规模化CNBC品牌和市场认可;增长轨迹获得独立验证

里程碑日期来自官方新闻稿、投资者公告和可信第三方报道。创立年份冲突(2016 vs 2017)未解决;见 evidenceGaps。2024 轮财务数字仅来自 TCV 新闻页标题——完整条款未公开披露。

[CO002, CO003, CO004, CO005, CO006, CO007]
FO001: Cognite 公司里程碑时间线(2016–2026)

从创立到 May 2026 的关键里程碑,涵盖融资轮次、领导层变化、产品发布和重大合作。

成立年份在 2016(关于我们页、CNBC)与 2017(2021 年新闻稿)之间存在冲突。2024 年 TCV 轮金额仅来自 TCV 新闻页标题;完整条款未披露。

[CO002, CO003, CO004, CO005, CO007, CO013]

1.6 图表

Chapter 02

02市场分析

2.1 市场边界与定义

Cognite 的主战场是工业 DataOps 与工业 AI 平台细分市场:这类软件平台面向资产密集型行业,将运营技术(OT)与信息技术(IT)数据接入、上下文化并投入运营,以支撑 AI 驱动的工作流自动化和决策支持。该细分属于更宽的工业物联网(IIoT)伞形市场,但边界明显更窄:不包括商品化传感器硬件、通用云存储、SCADA/DCS 控制系统和企业资源计划软件。现状替代方案包括传统历史数据库(尤其是 OSIsoft PI,现为 AVEVA/Schneider Electric 的一部分)、点状分析工具、工业自动化套件内捆绑的 OEM 分析能力(Siemens、GE Vernova、Honeywell、AspenTech),以及基于电子表格的工程工作流。该类别也不同于通用企业 AI 平台(Palantir AIP、C3.ai):其主要焦点是 OT/IT 融合、实时传感器数据和工业知识图谱,而不是业务流程 AI。 Cognite 服务三类主要终端市场垂直:石油与天然气(根据“40% 非 O&G”披露反推,约占 FY2025 收入 60%)、流程制造 / 化工 / 制药,以及电力与可再生能源。这些垂直共同具备资本密集、多站点运营、异构旧数据基础设施,以及通过运营优化获取经济和安全收益的强激励。Cognite 正积极利用的相邻机会包括超大规模云厂商数据平台市场(Databricks、Snowflake)和智能体 AI 工具(Microsoft Copilot Studio、AWS Bedrock);两者既有竞争性,也有互补性。IOGP(International Association of Oil and Gas Producers)是全球组织运营商标准和数字化转型计划的主要行业机构。 “Industrial DataOps”一词在 2020 年后被广泛采用,并在 2026 年初由 IDC 首次发布 MarketScape 评估时,被验证为独立分析师类别。在此之前,相关软件类别曾被称为“工业分析”“运营智能”或“IIoT 平台”,使历史规模比较在方法上不可靠。Cognite 同时面对专业纯软件厂商(Seeq、Aspentech Inmation/Mtell)和超大规模云原生服务(AWS IoT SiteWise、Azure Digital Twins、Google Cloud Manufacturing Data Engine)的竞争;后者以云定价提供商品化数据汇聚。 [CM001, CM002, CM003, CM004, CM005, CM006]

市场定义表
细分 / 类别纳入支出排除支出主要买方付款方与 Cognite 相关性
工业 DataOps 平台OT/IT 数据摄取、知识图谱、AI/ML 部署、语境化软件硬件传感器、SCADA/DCS、ERP、大宗云存储CDO / 数字化 VP / 工程 VP企业 IT/Ops 预算核心产品市场——CDF 和 Atlas AI
工业 AI 智能体与工作台低代码 AI 智能体构建器、面向工业数据的 LLM 编排通用 LLM API、消费者 AI 工具首席 AI 官 / 数字化 VP技术预算Atlas AI 智能体工作台——增长中的细分市场
工业分析与监控流程监控、预测性维护、运营智能软件商业智能、HR 分析、财务报告工具工厂经理 / 运营 VP维护 / Opex 预算相邻市场——Seeq、AspenTech 在此竞争
IIoT 平台(广义)完整 IIoT 堆栈,包括连接、边缘、云、分析消费者 IoT、楼宇自动化、非工业智慧城市CTO / IT 领导层资本项目预算Cognite 在这里是子细分,不是整个市场
资产绩效管理(APM)状态监测、可靠性工程、预测性维护反应式维护人工、备件库存可靠性 / 维护经理维护 Capex/Opex相邻市场——GE Vernova、Aspentech APM 在此竞争
油气数字化转型数字孪生、生产优化、远程监控平台上游勘探软件、地震处理上游运营 VP / CTO上游 IT/OT 预算Cognite 的锚定垂直行业(约 ~60% 收入)

纳入 / 排除支出分类基于 Cognite 产品定位、IDC MarketScape 范围描述(摘录)以及 Forrester 分析师对数字工业平台类别的描述。不同分析机构对市场边界没有标准定义;买方画像标签是代表性原型,并非穷尽。油气收入占比(约 ~60%)由已披露的「~40% 非 O&G」FY2025 收入数据推导。

[CM001, CM002, CM003, CM006, CM007, CM009]

2.2 市场规模:多重视角

公开可获取的分析师报告中,没有一份按 Cognite 定义直接界定“工业 DataOps 平台”或“面向资产密集型行业的工业 AI 平台”的 TAM。IDC MarketScape Worldwide Industrial DataOps Platforms 2026(Doc #US53013025)是最权威参考,但报告被付费墙挡住,只能通过 Cognite 自己的新闻稿看到节选。更宽 IIoT 市场有两个独立来源可作为规模锚点,但预测差异很大。 Grand View Research(GVR)估计,全球 IIoT 市场 2024 年为 $483.16B,到 2030 年将达到 $1,693.44B,2025–2030 年 CAGR 为 23.3%。Precedence Research 估计同一市场 2025 年为 $514.39B,到 2035 年将达到 $2,430.21B,2026–2035 年 CAGR 为 16.8%。两组估计在 2024/2025 基数($483B vs $514B)和 CAGR(23.3% vs 16.8%)上均冲突,反映市场边界定义不同——GVR 可能纳入了更多工业硬件 / 连接收入,而 Precedence 可能对软件 / 服务权重不同。两份报告都没有给工业 DataOps 平台提供专门子细分。 自下而上的规模视角,结合 Cognite 披露指标,指向一个更受约束的可服务市场:Cognite 收入 $170M,可能已经是全球第二或第三大纯软件工业 DataOps 厂商(位于 AVEVA/OSIsoft 之后,后者现为 Schneider Electric 的企业级资产),这意味着纯软件工业 DataOps 年合同价值市场也许只有 $500M–$2B——比 IIoT TAM 小几个数量级。C3.ai SEC 10-K 文件(2025 财年,2025 年 7 月提交)显示 C3.ai 收入约 $252M,为企业工业 AI 规模提供了一个公开类比,并确认该市场真实存在,但收入集中度还没有达到数十亿美元。制造业是最大的 IIoT 终端使用细分(据 Precedence Research,2025 年约 28.7%),超过单独的石油与天然气;Cognite 正积极跳出 O&G 基本盘,这一点相关。 [CM011, CM012, CM013, CM014, CM015, CM016]

TAM / SAM / SOM 规模测算视角表
发布方发布年份地理范围市场 / 细分基准值预测值CAGR方法置信度局限
Grand View Research2024/2025全球工业物联网(宽口径 IIoT)$483.16B (2024)$1,693.44B (2030)23.3% (2025–2030)需求侧调研 + 二手数据口径过宽:包括硬件、连接能力,不只是 DataOps 软件
Precedence Research2025全球工业物联网(宽口径 IIoT)$514.39B (2025)$2,430.21B (2035)16.8% (2026–2035)市场测算 + 一手访谈CAGR 低于 GVR;预测期不同;市场边界可能不同
Grand View Research2024/2025北美IIoT 区域子细分>$150B 估计(2024)~18% (2025–2030)由全球占比 31%+ 推导区域份额为推导值;DataOps 子细分未获独立验证
IDC MarketScapeMarch 2026全球工业 DataOps 平台(窄口径)未披露(付费墙)未披露未披露厂商评估矩阵(能力 + 战略)低(付费墙)付费墙;仅有 Cognite 摘录——无法提取市场规模数字
自下而上(研究估计)May 2026全球纯工业 DataOps 软件(估计)~$500M–$2B ARR 估计~20–30% 估计收入类比:Cognite $170M+,按纯工业 DataOps ARR 可能排第 #2–3无一手来源;仅由可见公开可比公司推导;高度不确定
C3.ai SEC 10-K 文件FY2025(财年 2025 年 7 月)全球企业工业 AI(公开可比对象)~$252M 收入按 GAAP 披露的收入高(SEC 文件)C3.ai 产品组合包括政府 / 国防业务,不是纯工业 DataOps

GVR 和 Precedence Research 的所有 IIoT 市场数字都覆盖宽口径市场,包括硬件、连接能力和边缘计算,并不只对应工业 DataOps 软件子细分。自下而上的估计来自研究团队基于公开收入可比对象的测算,置信度低。IDC MarketScape 的市场规模在付费墙后,也可能不提供整体市场数字。C3.ai 收入来自公开 SEC 文件,列入它是为了提供公开可比对象,不是对工业 DataOps 子细分的估计。

[CM011, CM012, CM013, CM014, CM015, CM016]
FM001: 工业 AI 市场规模金字塔

从宽口径 IIoT TAM 到 Cognite 估计可服务市场的三层市场测算,展示头部 IIoT 数字与实际 Industrial DataOps 软件子板块之间的数量级差距。

SAM 层级为研究团队估计;没有公开分析师来源免费给出 Industrial DataOps 平台市场规模。SOM 基于 Cognite 披露的 FY2025 收入及竞争组的公开收入类比估算。

[CM011, CM012, CM016, CM020, CM041]
FM002: 不同来源的 IIoT 市场 CAGR 估计区间

两家分析机构对宽口径 IIoT 市场给出的 CAGR 估计互相冲突,显示增长预期的分布,以及在可得情况下的区域拆分。

每个数值上下界为估计置信区间(所披露 CAGR 的 ±20%),因为 GVR 与 Precedence Research 都未公布预测的不确定性范围。两个中心值(23.3% vs 16.8%)直接来自来源报告;差异反映口径定义和预测期长度不同。

[CM011, CM012, CM013, CM014, CM044]

2.3 买方、用户与付款方分层

Cognite 客户群高度偏向大型工业企业——这些公司拥有显著的多站点 OT 基础设施、超过 $1B 的资本预算,并已在工业自动化上投入。典型买方是综合油气巨头、大型化工或流程制造企业,或国家能源公司。Cognite 具名客户包括 bp、Saudi Aramco、TotalEnergies、Aker BP、Cosmo Energy、TotalEnergies、NOVA Chemicals、B. Braun 和 Idemitsu Kosan;它们都是运营环境复杂、多站点的大企业,在这些场景里,数据上下文化和 AI 的经济性最强。 工业 DataOps 预算归属位于 IT 与运营的交汇处:首席数字官(CDO)、数字副总裁或工程副总裁通常推动投资,采购 / 财务则控制多年期 SaaS 协议机制。这意味着 Cognite 既要争夺可自由支配的“数字化转型”预算,也要争夺核心运营 IT 支出。采用触发因素通常是三类:(1)高管层基于运营效率目标下达数字化转型要求;(2)新大型资本项目需要现代数据基础设施;(3)ESG 披露要求带来压力,要求提升运营可视性。 落地后扩张是 Cognite 商业模式的核心:NOVA Chemicals 先在一座工厂部署 CDF,再计划年内扩展到 11 座工厂;TotalEnergies 用三年覆盖 39 个上游资产;Cosmo Energy 覆盖全部三家日本炼厂。这个模式意味着初始销售往往是范围有限的试点(一个站点、一个用例),多数收入增长来自扩张。Cognite 新客户 150% NRR 在商业上验证了该模式。用户画像包括数据科学家、流程 / 工厂工程师、IT/OT 集成架构师,以及——借助 Atlas AI——非技术一线员工。付款权通常位于企业级预算,而非单个工厂经理手中,因此企业采购周期很长(初始交易常为 12–24 个月),但多年期合同是常态。 [CM021, CM022, CM023, CM024, CM025, CM026]

细分市场与买方地图
垂直细分主要买方终端用户付款方预算负责人关键工作流采用触发因素
油气(上游)CDO / 数字化副总裁 / 工程副总裁油藏 / 生产工程师、操作员企业 IT / 上游 Opex首席数字官生产优化、远程运营、预测性维护运营效率要求 + 降本压力
油气(下游 / 炼化)运营副总裁 / IT 总监炼厂工艺工程师、可靠性团队炼厂 Opex 预算工程副总裁或 CTO资产可靠性、预测性维护、排放监测减少安全事故 + 合规要求
化工与流程制造工程副总裁 / 数字化负责人工艺工程师、工厂经理Capex + Opex 混合工程副总裁或 COO工艺优化、质量控制、良率提升降本 + ESG 披露要求
制药 / 生命科学制造副总裁 / CDO制造科学家、QA 工程师制造 IT 预算CTO / 制造副总裁批次工艺优化、合规数据管理FDA/EMA 数据完整性要求
电力与可再生能源资产管理副总裁 / CDO电网 / 电厂运营工程师资产管理预算资产管理副总裁资产绩效监测、电网优化、预测性维护能源转型 + 运营效率
矿业运营副总裁 / CDO采矿工程师、安全主管资本项目 + 运营预算CTO 或运营副总裁设备监测、安全合规、能源优化ESG 要求 + 偏远现场运营挑战

买方画像是基于 Cognite 客户案例(TotalEnergies、NOVA Chemicals、Cosmo Energy、Aker BP)、IDC MarketScape 范围,以及 Forrester 分析师对数字工业平台买方的描述抽象出的原型。预算归属会随公司规模和数字化成熟度显著变化;大型企业买方往往有专门的 Digital/CDO 组织,中型市场买方则经由 IT 或运营部门推进。制药和矿业对 Cognite 仍属新兴细分,公开资料记录较少。

[CM021, CM022, CM023, CM024, CM025, CM026]
FM003: 按垂直行业划分的买方—用户—付款方地图

矩阵将工业垂直行业映射到买方画像、终端用户、付款方与主要采用触发因素,展示 Cognite 必须穿越的企业采购复杂度。

预算周期估计是代表性原型,来自 Forrester 分析师对企业工业软件采购的刻画。实际周期会随公司规模、数字化成熟度和所在地司法辖区显著变化。

[CM021, CM022, CM023, CM027, CM028, CM029]

2.4 增长驱动与采用约束

Cognite 市场近期最重要的增长驱动,是生成式 AI 采用浪潮与工业数据集成数十年积压工作的汇合。Cognite Atlas AI 客户在 FY2025 同比增长近 700%,说明 AI 智能体能力已跨过工业买方的商业化门槛;此前这些买方一直等待技术成熟。ESG 与脱碳要求进一步放大了这股 AI 需求:TotalEnergies 等公司明确部署工业数据平台,同时改善健康安全监控、减轻排放跟踪负担,并优化生产。 结构性顺风还包括工业劳动力老龄化和用工短缺:资深工程师退休后,工业运营商面临“知识传递”危机,AI 驱动的知识图谱正对准这一约束。Cognite 对 Atlas AI 的定位——让 AI 智能体执行过去需要数周的根因分析——直接映射到这个痛点。OT 环境的云迁移(历史上比 IT 滞后 10–15 年)正在加速,AWS(IoT SiteWise)、Azure(Digital Twins)和 Siemens(Industrial Edge)等供应商让云连接的工厂和现场运营变得常规化。这种常规化降低了 Cognite 的基础设施摩擦,并扩大可部署基础。 采用约束真实且重大。Forrester 分析师将企业级公司的工业 AI 部署描述为天然“慢、难、贵”,并指出闪亮的 AI 应用只是“冰山上一小块,底下压着一座数据运营基础工程的大山”。切换成本很高:多年期工业 DataOps 实施会整合数百个 OT/IT 数据源,一旦知识图谱建立,就形成很深的供应商锁定。油气和流程制造的预算周期很长——大型 IT/OT 集成项目从批准到投产通常要 18–36 个月。OT 安全与数据主权要求增加合规负担,拉长销售周期,并要求投入专门客户成功资源。来自 Siemens、GE Vernova、Honeywell 和 AspenTech 的既有供应商竞争也会造成切换惯性;这些厂商在自动化套件中捆绑分析能力,且大型客户已在使用其硬件。 [CM031, CM032, CM033, CM034, CM035, CM036]

增长驱动因素与采用约束
驱动因素 / 约束方向时间对 Cognite 的影响尽调问题
生成式 AI 智能体采用浪潮驱动(正向)当前(2024–2026)Atlas AI 客户同比增长 700%;直接拉动收入核实 Atlas AI 增长来自存量客户增购还是新增 logo;新增 logo 增长才是耐久性指标
ESG 与脱碳要求驱动(正向)中期(2025–2028)为排放监测、可持续运营数据创造新增预算评估 ESG 驱动交易是增量预算,还是蚕食现有预算池
工业劳动力老龄化 / 知识传承危机驱动(正向)当前且继续扩大将 AI 智能体定位为专家知识沉淀工具——对保守买方叙事有力验证劳动力挑战是否尖锐到足以推动紧急预算分配,而不是停留在愿景路线图
OT 环境云迁移驱动(正向)持续(2022–2030)让工业数据接入云端变得常态化;降低 Cognite 的集成摩擦跟踪超大规模云厂商的通用产品(AWS IoT SiteWise、Azure)是否会先于 Cognite 拿走这笔预算
监管与合规要求(ISO 27001、SOC 2)驱动 + 约束当前安全认证降低采购摩擦;合规开销拉长销售周期核实 Cognite 所称认证是否全部仍有效;检查审计续期日期
实施复杂度 / 「慢、难、贵」约束(负向)长期存在需要大规模客户成功投入;将可触达市场限制在有耐心且资源充足的企业评估早期客户队列流失;Forrester 分析师将其标为系统性市场约束
既有工业软件厂商竞争约束(负向)长期存在Siemens、GE Vernova、Honeywell、AspenTech 都提供捆绑分析能力;存量账户切换成本高梳理多少目标账户已被既有套件锁定;评估 Cognite 在使用 Siemens PCS 7 或 ABB Ability 的账户中的差异化
企业预算周期长(18–36 个月)约束(负向)长期存在拖慢新增客户获取;让收入确认不均衡;增加预测难度索取按细分市场拆分的交易周期分析;核实是否已有或计划推出周期更短的 SME 项目
OT 安全与数据主权约束约束(负向)长期存在但在缓和在国防相关、政府监管或中国市场部署中尤其尖锐评估数据驻留要求是否阻断过重大交易;询问本地部署支持路线图
超大规模云厂商价格压力(AWS、Azure、Google)约束(负向)增强(2025+)云巨头以低边际成本提供通用工业数据服务;可能将 Cognite 的连接层商品化评估 Cognite 在超大规模云厂商原生服务之上提供差异化价值的策略;合作与替代的动态

时间分类(当前 / 中期 / 长期存在)是分析师基于 Cognite 新闻稿、Forrester 分析师博客(2025)和行业研究作出的估计。「慢、难、贵」这句引语逐字来自分析 Cognite 客户群的 Forrester 博客。竞争对手名称(Siemens、GE Vernova、Honeywell、AspenTech)来自公开产品目录;竞争强度会随细分市场和地域变化。超大规模云厂商价格压力来自对 AWS IoT SiteWise 和 Azure Industrial IoT 产品定位的推断;公开来源没有直接价格对比。

[CM031, CM032, CM033, CM034, CM035, CM036]
FM004: 工业 DataOps 采用漏斗

从初始认知到落地并扩张的端到端企业采用旅程,映射 Cognite 在已记录客户部署中呈现的商业动作。

采用漏斗阶段根据 Forrester 分析师博客(NOVA Chemicals、TotalEnergies、Cosmo Energy 案例研究)、Cognite FY2025 新闻稿与 Atlas AI 发布公告推断。时间线为代表性估计;实际从 POC 到落地的周期取决于账户复杂度和预算就绪度。

[CM024, CM025, CM026, CM027, CM031, CM039]

2.5 尽调缺口与矛盾估计

本章最重要的分析缺口,是缺少一份公开、独立、无付费墙的工业 DataOps 平台子细分估计。IDC 2026 MarketScape(付费墙)、Verdantix 两份 2025 年领导力报告(付费墙)以及 Gartner 相关 Market Guide(研究时被阻挡 / 无法访问)都确认该类别存在且 Cognite 是领导者,但没有任何来源给出可提取的市场规模数字。尽调买方在接受管理层隐含市场叙事前,应购买完整 IDC 和 Verdantix 报告。 现有两组公开估计——GVR(2024 年 $483B,CAGR 23.3%)和 Precedence Research(2025 年 $514B,CAGR 16.8%)——方法不一致,无法合并成一个权威数字。两者都针对宽泛 IIoT 市场,而非特定工业 DataOps;两者也都承认,随着不同出版物对软件和硬件组件计数不同,市场边界仍在演化。这里保留而不调和这一矛盾;除非做大幅调整,否则不应把任何一个估计当作 Cognite SAM 的代理。 Cognite 提出的 Moonshot 目标(到 2035 年为客户累计创造 $100B 价值)是愿景,不是市场规模估计——它反映管理层内部信心,但无法映射到可验证 TAM。工业 DataOps 在石油与天然气这一主导垂直中的市场渗透率,公开资料无法量化:全球约有 50+ 家大型油气运营商具备成为 Cognite 客户的规模,但目前有多少使用任何工业 DataOps 平台,公开来源未知。 [CM041, CM042, CM043, CM044, CM045]

2.6 图表

Chapter 03

03竞争格局

3.1 竞争格局概览

Cognite 所处竞争格局分为多层,没有单一直接孪生对手。主要竞争类别包括:(1)传统 OT 数据历史数据库 和工业软件既有厂商——AVEVA PI System(Schneider Electric)、AspenTech(Emerson)和 Siemens Industrial Edge—— 其深厚装机基础带来高切换摩擦;(2)横向企业 AI 平台——C3.ai、Palantir AIP——争夺大型工业运营商同一笔企业 AI 预算;(3)云超大规模厂商工业服务——AWS IoT SiteWise、Azure Digital Twins——将数据接入和连接层商品化; (4)相邻工业 IoT 厂商——Seeq(分析)、PTC ThingWorx/Velotic(IIoT 连接)、Hexagon(资产生命周期 / 3D 空间智能) 和 GE Vernova(聚焦电力)——在具体能力上重叠;以及(5)现状和内部自建替代方案——OSI PI 历史数据库加人工电子表格分析加孤立 SCADA/DCS 系统,或基于 Snowflake、Databricks 搭建的自定义云数据湖架构。没有单一竞争者复制 CDF 面向工业运营的全栈: 数据接入、知识图谱上下文化和低代码 AI 智能体工作台。因此,竞争威胁是不对称的:既有厂商掌握分销权,超大规模云厂商掌握基础设施杠杆, 横向 AI 平台掌握企业销售关系,而 Cognite 掌握领域深度和 AI 原生架构。 [CP001, CP006, CP009, CP014, CP021, CP025]

竞争对手画像表
竞争对手类别规模 / 融资目标细分市场核心差异化相对 Cognite 的主要局限
AVEVA PI System(Schneider Electric 旗下)OT 数据历史库 / 既有厂商被 Schneider Electric 以 ~$9.5B 收购(2023);SE 收入 €36B+油气、电力、流程工业;所有现有 PI 客户30+ 年装机基础;深度 OT 集成;广泛标签支持遗留架构;AI / 知识图谱有限;锁定效应有利于既有厂商
AspenTech (Emerson)流程仿真 / APM 既有厂商被 Emerson 以 ~$11B 收购(2022);Emerson 收入 ~$15B化工、炼化、发电、工程aspenONE 仿真套件;Mtell APM;与 Emerson OT 硬件捆绑更偏仿真而非运营数据;知识图谱有限;需要 OT 硬件捆绑
Seeq工业分析平台私营;C 轮来自 Energize Ventures + Insight Partners流程工业、炼厂、制药、公用事业;分析买方数据源无关的分析层;可跑在 PI、CDF 或其他历史库之上只有分析;没有数据摄取 / 上下文化堆栈;没有 AI 智能体工作台
PTC ThingWorx / VeloticIIoT 连接与应用PTC 收入 ~$2.1B(FY2024);ThingWorx 正更名为 Velotic车间现场、离散制造、智能产品成熟 IIoT 连接能力;低代码应用构建器;庞大伙伴生态聚焦工厂 OT;油气 / 流程适配有限;更名带来路线图不确定性
C3.ai(NYSE: AI,上市可比)企业 AI 平台(横向)上市公司;年收入 ~$310M(FY2025 估计);SEC 文件显示仍在亏损制造、能源、国防、医疗、金融服务上市公司透明度;广泛 AI 应用目录;国防行业认证横向 AI,不是工业专用;需要客户数据基础设施;收入增长受审视
Palantir AIP企业 AI / 数据运营上市公司;年收入 ~$2.9B(FY2024)大型企业、政府 / 国防、能源;交易规模 $10M+国防级安全;Foundry 数据本体;前线部署工程师重服务实施;非自助式 SaaS;预置工业垂直能力有限
AWS IoT SiteWise云基础设施 – 工业 IoTAWS(Amazon);云收入基数 $107B+工业设备监测;AWS 生态客户托管服务规模;AWS 生态集成;按量计价只有基础设施层;没有知识图谱或 AI 智能体工作台;客户需大量自建
Azure Digital Twins云基础设施 – 数字孪生Microsoft Azure;云收入基数 $135B+智慧建筑、工厂、能源电网;Microsoft 企业客户开放建模语言;Teams / Office 集成;Azure 生态捆绑偏建模 / 仿真而非运营时序数据;没有领域专用工业 AI
Siemens Industrial EdgeOT 边缘计算 / 工厂Siemens 收入 ~€77B;投资 Xcelerator 平台离散与流程制造;Siemens OT 客户边缘原生 OT/IT 集成;Siemens 硬件 + 软件捆绑;工厂自动化深度聚焦工厂 / OT;油气 / 流程知识图谱用例有限;生态锁定在 Siemens 硬件上
GE Vernova / GE Digital能源设备与电网软件GE Vernova 2024 年分拆;收入 ~$36B(聚焦电力设备)发电、电网、风电;历史 Predix 客户电力设备领域深度;电网优化软件;长期公用事业客户关系GE Digital(Predix)已分离;分拆后工业 AI 平台投入有限
Hexagon Smart Digital Reality测量 / 资产生命周期私营(由多数股东持有);年收入估计 ~$5B+资产密集行业;3D 空间数据、工厂设计、现实捕捉3D 空间 / 精密测量;现实捕捉;工厂设计生命周期管理3D 空间而非运营时序数据;买方画像不同;AI 智能体工作台有限
现状 / 内部自建不作为 / DIY 替代零许可成本(但人力成本高)所有尚未部署 DataOps 平台的现有工业运营商无新增软件支出;沿用现有 PI + 电子表格工作流或 Snowflake+Python没有知识图谱;分析师人力成本高;不可扩展;没有 AI 智能体能力

规模 / 融资:Cognite 指标来自 Cognite 新闻稿;竞争对手规模来自公开文件(C3.ai、Palantir)或分析师估计(AVEVA、AspenTech、Siemens)。标注 “估计”或“大约”的单元格是分析师估计;null 单元格表示公开资料不可得。主要局限列反映 Cognite 的竞争定位,不是独立评估。

[CP001, CP002, CP006, CP007, CP009, CP012]
FP001: 竞争定位图

用有证据支撑的顺序轴映射竞争对手:工业领域深度(x 轴:低到高)vs. AI/ML 平台成熟度(y 轴:低到高)。位置反映截至 2026 年 5 月审阅的官方产品主张和分析师评估;并非基于独立基准测试。

轴位置是有证据支撑的顺序估计,来自官方产品主张和 Gartner Peer Insights 评论模式;不是基于定量基准分数。X 轴:工业领域深度 = 预置工业语境化、垂直用例覆盖和 OT 生态集成程度。Y 轴:AI/ML 成熟度 = 是否具备生成式 AI、智能体工作台、ML 自动化和 AI 原生架构。

[CP001, CP009, CP014, CP019, CP021, CP023]

3.2 既有厂商与直接同行画像

AVEVA PI System 现由 Schneider Electric 持有,后者在 2023 年 1 月完成了约 $9.5B 收购。PI System 是占主导地位的工业时间序列数据 历史数据库,在石油与天然气、电力和流程工业中拥有 30 多年装机基础。PI System 的竞争优势在于无处不在的部署足迹:大型工业设施里数以万计的 历史数据库标签,形成深度运营依赖和多年切换摩擦。AVEVA 已推出云原生数据管理平台 AVEVA CONNECT,作为下一代产品与现代云替代方案竞争; 但它与 CDF 的知识图谱方法属于不同架构哲学。 AspenTech 在 2022 年由 Emerson Electric 以约 $11B 估值交易收购,面向化学工程、炼油、电力生产和资产绩效管理,核心产品包括 aspenONE 仿真套件和 Aspen Mtell 预测性维护产品。Emerson/AspenTech 组合把 OT 控制硬件和软件打包在同一个企业屋檐下,相比独立软件供应商形成渠道优势。 AspenTech 聚焦流程仿真,而不是运营数据上下文化,因此与 CDF 核心用例的直接重叠有限。 Seeq 将自己定位为面向工业时间序列数据的高级分析、机器学习和 AI 平台,以数据源无关层运行在 PI System、CDF 或其他历史数据库之上。 不同于 Cognite 的全栈路径,Seeq 解决的是分析和洞察层,而非数据接入或知识图谱建设,因此它对 CDF 的分析能力既可互补,也会在部分部署中竞争。 C3.ai(NYSE: AI)提供横向企业 AI 应用平台,标杆工业客户包括 Shell、Baker Hughes、Koch Industries 和 U.S. Air Force; 但其横向范围和对客户自备数据基础设施的依赖,与 Cognite 的纵向一体化路径形成对比。Palantir AIP 面向大型企业和政府 / 国防组织,采用高接触交付模式,交易规模通常高于标准工业 SaaS 合同。 [CP002, CP003, CP004, CP006, CP007, CP008]

功能 / 能力矩阵
采购标准Cognite CDF + Atlas AIAVEVA PI System / CONNECTAspenTech (Emerson)SeeqC3.aiPalantir AIPAWS IoT SiteWiseAzure Digital Twins
时序数据采集 / 历史库是 – OT 连接器、实时数据 + 历史库是 – PI 是标杆历史库部分 – 通过 PI/DCS 集成摄取否 – 只有分析层;依赖上游历史库部分 – 需要既有数据基础设施部分 – 需要既有数据基础设施是 – 托管云采集服务部分 – 只建模数据;依赖上游数据源
工业知识图谱 / 上下文化是 – 核心产品差异化部分 – PI AF 资产框架;语义图谱有限否 – 是仿真模型,不是运营知识图谱否 – 只有分析层部分 – 面向客户专属模型的数据本体部分 – Foundry 本体;通用而非工业原生否 – 原始数据采集与存储部分 – 数字孪生模型;运营上下文有限
低代码 AI 智能体工作台是 – Atlas AI 智能体工作台(Jun 2024)部分 – 无代码分析;不是智能体式部分 – C3 AI 应用;不是低代码自助式否 – 重服务、工程师主导部署
云原生 SaaS 架构是 – 从一开始就是云原生部分 – AVEVA CONNECT 是云端;PI 核心是本地遗留架构部分 – 正向云端迁移是 – 云原生是 – 云原生 SaaS是 – 云原生是 – 完全托管的 AWS 服务是 – Azure 云服务
开放 API / 开发者生态是 – REST 和 GraphQL API;SDK部分 – PI Web API;专有部分 – aspenONE API;生态相对云原生有限是 – 开放 API;连接器生态是 – API 优先部分 – Foundry API;受限是 – AWS SDK 和 API是 – Azure SDK 和 REST API
法规 / 合规文档是 – SOC 2、ISO 27001、GDPR;安全页面是 – 面向公用事业 / 电网的企业级合规是 – NERC-CIP、ISO,面向能源公用事业部分 – SOC 2;监管认证有限是 – FedRAMP、DoD;国防认证是 – 机密级;FedRAMP High是 – AWS 合规组合是 – Azure 合规组合
定价透明度否 – 私有定价;合同谈判否 – SI 渠道;谈判定价否 – 基于 token 的订阅;谈判定价部分 – 联系销售;可试用否 – 企业合同;虽为上市公司但无标价否 – 大型定制合同是 – 按用量计费;公开定价是 – 按用量计费;公开定价
预置工业垂直化能力是 – 油气、制造、化工、制药是 – 油气、公用事业、流程工业;PI 库很深是 – 化工、炼化、发电是 – 流程工业分析模板部分 – 能源 + 制造应用;整体偏横向否 – 通用本体;垂直场景由客户配置否 – 通用工业数据服务否 – 通用建模平台

标为「部分」的单元格表示相对该标准能力有限或不完整。标为「否」的单元格表示所审阅来源未记录该能力。竞品单元格证据来自 2026 年 5 月审阅的竞品官方产品页面;部分能力可能在页面审阅后新增。没有可用的独立基准测试;本矩阵反映基于来源的评估。

[CP001, CP004, CP009, CP013, CP014, CP016]
FP002: 功能广度 / 能力地图

截至 2026 年 5 月审阅的八项工业平台采购标准中,各竞争对手的能力覆盖。完整 = 有文档记录的能力;部分 = 能力有限或正在成形;无 = 未记录能力。基于官方产品页;未独立跑基准测试。

能力评级是基于证据的定性评估,来自 2026 年 5 月审阅的官方厂商产品页。“完整” = 作为核心产品能力记录在案;“部分” = 能力有限、正在成形,或需要客户额外配置;“无” = 审阅来源中未记录该能力。该矩阵从每个厂商覆盖广度切入,与 TP002 按标准逐一比较厂商的视角不同。

[CP004, CP009, CP013, CP016, CP021, CP023]

3.3 能力、定价与 GTM 对比

Cognite 相比竞争对手的关键差异化,来自工业知识图谱架构、Atlas AI 低代码智能体工作台和云原生开放 API 设计。工业知识图谱在统一语义层中上下文化资产、时间序列、文档和 3D 模型之间的关系。没有既有历史数据库(AVEVA PI、AspenTech)或横向 AI 平台(C3.ai、Palantir)复制这一组合。AWS IoT SiteWise 和 Azure Digital Twins 解决基础设施层,但缺少 Cognite 面向特定领域的数据建模和预置工业上下文化。 这一领域里,私营市场参与者的定价大多不透明。Cognite 采用 SaaS 订阅模式,包含按资产或按数据量计价的组件、实施服务和专业服务层,但具体标价未公开。AVEVA PI System 通过永久许可加维护以及订阅模式销售;AspenTech 的 aspenONE 工程套件使用基于 token 的订阅模式。C3.ai 使用按应用的订阅模式;Palantir 项目是大规模定制合同,金额常在 $10M+ 区间。AWS IoT SiteWise 和 Azure Digital Twins 使用云消费计价(按消息、按查询、按标签),小规模进入门槛更低,但企业规模下成本可能很高。 Cognite 的 GTM 策略强调高管级企业销售、Microsoft 联合销售伙伴关系,以及与 Databricks、Snowflake 的战略集成,用于分析和数据湖互操作。相对既有厂商,它的关键分销护城河是 Microsoft 伙伴生态:Cognite 连续三年至 2024 年获评 Microsoft 年度能源与资源合作伙伴,受益于 Microsoft 全球企业账户关系。这部分抵消了 AVEVA(处于 Schneider Electric 更宽工业自动化组合内)和 AspenTech(处于 Emerson 流程自动化套件内)所拥有的 OEM / 捆绑渠道优势。 [CP023, CP024, CP034, CP035, CP040]

定价 / 打包方式对比
厂商定价模式合同类型已披露价格打包 / 渠道尽调备注
Cognite CDF + Atlas AISaaS 订阅;按资产或数据量计费的组成部分 + 专业服务通常为多年期企业合同未公开披露Microsoft 联合销售;Databricks/Snowflake 生态集成;SI 合作伙伴网络私有定价;客户直接谈判或经 SI 谈判;部署需要较大的专业服务预算
AVEVA PI System / CONNECT本地 PI 为永久授权 + 维护;AVEVA CONNECT 云为订阅企业多年期;OEM/SI 渠道主导未公开披露;典型企业 SI 定价打包在 Schneider Electric 工业自动化组合内;大型 SI 生态(Accenture、Capgemini 等)SI 渠道导致实际定价差异很大;云端 CONNECT 定价不同于传统 PI
AspenTech (Emerson)基于 token 的订阅(aspenTech Subscription Suite)年度或多年期企业订阅未公开披露与 Emerson 流程自动化硬件和 OT 服务打包Emerson 打包可能压缩独立定价;续约率与 Emerson OT 关系绑定
Seeq按用户 / 服务器 / 数据源计费的 SaaS 订阅;分层计划年度;可试用联系销售;无公开标价SI 集成;PI 覆盖层;经主要工程公司渠道销售(ARC、Accenture)常部署在 PI 或其他历史数据库之上;相较替换历史数据库,是增量预算项
C3.ai按应用订阅 + 实施多年期企业合同;从试点到生产的路径未公开披露;大型企业交易规模直销企业客户;AWS、Microsoft 与 Google 云市场收入已连续多个季度低于分析师预期;需要审查交易结构
Palantir AIP大型定制合同;按平台或按价值定价多年期;Forward Deployed Engineers 纳入交易未公开披露;典型交易区间 $10M+直销;政府采购渠道;SI 有限ACV 很高;实施复杂;不适合中端市场
AWS IoT SiteWise按用量计费:按摄入消息、按资产(元数据)、按查询随用随付;预留容量折扣公开:标准层约 $0.00057/message(AWS 定价页面)AWS 市场;SI 合作伙伴;既有 AWS 企业协议仅为基础设施层;上层分析 / AI 需要大量开发;总成本随规模增长

定价数据来自官方产品页面、公开云定价计算器(AWS),以及公司披露的交易经济性(C3.ai、Palantir)。Cognite、AVEVA、AspenTech 与 Seeq 的定价未公开披露;注明处采用市场 / 分析师估计。所有 USD 数字均为近似值。空值单元格表示数据未公开可得。

[CP014, CP018, CP019, CP021, CP022, CP035]

3.4 切换成本、锁定与多平台并用

完整部署后,Cognite CDF 为客户制造多层切换成本。工业知识图谱模式一旦填入数千个资产模型、上下文关系和横跨企业设施的时间序列标签映射,就代表一项重大的自有数据结构投资。与 OT 源系统(DCS、SCADA、历史数据库)的自定义集成、基于 CDF 构建的 API 应用,以及训练好的 Atlas AI 智能体,都绑定在 CDF 的数据模型和 API 接口面上,抬高迁移运营成本。多平台并用——在竞争平台旁边运行 CDF——技术上可行,但运营复杂,除试点 / 评估场景外很少看到。 从传统 AVEVA PI System 部署迁移到 Cognite CDF,需要重建 OT 源连接器、将 PI 标签层级重新映射到 CDF 的知识图谱模式,并重新培训依赖 PI 工具做日常运营的用户。这类迁移通常需要系统集成商投入数月实施,形成有意义的壁垒;即便买方更偏好 CDF 能力,既有厂商(PI System)仍因此受益。 反过来,现状替代方案——PI 历史数据库、人工数据导出和基于电子表格的分析组合——初始替换的切换成本最低,但分析师生产率成本最高,也是 Cognite 最常见的新建场景机会。内部自建替代方案(自定义 Snowflake/Databricks 数据湖加 Python/ML 管道)正在成为拥有足够数据工程资源的工业 IT 团队中的竞争威胁;它以持续内部开发投入为代价,提供零许可成本。 [CP033, CP031, CP032, CP039]

3.5 护城河耐久性、商品化风险与反向证据

Cognite 最持久的竞争优势,是工业领域专长、工业知识图谱的语义数据架构,以及立在其上的 Atlas AI 智能体工作台。但三个结构性威胁值得持续尽调:第一,AVEVA 深入扎根的 PI System 装机基础,以及 Emerson/AspenTech 的 OT+软件捆绑渠道,都是无法仅靠产品优势快速打穿的分销优势——Cognite 必须在每次部署中证明数百万美元级 ROI,才能替换既有合同;第二,AWS 和 Azure 正积极扩展工业 IoT 服务组合,未来 3–5 年内,随着云原生连接和存储变得无差异,可能把 CDF 价值栈底层商品化;第三,Cognite 最后一次公开确认估值是 2021 年 5 月的 $1.6B,此后未更新,而公开上市竞争者(C3.ai、Palantir)有完整市值透明度,Schneider Electric 和 Emerson 旗下资金充足的企业事业部几乎拥有无限投资能力,这意味着 Cognite 相对同行的竞争资本化水平不透明。 Gartner Peer Insights 关于工业 IoT 平台市场的评论提供了反向证据:企业买方把实施复杂度、总拥有成本和专业服务依赖列为该类别最主要担忧,这些担忧直接适用于 Cognite 的高接触部署模式。Gartner 评论也反映出买方市场确实在多个供应商之间竞争,没有单一平台获得压倒性用户认可,凸显包括 Cognite 在内所有厂商的执行风险。截至 2026 年 5 月审阅的公开资料中,未发现 Cognite 输给具名竞争对手的具体记录;不过 Cognite 不公开披露赢 / 输数据。 [CP033, CP036, CP037, CP038, CP011, CP018]

护城河耐久度 / 竞争风险登记表
护城河主张威胁来源严重性支撑证据缓释措施 / 尽调追问
工业知识图谱的独特性:没有竞品能在统一语义层中复刻情境化资产模型 + 时间序列 + 3D 数据融合AspenTech PI AF、Azure Digital Twins、Palantir Foundry 都提供部分图谱 / 本体重叠能力;在 Databricks lakehouse 上自建正成为替代方案中 – 图谱架构尚未被规模化复刻,但构件正商品化AVEVA PI System 聚焦基于 tag 的历史数据库,不是语义图谱;Azure DT 偏建模而非运营;Cognite moonshot 公告称已将 80T+ 数据点置于上下文中(自研)通过 2–3 次客户访谈验证实际切换成本;评估 CDF 的图谱 schema 是否易于迁移,还是会形成真正锁定
Microsoft 战略合作与联合销售分发:三次获评 Microsoft Energy & Resources Partner of the Year,把联合销售管线接入 Microsoft 全球企业客户Microsoft 可能在 Azure 内自建工业 AI 能力,或偏向另一家 ISV;Azure Digital Twins 是同一合作伙伴生态内的直接竞品线中 – Microsoft 联合销售真实存在,但并非独家;Microsoft 仍在投资 Azure 工业能力Cognite 确认获得 2024 Microsoft Partner of the Year(Cognite 新闻稿);TotalEnergies 与 bp 部署涉及 Azure 基础设施评估 Microsoft 联合销售安排的排他性;核实管线贡献;关注 Azure Industrial Metaverse 投资
大规模 CDF 部署带来的数据引力:>80 万亿个已置于上下文的数据点,一旦知识图谱填充完成,迁移成本很高AVEVA PI 存在迁移路径;云超大规模厂商提升数据可迁移性;Open Industrial Data(Cognite OSIsoft 数据集)为公开数据低–中 – 数据引力真实存在,但开放标准降低了绝对锁定效应Cognite FY2025 业绩称数据建模实例增长 48%,知识图谱深度翻倍;TotalEnergies 多年部署覆盖 39 项资产核实 CDF 数据模型是否可通过开放标准导出;评估开放 PI 数据计划是否削弱 CDF 的图谱优势
借助 Atlas AI 的 AI 智能体差异化:低代码智能体工作台于 2024 年 6 月推出,Atlas 客户同比增长 700%横向 AI 平台(C3.ai、Palantir)正在加入低代码能力;Microsoft Copilot 面向工业场景;开源 LLM 工具降低智能体开发成本中 – 先发优势存在;2025 年订单中 70%+ 包含 Atlas AI;但 AI 开发速度可能在 18–24 个月内缩小差距Cognite FY2025 业绩:Atlas AI 被 59% ARR 基数使用;2025 年新增订单中 >70% 包含 Atlas AI;AI token 用量同比 +500%评估 Atlas AI 的工业专用预训练深度,而非通用 LLM 封装;评估非 Cognite 厂商能否在 12–18 个月内靠开源复刻
油气领域能力与锚定客户网络(Aker BP、Saudi Aramco、TotalEnergies、bp)在目标垂直行业形成密集参考案例Cognite 40% 非油气收入组合显示多元化正在推进,但如果能源转型压低 AI 平台资本开支,油气集中度仍是风险中高 – 油气行业风险具有结构性;能源转型可能在 3–5 年期压低油气 IT 预算Cognite FY2025 提到 40% 非油气收入;CNBC Disruptor 50 榜单强调油气起家的风险;GE Vernova / Siemens 在可再生能源中的竞争增强跟踪非油气 ARR 增长率;评估可再生能源 / 公用事业管线;核实 CDF 的价值主张能否顺畅迁移到传统油气之外

严重性是基于证据的定性评估(低 / 中 / 高),依据截至 2026 年 5 月审阅的来源。这不是量化风险分数。「支撑证据」引用已审阅的具体来源;没有可用的独立第三方护城河耐久度基准。

[CP034, CP035, CP036, CP037, CP038, CP040]
FP003: 护城河 / 就绪度 KPI

Cognite 相对于主要竞争对手类别的竞争耐久性指标,依据截至 2026 年 5 月审阅的证据评估。分数反映本分析师基于证据的判断;不是独立基准数据。

所有 KPI 值都是本分析师基于证据的定性评估;不是独立来源给出的定量分数。标为“未知(Unknown)”的项目反映 localEvidence.evidenceGaps 中记录的真实证据缺口。

[CP034, CP035, CP036, CP037, CP038, CP040]

3.6 图表

Chapter 04

04财务情况

4.1 收入模式与定价

Cognite 运行软件即服务(SaaS)订阅模式,由两大核心平台支撑:Cognite Data Fusion(CDF)是工业 DataOps 引擎;Cognite Atlas AI 是 2024 年 6 月推出的低代码工业智能体工作台。两者都以多年期订阅协议销售给企业级工业运营商,之上叠加专业服务和 Cognite Success Tracks。第三层平台 Cognite Flows 于 2026 年 5 月推出,把平台行动层延伸到一线员工,预计会随时间深化按席位或按工作流的变现。 截至 2026 年 5 月研究日期,Cognite 未公布任何产品标价。按这一规模下工业 DataOps 和 AI 平台的行业惯例,企业账户总合同价值(TCV)通常为每年数十万美元到数百万美元不等,由数据量、连接资产数量、席位数和专业服务范围驱动。TotalEnergies 为在全球全部上游资产部署 CDF 达成的 3 年合作,符合数百万美元级年合同价值(ACV)的特征;但没有披露具体金额。同样,B. Braun 合作和 Celanese 部署(通过 NVIDIA NV-Tesseract)暗示中大型企业 ACV,但具体条款未公开。 收入结构:FY2025 约 60% 收入来自石油与天然气客户,约 40% 来自石油与天然气以外(制造、化工、制药、可再生能源、公用事业)。Atlas AI 产品占 2025 年新预订超过 70%,说明平台正在深化:现有 CDF 客户叠加 Atlas AI 订阅,推动 NRR 超过 100%。公司“Moonshot”使命——到 2035 年交付 $100B 可衡量客户价值——是一个公开表述的北极星,既塑造其 GTM 叙事,也支撑其溢价定价姿态。 向 Brønnøysundregistrene 提交的挪威法定年度账目(Cognite AS, org. 918274758)记录 FY2024 收入约 NOK 1.67B(按 2024 年平均 NOK/USD 汇率约 10.55,约为 $158M USD)。这只代表挪威法人实体;纳入美国、日本和印度子公司的合并集团收入会不同。该数字与报告的 FY2025 全球收入 >$170M 一致,并为 FY2024 提供了一个收入下限。 [CI001, CI002, CI003, CI004, CI005, CI006]

收入流表
收入流机制单位 / 定价基础当前价值 / 状态收入质量尽调追问
CDF 平台订阅面向工业 DataOps 平台的年度 / 多年期 SaaS 订阅按连接资产、数据量或企业授权计费;标价未披露>60% 收入估计;核心成熟产品高 – 经常性、多年期;扩张模型的锚点单独确认 ACV 区间和续约率,不与 Atlas AI 附加销售混在一起
Atlas AI 订阅增购叠加在 CDF 之上的 AI 智能体工作台增购层按席位或用例计费;与 CDF 打包或独立销售;定价未公开2025 年新增订单中 >70%;2026 年 Q1 ARR 中 72%高 – NRR 信号强;证实先落地再扩张在跑通披露每个 Atlas AI 席位相对基础 CDF 的 ACV 增量;按队列年份披露 NRR
Cognite Flows 体验层2026 年 5 月推出的订阅 / 嵌入式工作流层可能按一线员工席位或工作流计费;定价未披露上线时 30%+ 客户已启用中 – 评估留存仍过早;采用信号强确认变现模式(打包 vs 增量收费);未来 ASP 影响
专业服务 / Success Tracks实施、集成、上线、持续成功服务按时间和材料或固定费用;列为名为「Success Tracks」的产品规模不小,但未披露占收入比例中 – 服务压缩毛利率;随平台扩张的动态披露 PS 收入占总收入比例;相对订阅的毛利率差异
油气垂直行业垂直行业收入组合;不是独立产品,嵌入以上收入流同一 SaaS + PS 模式,但用例具备行业特定性约 FY2025 总收入的 60%中 – 长期能源行业需求抵消集中度风险识别前 5 大油气客户及其 ARR 占比;确认 Aramco/bp 集中度
非油气垂直行业(制造、制药、可再生能源)油气之外扩张的垂直行业组合同一模式;较新队列具备高 NRR(150% 适用于此处)约 FY2025 总收入的 40%高潜力 – NRR 更高的队列显示新垂直行业适配度更好单独拆分非油气 ARR;识别头部客户和用例

定价未披露;收入流估计来自公司 ARR 组合披露和分析师信息。「收入质量」是分析师判断,不是公司披露。挪威法定账目(FY2024)确认仅挪威实体总收入为 NOK 1.67B(约 $158M USD),提供了一个与 FY2025 全球收入 >$170M 一致的 FY2024 数据点。

[CI001, CI002, CI003, CI004, CI005, CI008]
定价与变现表
产品 / 层已知标价实现价格代理合同结构折扣 / 未知项来源
Cognite Data Fusion (CDF)未公开披露企业客户每年数百万 USD(由 TotalEnergies 3 年交易范围推断)多年期企业订阅;TotalEnergies 39 项资产交易暗示 TCV 较大企业折扣为标准做法;合作伙伴 / 分销渠道未知Cognite 官方;TotalEnergies 新闻稿;OE Digital
Cognite Atlas AI未公开披露估计增购层带来 30–60% ACV 增量(分析师代理,未确认)可能与 CDF 打包,或作为按智能体 / 席位计费的增购项打包折扣相对独立销售不清楚Cognite newsroom;CNBC Disruptor 50;Cognite Flows BusinessWire 等来源
Cognite Flows未公开披露;2026 年 5 月推出评估仍过早;30%+ 客户已启用但未披露定价可能按一线员工席位计费,或嵌入平台上线定价可能带促销属性;长期 ASP 未知BusinessWire Flows PR;Cognite newsroom 等来源
Success Tracks / 专业服务未披露;「Quick Start」入门路径暗示存在命名分层与企业工业 SaaS 的 15–25% ACV 相符(分析师估计,未确认)年度或按项目;结构化上线计划差异很高;大客户可能谈判优先费率Cognite 招聘页面;BusinessWire Flows(实施速度基准)

所有定价均为推断或估计。截至 2026 年 5 月运行日,未发现公开可访问的 Cognite 定价页面。ACV 和 TCV 估计是分析师代理,来自部署规模和可比工业 SaaS 基准;并非公司来源。任何投资决策都必须通过资料室尽调获取实际合同条款。

[CI004, CI005, CI009, CI010]
FI001: Cognite 收入模式桥

客户工业运营活动如何经 Cognite 平台转化为订阅收入、专业服务收入,并最终转化为毛利。

收入流权重和毛利率未披露。订阅与服务之间的收入分配根据行业惯例估计;实际拆分需以 data room 确认为准。毛利节点不含数值,因为毛利率未披露。

[CI001, CI002, CI003, CI004]

4.2 GTM 动作与销售效率

Cognite 只靠直销型企业销售推进商业化。Paul Grenet 的头衔是总裁兼首席营收官,说明公司把一线运营和营收责任放在同一套架构里。区域总裁覆盖美洲(Bill Hendricks)、EMEA(Petteri Vainikka)、日本(Ryoichi Egawa)、北亚和东南亚(TS Park)以及印度(Guru Ananthanarayanan),形成完整的全球直销版图,也符合 $100M+ ARR 企业软件公司的配置。 工业 AI 和 DataOps 的销售周期很长;全新企业部署通常需要 9–18 个月,原因在于 OT 集成复杂、能源巨头采购治理严,以及需要概念验证(PoC)阶段。Cognite 用「Cognite Quick Start」降低首次落地门槛,并把 Atlas AI 嵌入 72% 的客户 ARR(截至 Q1 2026),意在加快初始签约、压缩扩张周期。BusinessWire 的 Flows 公告称,一家全球制药客户把见效速度提升 30X,UAT 从 6–9 个月降到两个月;这些公开生产率指标为定价谈判提供支撑。 关键共售路径是 Microsoft Azure 合作伙伴关系(2022–2024 连续三次 Energy & Resources Partner of the Year),它把 Cognite 导入 Azure Marketplace 商业协议,也接入 Microsoft 企业销售团队的共售动作。Snowflake 和 Databricks 生态合作(2026)把渠道触达延伸到以数据云为中心的采购流程。NVIDIA NV-Tesseract 集成(March 2026)又增加了 AI 基础设施共售维度。 获客成本(CAC)、销售周期长度、magic number 和回本周期均未公开披露。替代指标显示,Cognite 的 CAC 可能偏高,符合长周期企业销售的特征,但高 ACV 和较强 land-and-expand 经济性(新客户 NRR 150%)可以抵消一部分压力。客户数未公开披露。 [CI009, CI010, CI011, CI012, CI013, CI014]

4.3 成本结构与单位经济性

Cognite 是一家云原生软件公司,主要部署在 Microsoft Azure 上。其成本结构由几类开支主导:研发(平台工程、AI 模型开发、知识图谱基础设施)、全球销售和营销(覆盖 6 个地理区域的企业直销团队、面向小众工业垂直的市场投入),以及支撑 800+ 人全球组织的一般和行政费用。云基础设施成本(Azure 计算、存储和 AI 推理)在工业知识图谱规模下相当可观;截至 Q1 2026,该图谱已包含 85+ trillion 个时间序列数据点,但这些成本大多以平台成本形式转嫁给客户,或嵌入订阅定价。 毛利率未公开披露。带有重数据基础设施和嵌入式专业服务要素的工业 SaaS 平台,毛利率通常在 60–75% 区间,低于纯横向 SaaS 的 70–85%,原因是 OT 集成服务和逐客户的数据上下文化工作会吃掉毛利。Cognite 的「Success Tracks」以及多站点企业客户高实施强度的导入流程,意味着服务成分不小,综合毛利率应低于纯软件基准。 挪威法定账目(FY2024)披露,Cognite AS 报告运营开支(相对约 NOK 1.67 billion 收入,隐含约 NOK 2.08 billion)并产生显著净亏损,约 NOK -547M(约 -$52M USD)。这只是局部图景;挪威实体可能承担集中研发成本,而部分收入和成本在美国及其他子公司入账,但它确认了实体层面现金消耗的量级。集团完全合并后的亏损可能更高,也可能更低,取决于关联方转让定价和非挪威子公司的表现。 客户层面的单位经济性——LTV、CAC、回本周期以及按队列划分的净美元留存——均完全未披露。「新」客户 150% NRR 是唯一公开的单位经济性替代指标,也是强正面信号;但「新」的定义(年份、队列构成)没有说明,全量客户 NRR 也没有披露。 [CI015, CI016, CI017, CI018, CI019, CI020]

单位经济表
指标数值 / 区间置信度为什么重要尽调追问
毛利率(订阅)空值 – 未披露;估计 60–75%(分析师代理)长期盈利能力和定价权的关键驱动从经审计账目中披露混合毛利率和仅订阅毛利率
毛利率(含服务的混合口径)空值 – 未披露;估计 55–70%(分析师代理)服务压缩风险,相对纯 SaaS 基准分别拆分订阅与服务毛利率
NRR – 较新客户150%单位经济的最强公开信号;驱动 Rule of 40 分子定义「较新」队列年份;披露全基数 NRR 和总留存率
NRR – 全客户基数空值 – 未披露N/A没有全基数 NRR,净留存可能被流失的老队列掩盖披露全队列 NRR 和总收入留存率
ARR(绝对额)空值 – 未披露;隐含下限约 $170M(若约 100% 为经常性收入)增长率分母;SaaS 估值基准所必需提供最近季度末的绝对 ARR 基数
客户数空值 – 未公开披露N/A计算平均 ACV 和集中度风险所必需按层级和 ARR 四分位披露客户数
平均 ACV / 交易规模空值 – 推断企业账户为 $500K–$5M;未确认驱动 CAC 回本和销售效率基准提供过去 12 个月订单的 ACV 分布(P25/中位数/P75)
CAC(混合口径)空值 – 未披露N/A评估回本周期和 40 法则所必需按渠道披露全负荷 CAC 和回本周期
CAC 回本周期(月)空值 – 未披露;估计 18–36 个月(企业 SaaS 代理)衡量增长支出的资本效率披露混合回本周期,并按新增 ARR 与扩张 ARR 拆分
月度烧钱速度 / 经营现金流空值 – 未披露;Cognite AS FY2024 净亏损约 $52M USD(NOK 实体)低(部分代理)测算现金跑道和资本充足性提供集团合并经营现金流和月末现金余额
总美元留存率(GDR)空值 – 未披露N/A防止 NRR 头条掩盖流失披露 GDR,并与 NRR 区分
40 法则分数空值 – 估计:36% ARR 增长 + 未知 FCF 利润率SaaS 资本效率基准提供 FCF 利润率;与已知 36-57% ARR 增长组合后可能得到较高 R40

所有空值均反映缺少公开披露;并非对业绩差的评估。唯一确认的单位经济代理是公司 2026 年 1 月报告的「较新客户」150% NRR(初步 / 未审计)。所有估计都是基于工业 SaaS 同业基准的分析师代理,并已明确标注。挪威法定亏损(FY2024 约 $52M USD)是实体层面烧钱速度的代理,不能替代合并现金流披露。

[CI015, CI016, CI017, CI018, CI019, CI020]
FI002: 单位经济模型桥(指示性)

从获客到客户终身价值的定性单位经济模型流程,突出已确认的数据点(新客 NRR 150%)和仍为空白的节点。

CAC、ACV 和 LTV 仅为定性节点;本图没有引用任何数值估计,因为公司未披露。新客 NRR 150% 是唯一确认的单位经济模型数据点,并锚定扩张模型。销售周期长度(9–18 个月)是分析师基于可比工业企业 SaaS 给出的代理估计。

[CI016, CI017]

4.4 资本充足性与融资

Cognite 最近一次公开融资节点是 TCV 于 June 2024 领投的约 $300M 轮融资,TCV 新闻页标题为「TCV Leads $300M Investment in Cognite」。叠加 May 2021 的 $150M TCV 轮融资,以及更早未披露金额的 Accel 投资,已知累计融资至少约 $450M。2024 轮估值未公开披露。上一次确认估值 $1.6B 来自 2021 轮;到 May 2026 运行日期时已过时约五年。CNBC Disruptor 50(May 2026)引用的总融资额为 $300M,可能只反映 2024 轮,也可能是其对公开确认资本总额的计算口径。 本报告的公司概况章节(第 1 章)包含完整的逐轮融资时间线和投资人图谱;本章仅把它作为背景引用,并只提炼评估未来资本充足性所需的融资判断。 账上现金、月烧钱速度和现金跑道均完全未披露。挪威法定账目(FY2024)提供了唯一可用替代指标:Cognite AS 报告总资产约 NOK 2.26 billion(约 $214M USD),FY2024 单年净亏损约 NOK 547M(约 $52M USD)。如果假设挪威实体持有集团大部分现金(尚未确认),2024 年末资产负债表意味着现金垫较厚。不过,2024 轮约 $300M 于 June 2024 完成;到 May 2026 评估日时,这笔资本已部署多少并不清楚。以 FY2024 烧钱速度(仅挪威实体约 $52M/year)推算,集团层面 $60–90M/year 的烧钱速度是合理假设,意味着 2024 轮可支撑约 3–5 年现金跑道——缓冲充足,但不是无限期。 Aker ASA 作为控股股东提供隐性后盾:Aker 的 Net Asset Value 在 Q1 2026 超过 NOK 100 billion,创历史最强季度。Saudi Aramco 持有 7.4% 股权(自 February 2022 起,从 Aker BP 购入),带来资金实力雄厚的战略投资人,但也制造治理依赖:Aramco 同时是 Cognite 最大客户和重要股东,这种结构历来既带来进入优势,也带来产品路线图影响风险。 [CI021, CI022, CI023, CI024, CI025, CI026]

资本充足性表
项目数值 / 状态来源置信度尽调追问
现金余额(合并口径)空值 – 未公开披露私营公司N/A提供截至最近季度末经审计的现金及等价物
月度烧钱速度(集团)空值 – 估计 $5–8M/month($60–96M/year),基于挪威实体代理和员工成本基准SI007(挪威实体净亏损代理)低 – 分析师估计提供集团合并经营现金流出;按职能拆分
现金跑道(自 2024 年融资交割起)空值 – 按估计烧钱速度,从 2024 年 6 月算起约 3–5 年,意味着 2027 年中至 2029 年中SI007(烧钱速度代理)+ SI004(融资金额)低 – 分析师估计确认截至 2026 年 5 月的现金,并用已确认烧钱速度重算现金跑道
已知累计融资≥$450M(2021 年 $150M TCV + 2024 年约 $300M TCV + Accel 早前未披露金额)SI003(2021 轮)+ SI004(2024 轮)中 – 已确认轮次;Accel 金额未披露确认包括所有轮次和任何债务工具在内的累计融资
最近确认估值$1.6B(2021 年 5 月 TCV 轮)SI003(新闻稿)中高 – 同期新闻稿已过期约 5 年;无新轮估值披露;2024 年融资估值未披露
2024 年融资估值未公开披露;TCV 新闻页面仅确认「~$300M」融资标题SI004(TCV 新闻页面)低 – 金额已确认,估值未披露披露 2024 年融资 pre-money 估值、购入股权比例和 post-money 股权结构表
债务 / 项目融资义务空值 – 无信贷额度或项目融资的公开披露SI006(挪威登记处 – 未列示留置权)低 – 部分来源缺少证据披露任何债务、循环信贷额度、可转债或项目融资义务
Aker ASA 后盾定性 – Aker NAV 在 2026 年 Q1 超过 NOK 100B;多数股东SI013(Aker ASA IR 页面)中 – Aker NAV 已确认;无正式支持承诺确认来自 Aker ASA 的任何承诺性授信额度或强制支持义务
Saudi Aramco 投资方集中度自 2022 年 2 月起持有 7.4% 股权;同时是头部客户SI005(Reuters)+ SI002(CNBC)高 – Reuters 和 CNBC 独立交叉印证评估 Aramco 股权的董事会权利、信息权、排他条款和退出同意权

现金和烧钱速度数字是由公开代理推导的分析师估计;截至 2026 年 5 月运行日,公司没有现金披露。TCV 2024 年约 $300M 融资仅由轮次标题确认;完整条款、估值和交割后股权结构表未披露。融资历史参考:完整逐轮融资时间线见公司概况章节(本报告)。挪威法定账目(Cognite AS,组织号 918274758,向 Brønnøysundregistrene 提交)显示,FY2024 实体总资产约 NOK 2.26B(约 $214M USD),净亏损约 NOK 547M(约 $52M USD)。

[CI021, CI022, CI023, CI024, CI025, CI026]
FI003: 财务估计区间

估值承销所需关键财务输入的来源支撑或分析师估计区间,并按证据质量标注。

收入和烧钱估计来自公开披露、挪威实体法定申报和行业基准。凡非直接来源支持的内容均标为“分析师估计”。最后确认估值($1.6B)是 2021 年时点值,不应作为当前估值锚点使用。所有数值均应以后续 data room 披露为准。

[CI008, CI022, CI023, CI024, CI025]
FI004: 资本强度与现金流地图

Cognite 现金消耗和潜在现金产生的主要驱动因素,展示 800+ 员工规模下云原生工业 SaaS 平台的成本结构。

公司未披露 R&D、销售与营销、G&A 之间的成本分摊;节点标签按工业 SaaS 标准成本结构惯例处理。NOK -547M(约 -$52M USD)净亏损来自 Cognite AS FY2024 挪威法定账目,仅覆盖挪威实体。合并集团净结果未公开披露。

[CI019, CI020, CI021]

4.5 公开牵引力与私有信息缺口

Cognite 的公开财务披露方式符合一家主动管理投资人叙事的后期私营公司:它通过新闻稿和分析师验证释放部分 ARR 增长、NRR 和员工数指标,同时扣住毛利率、绝对 ARR、客户数、EBITDA 和资产负债表数据。以下牵引力指标均为公司自报且为初步 / 未审计数据,Cognite 在 January 2026 公告中也明确说明了这一点。 已确认公开指标(FY2025):年收入 >$170M;ARR bookings 同比增长 36%;新客户 NRR 150%;全球员工 >800 人(同比 +21%);40% 收入来自非油气客户;2025 年新 bookings 中 >70% 包含 Atlas AI;按 ARR 计,59% 客户使用 Atlas AI。 Q1 2026 加速信号:ARR bookings 同比 +57%;Atlas AI 按 ARR 渗透率 72%;知识图谱 85T+ 数据点(同比 +30%);AI token 使用量同比 +900%;MAU 同比 +54%。 挪威法定数据(FY2024,Cognite AS 实体):收入 NOK 1.67B(约 $158M USD);净亏损 NOK -547M(约 -$52M USD);总资产 NOK 2.26B(约 $214M USD);挪威员工 367 人(注册处截至 May 2026)。这只覆盖全球集团的一部分。 核心承销缺口包括:(1)毛利率和 EBITDA 走向盈利的路径;(2)绝对 ARR 基数(没有分母的增长率不足以承销);(3)客户数和集中度(基于战略定位,Saudi Aramco + bp 很可能贡献 >30% 收入——未确认);(4)账上现金和现金跑道;(5)2024 轮估值及由此产生的稀释。这些缺口是 pre-IPO 私营公司的系统性问题,只有正式尽调数据室访问才能补上。 [CI028, CI029, CI030, CI031, CI032]

公开财务缺口表
缺失指标对承销判断的影响精确尽调路径
毛利率(综合口径与仅订阅口径)无法判断盈利路径,也看不清平台与服务收入结构质量经审计 IFRS 合并利润表;分部层面毛利披露
ARR 绝对基数无法验证 36% 增速,也无法计算 SaaS 估值倍数董事会材料 ARR 瀑布图,或数据室 ARR 明细(最近四个季度)
全量客户 NRR 与总金额留存率「较新客户」150% NRR 未必反映成熟客群表现覆盖全部年份客群的 NRR 表;按年份客群披露总留存
客户数量与集中度无法判断收入集中度或流失敞口按客户层级拆分 ARR;前 10 大客户 ARR 占比;Aramco / bp 单独占比
在手现金(合并、经审计)无法确认现金跑道或资本充足性截至 2026 年 Q1 的合并资产负债表,列示现金、现金等价物和受限现金
月度经营现金消耗(集团)用于验证现金跑道估算和融资需求合并现金流量表;过去 12 个月经营现金流出
2024 年融资估值与股权结构表无法判断稀释、持股结构或当前股权入场价数据室:投资条款清单、股权结构表与现有投资人权利文件
EBITDA 与盈利路径无法评估 Cognite 是否有可信的 FCF 转正时间表董事会级财务模型,展示未来 3–5 年 EBITDA 桥
CAC 与回本周期无法判断销售效率或营销杠杆按分部拆分销售与营销费用;按渠道列示新增 ARR
债务与授信安排杠杆水平未知,可能影响股权价值和退出回款对所有债务工具、限制性条款和融资协议做法律审阅

所有缺口都来自私营公司未公开披露;不代表经营表现为负面。补齐这些缺口,需要 Cognite 及/或其多数股东 Aker ASA 开放正式尽调数据室。挪威法定账目(brreg)能补上一部分挪威实体 FY2024 信息,但不能替代合并集团财务。

[CI028, CI029, CI030, CI031, CI032]

4.6 财务结论与尽调阻断项

收入质量是 Cognite 财务图景中最清晰的正面因素。FY2025 ARR bookings 增长 36%、新客户 NRR 150%、Q1 2026 ARR 增长加速至同比 57%,三者合起来构成很有吸引力的复利曲线。Atlas AI 平台的 land-and-expand 动作也有支撑:到 Q1 2026,Atlas AI 覆盖 72% 客户 ARR,并驱动 >70% 新 bookings,说明装机基础内的扩张经济性有持续性。Forrester Total Economic Impact 研究(Q1 2026)引用 Cognite 客户 465% ROI,虽由供应商推动,但为溢价定价防守提供了第三方框架。 资本充足性看起来合理,但尚未确认。若按估计集团烧钱速度 $60–90M/year 计算,June 2024 约 $300M 融资自交割起可支撑 3–5 年现金跑道,对应 mid-2027 至 mid-2029 的跑道窗口。但这只是估计;实际烧钱速度、账上现金和任何债务融资额度均未披露。Aker ASA 这个控股股东后盾(NOK 100B+ NAV)降低了生存性融资风险。 主要财务阻断项包括:(1)$1.6B 估值停留在 2021 年,任何新投资都必须在不了解当前估值的情况下谈判;(2)缺少毛利率披露,无法承销平台业务经济性;(3)Saudi Aramco 同时是客户和投资人,形成利益冲突结构,可能影响产品路线图、数据访问条款和最终退出选择;(4)私营公司不披露绝对 ARR 和客户数,使收入质量无法与可比 SaaS 基准交叉验证;(5)公司未披露 IPO、SPAC 或二级流动性路径,因此最可能的退出场景仍是 Aker、Saudi Aramco 或大型科技公司进行战略收购。 [CI033, CI034, CI035, CI036]

4.7 图表与证据

Chapter 05

05产品与技术

5.1 产品模块图谱与定义

Cognite 的商业平台由三层具名产品和一组内置应用组成,全部共用同一套工业知识图谱数据底座。Cognite Data Fusion(CDF)是核心工业 DataOps 平台:它从分散的 IT、OT 和工程系统摄取、上下文化并存储数据——包括来自 SCADA 和 DCS 的时间序列、文档、事件、3D 模型和 P&ID——再通过版本化 REST API、Python SDK 和 JavaScript SDK 对外提供。CDF 是栈里唯一可以独立运行的产品;Atlas AI 和 Flows 都需要 CDF 作为数据底座。 Cognite Atlas AI 于 June 2024 推出,把 CDF 扩展成低代码工业智能体工作台。数字化团队和领域专家可以用预配置模板构建 AI agents,无需写代码。Agents 由 Context Augmented Generation(CAG)驱动,把 CDF 的图谱上下文注入 LLM 提示词;同时由 autoLLM 负责路由,为每个 agent 任务选择最合适的 Large Language Model、Small Language Model 或 Custom Language Model。截至 2025 财年,按 ARR 计 59% 客户已把 Atlas AI 投入运营,Atlas AI 客户数同比增长近 700%。 Cognite Flows 在 Cognite 的 IMPACT 大会上公布,并于 May 12, 2026 发布,是平台的行动层。它为一线操作员和工程师提供单屏、基于角色的自适应工作区,把 AI agents 嵌入其中,并同时呈现所有企业应用的数据。Flows 也重新定义开发者体验:开发者借助 Claude 或 Cursor 等 AI 原生编码工具,可以在数天而非数月内构建生产级工业应用。Flows 发布时,Cognite 超过 30% 客户基础已提前启用,50+ 客户正在活跃使用。Flows 打包的内置应用包括 Industrial Canvas(面向 P&ID 和时间序列的可视工作区)、InField(移动现场访问)、Maintain(维护工单管理)和 Charts(时间序列分析)。 [CE001, CE002, CE003, CE004, CE005, CE006]

Cognite 产品模块 / 资产矩阵
模块目标用户状态 / 成熟度核心差异化尽调缺口
Cognite Data Fusion (CDF)IT/OT 工程师;数据科学家;数据消费者GA / 生产可用(核心产品)工业知识图谱;为 80T+ 时间序列、事件、3D、文档建立上下文;不绑定云;RESTful API + SDK定价未披露;白标伙伴名单未公开;迁移成本高
Cognite Atlas AI领域专家;数字化团队;应用构建者GA / 生产可用(2024 年 6 月)低代码智能体工作台;CAG 提升领域准确性;autoLLM 模型路由;预置模板;智能体 API幻觉率与竞品工业 AI 未做基准比较;智能体质量 SLA 未发布
Cognite Flows一线操作员;工程师;开发者早期 GA(2026 年 5 月)统一一线 UX;工作流构建速度快 100 倍;自适应 AI 仪表盘;原生接入 CDF 知识图谱;通过 Claude / Cursor 提供开发工具链发布时 50+ 客户;更大规模验证仍待补充;独立定价未披露
Industrial Canvas工艺工程师GA / 生产可用在一个屏幕里处理 P&IDs、3D 模型和时间序列的可视化工作区功能深度与 AVEVA / Hexagon 2D/3D 工具未做独立基准比较
InField现场技术员 / 工人GA / 生产可用借助 CDF,在移动端优先访问工单、检查清单和设备数据低连接 OT 环境下的移动端可靠性未获独立验证
Maintain维修团队GA / 生产可用接入 SAP PM 和其他 CMMS,管理工单SAP 集成的双向能力和深度未获外部审计
Charts工艺工程师;分析师GA / 生产可用时间序列分析和图表能力直接建在 CDF 数据层上功能深度与 Seeq 或 OSIsoft PI Vision 未公开做基准比较

成熟度来自官方产品页和新闻室;Atlas AI 与 Flows 发布时间已确认。所有模块采用企业/定制定价,公开渠道未披露。 模块能否单独购买或必须打包,未披露。

[CE001, CE003, CE004, CE005, CE006, CE010]
FE001: Cognite 平台架构栈

三层架构展示 Cognite 工业 AI 与数据平台,从基础设施一路到 AI 和用户体验。

[CE001, CE003, CE004, CE018, CE019, CE021]

5.2 客户工作流与用例

Cognite 平台面向重资产行业——油气、能源、制造、制药和化工——这些行业的运营数据量大、形态复杂,且长期分散在孤岛中。平台的核心价值主张,是把原始传感器、设备和工程数据转成 AI-ready 知识,让一线员工、工艺工程师和数字化团队可以实时行动。 油气是 Cognite 最成熟的垂直领域。客户用 CDF 消除寻找运营数据的时间。Cognite 油气产品页称,凭借独特的 IT/OT/ET 上下文化,理解数据所需时间减少 90%,解决方案部署到实时生产的速度提升 10–25x,总体产量吞吐提升 0.5–1.5%。TotalEnergies 于 September 2024 扩大 Cognite 部署,覆盖全球所有上游资产,并称 AI-ready 数据改善了运营卓越性。Cognite 委托 Forrester 开展的 Total Economic Impact 研究引用了 465% ROI 和 USD 29.4 million 总收益。 制造领域,B. Braun(医疗技术)使用 Cognite Flows,在部署四周内实现所有站点资产健康的统一、实时可视化。Idemitsu Kosan(日本能源 / 化工)部署 Flows,把沉淀的专家知识数字化,并降低运营风险。一家全球前十制药公司用 Flows 把价值兑现速度提升到传统方法的 30x:过去需要 20+ 人团队、3–5 个月完成的工作,四天内交付,用户验收测试从 6–9 个月压缩到两个月。 化工领域,Celanese(Texas 特种化学品)与 Cognite 和 NVIDIA 合作,部署与 CDF 知识图谱集成的 NV-Tesseract NIM 时间序列预测,目标是消除反应水位监测中的预测偏差跳变。NVIDIA 合作于 March 16, 2026 公布,通过 Cognite 平台把基础时间序列 AI 落地到重工业。 [CE013, CE014, CE015, CE016, CE017]

Cognite 工作流 / 用例覆盖
用户任务当前 / 传统工作流Cognite 方案可衡量收益(披露口径)限制 / 缺口
预测性维护 / 设备健康人工查看传感器;临时告警;定期巡检CDF + Atlas AI 预测性维护智能体维护执行时间减少 30–80%(官网,公司口径)未找到对维护降幅主张的外部独立审计
根因分析多屏人工关联;靠专家电话定位数据Flows + Atlas AI RCA 智能体;CDF 知识图谱提供上下文假设验证更快;MTTR 降低(Flows 发布案例)MTTR 改善只在公司提供的证言中量化
知识数字化隐性专家知识散落在 PDF、流程文件和经验知识里Flows + Atlas AI 文档智能体(如 Idemitsu Kosan)将专家经验数字化,供 AI 驱动运营使用知识捕获的完整性和准确性未获独立测量
资产健康监测定期人工巡检;PI System 仪表盘;割裂的 ERP 数据CDF + Industrial Canvas;Flows 统一工作区465% ROI;USD 29.4M 收益(Cognite 委托 Forrester TEI)Forrester TEI 由供应商委托,仍需独立验证
运营透明度 / 报告SCADA、ERP、OT 数据割裂;人工拼报告CDF 知识图谱 + Flows 自适应仪表盘数据解读耗时减少 90%;NOVA Chemicals:数据定位从数小时缩短到数分钟节省时间的数据来自单一客户,且由公司提供
时间序列预测传统回归模型;基于样本的人工调整CDF + NVIDIA NV-Tesseract NIM 集成(Celanese 试点)消除了反应水位监测中的预测偏差跳变(单一部署)单一试点客户;更广部署尚无文档;模型准确率指标未发布

所列收益为公司声明,或来自供应商委托研究(Forrester TEI);未找到生产率指标的独立第三方验证。除非另有脚注, 收益区间来自营销文案。

[CE013, CE014, CE015, CE016, CE017]
FE002: Cognite 客户工作流——从数据到运营决策

端到端流程展示工业资产数据如何经过 CDF、Atlas AI 和 Cognite Flows,最终进入一线决策。

[CE001, CE003, CE004, CE007, CE008, CE026]

5.3 架构与集成模型

CDF 的核心架构创新是工业知识图谱——一个统一语义模型,把时间序列数据、事件、文档、视觉数据流、3D 模型和工程图纸连接进同一套上下文结构。截至 2025 财年,该图谱在上下文中存储超过 80 trillion 个时间序列数据点,同比增长 48%。同期,客户的数据建模实例翻倍,AI token 使用量增长近 500%,说明图谱之上的 AI 采用正在快速推进。 Cognite 的开发者界面刻意采用 API-first。REST API 使用版本化端点,并发布向后兼容政策和弃用 API 版本的生命周期计划。Python SDK(v8,当前主版本)通过新的 AsyncCogniteClient 引入完整 async 支持,为所有 API 操作提供非阻塞并发操作和原生 async/await 模式。该 SDK 与 pandas 深度集成。公司也提供 JavaScript SDK。Cognite 还发布 MCP(Model Context Protocol)server 集成,允许开发者把开发环境和 AI 编码工具直接连接到 Cognite 文档和 API。Python SDK 通过 PyPI 以 `cognite-sdk` 包分发。 基础设施层面,CDF 云无关:它可部署在 AWS、Azure 和 Google Cloud Platform,并通过全球分布式集群满足数据主权和延迟要求。身份和访问管理依赖 OIDC,支持 Microsoft Entra ID、Okta 等身份提供商。Cognite 也把 CDF 白标给其他软件提供商。 集成深度很广:Cognite 已宣布与 Snowflake(Energy Solutions 合作,January 2026)、Databricks(面向工业运营的企业 AI)、Microsoft Azure OpenAI Service 以及 NVIDIA NIM microservices 集成。客户通过 CDF 的 IT/OT 摄取连接器连接源系统——SCADA、DCS、ERP、OSIsoft PI 等。Flows 架构建立在 CDF 知识图谱之上,而不是另加一层数据层,确保洞察和新应用始终扎根在实时运营上下文中。 [CE018, CE019, CE020, CE021, CE022, CE023]

Cognite 技术 / 运营架构
层 / 组件作用关键依赖风险
工业知识图谱(CDF 核心)为所有 IT/OT 数据建立上下文并存储;80T+ 时间序列点;连接事件、文档、3D 和工程数据IT/OT 接入连接器;云对象存储;客户源系统供应商锁定:迁移需要重新为所有数据关系建立上下文;迁移成本高
Atlas AI 智能体工作台面向工业 AI 智能体的低代码构建器;智能体编排与部署CDF 知识图谱;LLM 提供商 API(Azure OpenAI、Anthropic 等);OIDC 认证LLM API 可用性和提供商定价属于外部依赖;autoLLM 质量未做独立基准比较
autoLLM 模型路由器为每个工业智能体任务选择最合适的 LLM、SLM 或 CLM多个外部 LLM/SLM 提供商 API 与端点LLM 提供商服务变更(模型弃用、定价)可能影响智能体行为和成本
上下文增强生成(CAG)将 CDF 知识图谱上下文注入 LLM 提示词,提高领域准确性CDF 数据模型质量;知识图谱完整性与新鲜度源 OT 系统数据质量差,会拉低 CAG 输出准确性;幻觉风险仍在
Cognite Flows(UX / 行动层)面向一线用户的单屏自适应工作区;用于定制应用的开发者 SDKCDF 知识图谱;Atlas AI 智能体;REST API;OIDC 认证不能独立运行,必须以 CDF 为底座;知识图谱数据陈旧或不完整时,UX 可能被打断
REST API 与 SDK(Python v8、JS)开发者访问层;版本化 API;异步 Python SDK;MCP 服务器集成OIDC 身份提供商(Microsoft Entra、Okta);用于 Python SDK 分发的 PyPIOIDC 配置错误可能打穿 API 访问控制;客户部署中的 SDK 版本碎片化
云基础设施多云执行;全球分布式集群;支持数据主权AWS、Azure 和 GCP 基础设施;区域可用区云提供商宕机会影响平台可用性;未找到 Cognite 公开披露的 SLA
OIDC 认证层为所有 CDF API 访问提供身份与权限管理第三方 IdP(Microsoft Entra ID、Okta 或其他 OIDC 提供商)IdP 服务中断或配置错误,可能把操作员挡在关键任务工作流之外

架构层级根据官方文档、开发者门户和 SDK 发布说明推断。截至研究日期,公开文档未找到具体 SLA 或正常运行时间保证数字。

[CE018, CE019, CE020, CE021, CE022, CE023]
FE003: Cognite 关键外部依赖图

按类别绘制 Cognite 关键外部依赖的有向无环图,展示云服务商和 LLM API 上的风险集中。

[CE019, CE021, CE022, CE023, CE024, CE025]

5.4 信任、安全与合规

Cognite 的安全计划面向那些安全事件不止造成数据丢失,还会导致停产、环境事故和安全事件的行业。作为正式 Secure Development Lifecycle(SDL)的一部分,平台在基础设施、应用和运营层面接受测试和第三方审计。公司把安全描述为「产品的功能性要求,而不是周边围栏」。 截至 May 2026,Cognite 持有以下认证和鉴证:ISO 27001(信息安全管理)、ISO 9001(质量管理)、ISO 27018(云端个人数据隐私)、SOC 2 Type 2(Security criterion)和 SOC 3(公开摘要安全报告)。CDF 还支持客户满足一组工业控制和关键基础设施标准:NIST CSF、IEC 62443(子标准 2-4、3-2、3-3 和 4-1)、CMMC、FIPs、NERC CIP v.5、GxP 和 CSA STAR Level 1。公司声明符合 EU AI Act,全球可用集群也支持受监管部署的数据主权。 trust.cognite.com 的 Trust Center 持续、实时展示安全态势、系统健康和合规文档。潜在客户可以获取 SOC 2 Type 2 和 SOC 3 审计报告。UpGuard 使用开源和自研威胁情报源,持续监控 Cognite 外部可见安全态势,评估网站安全、邮件安全、钓鱼和恶意软件、品牌声誉以及网络安全;不过截至 May 2026,UpGuard 尚未公开具体数字评分,也未指出重大漏洞。statusgator.com 服务监控 Cognite 平台中断状态,包括在中断公开前进行检测。研究期内未发现影响客户运营的重大公开记录中断。 尽调缺口仍在:SOC 2 Type 2 完整报告只在 NDA 下分享,并未公开发布;IEC 62443 合规被表述为可支持客户合规,但未发现独立第三方 IEC 62443 审计证书;高风险工业自主用例的 EU AI Act 分类可能需要额外文档;CMMC 认证等级未披露。 [CE027, CE028, CE029, CE030, CE031, CE032]

信任、质量与合规控制
控制 / 认证状态(截至 2026 年 5 月)范围尽调缺口
ISO 27001已认证(第三方审计)信息安全管理体系复审时间表和证书到期日未公开披露
ISO 9001已认证(第三方审计)质量管理体系QMS 具体范围边界与排除项未发布
ISO 27018已认证(第三方审计)面向个人身份信息的云隐私适用范围限于个人数据;工业 OT 数据在 GDPR 下不归类为个人数据
SOC 2 Type 2(安全)已审计 / 已认证安全信任服务准则;基础设施、应用和运营完整报告只在 NDA 下分享给潜在客户;无法公开下载
SOC 3(安全)已审计 / 已认证(公开摘要)面向公众的 SOC 2 安全控制摘要SOC 3 摘要只覆盖安全准则;可用性和保密性准则未鉴证
IEC 62443(2-4、3-2、3-3、4-1)支持 / 与客户合规对齐工业控制系统网络安全;OT 专用安全标准未发现独立第三方 IEC 62443 认证;合规可由 Cognite 支持客户实现,并非供应商自身认证
NIST CSF对齐面向关键基础设施的网络安全框架未公开披露相对于 NIST CSF 的成熟度等级
NERC CIP v.5 合规框架支持大电力系统关键基础设施保护NERC CIP 是否适用取决于客户部署场景;Cognite 是供应商,不直接受 NERC CIP 约束
EU AI Act合规 / 对齐(自述)面向欧盟部署 AI 系统的 AI 监管;全球可用集群支持数据主权工业自主用例若被归入 AI Act 高风险类别,可能需要客户侧补充合规文件
CMMC / FIPs支持(自述)美国 DoD CMMC 与联邦信息处理标准CMMC 认证等级(1、2 或 3)未披露;对美国政府或国防相关客户重要

认证状态来自 Cognite 官方安全页面和 Trust Center。「支持」表示 Cognite 提供控制,帮助客户合规,并不表示 Cognite 自身持有该认证。

[CE027, CE028, CE029, CE030]

5.5 路线图、发布与生态

Cognite 对 CDF、Atlas AI 和 Flows 能力采用季度产品发布节奏。最近最重要的里程碑横跨 June 2024 至 May 2026,包括两个重大新产品发布和多项合作伙伴集成。 June 2024 发布 Atlas AI,标志平台正式进入工业智能体 AI:它在 CDF 既有 DataOps 能力之上增加低代码 agent builder 和预构建 agent 库。January 2026 的 Snowflake Energy Solutions 合作,让双方共同客户可以在 Snowflake AI Data Cloud 中统一 IT/OT 数据,同时保留 Cognite 的知识图谱增强层。March 2026,Cognite 在 NVIDIA GTC 公布并演示 NVIDIA NV-Tesseract NIM 集成,Celanese 是锚定生产部署。March 2026,IDC 在 IDC MarketScape for Worldwide Industrial DataOps Platforms 2026 中把 Cognite 评为 Leader,理由包括 CDF 支持五项关键工业 DataOps 能力、白标生态以及通过 Atlas AI 提供的下游 AI 能力。Cognite Flows 于 May 12, 2026 发布,公司于 May 19, 2026 登上 CNBC Disruptor 50 榜单。 合作伙伴生态覆盖超大规模云厂商(Microsoft Azure、AWS、GCP)、专业数据平台(Snowflake、Databricks)、AI 基础设施(NVIDIA),以及不断扩大的 ISV/SI 渠道,其中 Radix、L&T Technology Services 和 RoviSys 是 Flows 发布合作伙伴。开发者社区可通过 hub.cognite.com 访问,围绕数据类型配置、API 认证和时间序列查询模式有活跃讨论。Stack Overflow 上带有 'cognite' 标签的问题显示从业者参与活跃,包括 CDF 数据类型管理和 SDK 认证流程方面的挑战。PyPI 上的 Python SDK(cognite-sdk)是开发者采用的主要分发渠道。cognite-sdk-python 的 GitHub 仓库显示项目仍在活跃开发,v8 版本引入了 async 支持。 [CE034, CE035, CE036, CE037, CE038, CE039]

路线图、发布与开发里程碑
日期 / 时期功能 / 里程碑状态含义来源
Jun 2024Cognite Atlas AI 发布 —— 带 CAG 与 autoLLM 的低代码工业智能体工作台GA平台正式切入智能体式工业 AI;前 18 个月客户增长 >700%Cognite 新闻室 / introducing-cognite-atlas
Q3–Q4 2024客户知识图谱中已上下文化的时间序列数据点超过 80T(同比增长 48%)已验证指标证明平台规模,也显示企业客户数据采用在加速Cognite Moonshot 报告,2026 年 1 月
Q3–Q4 2024AI token 使用量同比增长 ~500%;月活用户同比增长 26%已验证指标采用信号强;Atlas AI 正把 token 消耗推到被动 DataOps 使用之外Cognite Moonshot 报告,2026 年 1 月
Jan 2026Snowflake Energy Solutions 合作 —— 通过 Snowflake AI Data Cloud 统一 IT/OT 数据GA扩大平台互操作性;油气客户可将 CDF 数据产品导入 Snowflake,用于分析和 AICognite 新闻室;WorldOil
Mar 2026NVIDIA NV-Tesseract NIM 集成 —— 通过 NIM 微服务提供基础时间序列 AI(Celanese 试点)GA / 试点增加 AI 原生预测层;CDF 知识图谱为 NVIDIA 模型提供上下文;单一生产部署Cognite 新闻室 / NVIDIA GTC 公告
Mar 2026IDC MarketScape 领导者 —— 2026 年全球工业 DataOps 平台公开认可验证平台定位;IDC 提到 CDF 覆盖五项能力,并可通过 Atlas AI 承接下游 AIIDC MarketScape(经 Cognite 新闻室)
May 2026Cognite Flows 发布 —— 统一 UX 行动层;50+ 早期客户;发布前已覆盖 30%+ 客户基数GAUX 层显著补齐;Cognite 因此定位为端到端工业 AI 与数据平台BusinessWire、Automation.com
May 2026CNBC Disruptor 50 入选(#15)公开认可外部品牌可信度信号;有助于企业销售和人才吸引CNBC(2026-05-19)

指标数字(token 使用量、数据点、月活用户增长)均为公司报告的初步 / 未审计数据,除非归因于第三方来源。奖项认可不代表财务或运营表现。

[CE034, CE035, CE036, CE037, CE038, CE039]

5.6 产品与技术风险

Cognite 的产品与技术风险分为四类:数据依赖和上下文化复杂度;关键任务环境中的 AI 可靠性;供应商和平台锁定;以及监管 / 合规深度。 工业知识图谱虽然是战略差异点,也是主要锁定机制。所有下游 AI 能力——Atlas AI agents、Flows 工作流和内置应用——都需要 CDF 作为数据底座。客户若要迁离 Cognite,可能需要对其 IT/OT 资产中数亿条数据关系重新上下文化。前期数据上下文化投入也制造实施风险:如果没有专门的 OT/IT 集成经验,即便 Cognite 声称部署到可衡量影响只需 7 周,客户仍会面临较长见效周期。 Context Augmented Generation 提升了工业场景中的 AI 准确性,但 LLM 幻觉仍是关键任务决策的内生风险——包括设备停机、工艺参数调整或安全建议。Cognite 通过 CAG 架构和对可追踪、有来源输出的强调来承认这一风险,但尚未发现独立基准量化 Atlas AI 相对竞争方法的幻觉率。hub.cognite.com 和 Stack Overflow 上的开发者社区信号显示,CDF 数据类型管理(字符串与数值分类)、API 认证设置和时间序列查询优化反复出现集成挑战;这些都可能成为新客户摩擦点。 SOC 2 Type 2 审计范围可获得,但只在 NDA 下提供;IEC 62443 合规声明没有公开引用的独立认证支撑;EU AI Act 可能给部署高风险工业 AI(例如自主过程控制或安全关键建议)的 Cognite 客户增加文档负担。平台依赖第三方身份提供商(Microsoft Entra ID、Okta)的 OIDC,也带来依赖风险:IdP 中断或配置错误可能让操作员无法进入关键任务工作流。 [CE040, CE041, CE042, CE043, CE044, CE045]

FE004: Cognite 产品成熟度与能力图谱

从市场成熟度、功能完整度、集成深度和社区采用四个维度评估 Cognite 各产品模块能力。

市场成熟度评级根据产品发布日期、客户采用指标以及 IDC / Gartner 认可推断。功能完整度参照供应商披露的产品路线图,而非绝对行业基准。

[CE001, CE003, CE004, CE005, CE006, CE010]

5.7 图表与证据

Chapter 06

06客户情况

6.1 客户基础概览与分层

Cognite 的工业数据和 AI 平台服务于能源、制造、化工、制药、金属和矿业等重资产运营商。截至 FY2025,公司报告收入超过 $170M,ARR bookings 同比增长 36%,意味着客户基础横跨多个垂直和地区。管理层披露,2025 年约 40% 收入来自油气以外行业;相较公司作为挪威油气数据平台的起点,这是有意义的多元化。具名锚定客户包括 bp 和 Saudi Aramco 等一体化巨头,Cosmo Energy 等国家石油公司,TotalEnergies 和 Aker BP 等多元化能源公司,化学品生产商(NOVA Chemicals、Celanese),一家日本钢铁公司(JFE Steel),一家德国医疗器械制造商(B. Braun),以及一家日本炼油商(Idemitsu)。平台 go-to-market 强调企业数字化转型项目;公开引用的大多数部署覆盖数十到数百项资产、数千名用户。月活跃用户同比增长 26%;截至运行日期,按 ARR 计 59% 客户正在使用 Atlas AI 层,说明产品采用深度已经超出基础数据集成层。

客户分层表
垂直行业代表客户主要用例地理区域收入重要性
油气(上游 / 中游)bp、Saudi Aramco、Aker BP、TotalEnergies、Cosmo Energy 等客户生产优化、预测性维护、巡检全球(挪威、KSA、法国、日本)FY2025 收入约 60%(估算)
化学品 / 石化NOVA Chemicals、Celanese数据上下文化、数字工厂项目北美增长中;计入 40% 非油气
金属与采矿 / 钢铁JFE Steel资产绩效、数字孪生日本起步中;属于 40% 非油气
生命科学 / 医疗设备B. Braun通过 Flows 自动化生产工作流德国起步中;属于 40% 非油气
油气(炼化)Idemitsu、Cosmo Energy工程数据上下文化、用户导入日本计入约 60% 油气板块

垂直行业收入占比根据 Cognite 披露 FY2025 约 40% 收入来自油气以外来估算;更细分部颗粒度没有公开。 具名客户来自 Cognite 案例研究页面和新闻稿。

[CU001, CU004, CU007]
客户增长 / 采用轨迹表
指标数值日期来源置信度
总收入>$170MFY2025Cognite moonshot 文章
ARR 订单额增长同比 36%FY2025Cognite moonshot 文章
Atlas AI 客户数(增长)自发布以来增长 700%FY2025Cognite moonshot 文章
知识图谱时序数据80 trillion 个数据点FY2025(同比增长 48%)Cognite moonshot 文章
Flows 已启用客户群>30% 的客户群May 2026BusinessWire Flows 发布
月活用户增长同比 26%FY2025Cognite moonshot 文章
非油气收入约 FY2025 收入的 40%FY2025Cognite moonshot 文章

所有指标均来自 Cognite May 2026 moonshot 文章的公司披露;均未经过审计或独立验证。应将这些数值视为管理层披露,而非经审计财务数据。

[CU001, CU002, CU003, CU004, CU005, CU006]
FU001: 客户旅程图

Cognite 客户生命周期从最初认知到完全成为企业标准,并映射到代表性具名客户。

客户阶段分配基于公开证据推断;内部部署状态可能不同。具名客户放在公开证据最能支持的阶段。

[CU007, CU011, CU022, CU018, CU025, CU026]

6.2 具名客户部署与可衡量结果

十家具名客户的有记录生产部署,是现实采用最强证据。Aker BP 是 Cognite 合作时间最长的油气参考客户,报告称其在挪威北海六项资产上把目视检查时间最多减少 50%,维护执行时间减少 30-80%。该公司 CDO Dr. Paula Doyle 公开称平台让「下一代运营卓越」成为可能。这一结果由 Aker BP 自身投资者沟通独立佐证。Idemitsu 在不到六个月内上线四个炼厂站点,扩展到 3,000 多名用户,上传超过 10,000 份 P&ID,管理层估计每年节省 50,000 小时人工。TotalEnergies 在三年内把部署扩大到 39 项上游资产,集成超过 200 个关键生产系统,并由 TotalEnergies 自身新闻稿支撑。加拿大最大聚乙烯生产商 NOVA Chemicals 报告,在十一座工厂目标范围内把数据访问时间从数小时降到数分钟。Celanese 的「Digital Plants of the Future」项目把 Cognite 作为使能平台。B. Braun 将一项关键生产工作流四周内的改善归功于 Cognite Flows。Cosmo Energy 把三座日本炼厂与 250 billion 个数据点和 5 million 份工程文档集成。JFE Steel 也是具名 Flows 客户。证据集横跨五个垂直和四个地区,为一个定位为行业无关的平台提供了有意义的广度。

具名客户验证表
客户垂直行业部署范围状态报告结果证据质量
Aker BP油气上游挪威北海 6 项资产生产视觉巡检减少 50%;维护执行减少 30-80%高 —— Cognite 案例研究 + Aker BP 投资者沟通
TotalEnergies油气上游3 年覆盖 39 项资产;200+ 生产系统生产 / 扩展中多年扩展合作已确认高 —— TotalEnergies 新闻稿 + OE Digital
Idemitsu油气炼化<6 个月覆盖 4 个站点;3,000+ 用户;10,000+ P&IDs生产每年节省 50,000 小时(公司估算)中 — Cognite 案例研究 + BusinessWire
Saudi Aramco油气一体化旗舰油气部署;范围未披露生产环境成效未公开量化低 — Cognite 案例研究页仅 JS 可见;Aramco 也是控股股东
NOVA Chemicals化工目标覆盖 11 座工厂生产环境 / 扩展中数据访问时间从数小时降至数分钟中 — Cognite 案例研究
Celanese化工Digital Plants of the Future 项目生产环境支撑数字化转型的平台低 — 仅 Cognite 案例研究;无独立佐证
B. Braun生命科学Flows 部署在生产工作流中生产环境(Flows)关键工作流缩短 4 周(公司口径)低 — Cognite 通过 BusinessWire 发布的新闻稿
JFE Steel金属 / 钢铁具名 Flows 客户生产环境(Flows)成效未公开量化低 — Cognite 案例研究页仅 JS 可见
Cosmo Energy油气炼化3 座炼厂;250B 个数据点;5M 份工程文档生产环境数据集成规模得到确认中 — Cognite 在 Forrester 博客发文
bp油气一体化多资产;范围未完全披露生产环境客户页面和 moonshot 文章均点名中 — Cognite 官方来源;bp 未被独立引用

证据质量评级是分析师的编辑判断。按照 Cognite 或客户来源,「生产环境」指部署已超过试点阶段。成效栏为空值表示公开资料没有量化成效,不代表没有价值。

[CU007, CU011, CU012, CU013, CU017, CU018]
FU003: 客户证据矩阵

从四个维度评估具名客户证据质量:生产部署已确认、结果已量化、有独立佐证、留存可见。

证据质量评估为编辑判断。「部分」表示有一些证据,但并非完全独立。BusinessWire 新闻稿由 Cognite 发布,而非客户直接发布。

[CU007, CU011, CU012, CU022, CU023, CU018]

6.3 客户留存、扩张与商业耐久性

Cognite 披露其较新客户队列净收入留存率为 150%,说明扩张收入显著超过任何流失。公司报告 Atlas AI 自发布以来客户增长 700%,截至 May 2026,30%+ 的整体客户基础已经启用 Cognite Flows。这些数字由公司自报且未经审计,但 Forrester 分析师覆盖和具名客户部署模式提供了方向性佐证:TotalEnergies 从试点到 39 项资产的多年扩张、Aker BP 持续的多资产部署、Idemitsu 的全平台推广,都反映了耐久的企业承诺。知识图谱现在持有 80 trillion 个时间序列数据点,同比增长 48%,意味着客户数据活动持续发生,而不是静态部署。独立的总收入留存或流失率证据并未公开。未发现针对 NRR 或流失指标的监管文件或第三方审计。集中度风险存在:Saudi Aramco 同时是 Cognite 的大股东(2022 年取得 74% 股权)和最大锚定客户之一,形成所有权—客户依赖关系;这种结构并不寻常,且可能让正在评估该平台的竞争性石油公司担忧。

留存 / 重复使用 / 满意度表
指标数值分群置信度尽调问题
净收入留存率(NRR)150%(较新队列)全部客户中 — 公司披露,未审计通过客户访谈核验;按年份队列索取 NRR
ARR 订单额增长同比 36%(FY2025)全部客户中 — 公司披露交叉核对审计资料或债务契约披露
月活用户增长同比 26%全部客户中 — 公司披露厘清 MAU 定义;确认不包含 Cognite 内部用户
Atlas AI 客户采用率按 ARR 计,59% 客户使用 Atlas AI全部客户中 — 公司披露确认该比例反映活跃使用,而非仅开通许可
Flows 客户采用率>30% 的客户群已启用全部客户(May 2026)中 — 发布时公司披露核验发布后的实际使用量与已开通访问权限
总收入留存率未披露全部客户Unknown单独向 NRR 之外索取 GRR,以判断流失下限
独立评价评级G2 / Gartner / PeerSpot 无实质性评分企业买家低 — 平台显示 0 条评价或评价不可访问征求买家推荐访谈;登录后查看 Gartner Peer Insights

所有指标值均来自 Cognite 在 May 2026 的 moonshot 文章和投资人沟通中的公司披露。没有第三方审计或监管文件佐证这些数字。

[CU002, CU003, CU005, CU006, CU033]
FU002: 留存 / 重复队列

按初始部署年份示意客户留存队列,用已披露 NRR 和扩张信号作为年度留存代理。

所有队列留存值都是示意性估计,来自管理层披露的 150% NRR 和具名客户多年扩张模式;Cognite 未披露队列级留存数据。2024 和 2025 队列第 3 年数值仅为前瞻性估计。

[CU003, CU033]

6.4 竞争动态与 Go-to-Market 证据

Cognite 的客户赢单显示出大型企业 land-and-expand 模式:先在一两项资产上试点,再用多年时间扩展到数十个设施。公司与 Snowflake 和 Databricks 的生态定位,让它可以接入客户数据架构,而不是替换它们,从而降低采购摩擦。IDC MarketScape 将 Cognite 定位为全球工业 DataOps 平台 Leader,这是一项独立制作、具备第三方分量的分析师评估。CNBC Disruptor 50 入选(May 2026)释放了更广泛的科技媒体认可信号。Go-to-market 侧,Cognite 主要依靠面向工业运营商 C-suite 的直销;渠道伙伴(Snowflake、Databricks、AWS)扩大触达,但不是主要分销渠道。独立评论平台(Gartner Peer Insights、Slashdot、PeerSpot、SourceForge)相较其声称的企业客户数,用户评论稀少或几乎没有;这是独立社会证明中的重要缺口。平台强项在于大型具名客户的可引用生产部署,而不是自下而上的评论驱动采用。相对 AVEVA、OSIsoft(AVEVA PI)、AspenTech 和 Palantir 的竞争差异,落在工业知识图谱深度和 Atlas AI agent 层。

FU004: 采用 / 部署漏斗

Cognite 工业 AI 平台从行业活动认知到嵌入企业项目,形成发现到企业标准化的漏斗。

漏斗阶段转化率为示意性估计,基于行业常规和定性证据;Cognite 未披露销售管线转化指标。

[CU003, CU005, CU032]

6.5 集中度风险与证据质量评估

两项集中度风险很重要。第一,Saudi Aramco 74% 股权所有权制造结构性利益冲突:任何正在评估 Cognite 的竞争性一体化巨头(bp 也是具名客户)或国家石油公司,都必须权衡共享给 Cognite 的敏感运营数据是否会流向竞争对手控股实体。该风险由 Reuters 2022 年对收购的报道记录,Cognite 尚未公开解决。第二,油气仍约占收入 60%,使 Cognite 暴露于能源转型不确定性和行业资本纪律周期。O&G 具名客户(Aker BP、TotalEnergies、Idemitsu)的证据质量较强,Cognite 提供了带具体指标的案例研究页面,Idemitsu 和 TotalEnergies 还存在第三方或客户自有佐证。较新垂直进入者(B. Braun、Celanese、JFE Steel)的证据质量较薄:结果主要由 Cognite 声称,独立验证有限。净收入留存和客户数数据仅由公司披露,尚未被独立审计师或监管文件验证。

扩张和集中度风险表
风险 / 驱动因素性质影响当前证据尽调路径
Saudi Aramco 控股(74%)利益冲突高 — 可能阻碍与其竞争的油气巨头共享数据Reuters 2022 年报道;Cognite 未公开披露防火墙或数据治理保护措施索取正式数据治理政策;访谈非 Aramco 油气客户,了解担忧程度
油气收入集中(约 60%)垂直行业集中中 — 能源转型和油气资本周期带来需求风险公司披露;目前约 40% 来自其他垂直行业跟踪非油气 ARR 增速;按季度监测多元化进展
客户数未披露透明度不足风险中 — 妨碍独立评估队列规模未公开客户数;从 ARR 水平推断为数百家尽调时按年份队列索取客户数
流失率 / GRR 未披露留存透明度不足中 — 150% NRR 亮眼,但 GRR 下限未知已披露 NRR;GRR 和 logo 留存未披露按年份索取 logo 流失率;通过客户访谈核验
头部账户先落地再扩张集中客户集中低至中 — 前 3-5 大账户可能贡献显著 ARR 份额未披露 Herfindahl 指数或前 10 大客户 ARR 占比索取前 10 大客户 ARR 集中度;与工业 SaaS 同行对标

风险评级是定性编辑判断。Aramco 所有权带来的利益冲突风险有公开报道记录;其余风险来自公开披露缺失的推断,并非已确认的负面事件。

[CU034, CU035, CU036, CU038]

6.6 图表与证据

Chapter 07

07风险

7.1 风险概览

Cognite 位于工业运营、企业软件和生成式 AI 的交叉点;这种组合把多重风险维度同时集中在一起。截至 May 2026,公司报告 FY2025 ARR 为 $170M+、同比增长 36%,但仍是私营公司,没有经审计公开财务报表,因此烧钱速度、毛利率和现金跑道存在根本不透明。 风险热力图(FR001)按严重性和发生可能性绘制每个主要风险类别。剩余严重性最高的风险,是油气行业集中度,以及资本充足的既有厂商和超大规模云厂商带来的竞争替代。风险传导图(FR002)展示上游冲击——油价崩跌、Aker 治理变化或 AWS 中断——如何传导到收入、客户续约,最终传导到融资能力和估值。三类风险足以打破投资逻辑:(a)Aker BP 或 TotalEnergies 作为锚定客户流失且没有替代收入;(b)发生影响工业控制数据的重大安全事件;(c)油价持续跌破约 $60/bbl,导致大型 O&G 运营商推迟或取消数字化转型项目。 本章依据监管文本(GDPR、EU DSA、EU AI Act 条款)、第三方安全评估和行业媒体风险报道。未披露诉讼、具体客户合同赔偿条款和精确现金消耗方面缺少证据;这些缺口均在证据缺口登记中明确记录。 [CR023, CR024, CR025, CR033]

FR001: 风险热力图——可能性与剩余严重性

二维矩阵把 Cognite 主要风险类别映射到可能性(列)和缓释后的剩余严重性(行)。单元格值标出每个象限的主导风险。

可能性和严重性是基于截至 2026 年 5 月可用公开信息的定性分析师评估。未使用精算损失概率模型。

[CR004, CR008, CR015, CR023, CR028, CR033]
FR002: 风险传导图

有向无环图展示上游冲击如何穿过 Cognite 业务,影响收入、客户留存、利润率、融资和估值。

边标签为示意性因果机制。为便于阅读,R9(收入)与 R10(流失)之间的反馈循环已简化。

[CR009, CR016, CR023, CR028, CR034, CR039]

7.2 监管与法律风险

Cognite 作为数据处理方,依据 GDPR 为其欧盟客户处理工业运营数据,客户包括挪威、荷兰、德国和英国的主要运营商。挪威数据保护机构(Datatilsynet)是 Cognite 挪威实体的牵头监管机构;尽管公司在 2025 年把全球总部迁至 Tempe, Arizona,挪威实体仍是其很多欧盟业务的法律基础。 EU AI Act(Regulation 2024/1689,August 2024 生效)把用于关键基础设施的 AI 系统——包括炼油厂、化工厂和能源电网——归为高风险系统,要求部署前完成合格评定、风险管理文档和人类监督安排。Cognite 的 Atlas AI agents 会对生产运营和维护排程提出自主建议,具体是否落入该类别取决于运营商客户授予的自主程度。适用性取决于客户配置,以及 Atlas AI 输出在多大程度上未经人工复核就被执行。截至 May 2026,这仍是一个活跃监管解释风险,尚无明确先例。 NIS2(Directive 2022/2555)于 October 2024 生效,对基本服务运营商及其关键技术供应商施加网络安全义务。Cognite 作为能源和关键制造客户的 IT/OT 集成层,可能触发 Article 21 下的第三方 ICT 供应商义务,要求客户对 Cognite 部署进行安全审计,并通过合同约束 Cognite 遵守事件通知时限。 截至 May 2026 的检索未发现 Cognite 涉及公开披露的监管执法、诉讼或法律纠纷。没有披露并不等于法律状态干净;私营公司并不总是披露未决争议。这构成下文记录的重大证据缺口。 监管 / 法律风险登记(TR001)按严重性排列每项暴露,并映射到尽调动作。 [CR001, CR002, CR003, CR004, CR005, CR006]

监管 / 法律风险登记表
风险 / 规则法域状态可能性严重性缓解措施剩余敞口尽调路径
GDPR 数据处理者义务欧盟 / 挪威生效 – 持续确定客户合同包含 DPA;ISO 27018 认证低至中审查 DPA 模板;确认子处理者名单
关键基础设施中的 Atlas AI 智能体 EU AI Act 高风险分类欧盟监管过渡 – 新部署自 Aug 2026 起适用合规性评估推进中;产品内置人工监督功能中至高获取律师对 Atlas AI 风险等级的意见;审查合规性路线图
NIS2 第 21 条下第三方 ICT 供应商义务欧盟生效 – 指令自 Oct 2024 起生效SOC 2 Type II;ISO 27001;事件通知 SLA确认客户 MSA 中的 NIS2 审计权;审查通知时限
面向美国能源客户的 NERC CIP v.5 合规支持美国生效 – 客户驱动义务产品安全文档记录了 NERC CIP 支持低至中确认最新产品版本中的 NERC CIP 控制映射
面向沙特阿拉伯 / 阿联酋客户的技术出口管制美国(EAR)/ 欧盟生效 – 按法域监测低至中已对 Aramco 合同做法律审查;未标记双用途技术确认每个海湾部署均经过 OFAC / BIS 筛查
AI 驱动运营建议引发事故的专业责任多个法域潜在风险 – 无已知索赔合同责任限制;赔偿例外条款中至高审查 MSA 责任上限;确认 AI 专业责任保险可得性
Norwegian Datatilsynet 对 OT/HR 场景个人数据的监管挪威 / EEA生效 – 监管法域低至中已指定挪威 DPO;维护 GDPR 第 30 条记录确认迁址后 DPO 范围覆盖美国总部实体

状态截至 May 2026 研究;监管时间表可能变化。严重性按缺少缓解措施时最坏情形的财务和声誉后果评估。

[CR001, CR002, CR003, CR004, CR005, CR007]

7.3 运营与安全风险

Cognite 平台建立在横跨 AWS、Microsoft Azure 和 Google Cloud Platform 的多云基础设施上。这种架构降低了单一云中断暴露,但也让公司同时依赖三大云厂商。May 7, 2026 AWS 中断影响 150 多项云服务,说明超大规模云基础设施退化时,SaaS 平台会承受系统性风险;Cognite 的具体 SLA 承诺和冗余故障切换程序未公开记录。 Cognite 持有 ISO 27001:2022、SOC 2 Type II、ISO 27018 和 ISO 9001 认证,均在其 Trust Center 公开引用。平台还支持能源行业客户使用 NERC CIP v.5 和 IEC 62443 工业网络安全标准。UpGuard 持续监控 Cognite 外部安全态势;完整风险评级报告在付费墙后,本研究周期无法做独立第三方验证。没有任何公开披露的安全 breach 值得注意,但不能把它当作非公开事件记录干净的确认。 工业 OT 数据带来独特攻击面:Cognite 摄取来自 SCADA 系统、DCS 平台和 historian 数据库的数据,这些数据包含运营设施的实时工艺状态。若 breach 影响实时工业过程数据,后果会超出典型企业软件事件,触及安全关键层面。Cognite 的 IEC 62443 认证部分缓解了这一暴露,但客户侧集成层不在 Cognite 直接控制范围内,残余风险仍在。 企业级 CDF 部署通常需要 6–18 个月实施工作,拉长了集成风险窗口;这期间数据管道只经过部分验证。实施失败——数据模型不匹配、historian 连接器失败,或 CDF 在负载下性能退化——在工业 DataOps 平台的行业媒体报道中被作为延迟价值实现最常见原因反复提及。 [CR009, CR010, CR011, CR012, CR013, CR014]

运营和安全风险登记表
故障模式可能性严重性缓解成熟度剩余敞口未解决缺口
多云基础设施宕机(AWS / Azure / GCP 事件)中 – 已设计多云冗余SLA 承诺和故障切换 RPO/RTO 未公开披露
OT 数据管道入侵或外泄(SCADA / DCS 数据)低至中严重高 – 已具备 ISO 27001、SOC 2、IEC 62443中至高第三方渗透测试结果无法独立核验;UpGuard 完整报告需付费
企业 CDF 实施失败或长期延误中 – Success Tracks 专业服务项目实施失败率未公开披露;只有行业轶事数据
高容量 OT 数据负载下平台性能下降低至中中 – 知识图谱到 2025 年扩展至 80T+ 个时序数据点低至中未发布 SLA 正常运行时间指标;StatusGator 监测仅为第三方来源
AI 智能体安全事件(Atlas AI 建议导致运营故障)严重低至中 – 人工监督设计原则;EU AI Act 合规推进中无 AI 安全事件历史可查;合同赔偿范围未披露

可能性和严重性是基于企业工业 SaaS 平台的公开资料和行业基准做出的定性评估。未发现已确认的历史事故。

[CR009, CR010, CR011, CR012, CR013, CR014]

7.4 合作伙伴与依赖风险

Aker ASA 是 Cognite 的多数股东,也是事实上的战略后盾。Aker 的 Q1 2026 净资产价值超过 NOK 110 billion,说明它仍有财力支持 Cognite。但由工业集团控股也带来治理风险:战略方向、资本配置、退出决策都会受到 Aker 组合优先级的强影响,而这些优先级未必与纯企业软件增长策略一致。Aker 自身财务状况、领导层或战略重点一旦变化,Cognite 获取资本的能力和战略独立性都可能受到实质影响。 Saudi Aramco 在 February 2022 收购 Cognite 7.4% 股权。这带来与全球最大石油公司的高价值商业关系,也让 Cognite 在阿拉伯半岛拥有锚定客户;但它同样引入潜在治理复杂性:Aramco 的战略优先级可能影响 Cognite 产品路线图,尤其是 O&G 专用功能,从而拖慢向其他垂直行业多元化。与此同时,美国和欧盟制裁格局不断变化,Cognite 在沙特阿拉伯的业务也暴露在出口管制和地缘政治风险之下。 Cognite 与 NVIDIA 围绕 NV-Tesseract 集成的合作,直接依赖 GPU 可用性和 NVIDIA 的 NIM 微服务路线图。在 2025–2026,GPU 供应约束已被行业充分记录;NVIDIA 商业 API 可用性一旦中断,Atlas AI 的时间序列预测能力就会下降。Snowflake 和 Databricks 合作关系带来双向数据流依赖,定价争议或竞争冲突都可能影响联合销售管线。Microsoft Azure 和 AWS 既是平台合作伙伴,也可能在扩展自身工业 AI 产品时成为横向竞争者。 依赖图(FR003)把这些关键上游依赖及其向客户价值交付的传导路径可视化。 [CR016, CR017, CR018, CR019, CR020, CR021]

合作伙伴和依赖风险登记表
依赖项交易对手角色集中度失效情景严重性缓解措施剩余敞口
控股股东 / 战略后盾Aker ASA资金提供方和治理锚点关键 – 控股股权Aker 财务承压,或战略转向远离软件Aker NAV >NOK 110B(Q1 2026);Cognite 拥有独立管理层
战略少数股东 / 锚定客户Saudi Aramco7.4% 股权 + 企业客户高 – 战略一致性风险Aramco 施压产品路线图;地缘政治扰动中至高按公平交易原则签订商业合同;独立管理董事席位
工业 AI 模型推理 / 预测NVIDIA (NV-Tesseract)Atlas AI 时序数据的核心 ML 推理高 – 单一供应商 NIM 集成GPU 供应中断;NVIDIA API 废弃或调价中至高通过 autoLLM 回退 LLM 路由;标准 API 合同
联合销售和云基础设施Microsoft Azure / AWS / GCP平台基础设施和市场联合销售高 – 三项独立依赖云价格上调;联合销售计划变化;竞争对手进入多云架构;已签署云市场协议低至中
数据平台联合销售与互操作性SnowflakeData Cloud 集成与联合市场拓展Snowflake 扩张分析业务带来竞争冲突;合作关系可能重谈低至中合作于 Jan 2026 宣布;合同保护条款属标准配置
标杆客户与平台验证Aker BP / TotalEnergies / bp 等客户锚定企业客户;公开案例研究高 — 漏斗顶端信号合同不续签;公开披露负面绩效长期企业合同;多站点部署抬高切换成本

依赖严重程度按一个问题评估:如果交易对手关系发生重大中断,Cognite 维持收入增长和产品交付的能力会受多大影响。交易对手财务健康状况基于截至 May 2026 的公开来源评估。

[CR016, CR017, CR018, CR019, CR020, CR021]
FR003: 依赖图——关键外部依赖

有向无环图展示 Cognite 关键上游依赖,以及这些依赖通向客户价值交付和收入的路径。

[CR016, CR017, CR018, CR019, CR020, CR021]

7.5 财务、人员与执行风险

Cognite 的财务风险主要来自行业集中。直到 FY2024,油气客户仍贡献约 60% 收入;公司称 FY2025 收入中已有 40% 来自 O&G 之外。油价如果持续跌破约 $60/bbl ——历史上大型资本开支项目推迟的阈值——将立刻压缩 Cognite 核心客户群的数字化转型预算。公司没有公开披露客户行业分布或合同期限,外部很难独立分析集中度。 Cognite 上一次公开披露估值是 $1.6B,可追溯到 2021 独角兽里程碑;之后没有以新价格完成的股权融资。2024 年 TCV 领投的 $300M 投资属于新增资本,但估值未披露。任何未来融资轮或战略交易都会重置估值预期;平轮或下轮会通过期权稀释削弱员工留任,也会向评估供应商长期存续能力的企业买家释放竞争压力信号。 2025 年全球总部从奥斯陆迁至亚利桑那州坦佩,是一项重大人员风险。Cognite 的工程根基在挪威,创始技术团队(包括联合创始人、CTO Geir Engdahl)位于奥斯陆。总部迁移让公司战略(美国)和技术执行(挪威)形成地域切分,可能抬高协作成本,也可能导致未迁移的挪威资深工程师流失。员工数在 2025 增长 21%,超过 800 人;在新地域快速扩张,可能稀释组织知识和文化。 产品节奏快既是竞争优势,也是执行风险。Cognite 接连推出 Atlas AI(2024)和 Cognite Flows(May 2026);每次重大发布都需要大量一线销售赋能、客户导入能力和支持体系。如果商业化执行跟不上产品创新,成交率和价值兑现时间指标都会走弱。 [CR023, CR024, CR025, CR026, CR027, CR028]

人员与执行风险登记表
角色 / 职能依赖或缺口可能性严重程度缓释措施尽调路径
CEO(Girish Rishi)自 ~2022 起担任非创始人 CEO;组织知识集中在一位领导者身上领导梯队较深;董事长兼 CEO 结构增加问责确认 CEO 股权归属时间表和留任协议
创始人 / AI CTO(Geir Engdahl)联合创始人,也是 AI 架构锚点;总部迁址后仍驻 Oslo低至中Technical Fellows 项目(2026 届已宣布);AI 工程团队有纵深核验技术路线图归属不依赖单一创始人
挪威工程团队约 300+ 名工程师可能驻 Oslo;总部迁至美国带来流失风险中高Cognite 保留 Oslo 研发办公室;挪威薪酬具竞争力索取按地域划分的员工数;总部迁址前后流失率
市场拓展执行(美洲扩张)借 Tempe 新总部快速扩张美国市场,需要补齐新的 GTM 人才Bill Hendricks 出任美洲区总裁;一线销售领导经验充足审查美洲管线建设指标;按地区查看销售配额达成率
产品发布节奏(Atlas AI + Flows + 路线图)18 个月内推出三条主要产品线;一线赋能和支持扩容存在风险季度发布节奏已建立;新产品支持通道已建立核验 Flows 发布后支持工单 SLA 达成情况;客户上线指标

评估基于公开领导团队页面、新闻稿,以及 Series-C 至 IPO 前企业 SaaS 公司的行业基准。非公开留任数据不可得。

[CR029, CR030, CR031, CR032]

7.6 缓释措施、监控指标与投资逻辑击穿触发点

几项结构性缓释因素降低了最严重风险情景的发生概率。Cognite 较新客户 150% NRR、Atlas AI 覆盖 59% ARR,说明平台有黏性,近期非计划流失的概率较低。ISO 27001、SOC 2 Type II 和 IEC 62443 认证提供机构级合规支架,能够应对采购中最常见的安全门槛。多云架构(AWS/Azure/GCP)降低单一供应商故障风险。Aker ASA 的 NOK 110B NAV 证明其资产负债表厚度足以支持 Cognite 穿越较长增长阶段。 量化监控指标可以提前预警关键风险。油价期货如果超过 90 天低于 $65/bbl,应重新评估 O&G 客户续约概率。单季 NRR 读数低于 120%,将提示账户收缩先于流失出现。Gartner Peer Insights 分数若在两个季度内下滑超过 0.5 分,应标记客户满意度恶化。CTO、CPO、CFO 若在 12 个月内相继离职,应触发人员风险升级。 投资逻辑击穿触发点,是会从根本上挑战投资逻辑的事件:失去 Aker BP 或 TotalEnergies 客户且看不到替代管线,Cognite 平台被认定造成重大 OT 安全事件,油价持续 12+ 个月低于 $50/bbl,或发现未披露且有重大财务暴露的诉讼。若确认 Cognite 被大型既有厂商(AVEVA/Schneider、SAP、Microsoft)以低于 2021 $1.6B 估值标记收购,将意味着账面回报为负。 缓释与终止标准表(TR005)将每项主要风险映射到可监控触发点和建议行动含义。 [CR039, CR040, CR041, CR042]

缓释措施与否决标准表
风险可监测触发项阈值 / 事件行动含义
O&G 行业收入集中油价期货(Brent/WTI 90 天滚动均值);O&G 客户续约率Brent 低于 $65/bbl 持续 90+ 天,或 O&G 续约率低于 85%用 O&G ARR 减值 20-30% 压测模型;评估多元化管线推进速度
NRR 下滑季度 NRR 数据(所有客户和新近队列)NRR 连续两个季度低于 120%调查账户健康驱动因素;若趋势再持续一个季度,标记投资逻辑风险
Aker ASA 治理变化Aker ASA 公开公告;NAV 趋势;董事会构成变化Aker 宣布减持 Cognite、启动出售流程或战略转向立即尽调新买方 / 控制实体;重新评估治理风险
Cognite 平台重大安全事件StatusGator 中断报告;CVE 披露;工业网络事件媒体报道任何已确认、归因于 Cognite 平台的客户 OT 数据泄露立即上升风险等级;全面取证审查 SLA、责任和客户流失影响
被超大规模云厂商或既有厂商竞争替代Microsoft / AWS / GCP 工业 AI 产品发布;AVEVA / AspenTech 并购;Palantir AIP 扩张公告锚定客户为新站点选择竞品平台;Gartner MQ 排名下滑评估功能对等差距;判断价格护城河是否仍足以支撑增购
关键人物离任(CEO、CTO/AI)LinkedIn 资料变化;SEC 文件(若 IPO 后);新闻稿CEO 或联合创始人 CTO 在同一 12 个月窗口内离任审查继任计划;评估产品路线图连续性风险
EU AI Act 或 GDPR 下的监管执法EU DPA 公开登记;ENISA 执法通知;行业媒体法律报道任何针对 Cognite 的监管调查启动,或触发正式客户通知要求聘请 EU 监管律师;评估财务敞口和产品变更要求

触发项仅供指示,应由投资人结合被投公司汇报节奏和市场背景校准。阈值为分析师估计,并非 Cognite 披露的目标。

[CR028, CR033, CR039, CR040, CR041, CR042]

7.7 附录图表

Chapter 08

08估值

8.1 投资逻辑与建议

Cognite 处在两条长期趋势的交汇处——工业资产数字化和企业 AI 智能体采用——并拥有差异化的工业知识图谱,单个竞争对手很难端到端复刻。投资逻辑建立在四个支柱上:(1) 先落地再扩张模型已经被验证,较新客户 NRR 达 150%,Atlas AI 到 Q1 2026 覆盖 72% 客户 ARR;(2) 平台架构具备真实切换成本,来自多年数据语境化和嵌入式 AI 工作流依赖;(3) ARR 预订增长从 FY2025 的 36% 加速到 Q1 2026 的 57%,说明 Atlas AI 增购周期正在跑通;(4) 客户基础是蓝筹级,包括 Saudi Aramco、bp、TotalEnergies、Aker BP 和 B. Braun,其中数家是多年期锚定客户。 反向逻辑主要来自估值不透明和财务透明度缺口。最近一次确认估值为 $1.6 billion,定于 May 2021——当时 Atlas AI 尚不存在,June 2024 的 ~$300M 轮尚未发生,收入规模也只有当前的大约一半。投资者若不知道 2024 轮投后估值、毛利率结构(估计为 60–75%,但未披露)、绝对 ARR(增长率标题的分母),以及「较新客户」队列之外的全客户基数 NRR,就无法给持仓定价。Saudi Aramco 的 7.4% 股权和 Aker ASA 的多数持股带来退出悬置和潜在战略冲突风险;没有治理披露,很难定价。C3.ai 的公开市场轨迹——EV/Revenue 倍数从 2021 的 30x+ 压缩到 2025 的 8–12x——为工业 AI SaaS 类别提供了警示样本。 建议:观察。Cognite 在产品、客户验证和增长轨迹上获得强定性评分,但在毛利率、绝对 ARR、全客户基数 NRR、2024 轮估值和客户集中度数据披露前,无法落到具体入场价格。若入场价等于或低于 2021 $1.6B 标记,下行保护中等;若达到 2021 标记的 2–3×(与基准到乐观情景区间一致),在没有确认毛利率可持续性之前,风险 / 回报不够有利。 [CV001, CV002, CV005, CV009, CV010, CV011]

建议摘要表
维度评估支撑证据置信度
建议观察产品、客户、增长证据扎实;估值不透明,压住投资判断信心
风险评级估值陈旧、利润率未披露、O&G 集中、治理悬而未决
估值立场未知 / 偏高2021 年后无估值标记;按当前规模,2021 年 $1.6B 标记可能低于基准情景价值
收入增长质量强 — 已确认>$170M FY2025;ARR +36% FY2025,Q1 2026 加速至 +57%
利润率画像Unknown毛利率未披露;估计 60–75%(工业 SaaS 基准)
入场纪律以 2021 年估值标记(≤$1.6B)或更低价格进入,守住下行保护;若未披露利润率,避免 >$2.5B

建议为观察,前提是关键尽调披露尚待补齐。风险评级反映估值和透明度缺口,不反映产品或市场质量。置信度为作者基于可得证据的评估;关键财务输入(毛利率、绝对 ARR、2024 轮估值)未公开披露。

[CV001, CV005, CV009, CV010, CV013, CV016]
投资逻辑与反向逻辑表
论点强度证据哪些因素会改变判断
正向:150% NRR(新近队列)和 72% Atlas AI ARR 渗透率确认了落地扩张经济性公司新闻稿;Q1 2026 KPI 披露全量客户 NRR 低于 110%,或总留存率低于 85%
正向:差异化工业知识图谱,尚无直接全栈竞品中等IDC MarketScape 领导者;Verdantix Green Quadrant 领导者;Forrester TEI 465% ROIAVEVA CONNECT、Microsoft Fabric 或 Databricks 匹配全栈图谱 + 智能体能力
正向:ARR 预订增速加快(36% → 57% YoY),显示 Atlas AI 增购奏效中等January 2026 和 Q1 2026 KPI 新闻稿任意连续两个季度增长放缓至 25% 以下
正向:蓝筹客户群(Aramco、bp、TotalEnergies)且有多年部署证据TotalEnergies 39 项资产、3 年扩展;Forrester 博客;bp/Aramco 客户页面锚定账户出现重大客户流失,或单一客户集中度超过 35%
反向:估值不透明——2021 年后无标记、无毛利率、无绝对 ARR关键风险无公开披露;PitchBook 需付费;CNBC 未引用当前估值向投资人共享完整财务披露包(毛利率、ARR、股权结构表)
反向:油气行业集中(约 60% 收入)带来周期性需求风险重大风险Cognite FY2025 KPI 发布;ch4 财务分析非 O&G 收入占比持续超过 50%
反向:C3.ai 倍数压缩先例:30x+ EV/Rev(2021)→ 8–12x(2025)中等StockAnalysis.com C3.ai 财务数据;C3.ai SEC 10-KPalantir 式政府收入锚点,可隔离估值倍数;或收入持续增长 40%+
反向:Aker ASA 控股 + Aramco 7.4% 持股限制退出可选项重大风险Reuters Aramco 持股报道;TCV 新闻稿;Cognite 2021 公告Aker ASA 治理清晰,并证明愿意支持 M&A 退出

证据质量不一:正向论点部分由公司控制的披露支撑;估值不透明的反向论点是结构性的——没有公司披露,再多搜索也无法解决。投资逻辑和反向逻辑的证据质量并不对称。

[CV001, CV005, CV010, CV011, CV016, CV022]
FV001: 推荐逻辑流

从规模与增长证据,经风险与估值证据,串到跟踪建议;同时展示哪些输入强化投资论点,哪些输入形成核验缺口。

流程反映作者截至 2026 年 5 月基于证据作出的判断。所有优势输入都由公司新闻稿或第三方来源证实;风险输入来自已确认的结构性缺口或有记录的可比先例。

[CV001, CV005, CV009, CV010, CV011, CV016]
FV004: 投资 KPI 评分卡

截至 2026 年 5 月,基于可得证据,对 Cognite 七个投资维度给出可提交 IC 的评分。分数反映证据质量,而非愿景式轨迹。

定性评分是本分析师截至 2026 年 5 月基于证据作出的评估。没有公开可得的独立量化基准覆盖所有维度。标记为「未知」或「低」的项目反映真实证据缺口,而不是负面判断。

[CV009, CV010, CV011, CV016, CV027, CV036]

8.2 融资、估值与资本结构

Cognite 的融资历史包括两轮已披露的大额融资。May 2021,TCV 领投 $150M 投资,公司估值达到 $1.6 billion,TCV 的 Jake Reynolds 加入董事会。按公司新闻稿说法,这是当时欧洲最大的 SaaS 融资轮。June 2024,TCV 领投第二轮,金额被描述为约 $300M;该轮投后估值从未公开披露。再加上 Accel 早期投资(Cognite 投资者身份由 Accel 投资组合页面确认),已知累计融资至少 $450M。CNBC 的 May 2026 Disruptor 50 画像称融资额为 $300M,可能只反映 2024 轮,而非累计资本。 股权结构实质上不透明。Aker ASA 持有 Cognite 多数股权;截至 May 2026,确切比例未披露。Saudi Aramco 在 February 2022 从 Aker BP 收购 7.4% 股权,成为战略客户股东:它同时提供收入和股权上行,但也带来集中度和影响力风险。TCV 通过 Jake Reynolds 拥有董事会席位。Aker ASA 多数控制、Aramco 战略持股和 TCV 少数股权叠加,形成复杂治理结构,可能让标准收购退出更难,并可能需要多个收购方同意。 根据 Norwegian Registry(Brønnøysund Register Centre),Cognite AS(org. nr. 918274758)提交的账目显示,FY2024 收入约 NOK 1.67 billion(按 2024 平均汇率约 ~$158M USD),净亏损约 NOK 547M(约 ~$52M USD)。这些数字只覆盖挪威法人实体;按 Cognite January 2026 新闻稿,全球合并集团收入在 FY2025 超过 $170M。隐含集团层面烧钱速度估计为 $60–90M/year,说明在烧钱没有实质加速的情况下,June 2024 的 ~$300M 轮可提供到 mid-2027 至 mid-2029 的现金跑道。由于缺少合并现金流披露,这一估计高度不确定。 对任何投资者来说,过时估值标记是最有操作意义的单一缺口。按约 $200–220M 的 NTM 收入(由当前增长轨迹推导),如果 Cognite 以 2021 $1.6B 标记入场,意味着约 ~7–8x NTM Revenue 倍数,处在高增长工业 AI 平台适用可比区间的低端,提供一定下行纪律。任何高于 $2.5B 的入场价都意味着 11–12.5x NTM Revenue,需要确认毛利率可持续性和全客户基数 NRR 才能支撑溢价。 [CV001, CV002, CV003, CV004, CV006, CV007]

FV002: 估值对 NTM 收入倍数的敏感性

在 NTM 收入为 $200M 和 $220M 两种情况下,展示 Cognite 在一组 EV/NTM 收入倍数下对应的隐含企业价值,说明估值会如何随倍数扩张或压缩而变化。

根据 FY2025 披露的 >$170M 和 15–30% 的假设 NTM 增长,NTM 收入估算为 $200–220M。倍数来自可比分析;毛利率未披露,模型中假设不变。估值为稀释前口径;优先股堆叠未知。

[CV009, CV024, CV025, CV029, CV030, CV031]
FV003: 各情景估值与回报区间

列出悲观、基准、乐观各情景的企业价值低高区间,并标示 2021 年最后确认的估值标记,作为下行保护参照。

区间按估计 NTM 收入 $200–220M × 可比分析中匹配各情景的倍数区间计算。2021 年 $1.6B 标记是唯一公开参照点;2024 年融资轮的投后估值未披露。所有数值都是企业价值估算,未计入未知的优先股权利负担、期权池或债务。

[CV001, CV025, CV029, CV030, CV031]

8.3 可比公司分析

没有一家上市公司能完全对应 Cognite。最相关的可比公司分成两层:(1) 高增长企业 AI 和数据平台(Palantir、C3.ai),它们与 Cognite 共享 AI 原生定位和销售画像,但市场组合和客户画像差异很大;(2) 工业软件既有厂商(PTC)和已被收购的平台(AVEVA、AspenTech),它们与 Cognite 共享垂直行业焦点,但增长和利润率画像不同。 从产品定位看,C3.ai(NYSE: AI)是最接近的上市可比公司——面向工业运营的企业 AI 应用——但有重要限制。C3.ai FY2025 收入约 $310M,同比约 25% 增长。Cognite 的 ARR 预订增长(36–57% YoY)明显更快,支撑其相对 C3.ai 当前约 8–12x EV/NTM Revenue 倍数享有溢价。关键在于,C3.ai 的公开市场历史提供了警示数据点:EV/Revenue 倍数从 December 2020 超过 30x 的峰值,到 2025 降至 8–12x,说明当增长不及预期时,情绪驱动的工业 AI 倍数会迅速压缩。C3.ai 向 SEC 提交的 FY2025 10-K 年报确认净亏损约 $214M——绝对亏损高于 Cognite 估计的集团层面烧钱额,而收入规模约为 Cognite 的两倍。 Palantir Technologies(PLTR)截至 mid-2026 的 EV/NTM Revenue 倍数高于 30x,但不能作为有效的直接可比公司:Palantir 超过 40% 收入来自美国政府和国防合同,利润率结构不同,并受益于 Cognite 无法获得的美国国防 AI 支出顺风。Cognite 的可比倍数应较 Palantir 当前标记作实质折价。 PTC Inc.(NASDAQ: PTC)是规模化工业软件基准,FY2024 收入约 $2.3B,EV/NTM Revenue 倍数约 8–10x。PTC 增速较慢(有机增长 8–12%),意味着 Cognite 因更高增长应享有溢价;但这部分溢价要被 Cognite 私营公司不透明和油气集中度部分抵消。收购先例提供区间参考。AVEVA(Schneider Electric 以约 £9.5B 收购,January 2023 完成)隐含收购倍数约为 5–7x 过去十二个月收入。AspenTech(Emerson 在 2022 以约 $11B 估值交易收购)在交割时隐含 15–18x 过去十二个月收入——该溢价由 AspenTech 高毛利率和黏性流程仿真装机基础支撑。这些先例倍数夹住适用区间:Cognite 以 8–14x NTM Revenue 退出是可辩护的;考虑增长轨迹,低于 6x 将属于低于市场定价。 独立工业 AI 分析机构 Verdantix 在其 Industrial AI Analytics Green Quadrant(付费报告,2025)中将 Cognite 评为领导者,确认其在竞争格局顶部的第三方定位。PitchBook 截至 May 2026 的 Cognite 数据受访问限制;标准研究渠道无法获取更新的私人轮估值或老股标记,这本身就是负面的透明度信号。S&P Global Market Intelligence 对工业软件估值数据设置付费墙,阻止研究者从一手数据独立推导 NTM 中位倍数。 [CV017, CV018, CV019, CV020, CV021, CV022]

可比估值表
可比对象类型近似收入(最近 FY)EV / NTM 收入倍数(或收购倍数)收入增长(YoY)与 Cognite 的相关性关键限制
C3.ai(NYSE: AI,上市可比)上市公司 — 纯工业 AI SaaS~$310M (FY2025)8–12x NTM 收入~25% YoY最接近的上市可比:面向工业运营的企业 AI;垂直方向相似增长慢于 Cognite;没有工业知识图谱;偏重美国市场;倍数已从 2021 年峰值大幅压缩
Palantir Technologies (PLTR)上市公司 — 数据 / AI 平台~$2.87B (FY2024)>30x NTM 收入(AI 溢价)~27% YoY高增长 AI 数据平台;品牌认知强;可比炒作周期40%+ 收入来自美国政府;防务 AI 溢价不适用;Cognite 对 Palantir 倍数应打 >50% 折扣
PTC Inc. (NASDAQ: PTC)上市公司 — 工业软件~$2.3B (FY2024)8–10x NTM 收入~10% YoY(有机)规模化工业软件基准;IoT / 工业数据邻近增长率低得多;无 AI 原生架构;无知识图谱;Cognite 相对 PTC 的增长溢价需打折
AVEVA(Schneider Electric 于 Jan 2023 收购)M&A 先例 — 工业 DataOps 平台收购时过去十二个月收入 ~£1.5B(~$1.9B)~5–7x 过去十二个月收入中个位数垂直方向直接可比:工业 DataOps、资产密集型行业遗留架构;增长较低;无 AI 智能体层;收购发生在 AI 智能体浪潮前
AspenTech(Emerson Electric 于 2022 年收购)M&A 先例 — 工业仿真 / 分析 SaaS交割时过去十二个月收入 ~$600M(FY 约 ~$700M)~15–18x 过去十二个月收入~15% YoY高毛利工业软件溢价;粘性流程仿真装机基础毛利率高于 Cognite(估计);流程仿真与 DataOps 买方不同;溢价未必可迁移
Cognite(最近公开估值标记,May 2021)私有公司 — 最近确认估值>$170M(FY2025,当前规模)$1.6B(2021 年估值标记);当前估值未披露ARR 增长 36–57%(FY2025–Q1 2026)标的公司;2021 年估值标记对应约 $70–90M 收入,隐含 18–22x;按当前规模,2021 年估值标记隐含 8–9x估值标记已滞后 5 年;2024 轮投后估值未披露;无毛利率可供基准比较

上市公司倍数反映截至 May 2026 运行日期的 StockAnalysis.com 财务数据;EV/NTM 收入倍数为近似值,基于已发布财务和公开市场数据。AVEVA 和 AspenTech 的收购倍数来自交易公告(Schneider Electric;Emerson;BusinessWire)。Cognite 私有估值由作者根据披露收入规模与 2021 年估值标记估算。若缺少美国政府合同类比,Palantir 倍数不能作为 Cognite 的上限。

[CV017, CV018, CV019, CV020, CV021, CV022]

8.4 情景分析——乐观、基准与悲观情景

三个情景均用未来十二个月(NTM)收入乘以适配情景的 EV/NTM Revenue 倍数建模。NTM 收入基数估计为 $200–220M,来自 >$170M FY2025 数据,并保守延续当前 36–57% ARR 增长轨迹,同时假设 NTM 增长均值回归至 30–40%。毛利率未公开披露,各情景均沿用工业 SaaS 基准区间 60–75%;该输入一旦变化,投资逻辑会发生实质偏移。 悲观情景($1.0–1.4B):倍数压缩至 5–7x NTM Revenue,理由包括:(a) 油气市场持续下行,削弱 Aramco 和 bp 扩张;(b) Atlas AI 无法在既有装机基础之外商业化(渗透率跌破 ARR 的 50%);(c) ARR 增长放缓至 20% YoY 以下;或 (d) 工业 AI 倍数跟随 C3.ai 式轨迹整体压缩。按 5–7x × $200M NTM 收入,Cognite 隐含企业价值为 $1.0–1.4B,低于 2021 $1.6B 标记;对以 2021 价格或更高价格进入的投资者而言,这是下轮情景。 基准情景($1.6–2.6B):对 $200–220M NTM 收入应用 8–12x NTM Revenue 倍数区间。该情景假设:ARR 继续以 30–40% 增长;Atlas AI 渗透率保持在 ARR 的 60% 以上;混合毛利率为 65–70%;Aramco 或 bp 不发生重大客户集中度事件。$1.6B 底部对应当前规模下的 2021 估值标记(「无溢价」退出),而 $2.6B 需要通过毛利率和留存披露验证 12x 倍数。 乐观情景($3.0–4.4B):对 $210–220M NTM 收入应用 14–20x NTM Revenue 溢价倍数。该情景要求:(a) Atlas AI 建立可持续平台护城河,全客户基数 NRR 持续高于 130%;(b) 非 O&G 垂直扩张成功,在 24 个月内把 40% 非 O&G 收入占比推至 55%+;(c) 订阅收入占比改善,毛利率扩张至 70% 以上;以及 (d) IPO 或战略 M&A 按 AI 平台可比公司而非工业软件可比公司定价。没有美国政府合同等价物,Palantir 倍数不能作为上限。 Rule of 40 分数无法用公开数据计算。在 36–57% ARR 增长和估计 -20 至 -30% FCF 利润率(来自挪威实体 FY2024 ~$52M 净亏损和假设的集团层面 $60–90M 烧钱)下,估计 Rule of 40 分数大致横跨 6–37。这一区间很宽,符合这样一家公司的状态:增长已获确认,但盈利轨迹未知。这个不确定性是核心估值阻碍。 [CV009, CV010, CV011, CV013, CV014, CV015]

乐观 / 基准 / 悲观情景表
情景NTM 收入假设EV/NTM 收入倍数隐含企业价值关键假设下行触发项 / 概率信号
乐观$210–220M14–20x$3.0–4.4B全量客户 Atlas AI NRR >130%;非 O&G 占比 >50%;毛利率 >70%;AI 平台退出溢价概率信号:Q1 2026 ARR 增长 57% + Atlas AI 渗透率 72%;需确认毛利率
基准$200–220M8–12x$1.6–2.6BARR 增长 30–40%;Atlas AI 占 ARR >60%;毛利率 65–70%;无重大客户流失概率信号:当前披露指标与基准情景一致;未披露毛利率,不确定性很宽
悲观$180–200M5–7x$0.9–1.4B估值倍数沿 C3.ai 路径压缩;O&G 需求转弱;ARR 增长 <20%;Atlas AI 未能突破既有客户群触发项:O&G 资本开支削减周期、重大客户流失(Aramco 或 bp),或工业 AI 整体估值下修

NTM 收入按 >$170M FY2025 数据和 15–30% NTM 增长估算;精确 ARR 未披露。倍数来自可比公司分析(TV004);毛利率未披露,因此各情景中保持不变。隐含企业价值为稀释前区间;优先股堆叠和期权池未知。悲观情景下相对 2021 年 $1.6B 估值标记存在降价轮风险。

[CV009, CV010, CV013, CV022, CV024, CV025]

8.5 退出就绪度、最终尽调问题与否决触发点

Cognite 在 2026–2028 最可能的退出路径是:(1) 被大型工业自动化或企业软件既有厂商战略收购,后者需要 AI 原生工业 DataOps 能力;或 (2) 当公司达到盈利里程碑,或市场条件支持成长阶段工业 AI 上市时进行 IPO。May 2026 入选 CNBC Disruptor 50,与公司在潜在流动性事件前建立公众形象相一致。 潜在战略收购方包括 Schneider Electric(收购 AVEVA,可作为 DataOps 先例)、Emerson Electric(收购 AspenTech)、Siemens、Honeywell 和 ABB——它们都参与工业自动化竞争,也都有收购 Cognite 工业知识图谱和 Atlas AI 智能体工作台的战略理由。超大规模云厂商(鉴于 Energy and Resources Partner 关系的 Microsoft,或 AWS)也可能成为战略买方。AVEVA($9.5B)和 AspenTech($11B)两项先例 M&A 交易,为 Cognite 当前规模下战略退出的相关企业价值区间提供上下沿。 Aker ASA 的多数持股可能吓退收购方:战略买家若要取得控股权,需要 Aker ASA 同意;而 Aker ASA 可能更愿意保留与其更广泛能源投资逻辑一致的工业数据资产敞口。Saudi Aramco 的 7.4% 股权又增加一个否决维度。尽调应揭示这些治理复杂性,并在任何投资逻辑依赖近期 M&A 退出前解决。 投资前五个最关键的尽调问题:(1) 按收入流拆分毛利率(订阅 vs. 专业服务),否则无法进行 SaaS 基准比较;(2) 全客户基数 NRR 和总金额留存,以确认 150%「较新客户」NRR 是否掩盖老队列流失;(3) 绝对 ARR 和按层级拆分的客户数,以验证集中度和 ACV 分布;(4) June 2024 投后估值和当前股权结构,包括优先股堆叠和清算优先权;(5) 账面现金和 12 个月现金跑道,以评估下一次融资是否会稀释。 [CV036, CV037, CV038, CV039, CV042]

投资逻辑失效与否决触发项表
触发项阈值 / 事件对投资逻辑的传导行动含义
Atlas AI ARR 渗透率反转24 个月内按 ARR 口径,覆盖客户基础低于 50%AI 平台倍数溢价消失;乐观情景坍塌;基准情景受威胁退出 / 若投资前观察到该触发项则不投
主要锚定客户流失(Aramco 或 bp)单一客户流失超过 ARR 的 15%集中风险兑现;NRR 头部数字掩盖流失;150% NRR 指标可信度被打破立即重新评估;大概率触发退出
ARR 增长放缓连续两个季度 YoY 增长低于 20%基准情景倍数压缩至 5–8x;悲观情景被触发;现金跑道担忧上升按压缩倍数重新测算;大概率减仓或退出
披露毛利率低于 50%毛利率披露显示混合毛利率 <50%SaaS 基准失效;EBITDA 桥接需要重建模型;倍数压缩 30–50%立即复盘投资逻辑;除非收入加速抵消,否则大概率退出
降价轮融资或困境再融资下一轮投后估值低于 $1.6B,或债务带有惩罚性契约稀释风险兑现;管理层信誉受损;退出窗口变窄在最早流动性窗口退出
Aker ASA 被迫出售或战略重组Aker ASA 重组能源资产,要求以低于内在价值出售 Cognite 股份老股压力解除但价格风险上升;可能创造机会型入场作为潜在流动性事件跟踪;评估买方画像

触发项定义为可观察事件或阈值,会在不依赖其他因素的情况下实质损害投资逻辑。触发项评估是作者基于可得证据的判断;由于绝对 ARR 和客户集中度未披露,客户 ARR 阈值均为估计。

[CV011, CV016, CV033, CV034, CV038, CV039]
最终尽调问题表
主题缺失证据重要性负责人 / 尽调路径
按收入流划分的毛利率混合毛利率;订阅与专业服务拆分;FY2022–FY2025 毛利率趋势没有毛利率,就无法套用 EV/毛利润倍数;盈利路径无法建模;与 C3.ai 和 PTC 的基准比较也不精确向 Cognite CFO 索取;对照挪威登记机构成本结构交叉核验
全量客户 NRR 和总美元留存全队列 NRR(不只是新近客户);总收入留存率;按部署年份划分的流失新近客户 150% NRR 可能掩盖 Atlas AI 前部署的老 O&G 队列流失;总留存是扩张倍数的底线向 Cognite 销售副总裁 / CFO 索取;要求按部署年份做队列分析
绝对 ARR 和按层级划分的客户数当前 ARR(USD);按 ARR 层级划分的客户数(如 $1M+、$500K–$1M、<$500K);ACV 分布没有分母,增长率无法核验;没有客户数和层级分布,集中风险无法评估向 Cognite CFO 索取;对照 Forrester TEI 和客户案例研究中的收入信号交叉核验
June 2024 轮投后估值和股权结构表TCV June 2024 轮的投后估值;完整股权结构表,包括优先股堆叠、清算优先权和期权池不知道当前估值标记,就无法测算入场价格;优先权包袱影响下行保护和回报建模TCV 新闻稿或 Cognite 投资者关系;向法律顾问索取完整股权结构表
账面现金和 12 个月现金跑道最近季度现金及等价物;预计烧钱速度;如有,列明债务额度按估计集团烧钱 $60–90M 计算,June 2024 轮资金可能只剩约 2–3 年现金跑道;下一次融资时点影响入场纪律向 Cognite CFO 索取;对照挪威登记机构 FY2024 资产负债表核验实体层面现金
客户集中度确认前三大和前五大客户收入占比;Saudi Aramco 收入占总收入比例Aramco 7.4% 股权持股带来战略客户兼股东冲突;若 Aramco 超过 ARR 的 20%,集中风险应带来倍数折价向 Cognite CFO 索取;通过客户公告和 Forrester 客户访谈数据三角验证
IPO 或退出时间表董事会层面的 IPO 准备讨论(S-1 文件准备、审计师任命、CFO 上市公司经验);与明确买方的 M&A 对话退出路径就是回报机制;时间表不清会降低流动性风险定价能力;CNBC Disruptor 50 认可可能释放近期公开市场活动信号向 Cognite CFO/CEO 索取;通过 TCV 和 Accel 投资者关系渠道交叉核验
治理与 Aker ASA 退出条款Aker ASA 的治理权、拖售条款,以及对 Cognite 股份的任何优先购买权Aker ASA 多数控制可能阻止或推迟 M&A 退出;理解退出约束是建模任何收购结果的前提向法律顾问索取 Cognite 股东协议;审阅 Aker ASA 年报中的 Cognite 披露

尽调问题按对估值论证的重要性排序。第 1–4 项是投资论证的阻断项;第 5–8 项重要,但可通过替代数据源部分补齐。所有项目都需要直接接触公司或获得投资人级数据室访问权限。

[CV013, CV014, CV015, CV016, CV026, CV036]

8.6 附录图表

免责声明

本报告是基于截至 2026-05-20 公开信息生成的 AI 辅助尽调材料,不构成投资建议。Cognite 是私营公司,因此若干财务和治理结论依赖不完整披露、公司自称指标和情景分析,而非经审计的公开财务报表。

证据索引

结论
编号陈述可信度来源
CO001 Cognite is a B2B industrial AI and data platform company headquartered in Tempe, Arizona. SO002, SO012
CO002 Cognite's official about-us page states the company has been on its mission 'since 2016'; CNBC Disruptor 50 (May 2026) states Cognite was 'formed in 2016 in Oslo.' SO002, SO001
CO003 Cognite's May 2021 press release states the company 'was founded in 2017' and references 'four years ago' as the founding period. SO004
CO004 Cognite's founding year is disputed between 2016 (official website, CNBC) and 2017 (2021 press release); the discrepancy is unresolved in public sources. SO002, SO004
CO005 John Markus Lervik, a Norwegian serial entrepreneur, is the founder of Cognite and served as CEO until 2022, subsequently becoming Chief Strategy and Development Officer. SO001, SO003
CO006 Geir Engdahl co-founded Cognite and serves as CTO AI; he is a former Google software engineer. SO001, SO003
CO007 Girish Rishi was named CEO of Cognite in 2022 after co-founder John Markus Lervik stepped aside. SO001, SO003
CO008 Girish Rishi serves as both CEO and Chairman of the Board of Directors at Cognite. SO003
CO009 Girish Rishi previously served as CEO of Blue Yonder (a Panasonic-acquired supply chain software company) and held executive leadership positions at Motorola and Tyco. SO001
CO010 Cognite's corporate headquarters address is 40 E Rio Salado Pkwy, Suite 900, Tempe, AZ 85281. SO002
CO011 Cognite was formed as a strategic partner of Aker BP, a European oil company that became an early investor and major customer. SO001, SO004
CO012 Cognite is an independently run affiliate of Aker ASA, the Norwegian investment conglomerate controlled by billionaire Kjell Inge Roekke, which is the majority shareholder. SO001
CO013 Saudi Aramco purchased a 7.4% equity stake in Cognite from Aker BP in February 2022. SO015, SO001
CO014 Accel and TCV are institutional investors in Cognite; Accel led earlier rounds and TCV led the 2021 $150M and 2024 ~$300M rounds. SO004, SO021
CO015 Cognite's two flagship products are Cognite Data Fusion (CDF), an industrial DataOps platform, and Cognite Atlas AI, a low-code industrial agent workbench. SO006, SO012
CO016 Cognite Flows was launched in May 2026 as the action layer of the Cognite industrial AI and data platform, designed to serve frontline workers with unified AI-driven workflows. SO013, SO014
CO017 Cognite raised $150M from TCV in May 2021, achieving a $1.6B valuation and unicorn status; the round was described as one of the largest SaaS funding rounds in Europe. SO004, SO001
CO018 CNBC Disruptor 50 (May 2026) lists Cognite's total funding as $300 million and valuation as $1.6 billion. SO001
CO019 TCV led a ~$300M investment round in Cognite, as indicated by TCV's press release title 'TCV Leads $300M Investment in Cognite' and a concurrent TechCrunch reference; valuation for this round was not publicly disclosed. SO021, SO022
CO020 Cognite's FY2025 annual revenue exceeded $170 million, described as a record-breaking year. SO005, SO001
CO021 Cognite's annual recurring revenue (ARR) bookings grew 36% year-over-year in FY2025. SO005, SO001
CO022 Cognite's net revenue retention (NRR) for newer customers was 150% at end of FY2025. SO005
CO023 Cognite's global headcount exceeded 800 employees by early 2026, representing 21% growth during FY2025. SO005, SO001
CO024 Approximately 40% of Cognite's FY2025 revenue came from customers outside oil and gas, signaling industry diversification. SO001
CO025 Atlas AI was used by more than half of Cognite's customers (59% by ARR) and accounted for more than 70% of 2025 new bookings. SO005, SO001
CO026 The number of Cognite Atlas AI customers grew nearly 700% year-over-year in FY2025. SO005
CO027 Cognite's monthly active users grew 26% across the platform during FY2025. SO005, SO001
CO028 Cognite operates offices in Tempe AZ (HQ), Austin TX, Houston TX, Lysaker (Oslo) Norway, Tokyo Japan, and Bengaluru India. SO002
CO029 Cognite relocated its global headquarters from Oslo, Norway to Tempe, Arizona to be proximate to Arizona's high-tech and semiconductor manufacturing ecosystem. SO001, SO002
CO030 Cognite was named a Leader in the IDC MarketScape Worldwide Industrial DataOps Platforms 2026 Vendor Assessment (Doc #US53013025, March 2026). SO008
CO031 Microsoft named Cognite its Energy and Resources Partner of the Year for 2024, marking the third consecutive year of this designation. SO010, SO024
CO032 TotalEnergies expanded its strategic partnership with Cognite in September 2025 for a 3-year deployment of CDF across all TotalEnergies upstream assets worldwide. SO016, SO018, SO009
CO033 Cognite and Snowflake entered a partnership in early 2026 to integrate Cognite's industrial data platform with Snowflake's AI Data Cloud for oil and gas operations. SO017
CO034 Cognite's 2021 press release listed named customers including bp, Saudi Aramco, Alfa Laval, Statnett, and Mitsubishi. SO004
CO035 Cognite holds ISO 27001, ISO 9001, ISO 27018, SOC Type 2 (Security), and SOC Type 3 (Security) certifications. SO011
CO036 Cognite has stated a 'Moonshot' goal of delivering a cumulative $100 billion in business value to industrial customers by 2035. SO005, SO001
CO037 Cognite customers had already identified $2 billion in additional business value with Cognite's platform as of January 2026. SO005
CO038 Cognite's Industrial Knowledge Graph exceeded 80 trillion time series data points in context in 2025, up 48% year-over-year. SO005
CO039 Verdantix named Cognite a leader for Industrial AI Analytics and Industrial Data Management in two separate reports published in 2025. SO005
CO040 Cognite was ranked #15 on the CNBC Disruptor 50 list in May 2026. SO001
CO041 Cognite's target industries are oil and gas, energy, and process manufacturing, with the platform also serving chemicals, pharmaceuticals, and industrial manufacturing customers. SO002, SO012
CO042 UpGuard maintains an active vendor risk monitoring report on Cognite as of May 2026, evaluating security posture across website, email, network, and brand dimensions; the full risk rating is behind a paywall. SO019
CO043 Cognite Atlas AI uses a Semantic Industrial Knowledge Graph with Context Augmented Generation (CAG) and supports any cloud, any LLM, and any data source. SO006, SO007
CO044 Forrester Research documented Cognite enterprise-scale customer deployments at NOVA Chemicals (expanding to 11 plants), TotalEnergies (39 assets over 3 years), Aker BP (Yggdrasil remote-first asset), and Cosmo Energy (3 Japanese refineries with 250B data points). SO023
CO045 Cognite Flows was already adopted by more than 30% of Cognite's customer base and key partners at its May 2026 launch, including B. Braun and Idemitsu Kosan as customers. SO013
CM001 Cognite's primary addressable market is the Industrial DataOps and Industrial AI platform segment—software for OT/IT data contextualization, knowledge graph creation, and AI deployment in asset-intensive industries. SM003, SM018, SM023
CM002 The Industrial DataOps platform category is distinguished from general enterprise AI (Palantir, C3.ai) by its primary dedication to OT/IT convergence, real-time sensor data, and industrial knowledge graphs. SM007, SM009, SM018
CM003 Cognite's three primary end-market verticals are oil and gas (approximately 60% of FY2025 revenue, inferred from the disclosed ~40% non-O&G share), process manufacturing/chemicals, and power and renewables. SM017, SM023
CM004 Status-quo substitutes for Industrial DataOps platforms include legacy historian databases (OSIsoft PI/AVEVA), point analytics tools, OEM-bundled analytics from industrial automation vendors, and spreadsheet-based engineering workflows. SM019, SM013
CM005 Cognite's O&G industry page claims 90% reduction in data discovery time, 10-25x faster deployment, and 0.5-1.5% production throughput increase—these are company-stated metrics without independent third-party verification in available sources. SM003
CM006 Included spend in Cognite's market consists of industrial data management software, AI/ML deployment platforms, OT/IT integration middleware, and industrial knowledge graph infrastructure. SM018, SM014
CM007 Excluded from Cognite's relevant market are general-purpose business analytics (Tableau, Power BI), consumer IoT, IT monitoring (Datadog, Splunk), HR/ERP software, and SCADA/DCS control systems. SM018, SM019
CM008 The term 'Industrial DataOps' was adopted broadly post-2020 and was first recognized as a distinct analyst category by IDC in the inaugural MarketScape assessment published in March 2026. SM018, SM014
CM009 Adjacent markets include SCADA/DCS control systems, asset performance management (APM), industrial automation software, and hyperscaler-native industrial data services—each competitive and potentially complementary. SM006, SM010, SM011, SM013
CM010 Cognite competes against both specialized pure-play vendors (Seeq, AspenTech analytics) and hyperscaler-native services (AWS IoT SiteWise, Azure Digital Twins) that offer commodity industrial data aggregation at cloud pricing. SM004, SM005, SM010, SM011
CM011 Grand View Research estimates the global IIoT market at $483.16 billion in 2024, projected to reach $1,693.44 billion by 2030 at a CAGR of 23.3% for 2025–2030. SM001
CM012 Precedence Research estimates the IIoT market at $514.39 billion in 2025, projected to reach $2,430.21 billion by 2035 at a CAGR of 16.8% for 2026–2035. SM002
CM013 The GVR (23.3% CAGR) and Precedence Research (16.8% CAGR) IIoT market estimates are materially inconsistent, reflecting different market boundary definitions, base years, and forecast periods. SM001, SM002
CM014 No publicly accessible analyst report provides an unpaywalled TAM estimate for the 'Industrial DataOps platform' sub-segment specifically; IDC MarketScape 2026 is the most authoritative source but is sold at paywall pricing. SM014
CM015 The IDC MarketScape Worldwide Industrial DataOps Platforms 2026 (Doc #US53013025) confirms Industrial DataOps as a distinct vendor-assessed market category with Cognite positioned as a Leader. SM018, SM014
CM016 Cognite's $170M+ FY2025 revenue represents a fraction of even a narrow industrial analytics market; a bottom-up estimate suggests the pure-play Industrial DataOps software market at $500M–$2B in annual contract value. SM017, SM007, SM008
CM017 Manufacturing accounts for the largest IIoT end-use segment at approximately 28.7% of market share in 2025 per Precedence Research, exceeding oil and gas alone. SM002
CM018 North America leads IIoT market share at over 31–34% of global market per both GVR and Precedence Research, consistent with Cognite's decision to relocate its global headquarters to Arizona. SM001, SM002
CM019 C3.ai, a public-market comparable for enterprise industrial AI, filed a fiscal 2025 10-K with the SEC (fiscal year ending July 2025) showing approximately $252M in annual revenue, confirming the market supports companies at Cognite's revenue scale. SM008, SM007
CM020 The industrial AI platform market lacks a single widely-accepted TAM definition, with sizing estimates ranging from narrow ($3–10B software-only) to broad ($400–500B full IIoT stack) depending on scope. SM001, SM002, SM014
CM021 Cognite's customer base skews heavily to large industrial enterprises—typically companies with significant multi-site OT infrastructure, capital budgets exceeding $1B, and existing investments in industrial automation. SM019, SM017
CM022 Cognite's documented customer base spans three primary verticals: oil and gas (~60% of FY2025 revenue implied), process manufacturing/chemicals (~20-25%), and power/renewables/other (~15-20%), with the latter growing fastest. SM017, SM025
CM023 Budget ownership for Industrial DataOps typically resides with the Chief Digital Officer, VP Digital, or VP of Engineering/Operations—not traditional IT procurement—requiring Cognite to sell to operational leadership. SM019, SM027
CM024 TotalEnergies expanded its Cognite CDF deployment across all operated upstream assets worldwide over a three-year period, covering 39 assets and 200+ key production systems, demonstrating enterprise-scale adoption. SM020, SM022, SM027
CM025 NOVA Chemicals deployed Cognite CDF in one plant initially and planned expansion to 11 plants by year-end, illustrating the land-and-expand pattern that drives Cognite's 150% NRR for newer customers. SM019
CM026 Cosmo Energy deployed Cognite CDF across all three Japanese refineries, storing 250 billion data points and 5 million engineering documents—providing evidence of cross-site consolidation as the dominant expansion pattern. SM019
CM027 Industrial DataOps adoption triggers are consistently: (1) a C-suite digital transformation mandate, (2) a major capital project requiring modern data infrastructure, or (3) an ESG/sustainability reporting requirement. SM019, SM020, SM022
CM028 Cognite's user personas span data scientists, process/plant engineers, IT/OT integration architects, and—with Atlas AI—non-technical frontline workers who do not need to interact with the underlying knowledge graph. SM019, SM017, SM016
CM029 Payer authority for multi-year Industrial DataOps SaaS agreements typically sits at corporate level (CFO/procurement), separate from the CDO/engineering champion, requiring Cognite to navigate a two-track selling process. SM019
CM030 Cognite's partnerships with Snowflake and Databricks signal that data platform buyers who also use hyperscaler data lakehouses are a key target segment, requiring interoperability rather than data lock-in. SM021, SM016
CM031 The convergence of generative AI agent capabilities with accumulated industrial data infrastructure is the primary near-term growth driver; Cognite Atlas AI customer count grew nearly 700% year-over-year in FY2025. SM017, SM025
CM032 ESG and decarbonization mandates are creating new budget allocations for operational data visibility in energy and manufacturing sectors, opening an adjacent buying trigger beyond traditional operational efficiency. SM022, SM020
CM033 Industrial workforce aging and knowledge-transfer crises—as experienced engineers retire—are creating structural demand for AI tools that capture and operationalize expert knowledge in industrial knowledge graphs. SM019, SM017
CM034 ISO 27001, ISO 9001, SOC Type 2 compliance certifications both reduce procurement friction for Cognite (as requirements are met) and add compliance overhead to enterprise sales cycles (as buyers conduct extensive audits). SM023
CM035 Forrester Research characterizes industrial AI deployments as 'slow, hard, and expensive,' noting that the AI application is a 'tiny little piece of the iceberg, resting atop a mountain of data operations groundwork,' limiting addressable market to patient, well-resourced enterprises. SM019
CM036 Incumbent industrial software vendors (Siemens, GE Vernova, Honeywell, AspenTech, PTC ThingWorx) offer bundled analytics within their automation suites, creating switching-cost-driven competition at established accounts. SM006, SM013
CM037 Switching costs for Industrial DataOps platforms are severe—replacing a multi-year knowledge graph implementation requires re-integrating hundreds of OT/IT data sources and rebuilding semantic models—creating deep vendor lock-in. SM019, SM017
CM038 OT security and data sovereignty requirements slow adoption in regulated industries, government-adjacent applications, and markets with strict data residency laws, adding compliance overhead to Cognite's enterprise sales cycles. SM023, SM014
CM039 Budget cycles in oil and gas and process manufacturing for major IT/OT integration projects commonly span 18–36 months from approval to production, creating lengthy sales cycles and forecasting uncertainty for Cognite. SM019, SM020
CM040 Microsoft Azure Digital Twins and AWS IoT SiteWise actively promote competing industrial data services at commodity cloud pricing, creating downward pressure on Cognite's connectivity and data aggregation pricing. SM010, SM011
CM041 The exact size of the Industrial DataOps platform sub-segment is unavailable without purchasing IDC MarketScape 2026, Verdantix 2025 leadership reports, or Gartner Market Guide—none of which provide unpaywalled market size figures. SM014
CM042 Cognite's $170M+ FY2025 revenue cannot yield a meaningful market share estimate without a credible SAM denominator; the Industrial DataOps platform sub-segment has no publicly available total market size. SM017, SM014
CM043 Verdantix published two separate 2025 reports naming Cognite a leader in Industrial AI Analytics and Industrial Data Management, but the full report content and any market size estimates are paywalled. SM017
CM044 The GVR (23.3% CAGR) and Precedence Research (16.8% CAGR) IIoT sizing estimates are contradictory and use different market definitions, undermining confidence in either as a reliable proxy for Cognite's SAM. SM001, SM002
CM045 Cognite's stated Moonshot goal of delivering $100 billion in cumulative customer value by 2035 is an aspirational management target, not a market size estimate, and does not map to a verifiable TAM. SM017, SM025
CP001 AVEVA PI System is the industry-standard industrial time-series data historian, with a 30-plus-year installed base across oil and gas, power generation, and process industries, making it the primary incumbent competitor to Cognite CDF. SP001, SP019
CP002 Schneider Electric acquired AVEVA in a transaction that valued the company at approximately £9.5 billion (approximately $11.9 billion USD), completing the acquisition of all remaining publicly traded shares in January 2023. SP002
CP003 AVEVA PI System's competitive advantage against Cognite is its pervasive installed-base footprint: tens of thousands of historian tags at large industrial facilities create deep operational dependency and multi-year switching friction. SP001, SP002
CP004 AVEVA CONNECT is a cloud-native data management platform launched as AVEVA's next-generation offering to compete with modern cloud alternatives, aggregating and contextualizing industrial data across remote assets and third-party systems. SP003
CP005 AVEVA's competitive positioning versus Cognite reflects a legacy-architecture-vs.-cloud-native architectural divide: PI System excels on OT data depth and installed base while CDF excels on knowledge graph contextualization and AI agent workbench. SP001, SP003, SP023
CP006 AspenTech was acquired by Emerson Electric in a transaction valued at approximately $11 billion, with Emerson contributing its industrial software assets in exchange for a majority stake in a combined entity that subsequently became fully part of Emerson. SP004, SP005
CP007 AspenTech targets chemical engineering, refining, power generation, and asset performance management with its aspenONE engineering simulation suite and Aspen Mtell predictive maintenance product, serving asset-intensive industries where process simulation is a primary value driver. SP004, SP005
CP008 The Emerson/AspenTech combination bundles OT process control hardware and industrial software under one enterprise sales motion, creating a channel advantage over standalone software vendors like Cognite by enabling customers to procure AI and analytics software alongside Emerson control systems. SP004, SP005
CP009 Seeq positions itself as an advanced analytics, machine learning, and AI platform for industrial time-series data, designed to be data-source agnostic and operate as a layer on top of existing historians such as AVEVA PI System, Cognite CDF, or other data sources. SP006
CP010 Seeq addresses the analytics and insight layer of industrial operations rather than data ingestion, OT connectivity, or knowledge graph construction, making it partly complementary to and partly competitive with Cognite CDF's analytics capabilities depending on deployment architecture. SP006
CP011 Seeq has raised a Series C funding round led by Energize Ventures and Insight Partners, though the exact amount and current company valuation are not confirmed in primary sources reviewed through May 2026. SP006
CP012 PTC is transitioning its ThingWorx IIoT platform under the Velotic brand, describing it as a new chapter with the same underlying platform, an expanded focus, and a future-ready positioning. SP007
CP013 ThingWorx/Velotic targets manufacturing and factory-floor use cases with IIoT connectivity and application development, primarily competing with Cognite in discrete manufacturing contexts while having limited overlap with Cognite's core oil and gas and process industry focus. SP007
CP014 C3.ai (NYSE: AI) provides a horizontal enterprise AI application platform targeting manufacturing, energy, defense, financial services, and healthcare, and is the only pure-play industrial AI SaaS company listed on public markets with comparable positioning to Cognite. SP008, SP010
CP015 C3.ai's marquee customers include Shell, Baker Hughes, Koch Industries, Holcim, and the U.S. Air Force, demonstrating reach across energy, manufacturing, and defense verticals that overlap with Cognite's target segments. SP008, SP009
CP016 C3.ai's enterprise AI platform is architecturally horizontal—it requires customer-provided data infrastructure and custom model configuration—while Cognite provides a vertically integrated industrial-specific data contextualization layer and AI workbench, representing a fundamental difference in deployment model. SP009, SP022
CP017 C3.ai and Cognite compete most directly for enterprise AI budget allocation at large industrial operators, particularly in energy and manufacturing, where both vendors are pursuing multi-year SaaS subscription contracts for AI-driven operational intelligence. SP008, SP009
CP018 C3.ai's SEC-filed 10-K annual report (filed July 2025 for FY2025) confirms the company continues to report net operating losses, providing an adverse public-market signal that the horizontal enterprise AI SaaS category has not yet demonstrated clear paths to profitability. SP010, SP029
CP019 Palantir AIP is positioned as a large-enterprise and government/defense AI platform with a high-touch delivery model involving Forward Deployed Engineers, and reported FY2024 annual revenue of approximately $2.9 billion—a scale materially larger than Cognite's current revenue base. SP011, SP012
CP020 Palantir competes with Cognite primarily for large-enterprise AI budget at oil and gas, energy, and industrial companies, though Palantir's high-touch engineering model and typical deal size ($10M+) differ fundamentally from Cognite's SaaS subscription model targeting broad platform adoption. SP011, SP012
CP021 AWS IoT SiteWise is a managed cloud service from Amazon Web Services that collects, stores, organizes, and monitors data from industrial equipment at scale, operating as an infrastructure-tier industrial data service rather than an analytics or AI agent platform. SP013
CP022 AWS IoT SiteWise competes with Cognite CDF primarily at the data ingestion and storage layer, not at the knowledge graph contextualization or AI agent workbench layer, making it a lower-stack infrastructure threat rather than a direct full-stack replacement. SP013
CP023 Microsoft Azure Digital Twins provides an open modeling language and live digital twin execution environment for creating representations of connected industrial environments including factories, farms, energy networks, and cities. SP014
CP024 Azure Digital Twins and Microsoft's industrial cloud capabilities compete indirectly with Cognite CDF; Microsoft's competitive advantage is deep enterprise account relationships, Office 365/Teams integration, and Azure ecosystem bundling, while CDF's advantage is industrial domain expertise and pre-built operational contextualization. SP014, SP023
CP025 Siemens Industrial Edge is an edge computing platform that combines hardware, software, and OT/IT connectivity from Siemens and ecosystem partners, designed to optimize factory operations through scalable and secure edge solutions. SP015
CP026 Siemens competes with Cognite primarily in discrete manufacturing and factory automation contexts; Siemens has limited overlap with Cognite's core oil and gas and process industry knowledge graph use cases, making Siemens a partial adjacent competitor rather than a direct head-to-head rival. SP015
CP027 GE Vernova, which completed its spin-off from General Electric in April 2024, is focused on power generation equipment, grid technology, and energy transition rather than operational intelligence software platforms, making it a limited direct competitor to Cognite. SP016, SP017
CP028 GE Digital, which operated the Predix industrial IoT platform and APM software portfolio, was separated from GE Vernova during General Electric's corporate reorganization, leaving GE Vernova without a significant industrial AI or DataOps platform business as of May 2026. SP016, SP017
CP029 Hexagon is the global leader in measurement technologies and digital industrial solutions, offering asset lifecycle management, reality capture, and industrial simulation capabilities to asset-intensive industries through its Smart Digital Reality platform. SP018
CP030 Hexagon's competitive differentiation from Cognite is its focus on 3D spatial data, precision measurement, and reality capture for asset lifecycle management, while Cognite focuses on operational time-series data and AI-driven workflow automation—these represent overlapping but distinct buyer problems. SP018
CP031 The primary status-quo alternative to Cognite CDF at large industrial facilities is the combination of AVEVA PI System historian for operational data storage, manual data exports and spreadsheet-based analysis for engineering workflows, and siloed SCADA/DCS systems for real-time monitoring. SP001, SP020
CP032 Internal build alternatives using cloud data warehouse platforms (Snowflake or Databricks) combined with custom Python and machine learning pipelines are a growing competitive threat to Cognite among industrial enterprises with sophisticated IT and data engineering teams. SP025, SP027
CP033 Cognite CDF creates multi-layered switching costs through its proprietary Industrial Knowledge Graph schema, custom OT source connectors, API applications built on CDF, and trained Atlas AI agents—all of which are specific to CDF's data model and add migration friction for established customers. SP022, SP023
CP034 Cognite's Industrial Knowledge Graph architecture—contextualizing relationships between assets, time series, documents, and 3D models in a unified semantic layer—is not replicated by historian-centric competitors (AVEVA PI), horizontal AI platforms (C3.ai, Palantir), or infrastructure-tier cloud services (AWS SiteWise, Azure Digital Twins). SP001, SP013, SP014, SP023
CP035 Cognite's key differentiators versus incumbents include cloud-native architecture, open REST and GraphQL APIs, and the Atlas AI low-code agent workbench; versus cloud hyperscalers, the differentiator is industrial domain expertise and pre-built operational contextualization that AWS/Azure do not provide out of the box. SP022, SP023, SP013, SP014
CP036 AWS and Azure are actively expanding industrial IoT service portfolios—with AWS IoT SiteWise providing managed industrial data collection and Azure Digital Twins providing open modeling infrastructure—creating medium-term commoditization risk for CDF's lower data-collection and connectivity stack over a 3-to-5-year horizon. SP013, SP014, SP028
CP037 AVEVA's PI System entrenched installed base and Emerson/AspenTech's bundled OT-plus-software channel represent structural distribution advantages that constrain Cognite's displacement rate in existing facilities, as incumbent contracts typically have 3-to-7-year renewal cycles with embedded user workflows. SP001, SP004, SP019
CP038 Cognite's last publicly confirmed valuation of $1.6 billion (May 2021) has not been updated in public disclosures, while publicly traded competitors Palantir (~$100B+ market cap) and C3.ai (~$3–5B market cap range) provide full market transparency, leaving Cognite's competitive capitalization relative to its peers opaque for investment benchmarking. SP010, SP011, SP023
CP039 Migrating from AVEVA PI System to Cognite CDF requires rebuilding OT source connectors, remapping the PI tag hierarchy to CDF's knowledge graph schema, and retraining operational users, representing multi-month implementation friction that creates meaningful switching costs for existing PI deployments regardless of CDF's technical superiority. SP001, SP022
CP040 Cognite's partner ecosystem with Microsoft (co-sell, Azure cloud), Databricks (lakehouse integration), and Snowflake (data sharing) creates co-sell distribution pathways and ecosystem integration advantages that standalone competitors and internal-build alternatives lack. SP025, SP027, SP024
CI001 Cognite operates a SaaS subscription business model; primary revenue streams are CDF platform subscriptions, Atlas AI subscriptions, and professional services/Success Tracks. SI001, SI009
CI002 Cognite launched Cognite Flows in May 2026 as the experience/action layer of its platform; more than 30% of its customer base was enabled with Flows at launch. SI016
CI003 Cognite does not publish list pricing for any of its products (CDF, Atlas AI, or Flows) as of the May 2026 run date; no public pricing page is accessible. SI001, SI009
CI004 The TotalEnergies 3-year enterprise deal to deploy CDF across all upstream assets globally is consistent with a multi-million-dollar annual contract value; no specific dollar figure has been publicly disclosed. SI019, SI020
CI005 Atlas AI accounted for more than 70% of Cognite's new bookings in FY2025, indicating that the AI agent layer has become the primary growth driver of new ARR. SI001, SI009
CI006 Approximately 40% of Cognite's FY2025 revenue derived from customers outside the oil and gas sector, with the remaining ~60% from O&G. SI002, SI001
CI007 Cognite's 'Moonshot' mission is to deliver $100 billion in measurable customer value to its customers by 2035, which frames its premium pricing and GTM narrative. SI009
CI008 Cognite AS (Norwegian entity, org. 918274758) recorded revenue of approximately NOK 1,670,024,000 (~USD 158M at 2024 average NOK/USD ~10.55) in its FY2024 statutory annual accounts filed under IFRS with Brønnøysundregistrene. SI007, SI006
CI009 Paul Grenet serves as Cognite's President and Chief Revenue Officer, indicating a combined field-operations and revenue-ownership structure for global sales. SI010, SI009
CI010 Cognite sells exclusively via direct enterprise sales; regional presidents cover Americas, EMEA, Japan, North and Southeast Asia, and India. SI001, SI009
CI011 The Snowflake collaboration (January 2026) provides a co-sell channel into data-cloud-centric procurement workflows at energy companies, extending Cognite's GTM beyond direct sales. SI010, SI021
CI012 The NVIDIA NV-Tesseract integration (March 2026) adds an AI-infrastructure co-sell dimension to Cognite's GTM, with Celanese as a named joint customer. SI011
CI013 The Forrester Total Economic Impact study (Q1 2026) found that Cognite delivers 465% ROI to its customers, per Cognite's newsroom listing; the full study is not publicly accessible. SI026
CI014 A global pharmaceutical customer deploying Cognite Flows achieved 30X faster time-to-value in one engagement, reducing UAT from a typical 6–9-month cycle to two months, per the BusinessWire Flows press release. SI016
CI015 Cognite's blended gross margin is not publicly disclosed; no Cognite financial statement or press release has reported this figure as of the May 2026 run date. SI001, SI002
CI016 Cognite's net revenue retention (NRR) for newer customers was 150% at end of FY2025, as reported in the January 2026 company press release; full-base NRR is not disclosed. SI001, SI002
CI017 Cognite's customer count and average contract value are not publicly disclosed; the company has not published customer numbers in any accessible press release or filing as of May 2026. SI001, SI009
CI018 Cognite's CAC, payback period, and gross dollar retention are not publicly disclosed and represent central underwriting gaps in the unit economics profile. SI001, SI002
CI019 Cognite AS (Norwegian entity) reported a net result of approximately NOK -547,002,000 (~USD -52M) for FY2024 in statutory accounts filed with Brønnøysundregistrene. SI007, SI008
CI020 Cognite AS reported total assets of approximately NOK 2,263,650,000 (~USD 215M) in its FY2024 statutory accounts. SI007, SI008
CI021 The most recent publicly confirmed funding round is the ~$300M investment led by TCV in June 2024; no post-money valuation for this round has been publicly disclosed. SI004, SI002
CI022 The $150M round from TCV in May 2021 valued Cognite at $1.6 billion post-money; this is the last publicly confirmed valuation and is approximately five years stale by the May 2026 run date. SI003, SI002
CI023 Total known capital raised by Cognite is at least ~$450M: $150M (TCV, May 2021) + ~$300M (TCV, June 2024) plus an undisclosed earlier Accel investment; exact total is not confirmed. SI003, SI004, SI002
CI024 Cognite's cash on hand, monthly burn rate, and consolidated group runway are not publicly disclosed as of the May 2026 run date. SI001, SI002
CI025 At the FY2024 Norwegian entity burn rate (~$52M/year net loss), a group-level annual burn of $60–90M is an analyst estimate; at this rate, the June 2024 ~$300M round would imply approximately 3–5 years of runway from close. SI007, SI004
CI026 Saudi Aramco purchased a 7.4% equity stake in Cognite from Aker BP in February 2022 and is simultaneously one of Cognite's largest customers, creating a dual customer-investor role. SI005, SI002
CI027 No public disclosure of debt, revolving credit facilities, convertible notes, or project finance obligations by Cognite is available as of May 2026. SI001, SI006
CI028 Cognite's FY2025 annual revenue exceeded $170M (preliminary, unaudited), representing a year-over-year revenue increase consistent with the 36% ARR bookings growth reported for the same period. SI001, SI002
CI029 Q1 2026 new ARR bookings grew 57% year over year, an acceleration from the 36% FY2025 rate, per Cognite's May 2026 CNBC Disruptor 50 press release. SI009
CI030 As of Q1 2026, approximately 72% of Cognite's customer base by ARR had operationalized Atlas AI, up from 59% at FY2025 end, confirming deepening platform penetration. SI009
CI031 Cognite AS was incorporated in the Norwegian company registry on December 7, 2016 (registration date December 19, 2016), as evidenced by Brønnøysundregistrene entity record for org. 918274758. SI006, SI008
CI032 Cognite AS had 367 employees registered in Norway as of May 10–11, 2026 per the Norwegian business registry, out of a global headcount exceeding 800—indicating approximately 460 employees outside Norway. SI006, SI008
CI033 Cognite's ARR bookings growth has accelerated from 36% in FY2025 to 57% in Q1 2026 YoY; combined with 150% NRR for newer customers, this represents a compounding revenue quality profile materially above SaaS median benchmarks. SI001, SI009
CI034 The UpGuard vendor risk report for Cognite is active and paywalled as of May 2026; no public security incidents, breaches, or material outages have been identified in sources reviewed—but absence of evidence in a paywalled report is not evidence of absence. SI017
CI035 Cognite has not disclosed any IPO timeline, SPAC consideration, secondary liquidity mechanism, or strategic sale process as of the May 2026 run date; exit optionality is unclear. SI001, SI009
CI036 Aker ASA, Cognite's majority shareholder, reported Net Asset Value exceeding NOK 100 billion in Q1 2026—its strongest quarter on record—providing a significant financial backstop for Cognite's continued operations. SI013
CE001 Cognite Data Fusion (CDF) is the company's core industrial DataOps platform, providing an Industrial Knowledge Graph that contextualizes IT, OT, and engineering data from disparate sources into a unified, AI-ready data layer. SE010, SE020
CE002 As of fiscal year 2025, Cognite customers' Industrial Knowledge Graph held more than 80 trillion time-series data points in context, a 48% year-over-year increase. SE016, SE020
CE003 Cognite Atlas AI is a low-code industrial agent workbench launched in June 2024 that enables domain experts and digital teams to build AI agents without writing code using pre-configured templates. SE010, SE012
CE004 Cognite Flows, launched May 12, 2026, is the action layer of the Cognite platform, providing a single-screen adaptive workspace for frontline operators and a developer SDK for rapid custom application development. SE005, SE013
CE005 Cognite Flows enables frontline teams to build and scale production-ready workflows 100 times faster through adaptive, persona-based dashboards natively integrated with the CDF Industrial Knowledge Graph. SE005, SE014
CE006 Cognite bundles four built-in industrial applications within the Flows product layer: Industrial Canvas, InField, Maintain, and Charts. SE005, SE020
CE007 Atlas AI uses Context Augmented Generation (CAG), which injects the CDF Industrial Knowledge Graph's domain-specific context into LLM prompts to improve accuracy and reduce hallucination in industrial AI tasks. SE010, SE012
CE008 Atlas AI includes autoLLM, a routing mechanism that automatically selects the optimal Large Language Model, Small Language Model, or Custom Language Model for each specific industrial agent task. SE010, SE012
CE009 Atlas AI supports deployment of AI agents into Cognite Data Fusion, Microsoft Copilot, and any third-party application via published Agent APIs. SE010, SE012
CE010 As of fiscal year 2025, 59% of Cognite's customers by annual recurring revenue (ARR) had operationalized Cognite Atlas AI. SE016
CE011 The number of Atlas AI customers grew nearly 700% year-over-year in fiscal year 2025, and more than 70% of 2025 bookings included Atlas AI. SE016
CE012 More than 30% of Cognite's customer base was early-enabled with Cognite Flows prior to its May 2026 GA launch, and over 50 customers were actively using it at launch. SE013, SE005
CE013 Cognite's oil and gas product page states that CDF enables 90% less time spent interpreting data, 10–25x faster time to production deployment, and 0.5–1.5% increase in overall production throughput. SE026
CE014 A global top-10 pharmaceutical company achieved 30x faster time-to-value using Cognite Flows, delivering automated AI workflows in four days versus a 20-plus-person team requiring 3–5 months under traditional methods. SE013, SE014
CE015 B. Braun deployed Cognite Flows to achieve real-time, unified visibility into asset health across all global sites, refining the UX within four weeks through near-instant updates driven by user feedback. SE013
CE016 Idemitsu Kosan is deploying Cognite Flows to systematically capture and digitize accumulated operational knowledge, aiming to reduce operational risk and enable proactive AI-driven management. SE013, SE014
CE017 Celanese deployed the Cognite CDF and NVIDIA NV-Tesseract NIM integration at its Clear Lake, Texas facility to improve prediction of reaction water levels, targeting elimination of manual-sample-induced bias jumps. SE015, SE009
CE018 Cognite's Python SDK v8 introduced full async support via a new AsyncCogniteClient, enabling non-blocking concurrent operations, async/await patterns for all API operations, and significantly faster file uploads on Windows. SE002, SE001
CE019 Cognite provides an MCP (Model Context Protocol) server integration that allows developers to connect their development environments and AI coding tools directly to Cognite's documentation and API. SE001, SE002
CE020 Cognite's monthly active users grew 26% year-over-year in fiscal year 2025. SE016
CE021 Cognite announced integration of NVIDIA's NV-Tesseract NIM microservices with the Cognite AI and Data Platform on March 16, 2026, enabling foundational time-series AI forecasting for heavy industry. SE015, SE009
CE022 Cognite and Snowflake announced a collaboration in January 2026 supporting Snowflake's Energy Solutions, enabling joint oil and gas customers to unify IT, OT, and IoT data in the Snowflake AI Data Cloud enriched with Cognite's knowledge graph. SE022, SE021
CE023 Cognite supports a Databricks integration that enables enterprise AI for industrial operations by connecting CDF's knowledge graph with the Databricks lakehouse platform. SE020
CE024 CDF is cloud-agnostic and deploys on AWS, Azure, and Google Cloud Platform, with globally distributed clusters enabling data sovereignty for regulated industrial customers. SE010, SE004
CE025 Cognite white-labels CDF to other software providers, extending its industrial DataOps capabilities into third-party products and expanding the addressable market. SE017
CE026 CDF's REST API uses versioned endpoints with a published backward compatibility policy and a documented end-of-life schedule for deprecated API versions. SE001, SE002
CE027 Cognite holds ISO 27001, ISO 9001, ISO 27018, SOC Type 2 (Security criterion), and SOC Type 3 (Security) certifications as of May 2026, independently audited by third parties. SE011, SE004
CE028 Cognite's platform supports customer compliance with IEC 62443 sub-standards 2-4, 3-2, 3-3, and 4-1 covering industrial cybersecurity for control systems and OT environments. SE011
CE029 Cognite's Trust Center at trust.cognite.com provides real-time visibility into security posture, system health, and compliance documentation; SOC 2 and SOC 3 audit reports are available to prospective customers. SE004, SE011
CE030 Cognite states alignment with the EU AI Act and offers globally available clusters to support data sovereignty requirements for regulated EU deployments. SE010, SE004
CE031 Cognite's secure development lifecycle spans infrastructure, applications, and operations and is tested and audited by third parties. SE011, SE004
CE032 Cognite also supports customer compliance with NIST CSF, CMMC, FIPs, NERC CIP v.5, GxP, and CSA STAR Level 1 industrial and government security frameworks. SE011
CE033 UpGuard continuously monitors Cognite's externally visible security posture using open-source and proprietary threat intelligence across website security, email security, phishing, brand risk, and network security categories. SE018
CE034 Cognite follows a quarterly product release cadence, delivering new capabilities across CDF, Atlas AI, and Flows on a regular schedule. SE016, SE017
CE035 The IDC MarketScape Worldwide Industrial DataOps Platforms 2026 assessment, published March 2026, positioned Cognite as a Leader, citing CDF's coverage of all five key industrial DataOps capabilities and its downstream AI through Atlas AI. SE017, SE024
CE036 In fiscal year 2025, customers' CDF knowledge graphs doubled the number of data modeling instances and increased AI token usage by nearly 500% year-over-year. SE016
CE037 Cognite Flows launch partners include Radix, L&T Technology Services, and RoviSys, and the Snowflake Energy Solutions collaboration was announced in January 2026. SE013, SE022
CE038 The CNBC Disruptor 50 list published May 19, 2026 ranked Cognite at position 15, citing its leadership in industrial AI. SE023
CE039 Cognite offers a Forrester Total Economic Impact study that reports 465% ROI and USD 29.4 million in total benefits for a composite customer, though the study was commissioned by Cognite. SE025, SE020
CE040 LLM hallucination remains an inherent risk in Atlas AI industrial agent outputs; Cognite's CAG architecture reduces but does not eliminate hallucination risk in mission-critical settings. SE010, SE007
CE041 All Cognite applications — Atlas AI, Flows, Industrial Canvas, InField, Maintain, and Charts — require CDF's Industrial Knowledge Graph as their data foundation, creating high vendor lock-in and migration cost for customers. SE005, SE010
CE042 Developer community questions on hub.cognite.com and Stack Overflow reveal recurring integration challenges with CDF data type configuration (string vs. numeric), OIDC authentication setup, and time-series query optimization. SE006, SE007
CE043 Cognite's CDF relies on third-party cloud providers (AWS, Azure, GCP) for infrastructure and on external OIDC identity providers for all API authentication, creating dependency risks for platform availability and access control. SE004, SE001
CE044 Cognite's SOC 2 Type 2 audit report is shared only under NDA to prospective customers and is not publicly downloadable, limiting independent pre-engagement security assessment. SE018, SE011
CE045 No independent third-party IEC 62443 certification for Cognite's platform has been identified; the stated IEC 62443 support represents customer-supportable controls rather than a vendor-held certification. SE011, SE018
CU001 Cognite reported annual revenues exceeding $170M for FY2025, with 36% ARR bookings growth year-over-year. SU013, SU012
CU002 Approximately 40% of Cognite's FY2025 revenue derives from outside the oil and gas sector. SU013
CU003 Cognite reported 150% net revenue retention for its newer cohorts of customers as of FY2025. SU013, SU012
CU004 Cognite's named customer verticals include oil and gas, chemicals, metals and mining, life sciences, and refining, spanning Europe, North America, the Middle East, and Japan. SU008, SU001, SU002, SU003, SU004
CU005 Cognite's monthly active users grew 26% year-over-year as of FY2025. SU013
CU006 59% of Cognite customers by ARR were using the Atlas AI layer as of FY2025. SU013
CU007 Cognite's publicly confirmed named production customers include bp, Saudi Aramco, TotalEnergies, Aker BP, Cosmo Energy, NOVA Chemicals, B. Braun, Idemitsu, JFE Steel, and Celanese. SU008, SU001, SU002, SU003, SU004, SU005, SU006, SU007, SU015
CU008 Cognite's industrial knowledge graph holds 80 trillion time series data points, representing 48% year-over-year growth as of FY2025. SU013
CU009 Atlas AI customer count grew 700% since product launch, as disclosed by Cognite in its FY2025 year-end review. SU013
CU010 Cognite's Flows product had enabled more than 30% of its total customer base as of the May 2026 launch announcement. SU014, SU024
CU011 Aker BP deployed Cognite CDF across six Norwegian North Sea assets as of the published case study. SU001, SU010
CU012 Aker BP achieved up to 50% reduction in visual inspection time following Cognite CDF deployment across its offshore assets. SU001, SU010
CU013 Aker BP achieved 30-80% reduction in maintenance execution time following Cognite CDF deployment, according to the published case study. SU001, SU010
CU014 Aker BP CDO Dr. Paula Doyle publicly described Cognite as enabling 'the next generation of operational excellence.' SU001
CU015 Cognite's case study describes Aker BP as having deployed the platform in production, not as a pilot, with named CDO endorsement. SU001, SU010
CU016 Aker BP's company website independently references digital transformation investments, providing corroboration for Cognite deployment claims. SU010
CU017 NOVA Chemicals, Canada's largest polyethylene producer, reduced data access time from hours to minutes across an eleven-plant target scope. SU002, SU012
CU018 Idemitsu onboarded four refinery sites in under six months, scaled to over 3,000 users, and uploaded more than 10,000 P&IDs. SU003, SU012
CU019 Idemitsu estimates 50,000 hours of annual labor savings from its Cognite CDF deployment as reported in Cognite's case study. SU003, SU012
CU020 Idemitsu is named as a Cognite Flows design partner as of the May 2026 Flows launch announcement. SU014
CU021 Cosmo Energy deployed Cognite across three Japanese refineries with 250 billion data points and five million engineering documents integrated. SU012, SU013
CU022 TotalEnergies expanded its Cognite deployment to 39 upstream assets over three years, with more than 200 key production systems integrated. SU015, SU016
CU023 TotalEnergies issued a press release confirming expansion of its Cognite partnership, providing independent customer corroboration for the deployment scope claim. SU015, SU016
CU024 TotalEnergies and Cognite announced a further partnership expansion in 2026, documented by OE Digital. SU016, SU015
CU025 Celanese's Digital Plants of the Future program has adopted Cognite as an enabling platform, per the published Cognite case study. SU004, SU013
CU026 B. Braun attributes a four-week improvement in a key production workflow to the Cognite Flows platform, per the May 2026 BusinessWire announcement. SU014, SU007
CU027 JFE Steel is named as a Cognite Flows customer in the May 2026 Flows launch press release. SU014, SU006
CU028 B. Braun is named as a Cognite Flows design partner in the May 2026 Flows launch announcement. SU014
CU029 Cognite was positioned as a Leader in the IDC MarketScape for Worldwide Industrial DataOps Platforms, as announced by Cognite in April 2026. SU017, SU012
CU030 Cognite Flows launched in May 2026 as the company's first fully integrated AI-native industrial workflow experience, with B. Braun and Idemitsu as design partners. SU014, SU024
CU031 Compared to industrial AI competitors such as AVEVA, AspenTech, and Palantir, Cognite has more named production deployments with specific quantified outcomes in the public domain as of 2026. SU008, SU012, SU017
CU032 Cognite's land-and-expand mechanism operates through initial asset-level pilots, followed by multi-asset rollout as the knowledge graph accumulates contextualized data, with Atlas AI and Flows modules driving incremental ARR. SU013, SU012, SU001
CU033 Cognite's customer base shows consistent engagement depth: 26% MAU growth, 48% knowledge graph data growth, and 700% Atlas AI customer growth all indicate an expanding active use base, not merely a licensed installed base. SU013, SU012
CU034 Cognite's O&G vertical represents approximately 60% of FY2025 revenue, creating sectoral concentration risk if energy-transition capital discipline reduces digital investment. SU013
CU035 Saudi Aramco's 74% majority stake in Cognite, combined with Aramco's status as an anchor customer, creates a structural conflict of interest for competing O&G operators evaluating the platform. SU011, SU013
CU036 Reuters reported in February 2022 that Saudi Aramco acquired a 74% stake in Cognite for approximately $1.6 billion, establishing Aramco as the majority owner of a platform used by its oil and gas peers. SU011
CU037 Independent software review platforms including Gartner Peer Insights, PeerSpot, Slashdot, and SourceForge show no substantive enterprise user reviews for Cognite Data Fusion as of May 2026. SU009, SU021, SU022, SU023
CU038 Cognite has not publicly disclosed customer count, gross revenue retention rate, logo churn, or top-customer ARR concentration, making independent quantitative assessment of retention durability impossible. SU013
CR001 Cognite operates under GDPR as a data processor for EU-based industrial customers, with the Norwegian Data Protection Authority (Datatilsynet) serving as lead supervisory authority for its Norwegian entity. SR001, SR005
CR002 Datatilsynet oversees personal data processing by Norwegian-registered entities and provides compliance guidance to small and medium-sized organisations processing data under GDPR. SR001, SR005
CR003 As of May 2026, no publicly disclosed regulatory enforcement actions, lawsuits, or material legal proceedings involving Cognite were identified in searches across Norwegian, EU, or US sources; the absence of disclosure does not confirm a clean record for undisclosed matters. SR018, SR022
CR004 The EU AI Act (effective August 2024) classifies AI systems operating in critical infrastructure — including oil refineries, chemical plants, and energy grids — as high-risk systems subject to conformity assessment, risk management documentation, and mandatory human-oversight provisions. SR003, SR005
CR005 Cognite acts as a data processor under GDPR customer data processing agreements, not as a data controller for industrial OT data; specific applicability of EU AI Act high-risk classification to Atlas AI agents depends on customer deployment configuration and degree of autonomous operation. SR004, SR003
CR006 Cognite holds ISO 27001:2022, SOC 2 Type II (Security), SOC 3 (Security), ISO 9001, and ISO 27018 certifications, all independently audited and referenced on Cognite's Trust Center. SR017, SR020
CR007 Cognite's platform supports customer compliance with NERC CIP v.5 for US energy operators, IEC 62443 industrial cybersecurity standards, CMMC, FIPs, GxP, and CSA STAR Level 1. SR020, SR017
CR008 Cognite's customer contracts for industrial AI deployments create potential professional liability exposure if AI-driven recommendations contribute to operational incidents; contractual limitation-of-liability provisions and indemnification terms are not publicly disclosed. SR003, SR009
CR009 Cognite's platform is deployed on multi-cloud infrastructure spanning AWS, Microsoft Azure, and Google Cloud Platform, creating dependency on all three major hyperscaler providers simultaneously. SR020, SR012
CR010 StatusGator monitors Cognite's platform service status continuously; no major long-duration public outages were identified in public reporting through May 2026, though platform SLA uptime commitments are not publicly disclosed. SR019
CR011 UpGuard continuously monitors Cognite's external security posture; the full vendor risk report is behind a paywall, preventing independent third-party verification of security ratings without a subscription. SR018
CR012 Cognite's certifications include IEC 62443.2-4, IEC 62443.3-2, IEC 62443.3-3, and IEC 62443.4-1, providing industrial cybersecurity compliance for CDF deployments in operational technology environments. SR020, SR017
CR013 Industrial DataOps platform implementations of the scale Cognite targets typically require 6–18 months of integration work before production-grade data pipelines are fully validated, creating an extended period of implementation risk. SR008, SR014
CR014 Cognite ingests real-time process data from SCADA systems, DCS platforms, and historian databases that contain safety-critical operational parameters; a breach of this data layer carries consequences beyond typical enterprise software incidents. SR020, SR002
CR015 An AWS outage on May 7, 2026 disrupted over 150 cloud services, illustrating the systemic infrastructure risk for SaaS platforms deployed on major hyperscaler infrastructure; Cognite's multi-cloud design reduces but does not eliminate this exposure. SR019, SR009
CR016 Aker ASA holds a majority stake in Cognite and serves as its primary governance anchor; Aker's Q1 2026 NAV exceeded NOK 110 billion — its strongest quarter on record — confirming financial capacity to support Cognite. SR021, SR022
CR017 Saudi Aramco acquired a 7.4% strategic stake in Cognite in February 2022, creating both a commercial anchor relationship and potential governance complexity if Aramco's strategic priorities diverge from independent software-company growth strategies. SR022, SR006
CR018 Cognite's partnership with NVIDIA for NV-Tesseract integration creates a direct technology dependency on NVIDIA GPU availability and NIM microservice API continuity for Atlas AI's industrial time-series forecasting capabilities. SR028, SR012
CR019 Cognite maintains co-sell and platform-integration relationships with Microsoft Azure, AWS, Google Cloud, Snowflake, and Databricks; each partnership creates both go-to-market value and dependency risk if relationships change. SR027, SR029
CR020 TotalEnergies and bp are anchor enterprise customers with multi-site Cognite deployments; these relationships provide critical reference-customer value and represent material ARR concentration that would be difficult to replace quickly. SR007, SR024
CR021 Cognite's Snowflake partnership, announced January 2026, creates bidirectional data integration between CDF and Snowflake AI Data Cloud; potential competitive conflict may arise as Snowflake expands into industrial analytics. SR029, SR027
CR022 Aker BP, Cognite's original anchor customer, has deployed Cognite across its full Norwegian Continental Shelf operations; a non-renewal or reduction of this relationship would represent significant ARR loss and reputational signal risk. SR021, SR007
CR023 Oil and gas customers drove approximately 60% of Cognite's revenue as recently as FY2024; the company reports 40% of FY2025 revenue from non-O&G sectors, indicating diversification progress but ongoing concentration risk. SR025, SR024
CR024 Cognite's last publicly disclosed valuation of $1.6 billion dates to its 2021 unicorn milestone; the $300M TCV-led investment in 2024 did not publicly disclose a new valuation, leaving carry-level valuation uncertain. SR023, SR024
CR025 Cognite is privately held and does not file public financial statements; its FY2025 ARR of $170M+ and 36% YoY growth are explicitly described as 'preliminary and unaudited' in company communications. SR025, SR024
CR026 Cognite's revenue model includes professional services and Success Tracks alongside SaaS subscriptions; the services component may compress blended gross margins below pure SaaS benchmarks, reducing the quality of reported ARR. SR025, SR014
CR027 Enterprise industrial AI deal cycles typically require 12–18 months from initial qualification to contract signature, creating revenue timing risk and working capital intensity relative to mid-market SaaS benchmarks. SR008, SR013
CR028 A sustained decline in global oil prices below approximately $60/bbl has historically caused major O&G operators to defer or cancel digital transformation capital expenditure programmes, directly threatening Cognite's renewal rates and upsell momentum. SR022, SR015
CR029 Girish Rishi joined Cognite as CEO in 2022, replacing co-founder John Markus Lervik who transitioned to a founder/board advisory role; the leadership transition concentrated institutional knowledge in a non-founder executive. SR026, SR025
CR030 Cognite relocated its global headquarters from Oslo, Norway to Tempe, Arizona in 2025, creating a geographic divide between corporate strategy and the primary engineering centre, with potential talent attrition risk among senior Norwegian staff. SR025, SR026
CR031 Cognite headcount grew 21% in 2025 to over 800 employees globally; rapid headcount scaling in new geographies risks diluting institutional knowledge, culture, and execution quality. SR025, SR024
CR032 Cognite launched Atlas AI in 2024 and Cognite Flows in May 2026, a rapid succession of major product lines; each launch requires substantial field enablement, customer onboarding capacity, and support infrastructure that must scale with commercial demand. SR025, SR007
CR033 Cognite's 'Moonshot' goal to deliver $100 billion in cumulative customer value by 2035 creates expectation-setting risk; if early milestones are missed, the gap between ambition and delivery could damage customer and investor confidence. SR025
CR034 Microsoft's Copilot for Manufacturing, Google Cloud's industrial AI portfolio, and AWS industrial IoT services represent platform-level competition that could commoditize Cognite's CDF functionality over a 3–5 year horizon. SR008, SR013
CR035 Palantir AIP has expanded into industrial oil and gas markets with enterprise AI workflow deployments that directly compete with Cognite Atlas AI for the same enterprise AI deployment budget. SR030, SR014
CR036 AVEVA (Schneider Electric) and AspenTech (Emerson) hold deep installed-base advantages and existing maintenance contracts in the industrial software market, creating asymmetric switching-cost barriers that favour incumbents in competitive displacement scenarios. SR030, SR013
CR037 Industrial customers' growing familiarity with hyperscaler tools creates a DIY risk where large operators with strong internal IT teams build their own industrial data platforms on commodity cloud services, potentially reducing Cognite's total addressable market over time. SR008, SR014
CR038 C3.ai's public-company revenue misses and stock-price volatility in 2024–2025 illustrate the execution risk in enterprise industrial AI subscription models when deal velocity slows or customer ROI expectations are not met. SR030, SR013
CR039 Cognite's reported net revenue retention of ~150% for newer customer cohorts and 59% Atlas AI adoption by ARR as of FY2025 demonstrate platform stickiness that reduces the probability of unplanned churn in the near-term. SR025, SR024
CR040 Cognite's ISO 27001, SOC 2 Type II, IEC 62443, and dedicated Trust Center provide institutional compliance infrastructure that addresses the most common enterprise security procurement barriers and reduces regulatory non-compliance risk. SR017, SR020
CR041 Cognite's multi-cloud architecture across AWS, Azure, and GCP reduces single-provider outage risk; however, dependencies on all three simultaneously mean a coordinated multi-cloud event could still degrade platform availability. SR020, SR019
CR042 Aker ASA's Q1 2026 NAV of NOK 110 billion — its strongest quarter on record — confirms substantial balance-sheet depth sufficient to fund Cognite through an extended investment horizon without near-term capital stress. SR021, SR022
CV001 Cognite's last publicly confirmed valuation was $1.6 billion, established in May 2021 following a $150 million investment led by TCV. SV010, SV012
CV002 Cognite raised approximately $300 million in a growth equity round led by TCV in June 2024; the post-money valuation for this round has not been publicly disclosed. SV010, SV013
CV003 Total known capital raised by Cognite is at least $450 million ($150M in May 2021 plus approximately $300M in June 2024), excluding any undisclosed Accel investment. SV010, SV012, SV004
CV004 The $1.6B 2021 valuation implies an EV/Revenue multiple of approximately 18–22x trailing revenue, based on estimated revenue of approximately $70–90M at that time. SV012, SV011
CV005 As of May 2026, the 2021 $1.6B valuation mark is approximately five years stale and is the most significant underwriting gap for the chapter-8 valuation analysis. SV012, SV011
CV006 Saudi Aramco acquired a 7.4% equity stake in Cognite from Aker BP in February 2022; the exact transaction value and implied valuation were not publicly disclosed. SV014
CV007 Aker ASA holds a majority stake in Cognite; the exact ownership percentage is not disclosed in public filings available as of May 2026. SV025, SV012
CV008 TCV described the June 2024 $300M round as the largest investment in Cognite's history, according to the TCV press release. SV010
CV009 Cognite reported annual revenue exceeding $170 million for FY2025, representing the company's first year crossing this revenue milestone. SV011, SV013
CV010 Cognite's ARR bookings grew 36% year-over-year in FY2025 and accelerated to 57% year-over-year growth in Q1 2026. SV011, SV029
CV011 Net revenue retention for newer Cognite customers reached 150% as of FY2025, with Atlas AI penetrating 72% of customer ARR by Q1 2026. SV011, SV028
CV012 Cognite AS (Norwegian entity, org. nr. 918274758) reported FY2024 revenue of approximately NOK 1.67 billion (~$158M USD) per Norwegian registry filings. SV015, SV011
CV013 Cognite's gross margin is not publicly disclosed; industrial SaaS platforms with significant data-infrastructure and professional-services components typically carry gross margins of 60–75%. SV001, SV018
CV014 The Cognite AS Norwegian entity reported a net loss of approximately NOK 547M (~$52M USD) in FY2024, implying a group-level burn rate of $60–90M per year. SV015
CV015 At an estimated group-level burn rate of $60–90M per year and a June 2024 round of approximately $300M, Cognite's implied cash runway extends to mid-2027 through mid-2029, subject to undisclosed actual cash position. SV010, SV015
CV016 Cognite does not publicly disclose absolute ARR, customer count, gross margin, full-base NRR, or the 2024 round post-money valuation, preventing independent financial underwriting. SV005, SV011
CV017 C3.ai reported approximately $310M in FY2025 revenue and carries an EV/NTM Revenue multiple of approximately 8–12x as of early 2026, per public financial data. SV001, SV009
CV018 Palantir Technologies reported approximately $2.87B in FY2024 revenue and trades at an EV/NTM Revenue multiple above 30x as of mid-2026, reflecting US government contract premium and AI sentiment. SV002, SV021
CV019 PTC Inc. reported approximately $2.3B in FY2024 revenue and trades at an EV/NTM Revenue multiple of approximately 8–10x, representing a reference for scaled industrial software at lower growth. SV003, SV019
CV020 AVEVA was acquired by Schneider Electric for approximately £9.5 billion ($12B USD) in January 2023, implying an acquisition EV/Revenue multiple of approximately 5–7x on AVEVA's trailing revenue of approximately £1.5B. SV006, SV020
CV021 AspenTech was acquired by Emerson Electric in a transaction valued at approximately $11 billion in 2022, implying an acquisition EV/Revenue multiple of approximately 15–18x trailing revenue at closing. SV024, SV019
CV022 C3.ai's EV/NTM Revenue multiple declined from above 30x in late 2020 to approximately 8–12x by 2025, illustrating the multiple compression risk for high-growth industrial AI platforms. SV001, SV009
CV023 Palantir's EV/Revenue multiple above 30x is not a valid direct comparable for Cognite, given Palantir's more than 40% US defense and government revenue mix which commands a structural premium not available to commercial industrial software vendors. SV002, SV021
CV024 A blended NTM Revenue multiple range of 8–12x is defensible for Cognite in the base case, based on C3.ai (8–12x), PTC (8–10x), AVEVA (5–7x acquisition), and a growth premium for Cognite's faster ARR expansion. SV001, SV003, SV006
CV025 At NTM revenue of approximately $200–220M and a multiple range of 5–20x, Cognite's indicated enterprise value range spans approximately $1.0–4.4B across bear to bull cases. SV011, SV001
CV026 PitchBook data on Cognite is access-restricted; no updated private-round valuation or secondary transaction price is publicly accessible through standard research channels as of May 2026. SV005
CV027 Verdantix named Cognite a Leader in its Green Quadrant for Industrial AI Analytics in 2025, providing independent third-party validation of product positioning relative to commercial peers. SV007, SV016
CV028 S&P Global Market Intelligence paywalls industrial software valuation benchmarking data, preventing independent derivation of median EV/NTM Revenue multiples from primary source data without paid access. SV008
CV029 In the bear case, multiple compression to 5–7x NTM Revenue applied to $180–200M NTM revenue implies a Cognite enterprise value of approximately $0.9–1.4B—below the 2021 $1.6B mark. SV001, SV011
CV030 In the base case, a multiple of 8–12x applied to $200–220M NTM revenue implies a Cognite enterprise value of approximately $1.6–2.6B, contingent on continued 30–40% ARR growth and stable margin. SV001, SV011
CV031 In the bull case, a 14–20x AI-platform premium multiple applied to $210–220M NTM revenue yields an implied Cognite enterprise value of approximately $3.0–4.4B, requiring Atlas AI platform validation and gross margin confirmation. SV002, SV011
CV032 An investor entering Cognite at a post-money price at or below the 2021 $1.6B mark (approximately 8x estimated NTM Revenue) is likely paying at or below the base-case floor value given current revenue scale. SV012, SV011
CV033 The primary thesis-break trigger for Cognite's valuation is Atlas AI ARR penetration falling below 50% of the customer base by ARR within 24 months, removing the AI-platform multiple premium. SV011, SV028
CV034 A down-round financing risk exists for Cognite if the next capital raise occurs during a period of industrial AI multiple compression, given undisclosed gross margin and unclear profitability timeline.
CV035 Cognite's pattern of selectively disclosing growth metrics (ARR growth, NRR, Atlas AI penetration) while withholding gross margin, absolute ARR, and customer count is consistent with managing investor narrative ahead of a potential liquidity event. SV011, SV013, SV029
CV036 An IPO or strategic M&A exit in 2026–2028 is the most plausible liquidity path for Cognite investors; AVEVA ($9.5B) and AspenTech ($11B) acquisition precedents bracket the M&A exit range. SV006, SV024, SV013
CV037 The CNBC Disruptor 50 recognition in May 2026 is consistent with Cognite building public profile ahead of a potential liquidity event, as Disruptor 50 companies historically include pre-IPO companies building brand. SV013, SV029
CV038 Aker ASA's majority ownership of Cognite may deter strategic acquirers and constrain exit timing, as any acquisition requires Aker ASA's consent and potentially conflicts with Aker ASA's industrial data investment thesis. SV025, SV014
CV039 Saudi Aramco's 7.4% equity stake in Cognite creates a dual risk: Aramco churn would simultaneously reduce enterprise revenue and impair equity value, while Aramco's board influence may shape product prioritization toward O&G at the expense of vertical diversification. SV014, SV025
CV040 The Forrester Total Economic Impact study showing 465% ROI for Cognite customers provides a valuation-supportive proxy for customer value capture, though it does not directly address investor return or enterprise valuation multiples. SV018, SV026
CV041 Cognite's Rule of 40 score cannot be calculated from public data; at 36–57% ARR growth and an estimated FCF margin of -20 to -30%, the estimated Rule of 40 spans approximately 6–37—too wide to benchmark. SV011, SV015
CV042 The absence of gross margin disclosure is the single most important financial transparency gap: without it, EV/Gross Profit multiples cannot be applied and the long-run profitability bridge is unmodelable. SV005, SV009
CV043 C3.ai's FY2025 annual report filed with the SEC confirms revenue of approximately $306M and a net loss of approximately $214M, providing a directly observable comparable for industrial AI platform economics at scale. SV009, SV001
CV044 Accel is confirmed as an early investor in Cognite via its portfolio company listing, establishing that Cognite had pre-TCV institutional venture backing at an undisclosed amount and valuation. SV004, SV030
CV045 Palantir's customer mix—more than 40% US government/defense—commands a structural market premium that inflates its EV/Revenue multiple beyond what commercial industrial software benchmarks support; Cognite's applicable multiple ceiling is closer to 14–18x rather than 30x+. SV002, SV021
CV046 Cognite's non-O&G revenue reaching approximately 40% of FY2025 revenue reduces but does not eliminate the O&G concentration discount, as oil-and-gas customers likely still represent the majority of top-decile ARR. SV011, SV018
来源
编号出版方标题引文
SO001 CNBC 15. Cognite – CNBC Disruptor 50 2026 The company's fiscal year 2025 proved a record breaker. Revenue exceeded $170 million, while posting a 36% increase in annual recurring revenue.
SO002 Cognite About Us – Cognite
SO003 Cognite Senior Leadership – Cognite
SO004 Cognite Cognite Secures $150 Million Investment from TCV Cognite was founded in 2017 and is one of the fastest-growing industrial software companies in the world with over 500 employees.
SO005 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year Cognite exceeded $170 million annual revenue ending fiscal year 2025. Annual recurring revenue (ARR) bookings grew 36% year over year. Net revenue retention (NRR) for newer customers ended at 150%.
SO006 Cognite Cognite Atlas AI – Product Page
SO007 Cognite Introducing Cognite Atlas AI – June 2024 Launch Announcement
SO008 Cognite Cognite Positioned as a Leader in IDC MarketScape for Worldwide Industrial DataOps Platforms
SO009 Cognite Digital Transformation: TotalEnergies and Cognite Expand Partnership to Scale Industrial AI
SO010 Cognite Cognite Recognized as 2024 Microsoft Energy and Resources Partner of the Year
SO011 Cognite Security Built In for Critical Infrastructure – Cognite
SO012 Cognite Deploy and Scale Industrial AI – Cognite Homepage
SO013 BusinessWire Introducing Cognite Flows: The First Fully Integrated, AI-Native Industrial Experience
SO014 Automation.com Cognite Flows: The First Fully Integrated, AI-Native Industrial Experience
SO015 Reuters Saudi Aramco Buys 7.4% Stake in Norwegian Software Firm Cognite
SO016 TotalEnergies TotalEnergies and Cognite Expand Their Partnership to Scale Industrial AI Across All Upstream Assets This partnership with Cognite marks a new milestone in our digital transformation. By creating the data foundation which unifies our industrial data globally and makes it AI-ready, we are creating the conditions to accelerate AI-driven solutions.
SO017 World Oil Cognite Partners with Snowflake to Scale Industrial AI for Oil and Gas
SO018 OE Digital TotalEnergies to Roll Out Cognite AI Platform Across Upstream Assets
SO019 UpGuard Cognite Vendor Risk Report UpGuard continuously monitors the security posture of Cognite using open-source, commercial, and proprietary threat intelligence feeds.
SO020 StatusGator Cognite Status – Service Availability Monitoring
SO021 TCV TCV Leads $300M Investment in Cognite
SO022 TechCrunch Cognite Raises $300M
SO023 Forrester Research Cognite Lets Customers Demonstrate the Power of Industrial AI Cognite has real customer deployments — at scale — and used its recent Impact event in Houston to showcase NOVA Chemicals, TotalEnergies, Aker BP, and Cosmo Energy.
SO024 Microsoft Microsoft Announces 2024 Partner of the Year Awards Winners and Finalists
SO025 Databricks Cognite and Databricks Bring Enterprise AI to Industrial Operations
SM001 Grand View Research Industrial Internet Of Things Market | Industry Report, 2030 The global industrial internet of things market size was estimated at USD 483.16 billion in 2024 and is estimated to reach USD 1,693.44 billion by 2030, growing at a CAGR of 23.3% from 2025 to 2030.
SM002 Precedence Research Industrial IoT Market Size to Hit USD 2430.21 Billion by 2035 The global industrial IoT market size is valued at USD 514.39 billion in 2025 and is anticipated to reach around USD 2430.21 billion by 2035, expanding at a CAGR of 16.8% from 2026 to 2035
SM003 Cognite Transform oil and gas with Industrial DataOps | Cognite 90% less time spent making sense of data through unique IT, OT, and ET data contextualization. 10-25 times faster time to deploying solutions into live production. 0.5-1.5% increase in overall production throughput.
SM004 Seeq Seeq - Industrial Analytics and AI for Process Industries
SM005 Seeq About Us: Industrial Analytics & AI for Process Industries | Seeq Seeq is a trusted global leader in advanced industrial analytics, AI-driven insights, and enterprise monitoring software, empowering organizations to unlock the full value of time-series data.
SM006 PTC ThingWorx IIoT Platform | PTC
SM007 C3.ai C3 AI Applications | Enterprise AI Platform
SM008 U.S. Securities and Exchange Commission C3.ai Annual Report (Form 10-K) Filing Search — EDGAR Annual report [Section 13 and 15(d), not S-K Item 405] Acc-no: 0001628280-25-033813 (34 Act) Size: 8 MB — 2025-07-02
SM009 Palantir Technologies Palantir AIP — Artificial Intelligence Platform
SM010 Amazon Web Services AWS IoT SiteWise — Industrial Data Management Service
SM011 Microsoft Azure Azure Digital Twins — IoT Digital Representation Service
SM012 IOGP — International Association of Oil and Gas Producers About Us | IOGP
SM013 Siemens Industrial Edge | Siemens
SM014 IDC IDC MarketScape: Worldwide Industrial DataOps Platforms 2026 Vendor Assessment Cognite Data Fusion supports all five of the key industrial DataOps capabilities assessed in this document under a single product.
SM015 McKinsey & Company Capturing the true value of Industry 4.0 — McKinsey
SM016 Cognite Cognite Newsroom
SM017 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year Cognite exceeded $170 million annual revenue ending fiscal year 2025. Annual recurring revenue (ARR) bookings grew 36% year over year.
SM018 Cognite Cognite Positioned as a Leader in IDC MarketScape for Worldwide Industrial DataOps Platforms Cognite Data Fusion supports all five of the key industrial DataOps capabilities assessed in this document under a single product. Cognite Data Fusion is white labelled by other software providers and is supported by an ecosystem of both technology and service providers.
SM019 Forrester Research Cognite Lets Customers Demonstrate The Power Of Industrial AI Cognite's products (and those of its competitors) are for the firms that know this is slow, hard, and expensive — but ultimately rewarding. They know that someone has to think about data context, lineage, provenance, semantics, and all the rest of it.
SM020 Offshore Engineer Digital TotalEnergies to Roll Out Cognite AI Platform Across Upstream Assets
SM021 World Oil Cognite partners with Snowflake to scale industrial AI for oil and gas
SM022 TotalEnergies Digital Transformation: TotalEnergies and Cognite expand their partnership to scale industrial AI TotalEnergies has expanded its partnership with Cognite to scale the deployment of the industrial AI company's data platform across all of its operated upstream assets worldwide over the period of three years.
SM023 Cognite Deploy and Scale Industrial AI | Cognite
SM024 Business Wire Introducing Cognite Flows: The First Fully Integrated, AI-Native Industrial Experience
SM025 CNBC Cognite — CNBC Disruptor 50 2026
SM026 Automation.com Cognite Flows: The First Fully Integrated, AI-Native Industrial Experience
SM027 Cognite TotalEnergies Expands Industrial AI Partnership with Cognite
SP001 AVEVA AVEVA PI System™ | AVEVA AVEVA PI System is an integrated portfolio of solutions that enables industrial operations to collect, cleanse, and contextualize data from any source.
SP002 AVEVA AVEVA - Global Leader in Industrial Software
SP003 AVEVA CONNECT – Data management
SP004 AspenTech About AspenTech | Aspen Technology History | Industrial AI + Sustainability AspenTech is now part of Emerson.
SP005 AspenTech AspenTech | Process Engineering, Asset Performance Management, Sustainability
SP006 Seeq Corporation Advanced Analytics, ML & AI for Time Series Data | Seeq Because The Best Intelligence Isn't Artificial. It's Human Intelligence, Amplified.
SP007 PTC Inc. ThingWorx: Industrial IoT Software | IIoT Platform | PTC ThingWorx continues powering industrial innovation as part of Velotic™. Same platform, expanded focus, future-ready.
SP008 C3.ai Leading Enterprise AI Software Provider​ | C3 AI
SP009 C3.ai C3 Agentic AI Platform: Enterprise and IoT Applications | C3 AI
SP010 U.S. Securities and Exchange Commission EDGAR Company Search – C3.ai Inc. (CIK 0001807794) 10-K Filings Annual report [Section 13 and 15(d), not S-K Item 405] – Acc-no: 0001628280-25-033813 – Filing Date: 2025-07-02
SP011 Palantir Technologies Palantir Artificial Intelligence Platform
SP012 Palantir Technologies Palantir AIP for Defense
SP013 Amazon Web Services AWS IoT SiteWise Overview AWS IoT SiteWise is a managed service that makes it easy to collect, store, organize and monitor data from industrial equipment at scale.
SP014 Microsoft Azure Digital Twins – Modeling and Simulations | Microsoft Azure Easily model and create digital representations of connected environments with an open modeling language.
SP015 Siemens AG Industrial Edge | Siemens Industrial Edge is an edge computing platform designed to optimize factory operations through scalable and secure solutions.
SP016 GE Vernova GE Vernova | The Energy of Change
SP017 GE Vernova A unified and powerful Force | GE Vernova About
SP018 Hexagon AB About Hexagon | Global Leader in Measurement Technologies At Hexagon, we don't just measure the world – we define its future.
SP019 Gartner Peer Insights Best Global Industrial IoT Platforms Reviews 2026 | Gartner Peer Insights Gartner Peer Insights content consists of the opinions of individual end users based on their own experiences.
SP020 Grand View Research Industrial Internet of Things (IIoT) Market Size, Share & Trends Analysis Report
SP021 Grand View Research Industrial AI Market Size, Share & Trends Analysis Report
SP022 Cognite Industrial AI for Oil and Gas | Cognite
SP023 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year
SP024 Cognite Digital Transformation: TotalEnergies and Cognite Expand Their Partnership to Scale Industrial AI
SP025 World Oil Cognite partners with Snowflake to scale industrial AI for oil and gas
SP026 Precedence Research Industrial IoT Market Size, Share, Growth, Trends, Report 2034
SP027 Databricks Data intelligence for Manufacturing | Databricks
SP028 Gartner Top Strategic Technology Trends 2026 | Gartner
SP029 U.S. Securities and Exchange Commission EDGAR Filing Index – C3.ai Inc. 10-K Annual Report (FY2025)
SP030 International Association of Oil & Gas Producers About IOGP
SI001 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year Cognite exceeded $170 million annual revenue ending fiscal year 2025. Annual recurring revenue (ARR) bookings grew 36% year over year. Net revenue retention (NRR) for newer customers ended at 150%.
SI002 CNBC 15. Cognite – CNBC Disruptor 50 2026 The company's fiscal year 2025 proved a record breaker. Revenue exceeded $170 million, while posting a 36% increase in annual recurring revenue.
SI003 Cognite Cognite Secures $150 Million Investment from TCV TCV has invested $150 million in Cognite at a post-money valuation of $1.6 billion.
SI004 TCV TCV Leads $300M Investment in Cognite
SI005 Reuters Saudi Aramco Buys 7.4% Stake in Norwegian Software Firm Cognite
SI006 Brønnøysundregistrene (Norwegian Business Register) COGNITE AS – Enhetsregisteret Entity Record (Org. 918274758) organisasjonsnummer:918274758, navn:COGNITE AS, stiftelsesdato:2016-12-07, antallAnsatte:367, registreringsdatoEnhetsregisteret:2016-12-19
SI007 Brønnøysundregistrene – Regnskapsregisteret (Norwegian Accounts Registry) COGNITE AS – Annual Accounts FY2024 (Org. 918274758) FY2024 statutory accounts (IFRS): revenue field 1,670,024,000 NOK; net result field -547,002,000 NOK; total assets field 2,263,650,000 NOK; period 2024-01-01 to 2024-12-31.
SI008 Brønnøysundregistrene (Norwegian Business Register – Legacy Portal) COGNITE AS – Nøkkelopplysninger fra Enhetsregisteret (Org. 918274758) COGNITE AS; CEO: Girish D Rishi; Stiftelsesdato: 07.12.2016; Forretningsadresse: c/o Aker Tech House, John Strandruds vei 10, 1360 FORNEBU
SI009 Cognite Cognite Named to CNBC Disruptor 50 List for Leadership in Industrial AI Q1 new annual recurring revenue (ARR) grew 57% year over year. Roughly 72% of overall customers as measured by ARR have operationalized Atlas AI.
SI010 Cognite Cognite Partners with Snowflake to Accelerate AI-Driven Energy Operations By transforming raw, industrial engineering, operational, and IT data, like high-frequency time series and 3D models, into AI-ready data products, we enable energy leaders using Snowflake's AI Data Cloud to uncover high-fidelity insights, like the cost of downtime, to drive greater efficiency and value.
SI011 Cognite Cognite Integrates NVIDIA NV-Tesseract to Scale Industrial AI Forecasting
SI012 IDC IDC MarketScape: Worldwide Industrial DataOps Platforms 2026 Vendor Assessment (Doc #US53013025)
SI013 Aker ASA Aker ASA – Investor Relations and Q1 2026 Results Aker delivered its strongest quarter on record, with Net Asset Value surpassing NOK 100 billion for the first time.
SI014 Cognite Cognite Careers Page
SI015 Aker BP Aker BP Investor Relations
SI016 BusinessWire Introducing Cognite Flows™: The First Fully Integrated, AI-Native Industrial Experience While traditional methods required a 20+ person team and 3–5 months to build a prototype, Flows delivered automated AI workflows in just four days... achieving 30X faster time-to-value in one engagement.
SI017 UpGuard Cognite Vendor Risk Report UpGuard continuously monitors the security posture of Cognite using open-source, commercial, and proprietary threat intelligence feeds.
SI018 Forrester Research Cognite Lets Customers Demonstrate the Power of Industrial AI Cognite has real customer deployments — at scale — and used its recent Impact event in Houston to showcase NOVA Chemicals, TotalEnergies, Aker BP, and Cosmo Energy.
SI019 TotalEnergies TotalEnergies and Cognite Expand Their Partnership to Scale Industrial AI Across All Upstream Assets This partnership with Cognite marks a new milestone in our digital transformation. By creating the data foundation which unifies our industrial data globally and makes it AI-ready, we are creating the conditions to accelerate AI-driven solutions.
SI020 Cognite Digital Transformation: TotalEnergies and Cognite Expand Partnership
SI021 WorldOil Cognite Partners with Snowflake to Scale Industrial AI for Oil and Gas
SI022 Cognite Cognite Positioned as a Leader in IDC MarketScape for Worldwide Industrial DataOps Platforms
SI023 Cognite Security Built In for Critical Infrastructure – Cognite Security Page
SI024 IDC IDC MarketScape: Worldwide Industrial DataOps Platforms 2026 (from Cognite announcement)
SI025 Databricks Cognite and Databricks Bring Enterprise AI to Industrial Operations
SI026 Cognite Cognite Newsroom – All News Forrester Total Economic Impact™ Study Finds Cognite Delivers 465% ROI. Cognite Partners with Snowflake to Accelerate AI-Driven Energy Operations.
SI027 OE Digital TotalEnergies to Roll Out Cognite AI Platform Across Upstream Assets
SE001 Cognite Cognite Developer Portal Use the Cognite API and SDKs to retrieve data, add context, and develop applications that match your operational needs.
SE002 Cognite (GitHub) cognitedata/cognite-sdk-python — Python SDK v8 The SDK v8 introduces full async support with the new AsyncCogniteClient. This enables: Native async/await patterns for all API operations.
SE003 Python Package Index (PyPI) cognite-sdk — PyPI package listing
SE004 Cognite Cognite Trust Center Our security program is architected specifically for the industries we serve, where the consequences of a breach extend far beyond data loss to potential production shutdowns, environmental risks, and threats to human safety.
SE005 Cognite Cognite Flows™ — Product Page Cognite Flows™ is the action layer of the Cognite Industrial AI and Data platform. Rapidly build and scale production-ready workflows that empower frontline teams to turn expertise into enterprise-wide ROI.
SE006 Cognite Cognite Hub — Developer and User Community Help! I am currently facing an issue where the dataset data type is set as string = true, whereas it should be false (numeric). As a result, the chart is not displaying any data.
SE007 Stack Overflow Newest 'cognite' Questions — Stack Overflow
SE008 Gartner Product Information on Cognite Data Fusion — Gartner Peer Insights Cognite Data Fusion supports use cases such as asset monitoring, predictive maintenance, and process optimization by offering features like data modeling, time series analysis, and application orchestration.
SE009 Cognite Cognite + NVIDIA: Intelligent, Cost-Effective Industrial AI Solutions — Partner Page Cognite serves as the critical data refinery for an AI Factory for Industrial Data, accelerated by NVIDIA.
SE010 Cognite Cognite Atlas AI™ — Product Page Cognite Data Fusion's Industrial Knowledge Graph provides Atlas AI agents with industrial context specific to your industrial environment, ensuring trustworthy, reliable insights and delivering interactive-level user performance.
SE011 Cognite Cognite Security and Compliance Page ISO 27001, ISO 9001, ISO 27018, SOC Type 2 (Security), and SOC Type 3 (Security)
SE012 Cognite Introducing Cognite Atlas AI — Newsroom Announcement Cognite Atlas AI's low-code industrial agent builder will enable industrial organizations to use generative AI to carry out more complex operations with greater accuracy.
SE013 BusinessWire Introducing Cognite Flows™: The First Fully Integrated, AI-Native Industrial Experience that Accelerates Customer Value Today, more than 30% of Cognite's customer base and key partners have already been enabled with Flows, including Cognite customers B. Braun and Idemitsu Kosan.
SE014 Automation.com / ISA Cognite Flows: The First Fully Integrated AI-Native Industrial Experience That Accelerates Customer Value A global pharmaceutical leader is using Cognite Flows to accelerate operational use cases, achieving 30X faster time-to-value in one engagement.
SE015 Cognite Cognite and NVIDIA Operationalize NV-Tesseract to Transform Industrial Forecasting — Newsroom This integration leverages the context-rich data within Cognite's platform to power NVIDIA's advanced time-series AI, delivering unprecedented predictive accuracy for critical manufacturing processes.
SE016 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year — 2025 Annual Results Cognite customers' knowledge graph exceeded 80 trillion time series data points in context — up 48%; doubled the number of data modeling instances; and increased AI token usage by nearly 500%.
SE017 Cognite Cognite Positioned as a Leader in IDC MarketScape for Worldwide Industrial DataOps Platforms Cognite Data Fusion supports all five of the key industrial DataOps capabilities assessed in this document under a single product.
SE018 UpGuard Cognite Vendor Risk Report — UpGuard Security Monitoring UpGuard continuously monitors the security posture of Cognite using open-source, commercial, and proprietary threat intelligence feeds.
SE019 StatusGator Cognite Status — Check if Cognite is Down or Having an Outage
SE020 Cognite Cognite — Industrial AI and Data Platform Homepage 30-80% Less time executing maintenance; 7 weeks from deployment to measurable impact; 4 months to scale to 9 facilities.
SE021 WorldOil Cognite partners with Snowflake to scale industrial AI for oil and gas
SE022 Cognite Cognite Partners with Snowflake to Accelerate AI-Driven Energy Operations By transforming raw, industrial engineering, operational, and IT data, like high-frequency time series and 3D models, into AI-ready data products, we enable energy leaders using Snowflake's AI Data Cloud to uncover high-fidelity insights.
SE023 Cognite Cognite Named to CNBC Disruptor 50 List for Leadership in Industrial AI
SE024 IDC IDC MarketScape: Worldwide Industrial DataOps Platforms 2026 Vendor Assessment
SE025 Cognite Cognite Customer Stories Overview
SE026 Cognite Cognite Oil and Gas Industry Page 90% less time spent making sense of data through unique IT, OT, and ET data contextualization.
SU001 Cognite Aker BP Customer Story — Cognite Aker BP has achieved up to 50% reduction in visual inspection time and 30-80% reduction in maintenance execution time.
SU002 Cognite NOVA Chemicals Customer Story — Cognite NOVA Chemicals reduced data access time from hours to minutes across its manufacturing network.
SU003 Cognite Idemitsu Customer Story — Cognite Idemitsu onboarded 4 sites in under 6 months with 3,000+ users and 50,000 hours of annual savings.
SU004 Cognite Celanese Customer Story — Cognite
SU005 Cognite Saudi Aramco Customer Story — Cognite
SU006 Cognite JFE Steel Customer Story — Cognite
SU007 Cognite B. Braun Customer Story — Cognite
SU008 Cognite Cognite Customer Stories Resource Page
SU009 Gartner Cognite Data Fusion — Gartner Peer Insights Reviews
SU010 Aker BP Aker BP — Company Website
SU011 Reuters Saudi Aramco buys 74% stake in Norwegian software firm Cognite Saudi Aramco agreed to buy a 74% stake in Norwegian industrial software company Cognite for about $1.6 billion.
SU012 Forrester Cognite Lets Customers Demonstrate the Power of Industrial AI Cognite's customers demonstrate measurable ROI including 50,000 hours of annual labor savings at Idemitsu.
SU013 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year >$170M revenue FY2025; 36% ARR bookings growth; 150% NRR newer cohorts; 59% customers by ARR using Atlas AI.
SU014 BusinessWire Introducing Cognite Flows: The First Fully Integrated AI-Native Industrial Experience 30%+ of Cognite's customer base has been enabled with Flows; B. Braun and Idemitsu are named design partners.
SU015 TotalEnergies Digital Transformation: TotalEnergies and Cognite Expand Their Partnership TotalEnergies has deployed Cognite across 39 assets with 200+ key production systems integrated.
SU016 OE Digital TotalEnergies to Roll Out Cognite AI Platform Across Upstream Assets
SU017 Cognite Cognite Positioned as a Leader in IDC MarketScape for Worldwide Industrial DataOps
SU018 Databricks Cognite and Databricks Bring Enterprise AI to Industrial Operations
SU019 Cognite Cognite Partners with Snowflake to Accelerate AI-Driven Energy Operations
SU020 World Oil Cognite Partners with Snowflake to Scale Industrial AI for Oil and Gas
SU021 Slashdot Cognite Software Reviews — Slashdot
SU022 PeerSpot Cognite Data Fusion Reviews — PeerSpot
SU023 SourceForge Cognite Software Page — SourceForge
SU024 Automation.com Cognite Flows: First Fully Integrated AI-Native Industrial Experience
SU025 CNBC Cognite — CNBC Disruptor 50 Ranking 2026
SR001 Datatilsynet (Norwegian Data Protection Authority) Datatilsynet — Norwegian Data Protection Authority Are you unsure how to comply with GDPR? Do you process personal data about your staff, consumers, and business partners? Do you want to understand data protection rights?
SR002 European Union Agency for Cybersecurity (ENISA) ENISA Threat Landscape 2024 Seven prime cybersecurity threats were identified in 2024, with threats against availability topping the chart and followed by ransomware and threats against data.
SR003 European Union — Official Journal Regulation (EU) 2022/2065 — Digital Services Act
SR004 European Union — Official Journal Regulation (EU) 2016/679 — General Data Protection Regulation
SR005 GDPR.eu What is GDPR? — GDPR.eu
SR006 Natural Gas World Cognite and Saudi Aramco — Strategic Partnership
SR007 Oil and Gas Digital Cognite and bp Deploy Industrial AI Across Upstream Operations
SR008 Control Engineering Cognite: Challenges in Industrial AI Deployment
SR009 Cloudsmith Cognite Industrial IoT Platform Review
SR010 Verdantix Verdantix: Cognite Industrial AI Analysis
SR011 Oil and Gas Technology Cognite Industrial AI Risks and Challenges
SR012 StackShare Cognite — Tech Stack and Tools
SR013 IEEE Spectrum Industrial IoT in 2024: Security and Deployment Challenges
SR014 Nalpeiron Industrial AI Platform Comparison — Risks and Vendor Lock-In
SR015 Fierce Industry Cognite Oil and Gas Customer Concentration Risk
SR016 Hart Energy Cognite Customer Risk in Oil and Gas Technology Sector
SR017 Cognite Cognite Trust Center — Compliance and Security
SR018 UpGuard Cognite Vendor Risk Report — Security Rating UpGuard continuously monitors the security posture of Cognite using open-source, commercial, and proprietary threat intelligence feeds.
SR019 StatusGator Cognite Service Status — StatusGator Monitoring
SR020 Cognite Cognite Security — Certifications and Compliance Cognite's Secure Development lifecycle, spanning infrastructure, applications, and operations, are tested and audited by third parties, demonstrating compliance to: ISO 27001, ISO 9001, ISO 27018, SOC Type 2 (Security), and SOC Type 3 (Security)
SR021 Aker ASA Aker ASA — Proud and Active Owner Aker delivered its strongest quarter on record, with Net Asset Value surpassing NOK 100 billion for the first time.
SR022 Reuters Saudi Aramco Buys 7.4% Stake in Norwegian Software Firm Cognite
SR023 TCV TCV Leads $300M Investment in Cognite
SR024 CNBC Cognite Named to CNBC Disruptor 50 List for Leadership in Industrial AI
SR025 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year
SR026 Cognite Cognite Leadership — Senior Leadership Team
SR027 Databricks Cognite and Databricks Bring Enterprise AI to Industrial Operations
SR028 Cognite Cognite and NVIDIA Partnership
SR029 World Oil Cognite Partners with Snowflake to Scale Industrial AI for Oil and Gas
SR030 Gartner Cognite Data Fusion Reviews — Gartner Peer Insights
SV001 StockAnalysis.com C3.ai (AI) Financials & Income Statement C3.ai revenue and income statement data for benchmarking EV/Revenue multiples.
SV002 StockAnalysis.com Palantir Technologies (PLTR) Financials & Income Statement
SV003 StockAnalysis.com PTC Inc. (PTC) Financials & Income Statement
SV004 Accel Accel Portfolio Companies
SV005 PitchBook Data Cognite – Company Profile (PitchBook) PitchBook profile access-restricted; no updated valuation or secondary mark publicly accessible.
SV006 Schneider Electric SE Schneider Electric – Investor Relations (se.com)
SV007 Verdantix Green Quadrant: Industrial AI Analytics 2025 (Verdantix) Verdantix named Cognite a Leader in Industrial AI Analytics in 2025 Green Quadrant report.
SV008 S&P Global Market Intelligence S&P Global Market Intelligence – Industrial Software Coverage
SV009 U.S. Securities and Exchange Commission (SEC) C3.ai Inc. – Annual Report (10-K) Filings on EDGAR C3.ai FY2025 10-K confirms annual revenue and net loss for benchmarking purposes.
SV010 TCV TCV Leads $300M Investment in Cognite TCV Leads $300M Investment in Cognite—confirming the June 2024 round amount but not the post-money valuation.
SV011 Cognite Cognite's Moonshot and AI Drive Record-Breaking Year Cognite posted annual revenue over $170 million for the first time; 36% ARR bookings growth year-over-year.
SV012 Cognite Cognite Secures $150M Investment from TCV — $1.6B Unicorn Announcement (2021) Cognite raises $150 million from TCV at a $1.6 billion valuation, becoming Norway's largest SaaS unicorn.
SV013 CNBC Cognite Named to CNBC Disruptor 50 List for Leadership in Industrial AI (May 2026) CNBC Disruptor 50 lists Cognite with $300M in funding—suggesting the 2024 round figure only and not disclosing post-money valuation.
SV014 Reuters Saudi Aramco Buys 7.4% Stake in Norwegian Software Firm Cognite (February 2022) Saudi Aramco acquires 7.4% stake in Cognite from Aker BP—creates strategic customer-shareholder with potential conflict of interest in pricing and product decisions.
SV015 Brønnøysund Register Centre (Brreg) Cognite AS — Annual Accounts FY2024 (Norwegian Registry) Cognite AS (org. 918274758) filed FY2024 accounts: revenue ~NOK 1.67B (~$158M USD); net loss ~NOK 547M (~$52M USD).
SV016 Gartner Gartner Peer Insights — Cognite Data Fusion Reviews
SV017 IDC IDC MarketScape: Worldwide Industrial DataOps Platforms 2025 Vendor Assessment IDC MarketScape names Cognite a Leader in Worldwide Industrial DataOps Platforms 2025.
SV018 Forrester Research Cognite Lets Customers Demonstrate the Power of Industrial AI (Forrester Blog) Forrester TEI study finds Cognite delivers 465% ROI for customers—supporting valuation thesis on customer value capture.
SV019 AspenTech Aspen Technology — About AspenTech
SV020 AVEVA AVEVA PI System — Industrial Data Historian
SV021 Palantir Technologies Palantir AIP — AI Platform
SV022 McKinsey & Company Capturing the True Value of Industry 4.0
SV023 Grand View Research Industrial AI Market Size & Share Report, 2025–2030
SV024 Business Wire Emerson Completes Majority Stake Acquisition of AspenTech Emerson completes $11 billion acquisition of AspenTech—providing M&A precedent multiple for industrial software.
SV025 Aker ASA Aker ASA — About Aker
SV026 TotalEnergies TotalEnergies and Cognite Expand Partnership to Scale Industrial AI
SV027 Seeq Seeq — Industrial Analytics Platform
SV028 Cognite Cognite Atlas AI — Product Page
SV029 Cognite CNBC Disruptor 50 List — Cognite Named for Industrial AI Leadership
SV030 Accel Accel Portfolio — Cognite Investment