初创公司尽调
尽调报告 AI infrastructure / clean-energy data centers Series E private company 2026-07-24

Crusoe Energy Systems

覆盖超大规模园区、模块化 AI 工厂和 GPU 云的能源优先 AI 基础设施

Crusoe 已经搭起差异化的电力优先 AI 基础设施平台,但 Series E 后估值已经提前计入超大规模执行成功,公开证据还撑不起这个假设。

封面要素

成立时间 01
2018 [CO001]
Series D 02
600 USD M [CO014]
当前估值 03
10000 USD M+ [CO018]
累计融资 04
3900 USD M [CO021]
Abilene 园区 05
1.2 GW planned [CO028]
员工数 06
1217 employees [CR031]

公司概况

Crusoe Energy Systems 现在以 Crusoe 品牌运营,最早是一家甲烷减排创业公司,靠模块化数据中心把滞留火炬气变现;此后转向 AI 基础设施,成为最激进的私营建设者之一。公司如今把吉瓦级园区开发、模块化 Spark AI 工厂和 Crusoe Cloud GPU 服务拼在一起,背后有大额融资和 NVIDIA 战略协同支撑,但公开财务透明度仍有限。

官网
www.crusoe.ai
成立时间
2018-01-01
创始人
Chase Lochmiller, Cully Cavness
创立地点
Denver, Colorado, USA
总部
Denver, Colorado, USA
产品
Crusoe 通过三个相连的产品面销售 AI 基础设施:长期超大规模园区容量、Crusoe Cloud GPU 算力和托管推理,以及面向 10–100 MW 部署的 Crusoe Spark 模块化 AI 工厂单元。
客户
超大规模云厂商、前沿模型实验室、AI 原生软件公司、企业 AI 团队和主权 AI 项目。
商业模式
收入来自长期园区租约或容量协议、按 GPU 小时和预留云算力计费、托管推理用量,以及 Spark 部署可能带来的产品 / 托管服务收入。
阶段
Series E private company
融资情况
2024 年 12 月以 $2.8B 估值完成 $600M Series D,2025 年 10 月以 $10B+ 估值完成 $1.375B Series E;另有项目融资和信贷额度支撑数据中心扩张。
[CO001, CO002, CO003, CO006, CO014, CO018, CO021, CO028]

执行摘要

主要优势

  • 从电力采购、园区建设到 GPU 云的纵向整合,形成了差异化的能源优先护城河。
  • 融资厚度、与 NVIDIA 的协同,以及标志性园区和客户落地,让 Crusoe 的可信度超过典型 neo-cloud 初创公司。
  • 产品面已经覆盖超大规模园区、模块化 Spark 部署和云 / 推理服务,并有客户证据支撑。

主要风险

  • 收入集中在少数超大规模锚定客户身上,项目成败和再融资风险都更二元。
  • 毛利率、烧钱速度、利用率和债务契约缺乏公开透明度,$10B+ 估值很难承保。
  • 主要增长项目仍明显依赖天然气支撑的基础设施,ESG 叙事容易被审视。

未决问题

  • 绝对云 ARR、毛利率、烧钱速度和项目融资契约条款。
  • Microsoft、Oracle/OpenAI、Meta 和 xAI 相关容量的收入集中度、续约条款和取消权。
  • 已宣布 GW 级项目转化为有约束力签约容量和已投运容量的比例。

目录

Chapter 01

01公司概览

1.1 身份与商业模式

Crusoe Energy Systems(以 Crusoe 名义运营)是一家总部位于 Colorado 州 Denver 的垂直整合 AI 基础设施公司。公司成立于 2018 年,起步做法是在油气井口部署模块化数据中心,捕获原本会被燃烧浪费的天然气;这套专有方法注册为 Digital Flare Mitigation®(DFM)。DFM 把滞留井口气转成电力,给现场计算供能;燃烧效率最高可达 99.89%,标准火炬约为 93%,相较继续放空燃烧可减少约 63% 的 CO2 当量排放。每个 DFM 部署点估计每年消除 8,000 吨 CO2 当量。 Crusoe 最初把捕获的能源用于 Bitcoin 和加密货币挖矿,但 2023 年 GPU 算力需求激增后,公司根本性转向 AI 云基础设施。Crusoe 借入 $200M 购买 NVIDIA H100 GPU,并开始提供高性能 AI 和机器学习平台 Crusoe Cloud。到 2024 年 12 月 Series D 时,Crusoe 已把自己定义为「行业第一家垂直整合 AI 基础设施提供商」,把专有能源策略(滞留气、可再生能源)、面向 AI 优化的数据中心建设和全栈云平台合在一起。公司当前使命是「加速能源和智能的充裕」。 Crusoe 主要靠长期数据中心租约(由超大规模云厂商级客户的多年合同锚定)以及 Crusoe Cloud 订阅和用量费变现。Texas Abilene 的 1.2 GW 园区由 Oracle 长期租约锚定,支撑近期收入可见度。Denver 和 Tulsa 的制造设施先在受控环境中安装 GPU 服务器机架,再用卡车运往数据中心场地。 [CO001, CO002, CO003, CO004, CO005, CO006]

Crusoe Energy 快照 KPI 表
指标值 / 状态日期置信度缺口 / 尽调问题
最新估值$10B+2025-10私营公司;无审计数字;基于 Series E 交割
Series D 估值$2.8B2024-12-12公司公告;多家新闻来源确认
累计融资(股权 + 债务)~$3.9B2025-10CrunchBase 估计;无审计披露
Series E 金额$1.375B2025-10公司公告;DCD 确认
Series D 金额$600M2024-12-12公司公告;多方来源确认
客户数增长(同比)12 个月内增长超过 7x,截至 2024 年 12 月2024-12公司声称;未披露绝对数字
收入年化规模接近 ~$1B(估计)2026-06分析师估计;公司未披露
云区域4 个云区域(us-northcentral1、us-east1、us-southcentral1、eu-iceland1)2026-06官方产品页
Abilene 园区容量(规划)1.2 GW2026-06公司公告;Forbes 确认
清洁能源储备项目开发中 15+ GW2024-12Series D 时公司声称
已避免温室气体排放(DFM)累计避免排放 680,000+ metric tons CO2e2024公司 ESG 报告;未经审计
员工人数未公开披露直接索取员工人数;LinkedIn 作近似参考

标为 '低' 置信度的数值是分析师估计,或缺少第三方审计的公司主张;估值数字是特定时点的私募市场价格。

[CO014, CO018, CO021, CO023, CO025, CO026]
FO002: Crusoe 公司快照逻辑

Crusoe 商业模式中,身份定位、产品、客户、资本与能源依赖如何彼此连接。

[CO021, CO022, CO032, CO046, CO047]

1.2 创始人与领导层

Crusoe 由 Chase Lochmiller(CEO)和 Cully Cavness(总裁兼 COO)共同创立,两人分别带来技术和能源行业能力。Chase Lochmiller 拥有 MIT 数学、物理学位,以及 Stanford 计算机科学硕士学位,专攻 AI。创立 Crusoe 前,他做过量化交易员,也曾任加密货币投资机构 Polychain Capital 的普通合伙人,因此深度接触了计算金融,以及后来催生 Crusoe 原始构想的区块链 / 能源交叉地带。 Cully Cavness 拥有 Middlebury College 地质学学位和 Oxford MBA,并有油气投行背景。他对能源行业和上游油气经济的实操理解,支撑了 Crusoe 最初的 Digital Flare Mitigation 业务和监管推进策略,包括直接与 North Dakota 州议会合作,为部署火炬减排系统的公司推动税收减免立法。 管理层已大幅扩容。关键任命包括 Nadav Eiron 出任云工程 SVP、Erwan Menard 出任 Crusoe Cloud 产品 SVP,以及 Nader Pakfar 出任总法律顾问(房地产),后者于 2026 年 1 月获任。战略顾问包括 Bill Stein(Digital Realty 前 CEO)、Peter Gross 和 Matt Field。Lochmiller 和 Cavness 兼具联合创始人与运营角色,公司对二人的关键人依赖度高,接班或离任风险实质存在。 [CO010, CO011, CO012, CO013, CO043, CO044]

领导层与创始人表
人物职位背景创始人-市场匹配 / 覆盖能力关键人物依赖
Chase LochmillerCEO、联合创始人MIT(数学 / 物理);Stanford CS/AI 硕士;量化交易员;Polychain Capital GPML/AI 技术深度;资本市场经验;计算基础设施愿景关键——同时担任 CEO 和技术负责人;离职会造成重大扰动
Cully Cavness总裁兼 COO、联合创始人Middlebury(地质);Oxford MBA;油气投行能源行业运营;监管周旋;上游油气关系关键——掌握运营执行和能源合作网络
Erwan Menard产品 SVP(Crusoe Cloud)云基础设施产品领导背景云产品路线图;推理服务;企业 AI 平台中等——云收入增长的关键
Nadav Eiron云工程 SVP大规模云系统工程工程扩展和可用性;GPU 集群编排中等——停机事件后承担可靠性改进职责
Nader Pakfar房地产总法律顾问SPC LLP 创始人;房地产法土地收购;租赁结构设计;场址开发法律事务低——专业性强,但不是单点人风险
Bill Stein董事会顾问Digital Realty 前 CEO大规模数据中心建设与运营;超大规模云厂商关系低——顾问角色

董事会构成和独立董事名单未公开披露;投资人董事席位根据领投方角色推断。

[CO010, CO011, CO012, CO013, CO043, CO044]

1.3 融资历史与投资方

Crusoe 的融资路径显示,公司从气候科技种子项目一路演进为大规模 AI 基础设施投资标的。2018 年 12 月,公司完成约 $30M Series A,由 Founders Fund 领投——这家 Peter Thiel 的风险投资机构在之后每一轮都持续担任主要投资方。2021 年 5 月,公司又从 Bain Capital Ventures 等投资方获得约 $50M Series B,用于扩张 DFM 业务。 真正改变公司量级的是 2024 年 12 月的融资:2024 年 12 月 12 日,公司以 $2.8B 估值完成 $600M Series D。Founders Fund 再次领投,NVIDIA、Fidelity、Long Journey Ventures、Mubadala(Abu Dhabi 主权财富基金)、Ribbit Capital 和 Valor Equity Partners 参投。Blue Owl Capital 参与股权轮,并另外帮助从 JPMorgan 安排 $2.3B 项目融资,用于 Abilene 园区第一阶段。发布时 Peter Thiel 的评论——「AI 最大的风险,是我们做得还不够大。Crusoe 要把我们从野心有限的孤岛上解放出来」——凸显了这轮融资的战略意义。 2025 年 10 月,Crusoe 完成超额认购的 $1.375B Series E,估值超过 $10B,由 Valor Equity Partners 和 Mubadala Capital 共同领投,超过 20 家投资方参与,包括 NVIDIA、Fidelity、Founders Fund、T. Rowe Price、Tiger Global Management、Altimeter Capital、Franklin Templeton、Salesforce Ventures、Lowercarbon Capital 和 Spark Capital。公司在约 13 轮债务和股权融资中累计融资约 $3.9B。员工数和私营收入数字没有公开披露;公司尚未公开承诺 IPO 时间表,不过 CEO Lochmiller 曾表示公司最终会上市。 [CO014, CO015, CO016, CO017, CO018, CO019]

利益相关方或投资人图谱
利益相关方角色 / 轮次控制权或经济重要性尽调问题
Founders FundSeries A、D、E 领投方;Peter Thiel / Sean LiuSeries D 中的主要经济权益和董事会影响力;跨轮次长期关系确认董事席位、按比例认购权、治理文件
Valor Equity PartnersSeries E 联合领投方Series E 后按 $10B 估值持有重要经济权益投资论点;治理预期
Mubadala CapitalSeries E 联合领投方;Series D 参与方UAE 主权财富基金;资本体量和潜在区域扩张意义数据驻留要求;任何治理条件
NVIDIASeries D 和 E 参与方;战略伙伴关键技术和 GPU 供应关系;投资人身份强化供应优先级排他或优先定价条款;董事会观察员权利
FidelitySeries D 和 E 参与方机构级后期跨界投资人;潜在 IPO 锚定方二级市场活动;锁定期预期
Blue Owl Capital / JPMorganAbilene 项目融资($2.3B)对核心基础设施资产享有留置权的债务提供方;契约风险重大贷款契约;违约事件触发条件;再融资时间线
OracleAbilene 园区锚定租户长期租约支撑收入;据报道合同约 ~15 年租约条款;解约条款;Oracle 交易对手信用质量
OpenAI / Microsoft通过 Oracle 使用算力的最终用户间接但关键;Stargate 需求驱动和商业案例锚点Oracle 到 OpenAI 的合同链;Microsoft Azure 角色

精确持股比例和董事席位分配未公开;信息来自新闻稿、Forbes 报道和 DCD 覆盖。

[CO014, CO015, CO016, CO018, CO019, CO020]

1.4 里程碑与战略演进

Crusoe 的轨迹分为三个清晰阶段:(1)火炬气加密挖矿阶段(2018–2022),(2)转向并建设 AI 云阶段(2023–2024),以及(3)超大规模数据中心开发阶段(2024 年至今)。2020 年初,COVID-19 封锁压垮石油需求,公司一度接近生死线,被迫转向更贴近可再生能源的站点。这次被迫多元化最终为 AI 转型铺好了位置。 2024 年第一季度,Crusoe 拿下 Texas Abilene 未来 Stargate AI 数据中心园区的开发合同,成为关键战略节点。Elon Musk 的 xAI 据称是该场地第一位潜在客户,但后来退出,选择在 Memphis 独立建设「Colossus」超级集群。随后 Oracle 签署长期租约,设施设计目标是经由 Microsoft Azure 向 OpenAI 提供 100,000+ 颗 NVIDIA GPU。1.2 GW Abilene 园区第一阶段于 2025 年 9 月上线,是有记录以来推进最快的绿地超大规模数据中心建设之一。2025 年 3 月初一场 45 小时故障引发客户投诉和行业质疑;随后 2026 年 5 月 us-east1-a 区域又发生网络事件,说明运营可靠性仍是最需要跟踪的声誉风险。 2026 年的产品里程碑包括 Crusoe Edge Zones 上线(2026 年 3 月)、Crusoe Command Center 统一运营平台上线(2026 年 2 月),以及取得 ISO 27001 / ISO 42001 认证(2026 年 2 月)。2026 年 3 月 NVIDIA GTC 上,Crusoe 宣布加深合作,涵盖成为 NVIDIA Vera CPU 和 Rubin GPU 早期采用者、支持 NVIDIA Omniverse DSX Blueprint,并向 NVIDIA Dynamo 开源框架贡献专有 Rust BPE tokenizer。 [CO023, CO024, CO025, CO026, CO027, CO028]

里程碑表
日期事件类型金额 / 估值 / 状态参与方含义
2018公司在 Denver, CO 成立创立N/AChase Lochmiller、Cully CavnessDigital Flare Mitigation 概念源头;DFM 注册为商标
2018-12Series A 融资完成融资$30MFounders Fund(领投)早期机构背书;基金发起人 Peter Thiel 与 DFM 气候科技论点一致
2020COVID-19 导致石油需求崩塌;Crusoe 部分转向可再生能源负面N/A内部;油气合作方需求下降近乎生死存亡的威胁;转向可再生能源场址的韧性后来成为 AI 转型关键
2021-05Series B 融资完成融资$50M+Bain Capital Ventures 等用于扩张 DFM 机群;部署到 Bakken、Permian、Wyoming
2023转向 AI 云;借入 $200M 购买 NVIDIA H100 GPU产品$200M 债务NVIDIA(GPU 供应商);Digital Realty、AtNorth(Iceland,托管共址)定义了以 AI 为先的现代 Crusoe;退出加密挖矿路径
2023与 AtNorth 建立合作;Crusoe Cloud 服务落地 Iceland合作N/AAtNorth(Iceland 数据中心开发商)首个国际云区域;EU 布局;凉爽气候提升 GPU 效率
2024-Q1赢得 Abilene, TX 开发合同;从 Lancium 转租 1,000 acres产品N/AOracle、OpenAI、Lancium拿到关键 Stargate 场址;最快完成签约的超大规模园区 RFP
2024-06Oracle-OpenAI-Crusoe Abilene 合作公开宣布合作$300B Oracle-OpenAI 合同(Oracle 总额)Oracle、OpenAI、Microsoft Azure战略锚点;$300B 合同宣布后 Oracle 股价随后上涨 40%
2024-12-12Series D 交割;Crusoe Cloud GA 发布融资$600M,估值 $2.8BFounders Fund(领投)、NVIDIA、Fidelity、Mubadala、Ribbit、Valor、Long Journey转型级融资;Crusoe Cloud 正式 GA;验证 AI 基础设施论点
2025-03云服务中断 45 小时;客户和行业批评负面N/A未具名客户(Forbes)、未具名行业专家声誉风险;Series E 前可靠性担忧已被标记;可用性主张受质疑
2025-09Abilene 1.2 GW 园区一期上线规模化N/AOracle、OpenAI、高峰期 5,000 名现场工人最快的绿地超大规模数据中心建设;验证建设执行模型
2025-10Series E 按 $10B+ 估值交割融资$1.375B,估值 $10B+Valor、Mubadala(联合领投)、NVIDIA、Fidelity、Tiger Global、Altimeter、20+ 其他方10 个月估值上升 3.6x;广泛机构背书
2026-02ISO 27001 + ISO 42001 认证;Command Center 平台发布产品N/A内部企业可信度;安全与负责任 AI 治理认证
2026-03NVIDIA GTC 2026 深度合作公告;Edge Zones 发布合作N/ANVIDIA 合作(Vera CPU、Rubin GPU、Dynamo、Omniverse DSX)全栈 NVIDIA 原生定位;专有分词器向 Dynamo 开源
2026-05us-east1-a 区域网络中断负面N/A内部工程;影响部分计算主机14 个月内第二起重大可靠性事件;工程缓解已完成

融资金额来自公司新闻稿和 DCD 报道;负面事件来自 Forbes 和公开事故追踪;2020 年 COVID 事件来自 Keep Cool / Bain Capital Ventures 叙事。

[CO001, CO002, CO003, CO014, CO018, CO028]
FO001: Crusoe Energy 公司里程碑时间线

2018–2026 年,Crusoe 从 DFM 创立一路演进为超大规模 AI 基础设施公司。

[CO001, CO014, CO018, CO028, CO030, CO039]

1.5 产品组合与运营

Crusoe 横跨三层产品。第一层是实体 AI 工厂基础设施:公司获取土地,设计并建设吉瓦级数据中心园区,再向锚定租户(Oracle,以及潜在其他超大规模云厂商)提供长期租赁安排。Abilene 园区规划 1.2 GW;Wyoming 一个 1.8 GW 园区正在规划;Canada 的 Alberta 和 Norway 的 Polar 站点也在开发中。第二层是 Crusoe Cloud,一个四区域 AI 云平台(us-northcentral1-a、us-east1-a、us-southcentral1-a、eu-iceland1-a)。Crusoe Cloud 提供 GPU 实例、Managed Kubernetes,以及由专有 MemoryAlloy™ 技术驱动的 Crusoe Managed Inference Service——这是一张集群级 KV 缓存织网,相比标准推理引擎,time-to-first-token 最多快 9.9 倍。第三层是 Crusoe Intelligence Foundry 和 Edge Zones:Crusoe Edge Zones 于 2026 年 3 月推出,是模块化、工厂预制集群,可用于主权 AI、低延迟推理和企业专用部署。 Series D 公布时,公司在美国 9 个州和 3 个国家运营,过去 12 个月客户数增长超过 7 倍。值得关注的云客户包括 Decart(四天内扩容 5 倍,服务 200 万用户)、Codeium/Windsurf(700,000+ 开发者,集群正常运行时间 99.98%)等。Crusoe 也已剥离加密挖矿业务,彻底转向 AI 基础设施。根据 ESG 披露,DFM 技术帮助公司避免了超过 680,000 吨 GHG 排放。分析师认为,能源优先策略——提前多年锁定滞留气、风能、地热和可再生能源站点——就是 neocloud 市场的核心护城河。 [CO046, CO047, CO048, CO049, CO050]

FO003: Crusoe 快照 KPI

截至 2026 年中,概括 Crusoe 成熟度和牵引力的关键绩效指标。

收入和员工数未披露;估值为最后一轮收盘时的私募市场价格。

[CO018, CO021, CO025, CO026, CO027, CO034]

1.6 图表与证据

Chapter 02

02市场分析

2.1 市场定义与边界

Crusoe Energy 处在三个彼此不同但重叠的市场交汇处。第一是 neocloud GPU 云:面向 AI 训练和推理工作负载,按需、预留实例或长期合同提供的专用 GPU 加速 IaaS。第二是 AI 工厂开发:获取绿地、建设并长期出租吉瓦级数据中心园区,租给超大规模云厂商锚定租户。第三项战略性越来越强,即面向计算的能源采购和发电:捕获滞留气、可再生能源和替代电源,为 AI 数据中心打底;这个市场正从成本项迅速演化成数千亿美元级战略投资类别。 AI 计算基础设施的市场边界包括 GPU 加速 IaaS、裸金属 HPC、AI 工厂托管和托管 GPU 云服务。不包括通用云(广义 CPU-only IaaS)、AI 软件和应用层(SaaS、MLOps 工具)以及实体 ASIC 设计。相邻市场包括面向计算的能源基础设施(数据中心用天然气、核能、太阳能)和 CPU-only 机房托管;后者是较小或非 GPU AI 工作负载的现状替代方案。关键替代品还包括超大规模云厂商 spot 实例、大学裸金属 HPC 和本地 GPU 集群;相较专用 neocloud 供给,这些方案都存在明显成本、可用性或可扩展性限制。 GPU-as-a-Service(GPUaaS)是一个独立且快速增长的子领域。Fortune Business Insights 数据显示,该市场 2023 年规模为 $3.23B,预计复合年增长率为 36%,反映企业正从购买 GPU 转向租用 GPU 访问。广义 neocloud 类别——包括 CoreWeave、Lambda Labs、Nebius、Crusoe 和较小同业——据 Synergy Research Group 数据,2025 年收入约 $23B。 [CM001, CM002, CM003, CM004, CM005, CM006]

市场定义表
维度AI 算力基础设施(范围内)排除 / 相邻对 Crusoe 的意义
服务模式GPU-as-a-Service、裸金属 HPC、AI 工厂共址托管通用 IaaS(仅 CPU)、SaaS、PaaS、MLOps 工具Crusoe 提供 GPU 云 + 超大规模 AI 工厂 / 共址托管
硬件重点NVIDIA H100、H200、B200 GPU 集群;InfiniBand 网络仅 CPU 服务器;仅存储基础设施Crusoe 所有收入流都以 GPU 为中心或依赖 GPU
客户类型AI 实验室、AI-native 初创、部署 GenAI 的企业、超大规模云厂商、主权 AI 项目面向消费者的应用;SaaS 产品内嵌 AICrusoe 瞄准 B2B 企业、AI 实验室和超大规模云厂商锚定客户
能源栈替代 / 可再生 + 电网:滞留天然气、太阳能、地热、表后仅电网零售电力;无能源策略Crusoe 能源优先的出身是主要差异化
现状替代方案超大规模云厂商 spot 实例;本地 GPU 集群;大学裸金属 HPC替代方案贵 2–6x,或受容量 / 可用性限制

市场范围按 Crusoe 当前和规划产品界定;相邻市场(能源基础设施、AI 软件)仅作为战略背景跟踪。

[CM001, CM002, CM003, CM004, CM005, CM006]

2.2 市场规模测算:多视角分析

AI 基础设施市场的分析师预测差异很大,取决于定义边界,因此任何 SAM/SOM 测算都必须多视角交叉校准。最宽口径下,S&P Global Market Intelligence 估计 2025 年 AI 基础设施支出为 $337B,到 2030 年以 28% CAGR 增至 $1.2T。IDC 预测全球 AI 基础设施支出 2026 年达到 $487B,较 2025 年同比增长 53%。ARK Investment Management 的预测更高,认为市场将从 2025 年 $500B 增至 2030 年约 $1.5T。这些宽口径估计同时覆盖超大规模云厂商和专门厂商的硬件、软件与服务;Crusoe 能直接触达的只是其中一部分。 缩窄口径后,Gartner 预测 AI-optimized infrastructure-as-a-service(AI-optimized IaaS)——最接近 neocloud 和 Crusoe Cloud 板块的代理指标——2025 年为 $18.3B,2026 年同比增长 105% 至 $37.5B。这是 Crusoe 云业务最相关的单一 TAM 对标。S&P Global 进一步将 AI-as-a-Cloud-Service(ACaaS)列为增长最快的 AI 基础设施子板块,2025 年规模 $20B,并以 48% CAGR 增至 2030 年 $145B。 仅看 neocloud 板块,Synergy Research Group 估计 2025 年收入约 $23B,Forrester 对 2026 年的预测为 $20B(尽管口径可能不同)。SRG 认为该板块到 2030 年会达到 $180B;IndexBox 预测到 2031 年为 $236B,CAGR 为 46%。多组预测明显分歧,反映定义不一致(有的纳入超大规模云厂商专用 GPU 服务,有的剔除);这种分歧本身是尽调缺口,而不是模型错误。 测算 Crusoe 的可服务市场时,超大规模数据中心开发业务(AI 工厂)和 GPU 云(Crusoe Cloud)必须分开。仅 Abilene 园区全部阶段的建设和基础设施投资就超过 $10B,长期租约锚定 AI 工厂收入。Crusoe 云 SAM 则更贴近 Gartner 的 AI-optimized IaaS 区间。实际收入数字没有公开披露。 [CM007, CM008, CM009, CM010, CM011, CM012]

TAM/SAM/SOM 或规模测算口径表
口径 / 范围定义2025 估计2026E 估计2030E 估计主要来源
最宽 TAM:全部 AI 基础设施AI 硬件、软件和服务,包括超大规模云厂商和专业厂商$337–500B$487–1,000B+$1.2–1.5T数据来源:S&P Global、IDC、ARK Invest
AI 优化型 IaaS(Gartner)仅 GPU/AI 加速云 IaaS;最接近 neocloud 收入的近似指标$18.3B$37.5B(同比 105%)N/AGartner(2025 年 10 月)
ACaaS:AI-as-a-Cloud-ServiceGPU 云 + AI-native 托管服务$20B~$30B$145B(48% CAGR)数据来源:S&P Global
Neocloud 细分市场专用 GPU 云提供商(CoreWeave、Lambda、Nebius、Crusoe 等)~$23B$20–40B$180–236BSynergy/RCR、Forrester、IndexBox
GPU-as-a-Service按需 GPU 租赁(按小时或预留)~$8–10B(估计)~$12–15B(估计)~$50–67BFortune Business Insights(自 2023 基数起 36% CAGR)
Crusoe SAM(估计)Crusoe 目标地域内的 GPU 云 + 超大规模 AI 工厂租赁未披露未披露未披露无公开数据;收入未披露

各来源差异很大,来自定义范围不同;$487B(IDC)和 $18.3B(Gartner IaaS)都指 2026 年,但衡量对象不同。Crusoe SAM 是证据缺口。

[CM007, CM008, CM009, CM010, CM011, CM012]
FM001: 市场规模测算视角

从最宽 TAM 到 Crusoe 估算 SOM 的 AI 基础设施市场范围层级。

金字塔层级仅作示意;Gartner AI 优化 IaaS 与 neocloud 估算会因定义口径不同而部分重叠。

[CM007, CM008, CM011, CM012, CM014]
FM002: 市场估算区间

分析师对 2025、2026 和 2030 年 AI 基础设施与 neocloud 市场规模的估算区间。

估算来自多种方法;各来源的 TAM 定义不一致(仅硬件 vs. 硬件 + 软件 + 服务)。所有数字都按方向性区间处理,而非点估计。

[CM007, CM008, CM009, CM011, CM012, CM013]

2.3 买方分层

AI 云买方可分成五条线,每条线的预算归属、采购机制和 SLA 预期都不同。超大规模云厂商锚定租户——Oracle、Microsoft、Amazon 和 Google——代表最高价值账户,会与 neocloud 签署数十亿美元级容量承诺,常常提前多年预留产能。CoreWeave 截至 2026 年中 $66B 至 $99.4B 的合同积压就是这种机制的例子,主要由 Microsoft 和 Meta 锚定合同驱动。Crusoe 的 Oracle-Abilene 交易也遵循这一模式:一份长期租约锚定 AI 工厂业务,交易对手信用质量接近超大规模云厂商。 AI 基础模型实验室和前沿 AI 公司(OpenAI、Anthropic、xAI、Mistral)是第二层:这些公司的完整价值主张都高度依赖算力。这类买方每年在训练和推理 GPU 算力上消耗数亿到数十亿美元,通常采用预留集群安排。xAI 在 Abilene 的擦肩而过,以及 Crusoe 通过 Oracle Stargate 交易间接触达 OpenAI 需求,都说明 Crusoe 离这一层很近。 AI 原生创业公司是数量型买方,单客支出较小:每月 $5,000 到 $500,000,不过中型生产团队平均每月 $50,000 到 $500,000。这一板块推动 Crusoe Cloud 利用率和按需 GPU 产品收入。采用 GenAI 做业务应用的企业 IT 买方,是一个正在浮现且战略重要的群体:2025–2026 年 GenAI 从试点走向生产后,企业需要合规认证(SOC 2、HIPAA、FedRAMP),目前这些认证更有利于超大规模云厂商。Crusoe 的 ISO 27001 和 ISO 42001 认证是迈向企业就绪的步骤,但 SLA 历史仍令人担忧。 主权 AI 项目——由各国政府建设、带有数据驻留要求的本土 AI 基础设施——是快速出现的买方板块。Nebius 从 Microsoft 和 Meta 获得 $46B 欧洲容量承诺,并按 GDPR 条款执行,显示了该市场的规模。Crusoe 于 2026 年 3 月推出的 Edge Zones 产品正对准这一用例,用模块化、工厂预制集群服务主权和低延迟部署。 [CM019, CM020, CM021, CM022, CM023, CM024]

细分市场 / 买方图谱
属性超大规模云厂商锚定客户AI 基础模型实验室AI-native 初创企业 IT 买家主权 AI 项目
预算规模已承诺 $1B–$20B/yr$500M–$5B/yr 算力支出AI 初创预算:$5K–$500K/mo企业预算:$500K–$50M/yr政府资助,浮动
主要用例超大规模训练 + 推理前沿模型训练;高吞吐推理产品 AI、MLOps、微调业务应用中的 GenAI 部署国家 AI 能力;数据主权
采购模式长期 SPA(5–15 年)预留集群(1–3 年)按需 + 短期预留带 SLA 的托管服务长期主权合同
关键采购标准电力保障 + 建设速度GPU 密度 + InfiniBand 网络成本 + 可用性SLA + 合规认证数据驻留 + 本地控制
Crusoe 匹配度很高(Abilene-Oracle 先例)高(xAI 曾接近成交;OpenAI 通过 Oracle 间接触达)中等(Crusoe Cloud 定价有竞争力)仍在成形(ISO 认证;SLA 履约历史仍是疑点)增长中(Edge Zones 2026 年 3 月发布)

细分边界仅作示意;单一客户(如 Microsoft)可能跨多个采购通道。预算规模来自公开定价和分析师来源,只能作方向性估计。

[CM019, CM020, CM021, CM022, CM023, CM024]
FM003: 买方 / 细分市场地图

五类买方地图,展示各细分市场的预算归属、采购模式、关键标准,以及 Crusoe 的相对契合度。

契合度评分是基于公开产品定位和已披露客户关系的定性评估。

[CM019, CM020, CM021, CM022, CM023, CM025]

2.4 增长驱动因素

最主要的需求催化剂,是 AI 模型从批量训练转向持续的代理式 AI 推理。到 2025 年末,推理工作负载消耗的 GPU 总算力已经超过训练工作负载,背后是企业软件、代码助手和消费者应用中 AI agents 的铺开。Agentic AI 系统每个任务所需推理算力比单轮模型高出数量级,创造出持久的指数级需求。NVIDIA CEO Jensen Huang 在 2026 财年第四季度业绩会上指出,「有用的 AI agents 才刚开始推出」,且规模化后「极度吃 token」。 企业 AI 采用在 2025–2026 年进入结构性拐点,从受控试点转向任务关键型生产工作负载。neocloud 相对超大规模云厂商的成本套利提供了可量化驱动因素:领先 neocloud(Lambda Labs、CoreWeave)的按需 H100 GPU 价格为每 GPU 小时 $2.49 至 $4.25,而 AWS 为每 GPU 小时 $6.88,Azure 为每 GPU 小时 $12.29。下一代 B200 GPU 的价差更大:AWS 约每 GPU 小时 $14.24,而 Lambda Labs 为 $4.99 至 $5.29。2–6 倍成本差有量化依据,也已在 2026 年第一季度公开定价数据中记录。 电力可用性已经成为竞争定位最关键的单一驱动因素,2025–2026 年实际取代 GPU 硬件可用性,成为首要约束。2023–2024 年锁定电力容量的运营商,相比面对 4–10 年电网并网排队的新进入者,拥有结构性的多年优势。NVIDIA 用 2025 年第三季度(2025 年 10 月)$57B 收入验证了需求基本面,同比增长 62%;Amazon $34.2B、Microsoft $35B 和 Alphabet $24B 的 2025 年第三季度资本开支也相互印证,且主要投向 AI 基础设施。国家层面的主权 AI 政策要求也在扩大 TAM,在美国市场之外创造新的需求口袋。 [CM028, CM029, CM030, CM031, CM032, CM033]

增长驱动与约束表
因素类型方向影响幅度时间线机制
GenAI 智能体推理需求驱动因素顺风很高2025–2030AI 智能体每项任务所需 token 比单轮模型高 10–100 倍
企业从试点转向生产驱动因素顺风2026 年拐点超过 50% 的企业已有活跃试点;到 2026 年,25% 进入生产
新云厂商成本套利(相对超大云厂商便宜 2–6 倍)驱动因素顺风当前新云厂商 H100 $2.49/hr,AWS $6.88/hr;价差已量化且可复现
电网并网延迟(4–10 年)约束逆风关键多年结构性新进入者要排数年队;已锁定容量的既有玩家把竞争挡在门外
NVIDIA GPU 稀缺与集中驱动 + 约束双向2025–2027稀缺推高溢价;约 95% 的 AI 训练市场份额带来单一供应商风险
表后供电采用驱动因素顺风中高2026–2028燃料电池、太阳能 + 储能绕开电网延迟;capex 增加,但容量被释放
资本强度(每园区 $1B+)约束逆风持续竞争只剩资金充足的玩家;相较资本不足的新进入者,Crusoe 占优
超大云厂商的 SLA 与合规优势约束逆风中等当前SOC 2/HIPAA/FedRAMP 认证让受监管企业工作负载更偏向超大云厂商
定制芯片替代方案(Cerebras、Groq)约束(潜在)逆风低–中(2027+)正在出现下一代 ASIC 可能降低对 NVIDIA 的依赖;时间线仍不确定
主权 AI 政策要求驱动因素顺风中等2026+政府 AI 投资正在美国市场之外开出新的买方通道

影响幅度评级是相对且定性的。双向因素(NVIDIA 稀缺)对不同参与方影响不同:持有 GPU 的玩家受益;新进入者受限。

[CM028, CM029, CM030, CM031, CM032, CM033]

2.5 采用约束与竞争风险

供给扩张的结构性约束是电力。2025–2026 年进入电网并网队列的新数据中心项目,在美国主要市场通常要等待 4 到 10 年。Northern Virginia 是美国最大数据中心市场,但拥堵严重到公用事业公司已警告区域电力短缺,迫使新项目转向二级电网区域的二级市场。World Economic Forum(2026 年 5 月)把电网连接称为「AI 的战略瓶颈」,并指出 AI 园区开发商若能结合分阶段接入、现场储能和负荷控制,就能在电网增强完成前先启动运营。 资本强度形成高门槛护城河。吉瓦级 AI 数据中心每个园区需要超过 $1B 的建设和基础设施投资,把有效竞争者限制在资本充足的玩家中。表后电力方案(Bloom Energy 燃料电池、太阳能 + 储能、下一代核微反应堆)正在成为无法等待电网并网的运营商标配,但这些方案同样需要大量前期资本。 客户切换成本是中等但真实的市场进入和流失壁垒。生产 AI 工作负载在 GPU 云提供商之间迁移,需要数据迁移、应用重构、API 变更和合规重新认证。企业和主权买方已经投入集成,因此锁定效应更强。不过,AI 原生创业公司通常更价格敏感,也更愿意切换。 NVIDIA 在 AI 加速器中的主导地位,为所有 neocloud 玩家带来集中度风险。NVIDIA 约占 AI 训练加速器市场的 95%;Cerebras(wafer-scale engine)、Groq(最近在团队收购中被 NVIDIA 吸收)和 AMD MI 系列芯片等替代方案存在,但在大规模训练中的生产采用有限。可靠性担忧是 neocloud 类别特有的结构性风险:企业买方期待超大规模云厂商级 SLA 承诺(99.99%+),但早期 neocloud 在实践中难以达到,Crusoe 自身 2025 年 3 月 45 小时故障和 2026 年 5 月网络事件就是例证。 [CM037, CM038, CM039, CM040, CM041, CM042]

FM004: 采用漏斗或价值链地图

AI 云采用漏斗从初始认知走向锚定租户容量承诺,展示每个阶段的转化率。

漏斗转化率是基于 Flexential 试点数据和分析师评论综合得出的估算;没有单一来源发布端到端转化数据。

[CM029, CM019, CM020]

2.6 图表与证据

Chapter 03

03竞争对手

3.1 竞争格局概览

Crusoe 所处的 GPU 云市场至少有四个结构层级,每一层都带来不同竞争挑战。第一层是纯 neocloud GPU 云运营商——CoreWeave、Lambda Labs、Nebius AI 和 Vast.ai——它们在开发者、AI 实验室和企业 GPU 容量买方上,是 Crusoe 最直接的收入竞争者。CoreWeave 于 2025 年 3 月在 NASDAQ 上市,是这些运营商中按估计 ARR(2025 年初约 $5B)计算规模最大的一家,并拥有 NVIDIA 优先 GPU 分配资格,因此成为 neocloud 定价和可靠性的比较基准。 第二层是超大规模云厂商——AWS、Microsoft Azure、Google Cloud 和 Oracle OCI——也是全球 GPU 算力最大的整体供应方。AWS P5 实例在 96-GPU EFA 网络集群中部署 H100;Azure NDv5 VM 为分布式训练提供 H100;Google Cloud 提供按需 A100/H100 和承诺使用折扣;Oracle SuperCluster 在单个裸金属集群中打包 400 颗 H100 GPU。超大规模云厂商正重金投入 AI 基础设施(仅 Microsoft 就为 2025 年承诺 $80B+),并代表企业采购团队的「安全默认选项」。 第三层是 Lancium 这类贴近能源论点的运营商,在 Texas 运营可再生能源供电的 HPC 基础设施,与 Crusoe 最初的市场定位重叠。Lancium 的清洁园区已获 ERCOT 批准,可接入吉瓦级电网电力,但 Lancium 不提供通用 GPU 云产品。 第四层是现状替代:本地 HPC 集群、传统数据中心托管(Digital Realty、Equinix)以及大型企业内部建设项目。这一层最重要,因为今天企业 GPU 支出的大部分都在这里,也是 Crusoe 必须替代的基准。Crusoe 的 BYO Power 路线和 Spark 模块化工厂,正是用更快、功率密度更高的替代方案,打本地部署 / 托管机房这一层。 反向背景:neocloud 提供商的 GPU H100 现货市场价格从 2023 年约 $8/hour 降至 2024 年末 $2–3/hour,压缩了最初吸引企业买方离开超大规模云厂商的套利空间。Crusoe 曾发生一次有记录的连接故障,影响 us-east1-a 区域计算实例,暴露出 CoreWeave 和超大规模云厂商会在竞争销售中强调的可靠性风险。 [CP001, CP002, CP003, CP004, CP005, CP006]

竞争对手画像表
竞争对手类别规模 / 融资目标客群关键差异点相对 Crusoe 的主要短板
CoreWeave专注型新云厂商已融资 $1.1B;估算 ARR 约 $5B(2025);2025 年 3 月 NASDAQ IPOAI 实验室、企业 AI 平台NVIDIA 优先合作伙伴;最大 H100/H200 机队不拥有园区级电力;依赖第三方 colo
Lambda Labs专注型新云厂商已融资 ~$1B+开发者、研究人员、初创公司按需 H100/A100,无最低用量,快速开通无垂直整合;企业合同容量有限
Nebius AI专注型新云厂商(聚焦 EU)2024 年 9 月融资 $700M(NVIDIA 参投);NYSE: NBISAI 初创公司、EU 数据主权买方数据主权、EU 驻留;声称较 AWS TCO 改善 112%美国布局有限;近期从 Yandex 剥离,仍有声誉包袱
Vast.aiGPU 现货市场未披露突发 GPU 用户、成本敏感型开发者聚合现货市场;可扩至 20K 块 GPU无 SLA 支撑的预留容量;不拥有基础设施
AWS (P5)超大云厂商收入 >$100B;2026 年 AI capex $50B+企业、受监管行业、SageMaker 用户完整托管 ML 栈;EFA 网络;SageMaker HyperPod按需价格更高;大型 GPU 集群开通周期更长
Azure NDv5超大云厂商Microsoft Azure 收入 >$35B/年企业、Microsoft 技术栈客户与 Azure ML 深度集成,绑定 OpenAI 合作H100 可用性受 Crusoe/Abilene 容量依赖制约
Google Cloud (A3)超大云厂商Google Cloud 收入 >$40B/年研究机构、TPU + GPU 混合用户TPU pods、专有 AI 栈、研究合作伙伴关系大规模预留集群的 A100/H100 可用性比 AWS 更紧
Oracle OCI SuperCluster超大云厂商 / Stargate 锚定方Oracle 基础设施收入 >$20B/年Stargate/OpenAI 生态买方、企业400 块 H100 裸金属集群;RDMA 网络;Stargate 合作方同时也是 Crusoe 客户(Abilene);竞争者 / 客户双重关系
Lancium能源相邻型 HPC未披露;2024 年曾尝试 IPO 但未完成Texas 的 HPC 批处理计算、电网友好型工作负载ERCOT 批准的 GW 级园区;太阳能 / 电池储能无通用 GPU 云产品;只聚焦 HPC
内部自建 / 本地部署现状方案 / 替代品N/A — 企业 capex 项目大型 AI 实验室、具备 GPU 能力的企业数据本地化、无 GPU 租赁加价、可完全定制需要 3–5 年前置周期;电力获取是瓶颈;无托管服务

规模 / 融资来自公司官方页面、新闻报道和 SEC 文件(如有)。CoreWeave 估算 ARR 来自 computeforecast.com 分析师估计(2025);CoreWeave 未公开确认确切 ARR。 融资数字代表最后一轮已知融资时披露的股权融资。“现状方案”行捕捉企业 GPU 采购中最常见的竞争替代选项。

[CP001, CP002, CP003, CP004, CP005, CP006]
FP001: 竞争定位图 — 基础设施深度 vs 云产品广度

竞争对手按基础设施所有权深度(x 轴:低 = 仅云,高 = 电力 + 工厂 + 云)与云产品广度(y 轴:低 = 仅 GPU 租赁,高 = 全栈托管 AI 平台)绘制。Crusoe 在基础设施深度上领先;超大规模云厂商在云广度上领先;CoreWeave 与 Nebius 处于中间。

坐标轴采用 1–5 的序数评分,依据可获得的公开证据(官方页面、分析师报告、新闻)。x 轴(基础设施深度):1 = 仅云 GPU 租赁,5 = 全垂直整合(电力 + 工厂 + 云)。y 轴(云广度):1 = 仅现货 GPU 租赁,5 = 具备 ML 服务、存储、网络的全托管 AI 平台。评分是有证据支持的估算;未使用专有基准数据。

[CP001, CP002, CP003, CP004, CP005, CP006]

3.2 直接 neocloud 同业画像

CoreWeave 是最强大的直接竞争者。CoreWeave 成立于 2017 年,总部位于 New Jersey 州 Roseland,拥有 NVIDIA 优先合作伙伴资格,可优先获得 H100、H200 和 Blackwell(B200/GB200)GPU。CoreWeave 2025 年 3 月 NASDAQ IPO 为 neocloud 板块确立了公开市场可比估值,约为 5 倍收入;其客户群偏向前沿 AI 实验室和企业 AI 平台公司,与 Crusoe 最高端买方高度重叠。相较 Crusoe,CoreWeave 的关键结构性限制在于它不拥有发电资产,也不自建园区级数据中心;它依赖第三方机房托管和电力合同,因此随着 AI 基础设施市场收紧,更暴露在电力可用性约束下。 Lambda Labs(San Francisco,2012 年成立)面向开发者和研究 GPU 云板块,提供按需 H100 SXM5 和 A100 实例、快速开通、无最低消费和开发者友好工具。Lambda 不向电力或数据中心建设做垂直整合。它的销售打法偏自助和批量定价,因此对较小工作负载有竞争力,但对超大规模云厂商承诺或专用园区合同的相关性有限。 Nebius AI(Amsterdam;NYSE: NBIS)于 2024 年 6 月从 Yandex 拆分,提供 GPU 云基础设施,面向需要数据主权的 AI 公司,尤其是欧盟市场。Nebius 2024 年 9 月从包括 NVIDIA 在内的投资方募得 $700M。Nebius 声称推理 TCO 比 AWS 优 112%。它对 Crusoe 的主要竞争相关性在欧盟市场;美国重叠有限。作为上市公司,Nebius 提供了一定财务透明度,而 Crusoe 等私营 neocloud 缺少这一点。 Vast.ai 运营 GPU 现货市场模式,让买方按现货价格从独立提供商获取分布式 GPU 容量,更像聚合器,而不是基础设施运营商。Vast.ai 可按需扩展到 20,000 颗 GPU,因此能替代突发型工作负载,但不适合专用预留容量合同。 neocloud 板块正在承受整合压力:wetheflywheel 2026 neocloud guide 指出,只靠 GPU 租赁价格差异化的供应商正被挤压,长期成功取决于能否向上进入软件和编排层。 [CP011, CP012, CP013, CP014, CP015, CP016]

功能与能力矩阵
能力CrusoeCoreWeaveLambda LabsNebius AIAWS/Azure/GCPOracle OCI
按需 H100/H200 GPU是 — H100 SXM $4.29/hr是 — NVIDIA 优先配额是 — H100 SXM5是 — H100/H200是 — P5/NDv5/A3是 — SuperCluster
Blackwell (GB200) 获取能力是 — Crusoe 与 NVIDIA 合作是 — 优先配额部分 / 未知部分 / 未知是 — 2026 年放量Unknown
园区级电力所有权是 — 已投运 3.4 GW否 — 第三方 colo部分 — 租赁否 — Crusoe 建设
托管推理产品是 — Crusoe MemoryAlloy有限是 — SageMaker、Vertex是 — OCI AI
ISO 27001 认证是 — 2026 年取得UnknownUnknown是 — 所有超大云厂商
ISO 42001(AI 管理)认证是 — 2026 年取得UnknownUnknown部分 / 进行中Unknown
模块化预制 AI 工厂是 — Crusoe Spark
多区域云(US+EU)是 — 4 个区域部分 — 以 US 为主部分是 — EU + US是 — 全球是 — 全球
SOC 2 合规未知 — 未披露Unknown是 — 所有超大云厂商

能力覆盖基于截至 2026 年 6 月的官方产品页和新闻室。标记为“未知”的单元格表示未公开披露,不一定代表没有该能力。随着 NVIDIA 推进 GB200 量产,Blackwell 配额状态仍会变化。 在可得证据集中,Crusoe 的 ISO 42001 认证是新云同行中唯一的 AI 管理标准认证。

[CP015, CP016, CP017, CP018, CP026]
FP002: 能力覆盖矩阵 — Crusoe vs 关键竞争对手

绿色 / 是单元格表示有已确认的公开证据;未知单元格表示未见公开披露。Crusoe 的 ISO 42001 与模块化 Spark 工厂目前在 neocloud 同行中独一无二。

覆盖情况基于截至 2026 年 6 月的公开披露。未知单元格表示未披露,并不代表不存在。CoreWeave 和 Nebius 的 ISO 42001 状态在可获得的公开来源中未获确认。

[CP015, CP016, CP017, CP018]

3.3 超大规模云厂商与云替代格局

超大规模云厂商层——AWS、Azure、GCP 和 Oracle OCI——是 GPU 云市场的既有防守方,也是企业买方的默认采购路径。AWS P5 实例提供 8×H100 SXM 配置和 3,200 Gbps EFA 网络,用于分布式训练;P5 也通过 AWS SageMaker HyperPod 提供,用于托管 ML 集群部署。Azure Machine Learning 计算集群支持由 H100 支撑的 NDv5 GPU VM,用于大规模分布式训练。Google Cloud 提供 H100 和 A100 GPU 实例,并配套承诺使用折扣项目。Oracle OCI SuperCluster 在单个裸金属集群中提供 400 颗 H100 GPU 和 RDMA 网络。 超大规模云厂商握有的竞争优势是集成:已经在 AWS、Azure 或 GCP 上运行工作负载的企业,可以受益于存储共址、IAM、网络、托管服务和合规框架;独立 neocloud 不投入大量资源很难复制这套集成护城河。Crusoe 若要赢得企业替代交易,首先必须越过这道集成壁垒。 不过,超大规模云厂商在 GPU 供给上也受容量约束:Crusoe 的 Microsoft Abilene 900 MW 交易和 Oracle Stargate 关系之所以存在,正是因为超大规模云厂商建设基础设施的速度赶不上自身承诺。这一供给缺口是 Crusoe 最主要的收入机会。Crusoe 不只是与超大规模云厂商争夺终端用户,也在向它们提供容量。这种「竞争者 + 容量伙伴」双重角色,在 neocloud 板块并不常见,也形成差异化位置。 Oracle OCI 的 SuperCluster 产品尤其相关,因为 Oracle 是 Crusoe 的 Stargate 锚定客户之一;Crusoe 根据与 Oracle 和 Microsoft 签署的 15 年容量协议,建设并运营 Abilene 第一期园区。这意味着 Crusoe 一边与 Oracle OCI 争夺企业 GPU 买方,一边向 Oracle 供应原始容量。 现状下的内部建设仍是一股重要竞争力量。大型 AI 实验室(Meta、xAI)和资源充足的企业,正在本地或托管设施中建设自有 GPU 集群,绕开超大规模云厂商和 neocloud。Bain Capital Ventures 关于 Crusoe 创业论点的文章明确把「在电源地托管」模式视为对传统数据中心建设的切入口。 [CP020, CP021, CP022, CP023, CP024, CP025]

定价与打包对比
供应商H100 按需(每 GPU-hr)H100 预留 / 合同定价模式主要内置能力来源 / 置信度
Crusoe Cloud$4.29(SXM)未公开披露按需 + 预留订阅 + 超大云容量协议MemoryAlloy 推理、Command Center、Edge Zones官方定价页;高置信度
CoreWeave~$2–4(随配置变化)未公开披露按需 + 预留 + Kubernetes 集群(控制平面 $1,500/month)NVIDIA 优先;裸金属集群CoreWeave 定价页;中等置信度
Lambda Labs~$2.5–3.5(SXM5)未公开披露按需;无最低用量快速开通、开发者工具Lambda 官方;中等置信度
Nebius AI有竞争力 — 声称 TCO 较 AWS 改善 112%未公开披露按需 + 承诺使用数据主权、EU 驻留Nebius 价格页;中等置信度
AWS P5(按需)~$12–20+ 每 instance-hr(8x H100)1 年和 3 年预留节省计划按需、预留、节省计划、现货SageMaker HyperPod、EFA 网络、完整托管栈AWS 定价页;高置信度
Azure NDv5~$18–27 每 instance-hr(H100 VM)预留 VM / 承诺使用折扣按量付费、预留、节省计划Azure ML、OpenAI 集成、完整托管栈Azure 定价页;高置信度
Google Cloud A3 (H100)~$12–16 每 GPU-hr(A3 Ultra)承诺使用折扣(1 年 / 3 年)按需、承诺使用TPU pods 补充、Vertex AI、研究合作GCP 定价页;中等置信度
Oracle OCI SuperCluster按集群定价;非每 GPU-hrCrusoe/Stargate 已确认 15 年合同裸金属集群;企业合同RDMA、400-GPU 裸金属集群、Stargate 生态Oracle 云页面;中等置信度

所有价格均为 2026 年中挂牌价;预留或谈判合同下的实际价格会不同。Crusoe 和 CoreWeave 未公开披露预留 / 承诺容量价格。AWS、Azure、GCP 价格来自官方定价页;确切每 GPU-hr 取决于实例配置和区域。 超大云厂商价格按 GPU 实例的每 instance-hour 展示,而非每 GPU-hour(乘以 GPU 数即可换算每 GPU-hr)。价格反映按需 / 挂牌费率,仅具方向性。

[CP001, CP007, CP020, CP021, CP022, CP023]

3.4 差异化、护城河耐久性与替代风险

Crusoe 的竞争护城河建立在三条相互强化的主张上:(1)跨电力、AI 工厂和 GPU 云的垂直整合;(2)NVIDIA 战略协作为其提供早期模型和工具链接入;(3)ISO 27001/42001 认证把它与多数 neocloud 运营商区分开。每条护城河主张都有对应威胁。 垂直整合护城河资本密集、复制慢,但 Crusoe 自身扩张也慢。Abilene 的 $3.4B+ Blue Owl / JPMorgan 资本栈是一把双刃剑:它建立了规模壁垒,让纯 neocloud 难以匹配 Crusoe 的园区野心;但如果 Abilene 电力建设或客户承诺滑坡,也会把 Crusoe 暴露在集中风险下。Crusoe 的 Spark 模块化 AI 工厂产品代表一种当前没有 neocloud 竞争者能复制的制造能力——可在能源源头部署的预制单元;Redwood Materials 合作从 4 个单元扩至 24 个单元,是商业牵引力证据。 NVIDIA 协作护城河(早期模型访问、Nemotron 集成、Dynamo tokenizer)让 Crusoe Cloud 的托管推理产品区别于商品化 H100 租赁。不过,CoreWeave 同样拥有强势 NVIDIA 优先合作伙伴资格,也深度融入 NVIDIA 生态,因此这一护城河面对最直接竞争者时耐久性有限。 ISO 27001 和 ISO 42001 认证(截至 2026 年已取得)回应了企业安全和 AI 治理尽调要求,而很多 neocloud 还没有完成。这是面向受监管行业和大型企业采购团队的实质信任差异化。 反向信号:Heatmap News 报道称,Crusoe 的实际能源栈包括天然气涡轮机(Abilene 360 MW)以及通过 Engine No. 1 合资公司开发 4.5 GW 新天然气容量,使 Crusoe 的「清洁能源」品牌与运营现实出现张力。如果重视 ESG 的企业买方调查这一差距,Crusoe 来自可持续定位的定价溢价可能被侵蚀。isdown.app 记录的 us-east1-a 连接故障说明可靠性风险真实存在,AWS 和 Azure 这类云运营商会在竞争销售周期中引用。 neocloud 商品化威胁真实存在:H100 现货价格从约 $8/hr 降至 2024 年末约 $2-3/hr,随着 Blackwell 供应爬坡和新进入者加入市场,分析师预计压力会持续。Crusoe 的回应——长期企业合同和超大规模云厂商容量伙伴关系——战略方向正确,但执行风险高。 [CP027, CP028, CP029, CP030, CP031, CP032]

护城河耐久性与竞争风险登记表
护城河主张威胁严重性证据缓释措施 / 尽调问题
垂直整合(电力 + 工厂 + 云)CoreWeave 与公用事业公司签园区级电力协议;超大云厂商投资自有电力(SMR、太阳能)Microsoft $80B+ AI capex 承诺;NVIDIA 优先合作伙伴关系确认超大云厂商电力内包的速度,以及相对外包给 Crusoe 的节奏
NVIDIA 战略合作(Nemotron、Dynamo)CoreWeave 拥有同等 NVIDIA 优先地位;Blackwell 配额有限CoreWeave NASDAQ IPO 材料提到 NVIDIA;Crusoe 新闻室披露 NVIDIA 合作确认 NVIDIA 合作是否独家,还是与 CoreWeave 平级
ISO 27001 / ISO 42001 认证超大云厂商持有 ISO 27001 和更广合规框架;CoreWeave 也有 ISO 27001Crusoe ISO 认证新闻室;TP002 能力矩阵监测竞争对手是否取得 ISO 42001 — 目前这是 Crusoe 唯一独有认证
GPU 现货价格压缩护城河(企业锁定)H100 现货价格从 ~$8/hr 降至 ~$2–3/hr;Blackwell 会延续价格压缩computeforecast neocloud 分析;Sacra 的 Crusoe 画像核实 Crusoe 收入中长期合同占比,以及现货敞口
清洁能源 ESG 品牌 / 溢价定价Heatmap 报道 Abilene 有 360 MW 燃气轮机 + 4.5 GW Engine No. 1 燃气 JVHeatmap 负面报道;TCEQ 许可文件在 ESG 驱动交易前,审计实际能源结构与已披露清洁能源说法是否一致
Crusoe Spark 模块化工厂(部署速度)Amazon 和 Google 正在部署自有模块化数据中心;未点名新云厂商竞争对手Forbes 模块化数据中心文章;Crusoe 新闻室关于 Spark 工厂的报道跟踪企业采用 Spark 与传统园区建设周期的对比
超大规模云厂商容量协议构成收入底线超大规模云厂商最终会自建同等容量(Jevons 悖论两边都成立)Forbes Stargate 文章:Oracle/Microsoft 15 年合同构成底线监控 Abilene 项目融资契约条款;确认合同是否为照付不议结构

严重性评级(高 / 中 / 低)是分析师基于现有公开证据作出的判断;未使用专有尽调数据。GPU 价格压缩和 ESG 缺口的反向证据来自独立报道的事实,不是估算。「尽调事项」指需要公司披露私有数据才能解决的问题。

[CP027, CP028, CP029, CP030, CP031, CP032]
FP003: 护城河与竞争就绪 KPI

截至 2026 年中,Crusoe 关键竞争耐久性指标概览。GPU 价格压缩和 ESG 缺口等高风险因素,抵消了垂直整合与 NVIDIA 合作的优势。

电力容量来自 Crusoe 新闻室(2026 年 3 月里程碑更新)。ISO 认证来自官方新闻室。CSAT 来自 Crusoe Cloud 官方页面(公司自报;未经审计)。现货价格压缩风险基于 computeforecast neocloud 分析评为高。ESG 缺口基于 Heatmap 对燃气轮机使用的报道评为高。

[CP016, CP017, CP026, CP028, CP032]

3.5 图表与证据

Chapter 04

04财务

4.1 收入模式与来源

Crusoe 的收入模式经历了一次有意推动的结构性转型。2024 年,公司总收入为 $276M,其中约 45% 来自 AI 云($124M,同比增长 460%),55% 来自 Bitcoin 和数字火炬挖矿($152M)。2025 年 3 月,公司把挖矿业务——425+ 个模块化数据中心、250+ MW,以及约 135 名员工——出售给 NYDIG 后,转向纯 AI 云和数据中心基础设施收入模式。Crusoe 披露 2025 年云 ARR 增长 150%、总合同价值(TCV)增长 17 倍,但没有公布绝对分母或基础 ARR 数字。分析师对 2025 全年收入的估计介于 $500M 和 $1B 之间,但这些只是未确认估计,并非公司披露数据。 今天,Crusoe 的收入由三套机制驱动。第一,Crusoe Cloud 通过按需小时计费(H100 SXM 每 GPU 小时 $4.29)和预留容量订阅提供 GPU 算力。第二,长期超大规模云厂商容量协议——最典型的是 Microsoft Abilene 900 MW 承诺(2026 年 3 月)以及 Oracle/Stargate 锚定关系——提供大额合同收入分段,具备多年可见度和照付不议经济性。第三,Crusoe Spark 模块化 AI 工厂产品可能形成产品销售和托管服务收入线;Redwood Materials 合作从 4 个 Spark 单元扩至 24 个,显示初步商业牵引力,但收入规模未披露。 NYDIG 交易后,收入质量已经实质改善:AI 云收入具备经常性和订阅属性,旧 Bitcoin 挖矿收入实质上是商品挖矿,受减半周期和哈希率 竞争约束。不过,Microsoft 和 Oracle 两家超大规模云厂商集中度高,带来客户依赖风险,即便标价存在溢价,也会限制真实定价权。Crusoe Cloud 现货 / 预留 GPU 租赁与超大规模云厂商容量租约之间的收入结构没有公开披露,因此无法精确评估收入质量。 Redwood Materials 合作(从 4 个 Spark 单元扩至 24 个)以及类似 Spark 部署合作结构显示,模块化 AI 工厂收入线可能变得重要;但经济模型究竟是产品销售、租赁还是托管服务,目前仍是公司口径,尚未独立验证。 [CI001, CI002, CI003, CI004, CI005, CI006]

收入来源表
收入来源机制单位当前价值 / 状态收入质量关键尽调事项
Crusoe Cloud GPU 计算(按需)GPU 按小时计费$/GPU-hrH100 SXM:$4.29/hr 标价;用量未披露中 — 暴露于现货需求;存在价格压缩风险实际利用率、现货 / 预留组合;合约收入与现货收入拆分
Crusoe Cloud GPU 计算(预留)多月订阅容量$/month未公开披露高 — 经常性,流失率低于现货预留容量定价、期限和续约率;NRR
Crusoe Cloud 托管推理(MemoryAlloy)按 token 付费的推理服务$/1M tokens未公开披露若 token 量放大,收入质量高token 定价、月活模型数、推理毛利率与 GPU 租赁毛利率对比
超大规模云厂商容量协议(Microsoft、Oracle)长期园区容量租赁 / 照付不议$/MW 或 $/kWMicrosoft 900 MW(Abilene);Oracle Stargate 15 年合同;规模未披露高 — 长约,大型锚定租户合同照付不议条款、爬坡节奏、契约结构、终止条款
Crusoe Spark 模块化 AI 工厂产品销售、租赁或托管服务$/unit 或 $/kWRedwood Materials 扩展至 24 台(从 4 台起步);商业部署 2026 年 Q3 启动中 — 早期阶段;未公开披露收入收入模式(销售、租赁还是服务)、单台 Spark 毛利率、Redwood 之外管线

收入数据来自 Crusoe 官方新闻室、新闻稿和 SEC Form D 文件。2024 年总收入($276M)来自媒体报道中的公司披露。收入拆分(AI 云 $124M / 挖矿 $152M)为公司口径,未经独立审计。2025 年收入数字为分析师估算;公开渠道没有公司确认的 2025 年绝对收入。预留 GPU 定价和超大规模云厂商合同金额未公开披露。

[CI001, CI002, CI003, CI004, CI005, CI006]
FI001: Crusoe 收入模型桥 — 客户活动到收入

客户 GPU 使用如何经由 Crusoe 的三种收入机制转化为收入和毛利润,重点展示三条不同变现路径:现货云、预留 / 超大规模云合同,以及 Spark 工厂。

收入机制流基于公司官方描述、新闻室公告和分析师报道。利润率标签为定性,来自行业基准;未审阅 Crusoe 财务报表。边标签在可得情况下反映公司披露的合同结构。

[CI001, CI002, CI003, CI004, CI005]

4.2 单位经济与定价

Crusoe 公开披露的只是标价;实际每 GPU 小时收入、预留容量折扣结构和 GPU 利用率都未披露。H100 SXM 标价为 $4.29/GPU-hour,使 Crusoe 处在新云现货供给的高端。CPU 算力标价为 $0.04/vCPU-hour。GB200 NVL72 价格未披露。Crusoe 的托管推理产品 MemoryAlloy 采用按用量付费模式, 按每 100 万 tokens 计费,缓存 tokens 费率更低;绝对 token 价格未公布。 单位经济性的市场背景并不友好:H100 现货价格从 2023 年约 $8/hour,跌到 2024 年末的 $2–3/hour,背后是供给正常化和集群竞争。这种压缩迫使 Crusoe 不能只靠现货价格竞争,而要拼基础设施效率、产品差异化和企业长期合约粘性。 Crusoe 的 PUE 为 1.2–1.3(行业平均约 1.8),意味着相对典型数据中心可降低约 30–40% 的电力开销。按估算的混合能源成本 $0.03–0.06/kWh、H100 TDP 约 700W 计算,每台 H100 每小时电力成本约 $0.02–0.04,只占 $4.29 标价的一小部分。GPU 资本折旧才是主成本:H100 单价约 $25,000–30,000,若按 3–5 年折旧、80% 利用率计算,每 GPU 小时资本成本约 $0.60–1.20,按标价仍给毛利留下有意义空间。 同行对比看,新云公司披露的毛利率多在 20–40% 区间;Crusoe 在能源和冷却上的效率优势,意味着其云业务可能位于该区间高端,但公司没有公开披露任何毛利率数据。 实际利润率还会受 GPU 采购成本溢价影响:Crusoe 不是超大规模云厂商,采购 NVIDIA 时是否有折扣结构并未披露;Abilene 园区建设期的爬坡成本也会拖累利润率。 Crusoe 披露 2024 年收入为 $276M,2025 年估算区间为 $600M–$1B。按 Crusoe $10B+ 的 Series E 估值计算,隐含收入倍数为 10x–17x(基于 2025 年估算收入 $600M–$1B),与高端新云倍数一致,但高度依赖尚未验证的 2025 年收入基数。 [CI008, CI009, CI010, CI011, CI012, CI013]

定价与变现表
产品 / 服务标价(单位)定价模式实际成交价 / 标价来源
H100 SXM 按需 GPU$4.29/GPU-hr按需按小时未知 — 同行现货市场压缩到 $2–3/hrCrusoe Cloud 定价页(官方)
A100 SXM 按需 GPU$1.95/GPU-hr按需按小时UnknownCrusoe Cloud 定价页(官方)
CPU 计算$0.04/vCPU-hr按需按小时UnknownCrusoe Cloud 定价页(官方)
GB200 NVL72 集群未披露预留订阅不适用 — 尚未列入价格表Crusoe Cloud 定价页(官方)
Crusoe MemoryAlloy 推理每 1M tokens(分层)即用即付,按用量计费未披露Crusoe Cloud FAQ(官方)
Spark 模块化 AI 工厂单元未披露产品销售 / 租赁 / BOT未披露Crusoe 新闻室(官方)

所有价格均为 2026 年中期 Crusoe Cloud 标价;预留或企业合同的实际成交价未披露。H100 现货市场压缩基准($2–3/hr)来自 computeforecast 对新云厂商的分析;这是市场背景,不是 Crusoe 的实际成交价。FAQ 确认了 GPU token 定价(MemoryAlloy)结构;绝对的每百万 token 价格未发布。

[CI008, CI009, CI010, CI011]
单位经济摘要表
指标数值 / 状态置信度重要性尽调事项
H100 标价(每 GPU-hr)$4.29高(官方)设定按需 GPU 容量的收入上限确认折扣后的实际成交价;获取预留定价
估算 GPU 电力成本(每 GPU-hr)$0.02–$0.04(估算)低(估算)电力是资本开支之后第二大成本;PUE 优势会压低这项成本获取 Abilene 和其他园区每 kWh 的实际综合能源成本
估算 GPU 资本开支成本(每 GPU-hr)$0.60–$1.20(估算)低(估算)GPU 折旧是最大单位成本,决定毛利率底线确认 GPU 单台采购价和折旧周期
毛利率(云业务)未披露N/A — 私有公司数据承销最关键的财务输入;新云厂商同行:20–40% 区间要求披露私有财务数据;与 CoreWeave IPO 毛利率披露对比
净收入留存(NRR)未披露N/A — 私有公司数据反映客户扩张与流失;是 SaaS / 云业务质量的关键指标向 Crusoe Cloud 销售团队索取 NRR 或留存同期群数据
GPU 利用率未披露N/A — 私有公司数据利用率直接决定每块 GPU 收入;80%+ 才可能跑出正向经济性按地区和 GPU 类型索取机队利用率
电力使用效率(PUE)1.2–1.3(公司口径)中(公司口径)更低 PUE = 更低能源开销 = 更好毛利率;相对 1.8 行业均值有优势用第三方测量或 DCD 基准数据验证 PUE
2024 年总收入$276M(公司披露)高(媒体 / SEC 支撑)增长轨迹基线;挖矿($152M)+ AI 云($124M)组合获取 2025 年绝对收入;用基数验证 150% ARR 增长说法
2025 年 AI 云 ARR 增长150% YoY(公司披露,无基数)中(公司口径)增速高,但基数 ARR 未披露;可能是小基数上增长披露 2024 年和 2025 年绝对云 ARR 数字

估算值(电力成本、每 GPU-hr 资本开支成本)是分析师基于公开 TDP 规格、市场 GPU 价格和能源基准作出的估算;公司未确认。「未披露」字段反映这家私有公司真实存在的数据缺口。毛利率同行区间(20–40%)来自 computeforecast 对新云厂商的分析;Crusoe 的垂直整合经济性可能让结果不同。

[CI008, CI009, CI010, CI011, CI012, CI013]
FI002: 单位经济桥 — GPU 成本栈到毛利率

按 Crusoe Cloud 标价估算单个 H100 GPU-hour 的单位经济,从成本输入到估算毛利率。所有成本输入均为估算;毛利率未经公司确认。

所有成本数字均为估算,依据公开 GPU TDP 规格(H100:约 700W)、市场能源成本($0.03–0.06/kWh)、GPU 采购价(非超大规模云买方价格 $25K–30K)、3–5 年折旧、假设 80% 利用率,以及 Crusoe PUE 1.2–1.3 vs 行业 1.8。毛利率估算仅作方向性参考;实际 Crusoe 财务会因批量折扣、实际利用率以及合同 vs 现货组合而有重大差异。

[CI012, CI013, CI014, CI015]

4.3 资本结构与充足性

在 AI 基础设施领域的私营公司中,Crusoe 搭出了异常庞大且分层的资本结构。股权融资横跨六轮,合计约 $3.4B;最近两轮是 Series D($600M,估值 $2.8B,SEC Form D 于 2024 年 11 月 21 日提交,精确金额 $817,744,542,70 名投资者)和 Series E($1.375B,估值 $10B+,SEC Form D 于 2025 年 10 月 23 日提交,精确金额 $1,374,999,988,71 名投资者)。 Series D 投资者包括 Founders Fund、NVIDIA、Fidelity、Mubadala、Ribbit Capital 和 Valor。Series E 由 Valor Equity Partners 和 Mubadala Capital 共同领投。 股权之外,Crusoe 为 Abilene 数据中心建设拿到 Blue Owl Capital 的 $3.4B 合资安排,JPMorgan 正在为同一园区安排约 $9.6B 项目融资,仅 Abilene 园区承诺资本就超过 $15B。这套资本结构反映了 AI 超大规模数据中心开发的底层经济学:每 1 GW AI 优化容量,在安装第一台 GPU 机柜之前,就需要约 $10–15M/MW 投入土建、电力基础设施和冷却。 近期运营的资本充足性看起来较强:$1.375B Series E 股权加项目融资续航,应该覆盖 Abilene Phase 1。Michael Gordon 于 2025 年 12 月出任 COO/CFO(前 MongoDB CFO,曾在 MongoDB 收入约增长 50x 时主导 2017 年 IPO),说明公司正在积极准备潜在资本市场动作。关键风险包括项目融资契约敞口、资本开支集中在单一园区(Abilene),以及 JPMorgan 信贷额度提款时点带来的执行依赖。现金余额和月度烧钱速度均未公开披露。 公司的资本充足性高度依赖 Microsoft 900 MW 与 Oracle Stargate 15 年合约收入流,用来服务项目融资契约。 如果任一超大规模锚定客户修改承诺,项目融资结构就会在大规模层面暴露再融资风险。 [CI016, CI017, CI018, CI019, CI020, CI021]

资本充足性表
资本组成金额 / 状态用途风险 / 契约备注
Series E 股权融资(2025 年 10 月)$1,374,999,988(SEC Form D)一般运营、Abilene 一期、技术开发71 名投资者;Valor/Mubadala 共同领投;未披露优先股条款或清算优先权
Series D 股权融资(2024 年 11 月)$817,744,542(SEC Form D,包含此前分批融资)AI 云扩张、基础设施规模化70 名投资者;Founders Fund 领投;NVIDIA、Fidelity、Mubadala、Ribbit 参投
Blue Owl Capital 合资公司(Abilene)$3.4B 已承诺Abilene 数据中心建设(一期 + 二期)合资结构;Crusoe 保留运营控制;Blue Owl 持有股权;条款未披露
JPMorgan 项目融资(Abilene)~$9.6B(安排中)Abilene 园区长期项目融资长期项目债;以园区资产和 15 年超大规模云厂商合同作抵押;契约条款未披露
月度现金消耗率未披露N/A关键缺口:现金头寸和消耗率未知;没有它无法估算资金续航期
手头现金(Series E 后)未披露N/A考虑到近期 $1.375B 融资,估计充足;实际数字未公开
债务 / 信贷额度(其他)除 Abilene 项目融资外未披露N/A其他信贷额度、循环信贷或设备融资安排未披露
下一轮触发因素 / IPO 路径Michael Gordon(前 MongoDB CFO)2025 年 12 月入职;已有 IPO 信号Wyoming 园区(1.8 GW)和国际扩张资本Wyoming + 国际建设可能需要 IPO 或二级发行;时间未披露

Series D 和 Series E 金额来自 SEC Form D 文件(CIK 0001924674)。Blue Owl 合资公司和 JPMorgan 项目融资金额来自公司新闻稿和新闻报道。现金消耗率、现金头寸和契约条款未公开披露。所有股权轮金额均为 SEC 验证的精确数字;合资公司和项目融资金额为公司公告,未经独立验证。

[CI016, CI017, CI018, CI019, CI020, CI021]
FI004: 资本强度与现金流地图

瀑布图展示 Crusoe 已承诺资本栈:股权流入、项目融资流入,以及 Abilene 园区 capex 承诺支出,凸显 AI 工厂建设模式的资本强度。

所有数值均为百万美元。股权融资来自 SEC Form D 文件(精确)。Blue Owl JV 和 JPMorgan 项目融资来自公司新闻稿;据新闻报道,JPMorgan 总额为近似值。Abilene capex 估算为 $12B,基于每 MW $10–15M 区间应用于 1.2 GW 一期容量。运营储备为估算。净资本头寸不是实际现金;它反映已承诺资金与已承诺建设支出,不计入时间差、提款或拨付。

[CI016, CI017, CI018, CI019, CI020, CI021]

4.4 财务缺口、风险与结论

Crusoe 最重大的财务承销缺口,是没有毛利率、EBITDA、现金余额、烧钱速度、净收入留存率和经审计财报。 作为一家除 Form D 申报外不受 SEC 报告义务约束的私营公司,Crusoe 无需披露这些指标。MongoDB 老将 Michael Gordon 出任 COO/CFO,说明财务治理和 IPO 准备在改善,但财务表现仍无法被独立验证。 四个重大不利因素需要重点盯住。第一,GPU 现货价格压缩——H100 从约 $8/hour 降至 $2–3/hour——构成结构性价格天花板,限制按需云收入的上行空间。Crusoe 能否把客户锁进高于现货价格的标价预留容量协议,是关键价值驱动,但尚未验证。 第二,清洁能源财务风险:Crusoe 的 ESG 品牌或许能从重视可持续性的企业获得价格溢价,但 Heatmap News(2025 年 3 月)报道称,公司仍在使用天然气,包括与 Engine No. 1 合资开发 4.5 GW 新燃气容量。如果 ESG 敏感客户追查真实能源结构,客户流失风险和声誉损害可能把实际价格压到标价以下。第三,Microsoft 与 Oracle 客户集中带来收入依赖风险:合同修改或超大规模云厂商转向自建基础设施,都可能对收入产生重大影响。第四,高资本开支模式叠加 $9.6B 项目融资,在 JPMorgan 规模上制造契约和再融资风险。公开材料中没有可供尽调审阅的契约条款、LTV 比率或偿债覆盖要求。 公司 2025 年 TCV 增长 17x、云 ARR 增长 150%,是积极信号;但私营财务的不透明,尤其是公司从挖矿转向云业务时的毛利率轨迹不透明,意味着没有利润率可见性的收入增长很难承销。 Crusoe Spark 制造收入流是有意思的财务多元化来源,但目前只有 Redwood Materials 伙伴关系扩展作为证据。 财务结论是:收入增长轨迹有吸引力,但客户集中和利润率不透明风险显著;若没有包含毛利率、NRR、烧钱速度和项目融资契约条款在内的私营财务披露,$10B 估值无法承销。 [CI025, CI026, CI027, CI028, CI029, CI030]

公开财务缺口表
缺失指标对承销的影响尽调路径
毛利率(云业务)没有毛利率,无法验证收入增长到底有利润支撑,还是靠项目融资补贴索取私有财务报表;与 CoreWeave S-1/IPO 披露对比,建立新云厂商基准
净收入留存(NRR)NRR 决定 Crusoe 的云 ARR 增长来自新客户还是存量扩张;两者对估值的意义不同向 Crusoe Cloud 索取按同期群和客户分群拆分的留存数据与 NRR
现金头寸和现金消耗率无法估算资金续航期,也无法判断近期运营是自我造血还是依赖股权融资索取最近一个季度现金流量表;结合招聘数据和资本开支公告交叉验证
2025 年绝对收入(AI 云 ARR 基数)没有绝对基数,150% YoY 增长无法解读;$1 增到 $2.50 与 $200M 增到 $500M 完全不同索取 2024 年绝对云 ARR 和 2025 年估算云 ARR;与 2024 年 $276M 总收入对账
项目融资契约条款(JPMorgan)没有契约条款(LTV、DSCR、照付不议比例、补救期),无法评估超大规模云厂商合同变更下的下行情景索取条款清单摘要;查看公开 TCEQ/ERC 文件是否有项目融资披露
按地区拆分的 GPU 机队利用率利用率直接决定 Crusoe 的 GPU 机队是在产生经济性,还是背着闲置容量按 GPU 类型和地区索取机队利用率数据;观察开发者社区中的作业队列深度
Spark 收入模式细节不知道 Spark 是产品销售、租赁还是托管服务,就无法建模收入线或毛利贡献从投资人材料或管理层讨论中索取 Spark 单元经济性和商业结构

本表所有缺口都代表真实的私有公司信息不对称,且经所有已审阅来源缺乏公开披露所确认。这些指标都无法从 SEC Form D 文件获得,后者只披露发行金额和投资者数量。尽调路径项是尽调流程里的可执行下一步。

[CI025, CI026, CI027, CI028, CI029, CI030]
FI003: 财务估算区间 — 收入、估值与资本

基于公开证据、分析师估算和可比公司数据推导的 Crusoe 关键财务输入低 / 基准 / 高区间。缺少经审计财务时,区间反映真实不确定性。

2025 年 AI 云收入区间基于分析师估算,并将 Crusoe 披露的 150% 增长应用到 2024 年 $124M 基数($124M × 2.5 = $310M 低值,TCV 爬坡带来上行)。高值 $1B 反映 $124M × 8x,或若 2024 年 ARR 已在 Q4 爬坡则来自更高基数。收入倍数按 $10B / 2025 年收入区间计算。云毛利率区间来自 neocloud 同行基准(computeforecast:20–40%);Crusoe 效率可能把上限推至 45%。股权融资总额由 SEC Form D 文件确认。项目融资区间基于新闻稿,并考虑后续分批融资的上行。

[CI001, CI002, CI016, CI017, CI028]

4.5 附录

Chapter 05

05产品与技术

5.1 Crusoe Cloud——GPU IaaS 与托管服务

Crusoe Cloud 是一个全栈、NVIDIA 原生的 AI 云平台,用能源优先的成本结构,把自己定位为超大规模 GPU 云的替代方案。 产品在美国和欧洲区域提供多种 GPU 加速虚拟机 SKU:NVIDIA H100 SXM(80 GB)按需价 $3.90/hr,H200 SXM(141 GB)$4.29/hr,AMD MI300X(192 GB)$3.45/hr, 以及 B200(180 GB)、L40S(48 GB)和 A100(80 GB)。VM 配置从单 GPU 扩展到 8-GPU 节点,配备最高 3200 Gbps 的 RDMA InfiniBand 网络和感知拓扑的集群放置。IaaS 层之上,Crusoe 提供 Managed Kubernetes(CMK)、Managed Slurm 和 AutoClusters;AutoClusters 是自动化容错训练编排服务,会主动检查节点健康状态,并在无需操作员介入的情况下替换故障节点。 平台提供 VPC 网络、逻辑隔离环境、块存储、对象存储和共享存储,并配有 Crusoe Command Center 可观测性控制台。 开发者界面包括 api.cloud.crusoe.ai/v1 上的 REST API、CLI、Terraform provider,以及 Go、Python 和 TypeScript SDK。Crusoe Cloud 已连续 12 个多月保持 100% CSAT,平台承诺 99.98% 集群可用时间,并由 24/7 企业支持背书,平均首次响应时间低于 6 分钟。三到六个月期预留可在按需价格基础上折扣 30–50%,让企业在持续训练任务中获得预算可预期性。[CE001, CE002, CE003, CE005, CE006, CE007]

Crusoe 产品模块与资产矩阵
产品 / 资产主要买方 / 用户成熟度 / 状态关键差异化尽调缺口
Crusoe Cloud GPU IaaSAI 初创公司、企业 ML 团队GA / 生产环境能源驱动定价,比超大规模云厂商低 30–50%;99.98% SLANVML 驱动问题;SoK 未达企业级(SemiAnalysis 发现)
Crusoe 托管推理(MemoryAlloy)AI 应用开发者、生产环境 LLM 部署GA / 生产环境(2026)TTFT 9.9×、吞吐量 5×,相对 vLLM;全集群 KV 缓存基准为自报;架构未披露;无独立审计
Crusoe Intelligence Foundry尝试开源模型的 AI 开发者GA / 测试版(2026)统一枢纽;Nemotron 早期访问;API key 几分钟完成上手模型库广度相对 Fireworks、Together AI 尚不清楚
Crusoe Spark(模块化 1 MW 单元)企业、油田运营商、边缘 AI 买方早期 GA(2026 年 Q3 首批工厂单元)3 个月交付;能源无关;可搬迁;Redwood 电池微电网已验证工厂吞吐爬坡未确认;单元成本未披露
超大规模 AI 托管园区超大规模云厂商(Oracle、Microsoft、Meta)生产环境(Abilene 一期上线)GW 级 + 清洁能源;压缩交付周期;垂直整合Oracle/OpenAI 部分退出 Abilene 二期;Wyoming 项目暂停
Digital Flare Mitigation(DFM)油气运营商(Permian、Uinta Basin)生产环境 / 已部署99% 燃烧效率;运营商零资本开支;已部署 120+ 台AI 转型后 DFM 分部收入未披露;战略优先级不清楚

成熟度标签基于公开公告和第三方评测;公司未公开披露按产品线拆分的财务贡献。Crusoe Spark 工厂生产计划来自官方新闻稿(2025 年 6 月);2026 年 Q3 是公司目标,不是已确认里程碑。

[CE001, CE002, CE009, CE013, CE014, CE021]
客户工作流与用例图
用户任务 / 工作流当前痛点Crusoe 方案可衡量收益已知限制
LLM 预训练(多节点分布式)超大规模云厂商 GPU 成本过高;预留容量稀缺带 AutoClusters 容错的 Crusoe Cloud H100/H200 集群相对 AWS/Azure 节省 50% 成本(Windsurf 案例研究);VM 启动 <90 s多节点训练相对单节点有 8–15% 开销(DeployBase 评测)
生产环境 LLM 推理(低延迟服务)高并发下 TTFT 慢;预填充冗余Crusoe 托管推理,MemoryAlloy 全集群 KV 缓存TTFT 快 9.9×、吞吐量 5×,相对 vLLM(Llama 3.3 70B 基准)基准为自报;缺少外部审计
边缘 AI / 本地推理数据本地化约束;电网接入延迟Crusoe Spark 模块化单元;≤3 个月交付;能源无关3 个月交付;相对 18–36 个月超大规模建设,可快速部署生产工厂吞吐量尚未大规模证明(2026 年 Q3 目标)
超大规模 AI 工厂(模型训练 / 超大规模云厂商租户)电力约束;许可延误;ERCOT 可靠性暴露Crusoe 超大规模园区(Abilene 1.2 GW);共同制定电力策略Abilene 一期从破土到全站运营约 ~12 个月Wyoming 项目暂停;Oracle/OpenAI 部分退出园区

收益来自 Windsurf 案例研究(官方)、SemiAnalysis ClusterMAX 评测和官方产品文档。多节点开销估算来自 DeployBase 第三方评测。时间线基于 Abilene 一期建设启动(2024 年 6 月)和上线日期(2025/2026)。

[CE002, CE010, CE013, CE015, CE029, CE035]
FE002: 客户工作流与云运营流程

从识别计算需求,到预置、编排、运行工作负载、监控和扩张的端到端路径,展示 AI 团队如何使用 Crusoe Cloud 的产品界面。

流程代表典型云客户旅程;超大规模托管客户走单独的园区定制建设流程。

[CE003, CE007, CE035, CE037]

5.2 Crusoe Managed Inference 与 MemoryAlloy 引擎

Crusoe Managed Inference 于 2026 年初推出,是公司差异化程度最高的产品,由基于 MemoryAlloy 技术构建的专有推理引擎驱动。MemoryAlloy 是一种集群原生内存织物,实现了集群级键值(KV)缓存,让任意 GPU 节点都能即时从本地和远程节点获取 prefix cache,消除并发会话中的重复 prefill 计算。Crusoe 对 MemoryAlloy 的基准测试显示,相比 vLLM 运行 Llama 3.3 70B,其 time-to-first-token 快 9.9×、token 吞吐高 5×;该引擎支持持久会话、上下文连续性,并可在整个集群内无缝横向扩展。Crusoe Intelligence Foundry 是面向用户的入口,开发者可通过它访问 Managed Inference 端点、在几分钟内生成 API 密钥,并试用一套精选模型库:Llama 3.3 70B Instruct、DeepSeek V3 0324、DeepSeek-R1 0528、Qwen3 235B、Kimi-K2、Gemma 3 12B、GPT-OSS-120B、NVIDIA Nemotron 3 Super 和 NVIDIA Nemotron 3 VoiceChat。两款 Nemotron 模型都通过 Crusoe 与 NVIDIA 的早期采用者关系登陆 Foundry;该关系于 2026 年 3 月 GTC 公布。Crusoe 还把其高性能 Rust BPE tokenizer 贡献给 NVIDIA Dynamo 开源推理服务框架,相比 HuggingFace tokenizers 约提速 9×,长 prompt 上最高提速 31×;即便缺少传统 OSS 项目指标,这一动作也加深了 Crusoe 在工程实践者社区的可信度。MemoryAlloy 的内部架构和训练方法仍为专有,且未经过独立同行评审;这一缺口限制了外部对其性能主张的验证。[CE008, CE009, CE010, CE011, CE012]

技术与运营架构
层 / 组件作用关键依赖风险
物理 GPU 节点(H100/H200/B200/MI300X)训练和推理的加速计算底座NVIDIA 供应链(主导);AMD 作为次级对冲单一供应商 NVIDIA 集中度;GPU 价格波动
VM 虚拟机管理器(cloud-hypervisor)硬件虚拟化,支撑多租户 GPU 隔离开源 cloud-hypervisor 项目容器内 NVML 驱动不匹配(SemiAnalysis 发现)
RDMA InfiniBand 网络(3200 Gbps)分布式训练的高带宽低延迟节点间通信网络交换机供应商;光纤基础设施Iceland 站点 20,000 个光纤端被污染;需要物理修复
AutoClusters 编排容错多节点训练;自动节点替换Crusoe 内部工程;Slurm / Kubernetes 集成Slurm-on-Kubernetes 未达企业级(缺少默认配置、RBAC、SSO)
MemoryAlloy KV 缓存引擎全集群前缀缓存,用于低延迟推理Crusoe 专有 IP;NVIDIA Dynamo 集成分词器架构未披露;没有独立基准复现
能源与电力基础设施可靠、低成本电力支撑计算GE Vernova 涡轮机、Redwood EV 电池、Form Energy 铁空气、电网接入GW 级站点同步投运的执行风险;涡轮机交付周期

架构细节综合自 Crusoe 官方文档、ClusterMAX 技术评测(SemiAnalysis)和官方新闻稿。专有组件(MemoryAlloy、AutoClusters 内部机制)缺少独立架构文档。

[CE004, CE005, CE006, CE009, CE022, CE029]
FE001: Crusoe 平台架构栈

六层架构从实体 AI 工厂延伸到云 IaaS、托管推理和开发者入口,展示 Crusoe 从能源获取到模型服务的垂直整合。

层级顺序反映运营依赖关系(自下而上)。同一层内产品成熟度不同(Spark 工厂单元目标 2026 年 Q3;Vera/Rubin 部署在 2027 年)。能源结构因站点而异。

[CE001, CE005, CE013, CE018, CE022, CE024]

5.3 物理基础设施——Crusoe Spark 与超大规模园区

Crusoe 的物理基础设施业务有两种产品形态:Crusoe Spark,一种预制模块化 AI 工厂单元;以及大型超大规模托管园区。 Crusoe Spark 是交钥匙、类集装箱的模块化数据中心,提供约 1 MW 高密度算力,集成配电、液冷或风冷、消防,以及为最新加速器构建的 GPU 机柜。单元来自 Crusoe 位于科罗拉多州 Brighton 的 352,000 平方英尺 Spark Factory;该工厂投资超过 $200 million,年产能最高 100 个模块,工厂制造单元计划在 2026 年 Q3 首次交付。Crusoe Spark 可接入多元电源:在内华达州 Redwood Materials Tahoe Campus 的首个公开部署使用二次寿命 EV 电池微电网和太阳能,证明了 99.2% 可用时间。 与 Energy Vault 的合作让 Spark 可从 2026 年起在德州 Snyder 以 25 MW 规模部署。超大规模侧,德州 Abilene 的旗舰 1.2 GW 园区位于 Lancium Clean Campus 土地上,是 Stargate 计划的主场地;前两栋楼(980,000 sq ft,200 MW+)已上线,为 Oracle Cloud Infrastructure 服务 OpenAI 的算力需求。为 Microsoft 建设的第二个 900 MW Abilene 园区已于 2026 年破土。Crusoe 从一开始就共同开发电力和建设,并在科罗拉多、俄克拉荷马和路易斯安那设施中制造长交期电气组件;这种垂直整合深度,相比传统顺序开发压缩了部署周期。 截至 2026 年 6 月,Crusoe 已签约 4.9 GW AI 基础设施,项目管线超过 40 GW。[CE013, CE014, CE015, CE016, CE017, CE018]

路线图和开发阶段
日期 / 阶段里程碑 / 功能状态影响来源
December 2023SOC 2 Type I 认证已完成企业合规基线已建立Crusoe 官方博客
July 2024SOC 2 Type II 认证已完成控制措施运行有效性已验证;企业销售中的差异化因素Crusoe 官方博客
June 2025Crusoe Spark 发布(Redwood Materials Tahoe 试点)已完成首个边缘 / 模块化 AI 工厂部署,使用二次寿命 EV 电池官方新闻稿
Q3 2026Brighton, CO 工厂首批制造的 Spark 模块目标中验证制造扩产能力;加快 Spark 客户的模块化部署Crusoe 官方博客;Forbes
Late 2026 / 2027NVIDIA Vera CPU 和 Rubin GPU 在 Crusoe Cloud 部署已承诺(早期采用方)Crusoe Cloud 上的新一代智能体 AI 工作负载;为后 Hopper 周期占位Crusoe 官方新闻室(GTC 2026)
2027Form Energy 12 GWh 铁空气电池储能部署已签署协议为园区级可靠电力提供全天候长时储能Form Energy 新闻稿

路线图事项为公司声明目标或已签署协议,除非标注为“已完成”。NVIDIA 硬件时间表取决于供应链和 NVIDIA 制造排期。

[CE012, CE019, CE022, CE026]
FE003: 关键依赖地图

有向无环图梳理 Crusoe 的关键供应商、合作伙伴与监管方;从 GPU 供给、能源基础设施到合规,其可用性直接影响平台交付。

依赖强度未量化;NVIDIA 是压倒性的主导单点依赖。energy-vault 与 PROENERGY 关系已确认,但为便于阅读未纳入图中。

[CE018, CE019, CE022, CE023]

5.4 NVIDIA 伙伴关系深度与开发者生态

Crusoe 最深的护城河,是其与 NVIDIA 的垂直整合式对齐。在 GTC 2026,Crusoe 宣布将成为 NVIDIA Vera CPU 早期采用者;Vera CPU 专为 agentic AI 后训练环境打造,单机柜可支持超过 22,500 个并发 CPU sandbox 环境。 同时,Crusoe 计划在 2026 年末及 2027 年陆续部署 NVIDIA Rubin GPU 和 Vera Rubin NLV72 系统。 Crusoe 还采用 NVIDIA Omniverse DSX Blueprint 和 Vera Rubin DSX Reference Design,用于设计下一代 GW 级 AI 工厂,借助数字孪生做设施可视化,以及 AI 驱动的电力和冷却优化。这让 Crusoe 成为全栈、NVIDIA 原生 基础设施伙伴,而不只是 GPU 容量转售商。开发者侧,Crusoe 维护公开 GitHub 组织(github.com/crusoecloud),仓库包括 Terraform provider、CLI、Go client library、Slurm 配置、Helm charts、解决方案库,以及 fastokens Rust BPE tokenizer 仓库(截至 2026 年 6 月仍有活跃提交)。开发者中心托管了六个在线 AI 演示,覆盖在 CMK 上进行多节点 Llama 预训练、在 Slurm 上进行 PyTorch 分布式训练,以及用 vLLM 和 KubeRay 做 Ray 推理。Crusoe 没有运营一个具备可衡量社区采用指标(stars、contributors、fork counts)、可与云原生 OSS 项目相比的重要开源项目;因此,开发者信号证据更多锚定在文档深度、演示质量和 NVIDIA Dynamo 贡献,而不是公开仓库采用度指标。[CE022, CE023, CE024, CE025, CE033]

FE004: 产品成熟度与能力地图

围绕 Crusoe 五个主要产品界面,评估成熟度、开发者采用深度、差异化强度和主要尽调风险,区分哪些已经达到生产级,哪些仍在成熟。

开发者采用评级来自具名案例研究和 ClusterMAX 评测。成熟度评级反映已宣布的 GA 日期,并在可得时纳入第三方验证。

[CE001, CE009, CE013, CE015, CE028, CE029]

5.5 信任、合规、可靠性与已知缺口

Crusoe Cloud 于 2023 年 12 月获得 SOC 2 Type I 鉴证,并于 2024 年 7 月完成 SOC 2 Type II,完整报告可通过 Crusoe Trust Center 获取。平台符合 GDPR,99.98% 集群可用时间 SLA 由 AutoClusters 自动节点替换能力支撑。 截至 2026 年 6 月的状态页历史显示,可用性总体较强,但也有明显例外:ICAT 区域在 2026 年 6 月 17 日发生了两小时负载均衡器宕机。 SemiAnalysis 的 ClusterMAX 评测指出 Slurm-on-Kubernetes 产品存在有意义的可用性缺口——login pods 缺少 vim、nano、git 和 sudo;CMK 集群没有提供默认 ReadWriteMany StorageClass;容器化负载下可复现 NVML driver mismatch 错误。在 Crusoe 冰岛设施,20,000 个光纤端面因碎屑(归因于火山灰)需要物理清洁,碎屑导致大范围链路抖动和随机文件系统卸载;据报道 Crusoe 已解决这一具有地域特异性的可靠性风险。SemiAnalysis 指出,云部门工程师高流动率可能危及 Crusoe 的 Gold 评级。CCPA 合规状态未公开确认。公开记录中没有 MemoryAlloy 相比 AWS Inferentia、Azure AI 或 GCP TPU 的外部性能基准。Crusoe 任何设施的电力使用效率(PUE)都未披露。[CE026, CE027, CE028, CE029, CE030, CE031]

信任、质量与合规控制
控制 / 认证状态范围缺口 / 尽调事项
SOC 2 Type I已取得(2023 年 12 月)某一时点的控制设计已到期;由 Type II 取代;未公开披露持续续期节奏
SOC 2 Type II已取得(July 2024)审计期内控制措施运行有效下一次续期日期及 Trust Services Criteria 范围未公开说明
GDPR 合规称已合规欧洲数据处理未公开 DPA 模板或传输机制细节
CCPA 合规未公开确认加州消费者数据尽调问题:确认对加州居民客户数据的适用性及控制措施
99.98% 正常运行时间 SLA合同保证所有 Crusoe Cloud 集群(IaaS)有记录的 June 2026 ICAT 负载均衡器故障;2025 年冰岛光纤事件
数据驻留 / 隔离VPC 级隔离按组织做逻辑网络隔离GPU 节点物理多租户隔离模型(cloud-hypervisor)未经过独立审计

SOC 2 和 GDPR 状态来自 Crusoe 官方博客;正常运行时间和隔离信息来自产品文档。CCPA 缺口是基于公开材料作出的推断。尽调问题源于公开披露缺失,并不代表已确认不合规。

[CE026, CE027, CE031]

5.6 附录

Chapter 06

06客户

6.1 买方格局:初创公司、企业与超大规模租户

Crusoe 服务两个结构上不同的客户群,交易规模、采购机制和风险画像都不一样。云层服务 AI 原生初创公司和企业机器学习团队,提供按需 GPU 算力或年度预留,具备自助定价、低于 90 秒的 VM 开通,以及 SOC 2 Type II 合规姿态。数据中心园区层通过十年以上电力购买协议和定制建设项目服务超大规模运营商。 这两个群体不只是同一产品的不同体量版本——它们需要不同销售动作,流失动态不同,也让 Crusoe 暴露在本质上不同的集中度风险中。 在云业务中,主导买方画像是拥有 10–200 名工程师、需要高密度 GPU 访问但不想承受 AWS、Azure 或 GCP 预留摩擦的 AI 初创公司。 Crusoe 对这一群体的竞争定位靠三根支柱:公开透明定价(H100 $3.90/hr,H200 $4.29/hr,AMD MI300X $3.45/hr)、认证 NVIDIA Cloud Partner 背书以传达 GPU 供给可靠性,以及吸引 ESG 取向投资者和被投公司的可持续叙事。 企业 AI 群体——包括 Databricks、Sony 和 Together AI——通过直销使用 Crusoe Cloud,在标准 IaaS 目录 之上叠加定制 SLA 和专用集群选项。2026 年 4 月 Managed Inference 上线,打开了新的 API 优先消费层,把可触达市场延伸到无法承诺预留集群合同的 token 计费经济买方。 数据中心园区业务本质是地产与电力经纪:Crusoe 寻找场址,从电网或替代来源锁定电力容量,建设园区设施,再把托管容量租给需要确定性、多 GW 交付且时间线压缩的超大规模云厂商。 Oracle、Microsoft 和 Meta 是这一层已披露或报道的三大关系。第四个未具名客户(普遍报道称为 Google)在开发中途退出 Wyoming 园区项目,暴露了项目管线到合同之间的转化风险。xAI 是混合型客户——既把 Crusoe 当作基础设施建设伙伴,也把它作为 Grok 训练和推理的 GPU 云提供商。 云渠道以直接自助为主,辅以 NVIDIA NCP 伙伴渠道。园区渠道完全是直接企业销售和定制谈判。没有经销商或市场平台分销层来分散项目管线 风险。地域集中以美国为主(Texas、Wyoming、Nevada、California),欧洲容量位于 Iceland 和 Norway,服务数据驻留需求。

客户分群表
分群代表买家主要工作负载价值主张合同类型
AI 初创公司(云)AI 初创客户:Windsurf、Decart AI、BosonAI、Wonderful.ai、Yutori、OakletLLM 训练和实时推理GPU 定价透明;较超大规模云低 50%;<90 秒开通按月或按年预留 GPU
企业 AI 团队(云)企业 / AI 客户:Databricks、Sony、Together AI、Luma AI、Codeium分布式训练和批量推理企业级 SLA;专用集群;SOC 2 Type II 合规含预留容量的年度企业合同
超大规模数据中心租户Oracle、OpenAI、Microsoft、Meta(据报道)AI 园区托管;高功率密度 GPU 基础设施12 个月交付吉瓦级容量;能源优先选址10–20 年购电协议,分阶段扩张
AI 实验室 / 战略伙伴锚定项目:xAI(Colossus)、NVIDIA(NCP)下一代模型训练,exaflop 级规模GPU 供应连续性;基础设施建设伙伴关系混合模式——云预留加共同开发协议
研究机构(云)未披露的 HPC 和大学实验室高吞吐批量计算;模型评估相较国家 HPC 配额更具成本效率的突发容量现货或按月预留访问

分群综合 Crusoe 营销、案例研究、分析师覆盖(Sacra、Cheddarflow)及截至 June 2026 的新闻报道。超大规模租户细节反映已发布和报道的协议;Meta 在研究日期尚未确认。研究机构分群没有公开文档支持。

[CU001, CU002, CU003]

6.2 具名客户证明:云与园区的生产级胜利

Crusoe Cloud 文档最充分的客户胜利是 Windsurf(前 Codeium),这是一家服务超过 800,000 名开发者的 AI 编程助手平台。 根据 Crusoe 发布的案例研究,Windsurf 将完整 GPU 推理负载迁移到 Crusoe Cloud,相比 AWS 和 Azure 降低 50% 基础设施成本,并在 12 个月生产期内维持 99.98% 集群可用时间。Windsurf 的迁移覆盖训练和推理负载,随后在 Managed Inference API 于 2026 年 4 月上线后扩展使用该产品。这一引用构成 Crusoe 最强的公开证据:客户被点名、结果被量化、生产时长被说明,且来自官方案例研究文档。 Decart AI 和 BosonAI 是另外两个具名云客户。Decart 在 Crusoe Cloud H100 集群上运行面向人类模拟应用的实时 LLM 推理,并提到低于 200ms 的响应延迟。BosonAI 使用 Crusoe Cloud 进行多语言 LLM 预训练,并提到成本效率。两家公司都出现在 Crusoe 营销材料中,并由 Cheddarflow 分析佐证。其他具名云客户包括 Wonderful.ai、Yutori、Oaklet、Sony、Databricks、Together AI、Luma AI 和 Codeium(现 Windsurf),覆盖训练和推理。 在超大规模园区层,Oracle 和 OpenAI 是德州 Abilene 1.2GW 园区 Phase 1 的锚定租户;该园区已于 2026 年 5 月上线。 Oracle CEO Safra Catz 公开称 Crusoe 12 个月从破土到投运的周期是「无可匹敌的执行力」。Microsoft 是剩余 900MW 的 Phase 2 锚定租户,截至 2026 年 6 月在建。根据 Data Center Dynamics 报道,Meta 据称签署了覆盖德州 Childress 和密苏里 Warrenton 的 1.6GW 协议,但双方均未正式确认。xAI 的 Colossus 2 扩建通过多家媒体与 Crusoe 基础设施相连,使 Crusoe 成为 Grok 模型训练的建设和 GPU 云伙伴。 SemiAnalysis 的 ClusterMAX Gold 评级在网络、硬件质量、企业支持响应等八个评分维度上提供了第三方产品验证。 ClusterMAX 评测者指出,首次支持响应低于 6 分钟,GPU 可用性高于平均水平——与 Windsurf 报告的 99.98% 可用时间一致。 同一评测也指出 Slurm-on-Kubernetes 产品存在重大运营缺口(缺少开发者工具、缺少默认存储类),为整体强背书加入了必要的细节。

具名客户证明表
客户分群工作负载或部署证据类型结果或规模
Windsurf(Codeium)AI 初创公司 / 云面向 800K+ 开发者的 LLM 推理;训练从 AWS/Azure 迁移官方案例研究(Crusoe 新闻室)成本节省 50%;12 个月生产环境正常运行时间 99.98%
Decart AIAI 初创公司 / 云H100 集群上的实时人类模拟 LLM 推理Crusoe 营销材料;Cheddarflow 分析师覆盖Llama 级模型推理延迟低于 200ms(自报)
BosonAIAI 初创公司 / 云大规模多语言 LLM 预训练Crusoe 营销材料;Cheddarflow 分析师覆盖高性价比大规模分布式训练(定性)
Oracle / OpenAI(Stargate)超大规模租户 / 园区Abilene TX AI 训练园区 Phase 1(200MW+);May 2026 上线Data Center Dynamics 报道;Oracle CEO 公开声明园区已运营;12 个月建设速度无可匹敌(Oracle CEO 归因)
Microsoft超大规模租户 / 园区Abilene TX AI 园区 Phase 2(900MW);建设中Crusoe 新闻室和 DCD 报道Phase 2 建设推进中;预计 2027 完工
Meta超大规模租户 / 园区(据报道)Childress TX 和 Warrenton MO 的 1.6GW 数据中心容量Data Center Dynamics 报道(双方均未确认)合同价值未确认;若执行,将成为报道中最大园区交易
xAI(Colossus 2)AI 实验室 / 战略伙伴为 Grok 模型训练和推理建设 AI 基础设施Heatmap News 和 Cheddarflow 分析师覆盖报道称合作规模 1GW+;Crusoe 为主要基础设施伙伴
客户名单:Databricks、Together AI、Luma AI企业 AI 团队 / 云混合 ML 训练和推理工作负载Crusoe 营销材料具名云客户;部署规模未披露

没有官方案例研究的行依赖 Crusoe 营销说法或第三方报道。Meta 和 xAI 交易来自二手报道;双方均未发布新闻稿。客户结果数据反映自报或公司归因数字。

[CU007, CU008, CU009, CU010, CU011, CU012]
FU001: 客户旅程地图

六阶段旅程从发现 GPU 需求延伸到长期锚定租户,映射 Crusoe Cloud 关系中每个阶段的买方动作、摩擦点和价值增益。

旅程阶段基于 Crusoe 营销材料、案例研究和 ClusterMAX 评测者体验的定性综合。具体时间点与痛点 / 收益强度因买方而异。

[CU023, CU027, CU029, CU030]

6.3 采用轨迹:ARR 增长、合同扩张与收入爬坡

Crusoe Cloud 的 ARR 从 2024 到 2025 年同比增长 150%,同期平均合同价值扩大 17×;这一组合同时显示了量的扩张和交易规模成熟。 17× 合同价值增长反映了从小型初创公司月度 GPU 预留,转向企业级年度集群承诺;单个企业交易贡献的 ARR 可能超过数十个初创客户合同。 2025 年新客户增长同比 70%,说明顶线 ARR 增长不只是来自存量客户扩张,也包含可观的新客户成分。 Sacra Research 预计 Crusoe 2026 年总收入约 $2 billion,而 2024 年为 $276 million。这个 7× 复合增长几乎完全由数据中心园区业务驱动,具体是 Abilene 园区 Phase 1 和 Phase 2 租户协议。Cheddarflow 分析估计,截至 2026 年初,云平台独立 ARR 在 $150–200 million 区间。这条轨迹值得关注,但必须放在背景中看:Crusoe 总收入仍包含历史 Digital Flare Mitigation(DFM)能源收入和园区建设费用,经常性云 ARR 是重要但少数的一部分。 Managed Inference 于 2026 年 4 月上线,打开了 token 计费经济客户层:API 优先买方需要推理容量,但不愿承诺预留 GPU 集群。这一层服务的是此前由超大规模推理 API(AWS Bedrock、Azure OpenAI)和推理专业初创公司覆盖的市场。若基准在多样化生产负载下仍成立,MemoryAlloy 的 9.9× TTFT 优势会成为吸引这一群体的性能钩子。整体增长图景——云 ARR 增长 150%、合同价值 17×、新客户增长 70%——对基础设施业务来说很强,反映了真实市场需求。关键尽调问题是,这条轨迹能否持续,还是由两三个不会重复的超大规模承诺前置拉动。

客户增长 / 采用轨迹表
指标数值期间来源依据
云业务 ARR 增长(YoY)150%2024 至 2025Crusoe 2025 Impact Report(自报)
新客户增长(YoY)70%2025Crusoe 2025 Impact Report(自报)
平均合同价值增长(YoY)17×2025Crusoe 2025 Impact Report(自报)
已签约 AI 容量4.9 GWJune 2026Business Insider 和 SiliconAngle(据报道)
开发储备40 GW+June 2026Business Insider 和 SiliconAngle(据报道)
总收入(预测)$2 billionFY 2026(估计)Sacra Research(分析师估计;付费墙)
总收入(实际)$276 millionFY 2024Sacra Research(分析师估计;付费墙)

2025 年增长指标来自 Crusoe Impact Report 自报,未经独立审计。收入数字为第三方分析师估计;Crusoe 未公开披露财务。容量数字包括已签租约和报道协议;已确认容量与储备容量的拆分不可得。

[CU016, CU017, CU018, CU019, CU020]
FU002: 采用 / 部署漏斗

客户获取流程展示从最初 GPU 需求信号到多产品扩张的六个阶段;阶段之间的转化说明 Crusoe 如何把市场需求转成生产云客户。

阶段描述来自产品和客户信息的定性整理;Crusoe 未公开披露漏斗转化率。

[CU001, CU004, CU016, CU022]

6.4 留存与耐久性:SLA 表现、CSAT 与切换经济性

Crusoe Cloud 合同承诺 IaaS 层集群可用时间 99.98%,VM 启动时间低于 90 秒。99.98% 可用时间折算为每个集群每年允许约 105 分钟计划外停机。Windsurf 12 个月生产使用期内未报告 SLA 违约,是这一承诺最强的独立验证。SemiAnalysis ClusterMAX 评测指出支持响应速度快于平均水平,Crusoe 也报告过去 12 个月 CSAT 为 100%——但这些数字都来自自报,除 Windsurf 案例研究 外,未由具名客户独立审计或确认。 Crusoe 状态页记录,ICAT 区域在 2026 年 6 月 17 日发生负载均衡器宕机,使端点约两小时无法访问。该事件仍处在 SLA 余量内(单次事件在 99.98% 容忍度内),但说明 Crusoe 并不能免疫所有云提供商都会遇到的运营问题。冰岛设施在 2025 年也发生过大范围光纤污染事件(火山灰碎屑影响 20,000 个光纤端面),需要物理修复。两起事件似乎都没有触发公开 SLA 赔付或报告客户流失,与自报 100% CSAT 数据一致。 Crusoe Cloud 未公开披露 Net Revenue Retention 和 Gross Revenue Retention。Cheddarflow 根据定性扩张信号,估计 2025 年前客户队列的 NRR 在 110–130% 区间。该估计无法独立验证。不同层级的合同期限差异很大:超大规模数据中心租户签署 10 到 20 年电力购买协议,收入极其耐久;云客户签署月度或年度 GPU 预留,迁移性更强,但也受益于适度切换成本(负载集成、SLA 熟悉度、网络配置)。Windsurf 从 GPU IaaS 扩展到 Managed Inference,展示了推动 Crusoe Cloud 估算高 NRR 的自然追加销售模式。 缺少公开 NPS、独立 CSAT 审计或披露的流失率,是潜在企业买家评估长期承诺时的重大尽调缺口。

留存 / 重复使用 / 满意度表
信号报告值期间或范围反向备注
集群正常运行时间 SLA合同保证 99.98%所有 IaaS 集群(持续)2026 年 6 月 17 日 ICAT 负载均衡器故障(约 2 小时);2025 年冰岛光纤事件
客户满意度(CSAT)100%(自报)过去 12 个月(截至 June 2026)自报;无独立审计或 NPS 披露;方法未公开
首次支持响应时间<6 分钟平均持续(SemiAnalysis ClusterMAX 评测)基于评测者测试工单;并非具统计代表性的样本
净收入留存(NRR)110–130%(Cheddarflow 估计)2025 前队列仅为分析师估计;Crusoe 未公开披露 NRR
VM 启动时间 SLA<90 秒所有 GPU VM 类型持续突发负载下的性能未经独立验证
Windsurf 使用期生产环境 12 个月,未报告 SLA 违约2025–2026单一具名参考,不代表全部客户群

质量信号主要由 Crusoe 自报或第三方分析师估计。没有公开 NPS、GRR 或流失披露。Cheddarflow 的 NRR 估计是定性的。

[CU023, CU024, CU025, CU026, CU027]
FU003: 客户验证矩阵

从五个维度跨细分客群评估客户验证质量,突出 Crusoe 哪些地方有强公开证据,哪些地方仍需索取一手数据。

证据质量评级是基于公开来源的分析师判断。每个细分客群中的单个客户情况,可能不同于细分层面的刻画。

[CU007, CU008, CU009, CU013, CU015]

6.5 采购模式、集中度风险与负面信号

Crusoe 的超大规模数据中心采购遵循以分阶段园区承诺为锚的先落地再扩张模式。Oracle 和 OpenAI 拿下 Abilene Phase 1(200MW+),并充当建设催化剂;Microsoft 拿下 Phase 2(900MW),把园区扩展到总计 1.2GW。 这种分阶段结构对双方都合理:客户不用一开始承诺完整阶段,从而对冲建设风险;Crusoe 则拿到锚定客户,可在完整园区被订满前启动建设融资。 风险在于,后续阶段需要同一租户重新承诺或找到替代者;Wyoming 项目暂停说明,Crusoe 不能假设 Phase 2 会自然跟随 Phase 1。 云采购走两条路径。自助买家通过 crusoe.ai/cloud 注册,初始试用后转为月度或年度 GPU 预留合同。企业买家与 Crusoe 直销团队接触,谈定制集群配置、专用私有网络和协商 SLA。Managed Inference 用户采用 API-key 模式,没有集群预留,因此是迁移性最强、也最容易流失的群体。客户集中是 Crusoe 客户画像中最重大的风险信号。园区层有三个已确认或报道的锚定租户(Oracle、Microsoft、Meta),每个规模 200MW–900MW,另有 xAI 1GW+。如果其中任何两个无法从报道转为已执行合同,Crusoe 的签约容量指标将被重大削减,园区建设融资也可能承压。 Wyoming 项目暂停——一名未具名客户要求停止——是最清楚的证据:超大规模项目管线客户确实可能退出。Crusoe 对客户指标的公开披露很少:没有客户数量、收入拆分或客户队列 NRR。Worldmetrics 汇总的统计显示可能有数百个云客户,但引用的是二手来源,没有原始数据。根据 Crusoe 新闻室,Upstream Data 专利诉讼已于 2025 年通过许可协议解决;blockspace.media 报道称裁决「混合」,认定侵权但损害赔偿存在争议。Crusoe Cloud 核心业务是软件和基础设施,不是 DFM 硬件,因此 Upstream Data 案件的残余 IP 风险范围有限;但如果历史业务增长,该案仍为 DFM 技术 IP 挑战提供了先例。

扩张和集中度风险表
风险因素当前信号严重性尽调问题
超大规模租户集中(园区)前三大租户(Oracle / Microsoft / Meta)占园区容量大多数严重索取各租户已签约与 LOI 的拆分;确认 Meta 交易交割状态
Wyoming 储备项目撤回1.8GW 园区因未具名客户要求暂停;项目时间表未知重大识别撤回客户;评估该站点能否吸引替代租户
xAI Colossus 依赖据报道 xAI 为 1GW+ 伙伴;双方均未确认关系重大确认 xAI 关系的合同状态;评估续约和排他条款
云客户数量披露缺口云客户总数未公开披露;估计为数百家轻微索取按合同价值分桶的队列级云 ARR 和客户数
IP 残余风险(Upstream Data)专利案通过许可解决;损害赔偿裁决结果不一轻微审查许可范围;确认没有未决上诉或与 DFM 有关的 IP 主张
Managed Inference 流失风险API key 消费模型在 Crusoe 所有产品中切换成本最低轻微索取 Managed Inference(April 2026 发布)的月活用户留存队列

严重性评级是基于公开信息的分析师判断。园区集中度风险基于报道合同价值;实际绑定承诺排期未公开。

[CU031, CU032, CU033, CU034, CU035, CU037]
FU004: 留存 / 重复购买队列

按客户队列估算季度留存率,比较早期采用云客户与较新的企业和推理层新增客户的持续性。数值 100 代表纵向跟踪开始前的队列基准期。

所有留存值均为分析师估算。数值 100 代表该队列活跃的首个期间(基准);后续数值展示相对起点的估算留存。Crusoe 未披露队列级留存数据。超大规模租户留存反映合同锁定,不代表有机留存。2024 年前队列起点低于 100,是因为从 DFM 商业模式转出时推断出部分历史流失。

[CU025, CU026, CU027, CU028]
Chapter 07

07风险

7.1 风险概览与排序

Crusoe 已从放空天然气起点,成长为北美扩张最快的 AI 基础设施开发商之一。规模上来后,风险画像也发生质变:公司主要暴露不再是油田准入和甲烷信用 的波动,而是能源许可的政治经济学、少数超大规模客户的需求集中,以及一条由 CEO 本人形容为可能超过承诺方实际交付能力的项目管线执行风险。 本章按残余严重性从高到低,对七类重大风险进行排序,用证据逐一审视,并以可监控的终止标准收尾。风险簇包括:(1)电力建设执行与电网监管风险;(2)客户集中与超大规模客户依赖;(3)GPU 供应链与供应商集中;(4)甲烷与环境监管不确定性;(5)声誉与 ESG 矛盾;(6)IP 与法律遗留;(7)竞争替代。每项风险都有传导机制,会影响 Crusoe 的收入、利润率和融资可得性,投资者必须持续监控。 [CR001, CR002, CR003, CR004, CR005]

FR001: 风险热力图 — 可能性 vs. 剩余严重性

将 Crusoe 七类风险簇按可能性(低 / 中 / 高)和当前缓释后的剩余严重性(低 / 中 / 高 / 关键)绘制出来。

基于公开证据的定性位置;没有可用的量化概率估计。

[CR001, CR002, CR003, CR004, CR005]

7.2 监管、环境与 ESG 风险

Crusoe 的法律与监管风险是多层的。甲烷侧,虽然 Crusoe 已在 2025 年剥离 Digital Flare Mitigation 业务,但最初定义其身份的监管格局仍在变化。 EPA 2023/2024 年 OOOOb/OOOOc 规则禁止 2026 年 5 月 7 日后新油气井常规放空燃烧,同时 Trump 政府又提出放松监管并停止执行现有规则——对任何竞争定位依赖放空燃烧经济继续作为降本路径存在的公司,都形成政策急转风险。 BLM Waste Prevention Rule 对若干放空气计量和 LDAR 条款的执行,在 2025 年 11 月再次推迟至 2026 年 12 月;2026 年 6 月发布的拟议修订则会回滚废弃物最小化计划要求。五个州(North Dakota、Texas、Montana、Wyoming、Utah)已在初步禁令下阻止 BLM 执行。对 Crusoe 而言,当前最大的监管敞口在电力交付监管:Texas Senate Bill 6 由 Governor Abbott 于 2025 年 6 月签署并立即生效,要求 ≥75 MW 数据中心在强制削负荷事件中接受限电,要求 2025 年 12 月 31 日后新接入具备远程断开能力,并把互联升级成本转嫁给大负荷客户。ERCOT 预计,到 2030 年其电网上大型负荷将达 138 GW,而 2025 年中为 87 GW——维护季的可靠性风险很重大。ESG 侧,Heatmap News 记录称 Crusoe 计划在 Abilene Stargate 园区部署 360 MW 现场天然气涡轮机(占已宣布 1.2 GW 容量的 30%),并通过与 Engine No. 1 和 Chevron 的合资取得 4.5 GW 新天然气容量——这些动作与公司的气候科技自我描述并不协调。 EDF 报告称,Trump EPA 停止执行甲烷监管后的第一年,约 $4.2 billion 天然气被浪费,凸显了 Crusoe 原始模式所处监管环境的争议性。 [CR006, CR007, CR008, CR009, CR010, CR011]

监管 / 法律风险登记表
规则 / 案件司法辖区当前状态不利变化可能性对 Crusoe 的严重性缓释措施残余暴露尽调路径
EPA OOOOb/OOOOc 甲烷规则(May 7 2026 常规放空燃烧禁令)美国联邦规则生效,但 Trump 政府暂停执行;拟议复议正在推进高——放松监管回滚可能性高,但下一届政府可能重新执行低直接影响——DFM 已剥离;间接 ESG 标签风险仍在剥离 DFM;碳捕集 JV 规划中如果天然气数据中心与未来规则冲突,声誉风险会上升按季度跟踪 EPA OIRA 案卷最终规则状态
BLM 废气防治规则(甲烷放空计量和 LDAR)美国联邦 / 州两项条款执行延至 Dec 2026;5 个州有禁令中——Trump 修订案已于 June 2026 提出;诉讼暂缓DFM 剥离后直接影响低;若 Crusoe 重返油田领域则为中已剥离 DFM 业务;当前没有联邦租赁暴露未来政策反转可能影响任何从联邦土地获取能源的重返动作跟踪 North Dakota v. DOI 诉讼(8th Cir.);监测 BLM 规则制定案卷
Texas Senate Bill 6(ERCOT 电网限电和大负荷成本分摊)Texas / ERCOTJune 2025 生效;Jan 2026 起新连接须支持远程断开确定——法律已颁布;2026 年内规则继续制定高——Abilene 园区 ≥75 MW 必须接受限电;前期并网费用场内天然气备用;参与需求响应;与 PUCT 沟通电网紧张时的限电事件可能干扰对锚定客户的 SLA 承诺审查 PUCT 关于强制限电阈值的规则制定;核实备用发电规模
Crusoe v. Upstream Data 专利诉讼(US10862307 等)美国(EDTX 和 USPTO PTAB)November 24 2025 以保密许可和解;所有案件驳回已和解权利要求为低;Upstream 剩余 13 项创始人专利为中历史上为中(法律费用、放空气挖矿系统禁令风险);和解后为低保密许可;DFM 已剥离;重心转向 AI 基础设施若 Crusoe 重返该领域,Upstream Data 剩余油田系统设计专利可能再度浮现审查许可范围;监测 Upstream Data 新主张的备案
AI 基础设施和推理引擎的一般 IP / 专利风险美国 / 国际未发现活跃诉讼;竞争性 IP 格局加剧中——AI 推理工具和内存优化(Atero 收购)存在专利暴露中——MemoryAlloy 推理引擎和 Crusoe tokenizer 可能遭遇挑战ISO 认证;法务团队扩张;Atero 收购专利商业化后发现冲突专利,可能迫使许可或绕开设计对 MemoryAlloy 和 NVIDIA Dynamo tokenizer 集成出具专利清障意见

严重性和可能性评估是基于公开监管文件、法院案卷和新闻报道的定性推断。Crusoe 未披露直接财务暴露。

[CR006, CR007, CR008, CR009, CR010, CR011]
FR002: 风险传导图 — 风险如何流向收入、利润率和融资

有向无环图追踪 Crusoe 的主要风险触发因素如何经由运营指标层层传导到估值和融资可得性。

[CR001, CR002, CR003, CR004, CR005, CR019]

7.3 运营、技术与供应链风险

Crusoe 的运营风险集中在三条线:GPU 可得性、建设执行和供电可靠性。GPU 供给方面,公司在 2025 年 6 月承诺投入 $400 million 购买 AMD MI355X 加速器(约 13,000 台),降低对 NVIDIA 的纯粹依赖;但 AMD 的 ROCm 软件栈成熟度仍弱于 CUDA,会给推理客户带来运行时风险。NVIDIA 持续参股(Series D 和 E 投资者)带来供应链对齐,但也意味着 NVIDIA 的战略优先级会塑造 Crusoe 的硬件路线图。到 2025 年中,美国半导体关税平均为 18.2%,直接推高 GPU 采购成本并压缩利润率。 建设执行方面,CEO 本人公开表示——「这个领域已经许下了很多承诺,但我无法担保它们真的会发生」——说明 45 GW+ 项目管线带有相当大的投机性悬空。Wyoming 1.8 GW 园区因客户要求暂停,以及 Oracle/OpenAI 退出 Abilene 后续阶段,是这一风险的早期证据。公司被迫寻找替代租户(Microsoft 在 Abilene 接手),增加了时间线和收入确认不确定性。 Crusoe Cloud 达到 99.98% 可用时间,并持有 ISO 27001 + ISO 42001 认证(2026 年 2 月),这缓解了云平台运营风险。 但以天然气备份运行的 GW 级绿地数据中心,会引入纯云玩家不承担的单点能源供给故障。 [CR016, CR017, CR018, CR019, CR020, CR021]

运营和技术风险登记表
失效模式可能性严重性缓释成熟度残余暴露未解决缺口
GPU 供应中断——NVIDIA/AMD 短缺、出口管制、关税部分——AMD 多元化推进中;NVIDIA 供应协议如果关税恶化,GPU 单位经济性承压AMD MI355X ROCm 软件成熟度相对 CUDA 对推理工作负载的差距
多 GW 园区建设时间表滑坡低——Abilene Phase 1 按时交付,但 Wyoming 暂停;储备项目具投机性收入确认延迟;部分建成园区的资本开支搁浅未公开已承诺与投机性储备项目拆分
天然气支持数据中心供电中断低-中部分——Abilene 现场 360 MW 燃气轮机;电网互联若天然气供应中断与电网紧急情况重叠,存在 SLA 违约风险燃气机组碳捕集计划尚未落地或签约
影响 Crusoe Cloud 客户的软件 / 云平台故障强 — 声称正常运行时间达 99.98%;ISO 27001;已取得 SOC 2 Type II云服务残余风险低;裸金属 GPU 集群客户风险更高未审阅公开事件日志;未检查 status.crusoecloud.com 的历史事件
AI 工厂层面的网络安全入侵或数据外泄推进中 — ISO 27001 + ISO 42001 双认证;2025 年更新 Code of Conduct若超大规模云厂商数据被攻破,声誉和客户流失风险上升ISO 42001 在 AI 推理工作负载中的范围和覆盖度尚未独立核验
能源成本上涨削弱相对超大规模云厂商的价格优势部分 — VPPA 和搁浅能源采购提供低于市场的电价一旦能源套利优势收窄,Crusoe 只能靠执行力竞争VPPA 定价和交易对手信用质量未公开披露

发生概率和严重性为定性估计,依据公开报道事件和 CEO 披露。缓释成熟度评分仅反映已发布认证和公开表态。

[CR016, CR017, CR018, CR019, CR020]

7.4 客户集中、伙伴依赖与竞争风险

Crusoe 的需求集中是领先风险指标。Abilene Stargate 园区是其旗舰收入项目,后续阶段已出现 Oracle/OpenAI 退出,随后由 Microsoft 承接该容量。Wyoming 1.8 GW Project Jade 因单一客户要求暂停,客户和原因均未披露。第二项重大安排据称是 Meta 与 Crusoe 签署 1.6 GW 容量协议;这虽有助于多元化,却也放大了超大型单客户风险事件的模式。相比之下,云服务侧的 Crusoe Cloud 有 Cursor、Decart、Fireworks、Odyssey 和 Together AI 等具名客户——这些企业规模较小,单个客户集中风险较低,但其合计 ARR 未披露。竞争侧,CoreWeave 2025 年 3 月 IPO 时市值 $23 billion(企业价值约 $30 billion),2024 年收入 $1.9 billion(同比增长 730%),建立了清晰标杆:Crusoe 按收入倍数相对 CoreWeave 折价,但缺乏同等公开可比性。 AWS、Google Cloud 和 Azure 继续扩张 AI GPU 容量,资产负债表深度远超 Crusoe。Lambda Labs 和其他新云厂商 仍是价格上有竞争力的替代选项。伙伴依赖方面,Crusoe 与 Tallgrass 的 Wyoming 电力供应合作现在实际上已暂停;Blue Owl Capital 对 Abilene 园区的 $3.4 billion 合资制造项目融资相互依赖;NVIDIA 股权持股既带来供应链对齐,也可能在 NVIDIA 开发竞争性云服务时制造潜在利益冲突。 [CR024, CR025, CR026, CR027, CR028, CR029]

合作伙伴和客户依赖风险登记表
依赖项交易对手角色集中度失效情景严重性缓释措施残余风险敞口
Abilene Stargate 园区锚定租户Microsoft(Oracle/OpenAI 阶段性退出后)1.2 GW Abilene 园区 Phase 2+ 的收入和利用率锚点关键 — 已知最大单一收入驱动因素Microsoft 延期或退出;园区出现收入缺口关键Blue Owl JV 提供项目融资;可模块化转租高 — 未公开确认备选租户
Wyoming Project Jade 园区未披露客户1.8 GW Cheyenne 园区锚定租户高 — 整个园区按客户要求暂停永久取消;$12B+ 园区被放弃未公开宣布缓释措施;Tallgrass 能源伙伴仍在参与高 — 搁浅资本风险未解决
Meta 1.6 GW 容量协议Meta Platforms未指定园区的超大规模锚定客户高 — 已确认的最大单一多元化进展Meta 重新优先投入内部基础设施;协议未激活合同条款未公开;集中风险转移至 Meta中 — 即便分散 Microsoft/OpenAI 风险,也新增集中风险
NVIDIA 供应和股权合作NVIDIA CorporationGPU 供应、联合工程、云软件集成高 — NVIDIA 是战略伙伴和股权持有人NVIDIA 限制 Crusoe 获取 GPU,或偏向 CoreWeaveAMD MI355X 双供应商策略提供部分对冲中 — AMD ROCm 技术栈成熟较慢;短期内仍依赖 NVIDIA
融资伙伴:Blue Owl / Primary Digital Infrastructure JV($3.4B)Blue Owl Capital为 Abilene 最多 100,000 块 GPU 部署提供项目融资高 — 最大单一基础设施 JV利用率不佳时,Blue Owl 要求调整范围或退出Microsoft 长期锚定租约;合同条款保密中 — JV 结构部分隔离 Blue Owl 的资本承诺
Brookfield 信贷额度($750M)Brookfield Asset Management为 GPU 采购和数据中心建设提供债务融资对短期流动性影响高违反契约触发提前到期;高利率环境下再融资风险上升Crusoe 拥有多轮股权融资作为替代资金来源中 — 契约条款未公开;利率敞口未量化

依赖项严重性和残余风险敞口由公开报道推断;合同条款、契约和客户身份未完全披露。

[CR024, CR025, CR026, CR027, CR028]
FR003: 依赖地图 — 关键伙伴、监管方与融资

Crusoe 在电力、GPU、项目融资、云收入和监管合规上依赖的关键交易对手。

[CR024, CR025, CR026, CR027, CR028, CR029]

7.5 人员、执行与财务风险

Crusoe 2025 年员工数翻倍以上,达到 1,217 人,这是连续第二年翻倍。技术和建设管理岗位同时以这种速度超高速招聘,会制造组织风险:整合与文化压力、对联合创始人 Chase Lochmiller(CEO)和 Cully Cavness(COO)的关键人依赖,以及多个 GW 级项目同步建设下的执行质量风险。公司 2025 年债务融资约 $975 million,另有 Brookfield Asset Management 于 2025 年中安排的 $750 million 信贷额度。所有轮次合计融资约 $3.9 billion。这种资本深度对私营公司来说异常强,但债务服务义务叠加 GPU 采购和建设的激进资本开支,会形成高烧钱动态:客户激活或收入确认出现任何有意义延迟,都可能压迫资产负债表。 Abilene 全园区成本估算约 $12 billion(Blue Owl 新闻报道),意味着 Crusoe 的股权与债务合计覆盖不到已承诺建设价值的三分之一——剩余部分需要项目融资或额外股权。 最近 Wyoming 暂停说明,客户驱动的范围变化可能在收入锁定前就让沉没成本风险结晶。最后,2025 年把 Bitcoin mining 业务剥离给 NYDIG,移除了一个能产生现金、但周期性的收入支柱,而 AI 基础设施业务仍处在利用率爬坡期。 [CR031, CR032, CR033, CR034, CR035, CR036]

人员和执行风险登记表
角色 / 职能依赖或缺口发生概率严重性缓释措施尽调路径
CEO / 联合创始人 Chase Lochmiller愿景制定者和对外门面;重大战略和交易几乎都归因于他短期离职概率低关键 — 战略方向、投资者关系和交易撮合高度集中COO Cully Cavness 提供运营连续性;Series E 显示董事会有一定厚度继任计划;股权归属时间表;核验董事会构成
COO / 联合创始人 Cully Cavness日常运营和建设执行短期离职概率低高 — GW 规模建设项目管理依赖他的监督扩充高管团队;2026 年 1 月新增 Nader Pakfar 任房地产总法律顾问核验创始人之下 VP/Director 层的厚度,判断运营韧性
员工数快速增长(连续两次翻倍至 1,217 人)整合风险;文化稀释;建设项目管理招聘质量高 — 任何连续翻倍式招聘都会带来结构性风险高 — 在 GW 规模下,建设管理或云运营错误可能造成灾难性后果人均培训时长增至 53 小时;参与度评分 85%;结构化入职独立审查 Wyoming 和 Abilene 园区项目管理流程
能源 / 电力监管团队Texas SB6 下的新法律义务和持续变化的 ERCOT 规则要求专项能力中 — 监管复杂度上升高 — SB6 限电不合规或并网测算失误2025 Impact Report 称公司积极参与 PUCT 和公共政策沟通核验监管事务团队人数和深度;外部律师名单

风险发生概率和严重性依据公开披露及同阶段公司行业常态评估;个人离职概率带有推测性。

[CR031, CR032, CR033]

7.6 缓释措施、终止标准与尽调要求

Crusoe 已实施几项有意义的风险缓释。监管方面,剥离 Digital Flare Mitigation 业务,移除了对 BLM 和 EPA 甲烷执法周期的直接敞口; Crusoe Cloud 100% 可再生能源匹配(通过 VPPAs、energy attribute certificates,以及 Iceland/Norway 直接供电)部分回应了 ESG 关切。ISO 27001 + 42001 双认证(首批同时持有两项认证的 AI 云提供商之一)应对了信息安全和负责任 AI 治理风险。客户集中方面,2026 年 3 月宣布的模块化「Crusoe Spark」产品线面向较小 AI 运营商,在大型园区和模块单元之间形成杠铃策略。 通过 AMD MI355X 承诺实现 GPU 供应商多元化,对 NVIDIA 供应链事件提供部分对冲。尽管如此,若干风险仍未缓释或监控不足。 Crusoe/Chevron/Engine No. 1 天然气电厂规划的碳捕集技术,没有确认设计伙伴、没有已部署实例,却要在 2027 年交付电力——解决时间很紧。 锚定园区客户流失(Wyoming 暂停、Abilene 阶段变化)披露透明度不足,无法支持监控。油田计算 IP 图谱中,Upstream Data 创始人还持有至少 13 项专利,未被 2025 年 11 月和解覆盖。投资者应坚持获得签约利用率与已投运容量的季度报告,再把任何 Series E 后估值视为可支持。 [CR037, CR038, CR039, CR040, CR041, CR042]

缓释、否决标准和监控指标
风险簇可监控触发项阈值 / 事件行动含义
客户集中 / 园区暂停客户要求对 >500 MW 园区变更范围或暂停Wyoming 暂停后 12 个月内出现第二个大型园区暂停重新定价集中风险;要求提供完整客户管线、合同状态和客户名称
电力建设执行已宣布管线与已签约 / 已投运容量之间出现缺口管线 / 已签约比率低于 3x,或已签约 / 已投运比率低于 0.5x重新评估 CEO 的管线说法;要求独立建设进度审计
GPU 供应链AMD MI355X 交付延迟超过 Q4 2025 发布承诺,或 NVIDIA 关税升级 >25%任一事件发生量化 capex 上涨;核验算力容量延迟对客户 SLA 的影响
ESG / 监管矛盾新气候披露规则、诉讼,或 ESG 投资者因天然气数据中心退出至少一只主要可持续挂钩债券或 ESG 基金因环境理由退出重新评估可持续挂钩资本可得性,以及气候科技品牌是否站得住
财务 / 流动性在没有足够签约收入覆盖偿债的情况下,动用 Series E + Brookfield 额度的 >80%按未提取承诺口径计算的偿债覆盖率 <1.2x投资逻辑破裂 — 要求立即披露实际利用率与预测利用率
法律 / IP新专利侵权诉讼主张 Upstream Data 剩余 13 项创始人专利USPTO 文件或 EDTX 起诉书引用 Barbour 剩余 13 项专利中的任何一项评估 2025 年 11 月和解的许可范围;取得专利律师意见

阈值仅作示意,依据类似基础设施项目基准;实际触发值需要公司专属财务披露。

[CR037, CR038, CR039, CR040, CR041, CR042]

7.7 附录

Chapter 08

08估值

8.1 投资正反论点

Crusoe 的投资论点立在三根结构性支柱上。第一,AI 基础设施需求长期上行:超大规模云厂商和 AI 实验室都缺算力容量,而贴近电网电力、拥有专用 GPU 集群的独立云提供商最适合填补这一缺口。第二,Crusoe 的电力-算力一体化模型构成差异化——公司把发电(Abilene 的天然气轮机、规划中的可再生能源、储能合作)与 GPU 集群共址,边际电力成本更低,部署节奏也比传统托管机房到云迁移更快。第三,公司已拿到一线超大规模云厂商(Microsoft、Meta、xAI)的锚定关系和战略股权伙伴(NVIDIA),需求端和供应链接入都被部分去风险。三者共同支撑 Series E 的 $10B+ 估值。 反论点同样尖锐。客户集中风险很重:xAI 和 Meta 可能占已承诺容量的大部分。Oracle 和 OpenAI 退出 Abilene 后续阶段说明,即便大客户也可能比 Crusoe 重新部署资本更快地改计划。CEO Chase Lochmiller 公开承认,销售管线中相当一部分可能不会落地。Cheyenne Wyoming 1.8 GW Project Jade 在 2026 年 6 月按未具名客户要求暂停——这是集中风险实时变成现实的直接数据点。同时,Abilene 360 MW 天然气轮机建设制造了 ESG 身份矛盾;如果可持续主题 LP 退出,或新增贷款条件收紧,融资成本可能上升。估计约 $975M 的债务栈已约等于 FY2025 预计收入的 100%,限制自由现金流,也削弱公司承受收入缺口的韧性。 建议立场是有条件建设性:Series D $2.8B 入场的早期投资人仍然可看多;新资金若按 $10B+ 入场,必须严格核验合同条款、管线中具约束力的承诺和毛利率确认。Crusoe 相对 CoreWeave IPO 倍数的估值溢价,只有在 2025 收入假设和 2026 合同化爬坡被验证准确时才站得住。 [CV001, CV002, CV003, CV014, CV015, CV016]

建议摘要表
维度评估置信度关键证据
立场有条件投资(Series D),观望(Series E)CV001, CV002, CV025, CV028
估值锚点Series E 投后估值 ~$10B(2025 年 10 月);Series D ~$2.8B(2024 年 12 月)CV001, CV002
$10B 估值下的远期倍数约 10x FY2025E 收入(Sacra 估算 $998M);36x FY2024 收入($276M)CV006, CV007, CV008
CoreWeave 基准CoreWeave IPO 估值 $23B / 约 12x 2024 收入;按远期口径,Crusoe 落在区间内CV004, CV005, CV025
风险评级高 — 客户集中、建设执行、债务 / 收入比CV018, CV044, CV045
退出窗口2027–2028 年 IPO 目标或战略 M&A;二级市场流动性不足CV030

评估是基于公开来源的分析师判断;无法取得管理层预测或合同条款。

[CV001, CV002, CV004, CV005, CV006, CV007]
投资逻辑 / 反逻辑表
维度投资逻辑(牛市)反逻辑(熊市)
市场需求AI 推理 / 训练算力缺口延续至 2028+;Crusoe 拿下独立云 TAM 的 5–10%超大规模云厂商自建速度快于市场共识,降低对独立供应商的依赖;AI capex 进入寒冬情景
产品 / 技术电力与算力一体化带来 30–81% 价格优势;Crusoe Spark 支持快速模块化扩张AMD ROCm 成熟度缺口限制可服务的推理用例;CUDA 生态锁定让 CoreWeave/Lambda 占优
客户xAI + Meta + Microsoft 多年 take-or-pay 合同锚定已签约 4.9 GW 的 80%Oracle/OpenAI 退出释放客户波动信号;xAI 或 Meta 转向大规模自建
财务2025 年收入接近 $1B,EBITDA 利润率 50%+;偿债加速燃气轮机 + SB6 成本将毛利率压到 40% 以下;债务 / 收入 >1x 限制 FCF
竞争靠近电力的 GPU 集群具有先发规模优势;NVIDIA 合作带来供应优先级CoreWeave、Nebius 和垂直整合的超大规模云厂商在 12–18 个月内追平定价
ESG / 风险DFM 剥离和碳捕集 JV 修复 ESG 资质;守住可持续挂钩融资渠道Abilene 360 MW 燃气轮机,以及与 Engine No. 1 的天然气 JV,将 Crusoe 重新归类为传统 AI 基础设施; ESG 融资溢价被侵蚀

投资逻辑和反逻辑是分析情景,不是管理层指引。概率权重:牛市 25%,基准 50%,熊市 25%。

[CV009, CV015, CV016, CV023, CV024, CV026]
FV001: 建议逻辑

投资决策流:结构性支撑与约束导向有条件 INVEST 建议。

[CV001, CV002, CV009, CV010, CV026, CV027]

8.2 融资背景与当前估值标记

Crusoe 迄今通过股权和结构化融资累计募资约 $3.9B。最近一轮定价股权是 2025 年 10 月 Series E:募资 $1.375B,投后估值超过 $10B,Valor Equity Partners 和 Mubadala Investment Company 等领投。此前是 2024 年 12 月 Series D:募资 $600M,投后估值 $2.8B,由 Founders Fund 领投,NVIDIA 等参与。12 个月内从 $2.8B 抬到 $10B+(3.6x),与期间完成的执行里程碑一致:Abilene 上线、4.9 GW 签约、Meta 1.6 GW 容量协议宣布。 债务端,Crusoe 从 Brookfield Asset Management 获得 $500M 信贷额度,用于电力基础设施开发;又与 Blue Owl Capital 建立 $450M 合资安排,用于 Abilene 园区二期。Reuters 2025 年 6 月报道称 NVIDIA 参与一轮 $750M 融资,这大概率指 Series D 和 E 分批加结构化债务的组合,而不是另一轮单独定价融资。包含债务后的总资本估计约 $5.5B。 收入轨迹是基本估值锚。Sacra Research 估计 FY2024 收入约 $276M,对应 $10B 企业价值 / 收入约 36x(过去十二个月口径)——显著高于 CoreWeave IPO 约 12x 的倍数。不过 Crusoe 的增长曲线正在急速抬升:Sacra 预测 FY2025 收入约 $998M,按未来收入口径 EV / revenue 降至约 10x——大体接近 CoreWeave IPO。只有已签约 4.9 GW 按期转化为收入,且 xAI / Meta 关系贯穿 2025 年保持稳定,这一未来口径倍数才能实现。本章始终将已观察事实和模型预测分开;任何来自第三方估计的收入或估值数字均会明确标注。 [CV001, CV002, CV003, CV004, CV005, CV006]

牛市 / 基准 / 熊市情景表
情景FY2026 收入EV/收入倍数隐含 EVSeries D MOIC(2.8B 入场)Series E MOIC(10B 入场)
牛市(25%)~$1.7B13–15x 远期$22–26B7.9–9.3x2.2–2.6x
基准(50%)~$1.1B8–12x 远期$8–13B2.9–4.6x0.8–1.3x
熊市(25%)~$500M5–7x 远期$2.5–3.5B0.9–1.3x0.25–0.35x
概率加权~$1.1B(预期)~9–11x~$10–14B~3.5–5.0x~1.0–1.4x

收入和倍数为分析师情景。MOIC 假设 2027–2028 年退出。未建模分配或优先权。

[CV026, CV027, CV028, CV030, CV044, CV045]
FV002: 估值敏感性

在五档 EV/收入倍数(行)× 五档 FY2026E 收入情景(列)下推算 EV。CoreWeave IPO 约 12x。

FY2026E 收入情景为分析师估算。EV 以十亿美元(USD)计。

[CV004, CV007, CV008, CV025, CV026, CV027]

8.3 可比估值与同业样本

最相关的上市可比公司是 CoreWeave,2025 年 3 月 26 日 IPO 时企业价值 $23B。CoreWeave 向 SEC 提交的 S-1 披露 FY2024 收入 $1.92B(同比增长 737%),已承诺收入储备为 $8.7B。IPO 隐含约 12x 过去十二个月收入和 4–5x 未来收入。CoreWeave 上市后二级市场市值一度低于 IPO 水平,反映市场对 AI 基础设施倍数和高客户集中度(申报时 Microsoft 约占收入 60%)的不确定。 私有可比更不透明。Lambda Labs 是竞争性的推理优化云提供商,分析师报道其二级交易约 4–7x ARR。Scale AI 是相邻 AI 基础设施公司,聚焦数据标注和模型评估,2025 年初在约 $1B ARR 上取得约 $14B 私有估值——约 14x ARR——但商业模式明显不同。Anthropic 是有大量基础设施的基础模型公司,据报以约 $3B ARR 支撑 ~$18–25B 估值,但其垂直整合和护城河属性使其不可比。 本分析将 Crusoe 的主要同业界定为 IaaS 型 GPU 云公司。代表样本包括 CoreWeave(上市)、Lambda Labs(私有)、Voltage Park(私有、规模更小)和 Nebius(原 Yandex Cloud,在 Amsterdam 上市)。下表列出这一可比样本。所有私有公司数字均为公开来源中的分析师估计,可能不同于未披露的实际财务;投资人应把它们当作方向性信号,而非精确数字。 [CV004, CV005, CV006, CV007, CV008, CV023]

可比估值表
公司状态收入 / ARR(最新披露)EV / 收入倍数隐含 EV相对 Crusoe 的关键差异
CoreWeave上市(2025 年 3 月 IPO)$1.92B FY2024(+737% YoY)约 12x 2024 往绩IPO 时 $23B单租户裸金属 GPU;MSFT 收入集中度约 60%;无发电资产所有权
Lambda Labs私有(二级市场)~$400–600M ARR(估算)约 4–7x ARR(估算)~$2–4B(估算)推理优化;规模更小;未确认超大规模锚定合同
Voltage Park私有~$100–200M ARR(估算)约 4–6x ARR(估算)~$0.5–1B(估算)较小规模 GPU 云;位于 Utah;无发电集成
Scale AI私有(Series F)~$1B ARR(数据标注 + 模型评估)约 13–15x ARR(估算)~$13–15B聚焦数据 / 标注;AI 基础设施位于上游;不是直接可比公司
Nebius Group上市(AMS)~€200–300M ARR(估算)约 5–8x ARR(估算)~€1.5–2.4B欧洲 GPU 云;前身为 Yandex Cloud NV;俄罗斯背景带来客户风险折价

私有公司数据来自 Sacra、Cheddar Flow 和公开来源中的分析师共识估算。SEC S-1 数据仅适用于 CoreWeave。不要依赖私有公司数据做投资决策。

[CV004, CV005, CV007, CV008, CV023, CV024]

8.4 乐观、基准与悲观情景

乐观情景假设 Abilene 到 2026 年中完全爬升至 1.2 GW,Project Jade(Wyoming)恢复并在 2026 年 H2 交付初始容量,xAI 和 Meta 维持多年容量承诺,Crusoe Spark 模块化部署加速来自中小企业客户的收入。在这些假设下,FY2026 收入可达 $1.5–2.0B,按 10–13x 未来口径倍数 隐含 EV 为 $15–25B。Series D 投资人($2.8B 入场)可实现 5.4–8.9x MOIC;Series E 投资人($10B 入场)可实现 1.5–2.5x。该情景要求不再出现重大客户集中事件,ESG 重新定位成功,且 Texas 电网政策继续有利。 基准情景假设 Project Jade 恢复在建但延迟 12 个月,Abilene 的 Oracle 和 OpenAI 容量被 Microsoft 及新企业客户替代但利用率降低,Crusoe Spark 增量增加 200–300 MW。FY2026 收入约 $900M–$1.2B,按 8–12x EV/revenue 隐含 $7–14B 企业价值。Series D 投资人实现 2.5–5x MOIC;Series E 投资人在 2–3 年持有期内获得温和的 0.7–1.4x 回报。 悲观情景假设其他园区至少再出现两次 Wyoming 式暂停,xAI 或 Meta 因转向自建或竞争替代方案而将合同承诺削减 30–50%,电力成本上升(天然气轮机 + SB6 限电成本)使毛利率压到 40% 以下,且 2028 年前没有 IPO 窗口。FY2026 收入 $400–600M、EV/revenue 6–8x 时,隐含 EV 为 $2.4–4.8B——低于 Series E 投后估值,对 Series E 投资人意味着 0.24–0.48x MOIC,对 Series D 为 0.86–1.71x。 截至 2026 年 6 月,基于可得证据,三种情景概率权重为:乐观 25%、基准 50%、悲观 25%。概率加权 EV 约 $9.6–13.5B。该区间略高于当前 $10B+ 标记,说明 Series E 估值大体有效,但晚期入场者下行保护有限。 [CV026, CV027, CV028, CV029, CV030, CV031]

FV003: 估值 / 回报区间

悲观、基准、乐观、概率加权和 CoreWeave IPO 锚定情景下的企业价值(十亿美元,USD)。

区间仅代表分析师情景估算;Crusoe 没有可用的审计财务数据。列出 CoreWeave 是为了相对校准。

[CV004, CV025, CV026, CV027, CV028, CV030]

8.5 建议、置信度与信念框架

立场:有条件 INVEST,且需完成重要尽调前置项。置信度:中。风险评级:高。本建议适用于按 $10B+ Series E 标记进入的新资金。较早按 $2.8B Series D 标记进入的投资人已有可观账面增值,应谨慎建模退出时点。 信念框架有三道门。Gate 1(客户合同):投资人必须拿到前三大客户合同的脱敏版本,确认最低 take-or-pay 金额、取消通知期和罚则。Oracle / OpenAI 短通知退出 Abilene 后续阶段——据称已由 Microsoft 替代——说明非正式承诺执行风险很高。Gate 2(毛利率):投资人必须按客户、按园区确认毛利率路径高于 45%。Crusoe 的天然气轮机和 SB6 限电敞口带来电力成本不确定性,收入端预测未反映这一点。Gate 3(建设控制):投资人必须拿到至少两个非 Abilene 园区的详细项目管理计划,显示带里程碑和罚则的约束性客户承诺,并将其绑定到建设进度,以避免 Cheyenne 暂停重演。 否决触发条件是任何出现即无视当前标记、把建议转为 PASS 的事件:(1) 失去 xAI 或 Meta,且该客户代表 >30% 已签约容量;(2) 十二个月内第二次出现 Wyoming 式客户驱动暂停;(3) Brookfield 信贷额度触发债务契约违约;(4) 发现 DFM 剥离让 Crusoe 背负重大甲烷合规赔偿责任。 从退出准备度看,Crusoe 还没具备 IPO 条件,但如果收入超过 $1B 且 EBITDA 利润率扩张,2026–2027 年有望达到 IPO 门槛。来自超大规模云厂商(Microsoft、Meta、Google)的战略并购是可行替代退出,CoreWeave 的客户关系提供了先例。截至本文写作时,二级交易定价尚未独立确认。 [CV009, CV010, CV015, CV017, CV018, CV019]

逻辑破裂和否决触发表
触发项领先指标阈值投资动作
客户集中风险落地客户正式暂停或减少已签约容量>30% 已签约 GW 受单一客户动作影响立即 PASS;若 90 天内锁定替代客户,再重新评估
债务契约违约Brookfield 额度季度契约报告显示违约或豁免请求任何违约或豁免请求在契约解决细节明确前立即 PASS
≥2 个站点建设延误两个或更多合计超过 1.2 GW 的园区报告暂停或延误 >6 个月≥2 个园区;延误 >6 个月降低信心;要求项目管理整改计划
ESG 融资成本上升可持续挂钩信贷利差扩大 >100bps,或 ESG 挂钩分层被赎回>100bps 绿色融资利差扩大监控 WACC 影响;重新归类为传统基础设施投资
GPU 供应中断AMD MI355X 或 NVIDIA H200/H100 配额较既定计划削减 >20%>20% 配额削减或 >90 天交付延迟暂停;核验管线利用计划
IPO 窗口关闭CoreWeave 连续 ≥6 个月交易在 <8x 收入;可比 AI 基础设施 IPO 撤回CoreWeave EV/收入持续 <8x延长持有期;重新评估二级退出选项

触发阈值是投资组合监控的定性指引,不是合同承诺。投资者必须核验实际合同条款。

[CV009, CV010, CV011, CV012, CV015, CV016]
FV004: 投资 KPI

Crusoe Energy 截至 2026 年 6 月的关键投资指标,以 CoreWeave IPO 为基准。

收入数字来自第三方分析师估算(Sacra)。已签约容量和管线来自公司公告。

[CV001, CV002, CV003, CV004, CV006, CV007]

8.6 最终尽调清单

以下尽调项目是按 $10B 或更高标记投资的前置条件。项目按重要性排序:高优先级事项围绕客户集中和财务核验;中优先级覆盖法律、技术和运营尽调;低优先级为信息性、确认性事项。所有项目都应在结构化 资料室流程中审阅,并取得经审计财务、已签署客户合同和购电协议。投资人承诺资本前,应要求提供 xAI 已承诺容量的当前运行水平、按产品线拆分的毛利率,以及 DFM 油田业务剥离当前状态的证据。本尽调清单只是对本分析识别出的最重大缺口的部分列举;无论二级还是一级市场价格,作出任何投资决策前仍需完成完整商业、法律、技术和财务尽调。 [CV043, CV044, CV045]

最终尽调要求表
尽调事项理由优先级数据室要求
前三大客户合同审阅确认 xAI、Meta 和 Microsoft 的 take-or-pay 金额、取消通知期和罚则按已签约 MW 排名前三客户的脱敏 MSA + 订单表
收入和 ARR 核验Sacra 对 FY2024 $276M 和 FY2025 $998M 的预测未经确认,只是分析师估算;需要实际披露或经审计收入经审计或审阅的 FY2024 财务报表;FY2025 YTD 管理账
债务契约和条款披露Brookfield $500M 信贷额度和 Blue Owl JV $450M 条款未公开;需要 DSCR、契约测试和违约事件条款Brookfield 额度协议;Blue Owl JV 运营协议
DFM 剥离赔偿范围DFM 出售中的买方身份、赔偿范围以及环境 / 监管陈述未披露资产购买协议(关键陈述与保证);买方财务健康披露
园区层面建设进度Cheyenne Wyoming 暂停和 Abilene 分阶段时间线需要独立核验;所有在建园区的项目管理控制每个 >100 MW 园区的建设时间表 + 里程碑;Wyoming 客户重新承诺证明
分业务毛利率电力成本(燃气轮机摊销 + 燃料 + SB6 限电成本)与云收入利润率未公开披露;估值模型需要这项数据2024 年下半年和 FY2025 按园区拆分的分部 P&L 或贡献利润率

该清单并不完整。除上述项目外,投资人还必须完成完整的商业、技术、法律和财务尽调。

[CV001, CV002, CV006, CV007, CV010, CV011]

8.7 附录

免责声明

本报告是基于公开证据的尽调快照,不构成投资建议。重要的财务、法律、技术和合同事实仍未公开;做出任何投资决定前,应直接向管理层和一手文件核验。

证据索引

结论
编号陈述可信度来源
CO001 Crusoe Energy Systems was founded in 2018 and is headquartered in Denver, Colorado. SO001, SO006
CO002 Crusoe was co-founded by Chase Lochmiller (CEO) and Cully Cavness (President and COO). SO001, SO006
CO003 Crusoe originally deployed modular data centers at oil and gas wellheads to capture stranded flared gas and power on-site computing, a technology it trademarked as Digital Flare Mitigation® (DFM). SO003, SO007
CO004 Crusoe's original computing workload was Bitcoin and cryptocurrency mining using captured flared gas energy. SO009, SO014
CO005 Crusoe's DFM technology achieves approximately 99.89% combustion efficiency, compared to roughly 93% for standard oil field flares, reducing CO2-equivalent emissions by about 63% relative to continued flaring. SO007, SO005
CO006 In 2023, Crusoe pivoted from cryptocurrency mining to AI cloud infrastructure, borrowing $200 million to purchase NVIDIA H100 GPUs and launched Crusoe Cloud. SO010, SO014
CO007 Crusoe's stated mission is to 'accelerate the abundance of energy and intelligence' as the AI factory company. SO002, SO016
CO008 Crusoe Energy Systems is headquartered in Denver, Colorado, with manufacturing facilities also in Tulsa, Oklahoma. SO010, SO001
CO009 Crusoe manufactures server racks and installs GPUs at facilities in Denver and Tulsa before trucking to Abilene and other data center sites for installation. SO010
CO010 Chase Lochmiller holds degrees in mathematics and physics from MIT and a master's degree in computer science with an AI specialization from Stanford. SO007, SO011
CO011 Before co-founding Crusoe, Chase Lochmiller worked as a quantitative trader and was a general partner at Polychain Capital, a cryptocurrency investment firm. SO007, SO011
CO012 Cully Cavness has a geology degree from Middlebury College and an MBA from Oxford University. SO007, SO011
CO013 Cully Cavness has a background in oil and gas investment banking, giving Crusoe deep upstream energy sector expertise and regulatory relationships. SO007
CO014 Crusoe closed a $600 million Series D funding round at a $2.8 billion valuation on December 12, 2024. SO001, SO006, SO004
CO015 Founders Fund led Crusoe's Series D round; Sean Liu is the Partner at Founders Fund involved. SO001, SO006
CO016 Series D investors alongside Founders Fund included NVIDIA, Fidelity, Long Journey Ventures, Mubadala, Ribbit Capital, and Valor Equity Partners. SO001, SO006, SO004
CO017 Peter Thiel, Partner at Founders Fund, said at the Series D announcement: 'The biggest risk with AI is that we don't go big enough. Crusoe is here to liberate us from the island of limited ambition.' SO006
CO018 Crusoe closed a $1.375 billion Series E funding round at a valuation exceeding $10 billion in October 2025. SO014, SO015
CO019 Valor Equity Partners and Mubadala Capital co-led Crusoe's oversubscribed Series E round. SO014, SO015
CO020 Series E investors included NVIDIA, Fidelity, Founders Fund, T. Rowe Price, Tiger Global Management, Altimeter Capital, Franklin Templeton, Salesforce Ventures, Lowercarbon Capital, Spark Capital, and more than 20 firms in total. SO014, SO015
CO021 Crusoe has raised approximately $3.9 billion across approximately 13 rounds of debt and equity financing as of the Series E close. SO014, SO015
CO022 Blue Owl Capital helped arrange $2.3 billion in project financing from JPMorgan for the first phases of the Abilene campus. SO010
CO023 Crusoe operates across nine U.S. states and three countries including Iceland as of the Series D announcement. SO001, SO004
CO024 Crusoe Cloud operates four cloud regions: us-northcentral1-a, us-east1-a, us-southcentral1-a, and eu-iceland1-a. SO014, SO020
CO025 Crusoe's customer count grew more than 7x in the 12 months prior to December 2024. SO001, SO004
CO026 Crusoe has over 15 gigawatts of clean energy projects in development as of the December 2024 Series D announcement. SO001, SO004
CO027 Crusoe averted over 680,000 metric tons of GHG emissions through its Digital Flare Mitigation technology as reported in its ESG disclosure. SO005
CO028 Crusoe is developing a 1.2 GW AI campus in Abilene, Texas, anchored by a long-term Oracle lease for the Stargate OpenAI project — one of the largest planned high-performance computing clusters in the world. SO001, SO010, SO009
CO029 Elon Musk's xAI was reportedly the first potential tenant for the Abilene site, but Musk decided to build his own megacluster (Colossus) in Memphis independently, after which Oracle-OpenAI took the site. SO009, SO010
CO030 The first phase of the Abilene 1.2 GW campus went live in September 2025, one of the fastest greenfield hyperscale data center builds on record. SO014, SO010
CO031 The Abilene campus is projected to add $1 billion to the local economy over 20 years and create more than 800 jobs. SO001, SO004
CO032 At NVIDIA GTC 2026 in March 2026, Crusoe announced early adopter status for NVIDIA Vera CPU and deepened strategic collaboration spanning models, inference, and physical infrastructure. SO002, SO009
CO033 Crusoe is integrating its proprietary Rust BPE tokenizer with NVIDIA Dynamo open-source inference serving library and contributing it to the community; the tokenizer delivers up to 31x speedup on long prompts. SO002
CO034 Crusoe's Managed Inference Service achieves up to 9.9x faster time-to-first-token and 5x higher throughput than standard inference engines through its MemoryAlloy technology. SO002, SO016
CO035 Crusoe launched Crusoe Edge Zones on March 12, 2026, offering modular, factory-built AI clusters for sovereign AI deployments, low-latency inference, and dedicated enterprise use cases. SO016, SO002
CO036 Crusoe launched Crusoe Command Center in February 2026 as a unified operations platform for high-performance AI workloads. SO017
CO037 Crusoe achieved ISO 27001 (information security) and ISO 42001 (responsible AI governance) certifications in February 2026. SO017, SO002
CO038 Crusoe's MemoryAlloy technology is described as a cluster-wide KV cache fabric that enables the inference speed improvements claimed in Crusoe Managed Inference. SO016, SO002
CO039 In early March 2025, one of Crusoe's data centers experienced an outage lasting 45 hours that affected at least one enterprise customer. SO010, SO013
CO040 An unnamed Crusoe Cloud customer told Forbes following the March 2025 outage: 'This was a big deal. We were pretty unhappy about it. They're not as reliable.' SO013, SO010
CO041 An unnamed industry expert told Forbes: 'There's less differentiation in Crusoe than they would want you to believe,' though the expert acknowledged current demand means 'anybody that can provide power on site' has an advantage. SO010, SO013
CO042 In May 2026, Crusoe experienced a networking incident in its us-east1-a cloud region, making a subset of compute hosts unreachable; engineering identified a probable cause and reached partial restoration within hours. SO012
CO043 Bill Stein, former CEO of Digital Realty, joined Crusoe's board of advisors alongside Peter Gross and Matt Field. SO001, SO004
CO044 Nadav Eiron was appointed Senior Vice President of Cloud Engineering at Crusoe to scale its Cloud platform. SO001
CO045 Nader Pakfar, founder of SPC LLP, was appointed General Counsel for Real Estate at Crusoe in January 2026. SO017
CO046 Crusoe Cloud customers include Decart (scaled capacity 5x to serve 2 million users in four days) and Codeium/Windsurf (700,000+ developers, 99.98% cluster uptime). SO001, SO004
CO047 Crusoe has divested its cryptocurrency mining operations, completing its full pivot to AI infrastructure and cloud services. SO014, SO009
CO048 Crusoe CEO Chase Lochmiller stated the company will eventually go public but has not committed to a specific IPO timeline. SO018
CO049 Crusoe's revenue run-rate is approaching approximately $1 billion as of mid-2026, according to analyst estimates; the company has not disclosed audited revenue figures. SO018, SO011
CO050 Crusoe's headcount and board composition details are not publicly disclosed; the organizational scale and governance structure are diligence gaps.
CM001 Crusoe Energy operates at the intersection of three distinct market segments: neocloud GPU cloud, AI factory development and leasing, and alternative energy procurement for computing. SM024, SM023
CM002 Neoclouds are purpose-built GPU cloud providers — CoreWeave, Lambda Labs, Nebius, Crusoe, and peers — that differ from hyperscalers by offering tightly focused GPU-accelerated compute at significantly lower cost without the full cloud service stack. SM023, SM008
CM003 The AI compute infrastructure market boundary includes GPU-accelerated IaaS, bare-metal HPC, and AI factory co-location, and excludes general-purpose CPU cloud, AI software layers, and ASIC design. SM020, SM021
CM004 GPU-as-a-Service (GPUaaS) is a distinct and rapidly growing subsegment valued at $3.23 billion in 2023, projected to grow at approximately 36% compound annual growth rate. SM005
CM005 Energy infrastructure for computing — on-site power generation, alternative energy procurement, behind-the-meter assets for AI data centers — is evolving from a cost line item into a multi-hundred-billion-dollar strategic investment category by 2030. SM003, SM004
CM006 Status-quo substitutes for neocloud GPU cloud include hyperscaler spot instances, on-premises GPU clusters, and bare-metal HPC at universities, all of which carry significant cost or capacity limitations relative to dedicated neoclouds. SM009, SM012
CM007 S&P Global Market Intelligence forecasts the AI infrastructure market at $337 billion in 2025, growing to $1.2 trillion by 2030 at a 28% compound annual growth rate. SM020, SM001
CM008 IDC projects global AI infrastructure spending will reach $487 billion in 2026, a 53% year-over-year increase from 2025 levels. SM021, SM020
CM009 ARK Investment Management forecasts total AI infrastructure spending to grow from approximately $500 billion in 2025 to nearly $1.5 trillion by 2030, driven by AI agent proliferation and enterprise adoption. SM001, SM020
CM010 S&P Global identifies AI-as-a-Cloud-Service (ACaaS) as the fastest-growing AI infrastructure subsegment, growing from $20 billion in 2025 to $145 billion by 2030 at a 48% CAGR. SM020
CM011 Gartner forecasts AI-optimized infrastructure-as-a-service revenue at $18.3 billion in 2025, growing 105% year-over-year to $37.5 billion in 2026 — the most specific proxy for the neocloud and GPU cloud segment. SM019, SM021
CM012 The neocloud segment is estimated at approximately $23 billion in 2025, with Synergy Research Group projecting $180 billion by 2030 and IndexBox projecting $236 billion by 2031 at a 46% CAGR. SM008, SM016
CM013 Multiple analyst forecasts for 2026 AI infrastructure spending diverge significantly — from Gartner's $37.5 billion (AI-optimized IaaS) to IDC's $487 billion (all AI infrastructure) — reflecting inconsistent scope definitions that make direct comparison misleading. SM019, SM021, SM020
CM014 McKinsey estimates $6.7 trillion in cumulative global data center infrastructure investment through 2030, with power access identified as the primary bottleneck to realizing this investment. SM003, SM010
CM015 Goldman Sachs projects U.S. data center power demand will grow at a 15% compound annual growth rate through 2030, with data centers forecast to consume 8% of total U.S. electricity by 2030. SM003, SM025
CM016 Forrester Principal Analyst Lee Sustar projected neoclouds including CoreWeave, Lambda Labs, and Nebius would take in $20 billion in revenue in 2026, citing 205% year-over-year growth in Q2 2025. SM017, SM008
CM017 U.S. data center power demand is forecast to double from approximately 31 GW in 2025 to 66 GW by 2027, driven by AI training and inference workload growth. SM003, SM025
CM018 CoreWeave reported $5 billion in full-year 2025 revenue (737% year-over-year growth from 2024) after going public at $35 billion valuation on March 28, 2025 at $40 per share. SM007, SM017
CM019 AI cloud buyers segment into five distinct lanes: hyperscaler anchor tenants, AI foundation model labs, AI-native startups, enterprise IT buyers deploying GenAI, and sovereign AI programs — each with distinct budget ownership and procurement models. SM009, SM012, SM024
CM020 Hyperscaler anchor tenants including Microsoft, Meta, Amazon, and Google sign multi-billion dollar capacity commitments with neocloud providers, often reserving capacity years in advance under long-term supply agreements. SM017, SM009
CM021 CoreWeave's contracted backlog reached $66 to $99.4 billion by mid-2026, primarily from Microsoft, Meta, and AI foundation model lab anchor contracts, providing multi-year revenue visibility. SM009, SM016
CM022 AI-native startups typically spend $5,000 to $500,000 per month on GPU cloud; mid-sized production teams average $50,000 to $500,000 per month, with serious training runs easily exceeding $100,000 monthly. SM009, SM012
CM023 Enterprise IT buyers in regulated industries require compliance certifications (SOC 2, HIPAA, FedRAMP) for production AI workloads, which currently advantage hyperscalers over neoclouds for regulated enterprise procurement. SM009, SM012
CM024 The U.S. accounts for approximately 70% of global neocloud revenues as of 2025, with APAC and Europe expected to grow in relative importance as AI infrastructure investment globalizes. SM008
CM025 Sovereign AI programs at the national government level represent a fast-emerging buyer segment, with data residency and local control requirements creating demand for localized AI infrastructure distinct from hyperscaler offerings. SM004, SM009
CM026 Microsoft and Meta signed approximately $46 billion in AI cloud capacity commitments with Nebius under European GDPR data-residency terms, illustrating the scale and requirements of sovereign AI procurement. SM009
CM027 Even hyperscalers outsource significant GPU capacity to neoclouds rather than building exclusively in-house: CoreWeave reported $6.5 billion in OpenAI and Microsoft orders for GPU compute in 2025. SM017, SM007
CM028 By late 2025, AI inference workloads surpassed training workloads in total GPU compute consumed, driven by the rollout of AI agents in enterprise software, coding tools, and consumer applications. SM001, SM009
CM029 Enterprise AI adoption reached a pilot-to-production inflection in 2025–2026, with approximately 50% of enterprises running active AI pilots and roughly 25% reaching production deployment. SM022, SM009
CM030 Power availability has displaced GPU hardware access as the primary competitive constraint in the AI data center market as of 2025–2026, with operators who secured capacity in 2023–2024 holding a structural multi-year moat. SM002, SM004, SM025
CM031 On-demand H100 GPUs cost approximately $2.49 to $4.25 per GPU-hour on neoclouds (Lambda Labs, CoreWeave) versus $6.88 per GPU-hour on AWS and $12.29 per GPU-hour on Azure as of Q1 2026, representing a 2–5x cost differential. SM007, SM009, SM012
CM032 Next-generation B200 GPUs show an even wider hyperscaler-neocloud gap: AWS B200 instances cost approximately $14.24 per GPU-hour versus $4.99 to $5.29 per GPU-hour at Lambda Labs, a nearly 3x differential. SM009, SM007
CM033 NVIDIA reported $57 billion in revenue for the three months ending October 2025, a 62% year-over-year increase, with CEO Jensen Huang noting that 'useful AI agents have just begun to roll out' and are 'token hungry.' SM017, SM001
CM034 National AI policy mandates and government-funded sovereign AI infrastructure programs are expanding the total addressable market for AI compute globally beyond the commercial enterprise segment. SM004, SM009
CM035 Agentic AI systems require orders-of-magnitude more inference compute per task than single-turn models, creating durable exponential demand growth for GPU cloud capacity from 2026 onward. SM001, SM009
CM036 The 2–6x GPU cost arbitrage of neoclouds versus hyperscalers makes the economic case for migrating AI training and inference workloads off AWS and Azure compelling at scale, effectively expanding the neocloud TAM. SM009, SM023, SM024
CM037 New data center projects entering grid interconnection queues in 2025–2026 face typical waiting periods of 4 to 10 years in major U.S. markets, creating durable structural barriers for new entrants. SM002, SM025, SM018
CM038 Data center operators who secured power and grid access in 2023–2024 have a structural multi-year competitive advantage over new entrants entering interconnection queues in 2025–2026. SM002, SM004, SM025
CM039 Northern Virginia, the largest U.S. data center market, faces severe power constraints, with utilities warning of regional shortages and forcing new data center projects to secondary markets. SM003, SM010
CM040 The capital intensity of large AI data center development exceeds $1 billion per campus for gigawatt-scale deployments, limiting the effective competitive set to well-capitalized players. SM002, SM018
CM041 Customer switching costs in GPU cloud include data migration, application refactoring, API changes, and compliance re-certification, creating moderate lock-in for enterprise production workloads. SM009, SM012
CM042 NVIDIA commands approximately 95% of the market for AI training accelerators, creating a single-vendor supply chain concentration risk for all neocloud providers dependent on GPU availability. SM001, SM007
CM043 Neocloud providers face structural reliability challenges relative to hyperscalers: enterprise buyers expect 99.99%+ uptime SLAs that early neoclouds have struggled to consistently deliver, creating hesitancy for production workloads. SM009, SM012
CM044 Behind-the-meter power solutions — fuel cells from Bloom Energy, solar-plus-storage, next-generation nuclear microreactors — are becoming standard for AI data center operators who cannot wait for grid interconnection queues. SM004, SM018
CM045 Key competitors in the neocloud segment include CoreWeave (IPO March 2025, $5B 2025 revenue), Lambda Labs, Nebius ($3B+ 2026 revenue guidance, 684% YoY growth in Q1 2026), Voltage Park, Fluidstack, Applied Digital, Core Scientific, and Together AI. SM007, SM008, SM009
CP001 Crusoe Cloud lists H100 SXM GPU instances at $4.29/GPU-hour as of 2026. SP002, SP015
CP002 CoreWeave listed on NASDAQ in March 2025, the first major neocloud GPU cloud operator IPO. SP003, SP018
CP003 Lambda Labs offers H100 SXM5 GPU cloud instances with on-demand, no-minimum pricing targeting developers and researchers. SP005
CP004 Nebius AI (NYSE: NBIS) was spun off from Yandex in June 2024 and operates GPU cloud infrastructure in EU and US markets. SP006, SP007
CP005 Vast.ai operates a spot GPU marketplace scalable to 20,000 GPUs on demand, aggregating capacity from distributed independent providers. SP008
CP006 Lancium operates ERCOT-approved clean campuses in Texas for gigawatt-scale grid-connected power, targeting AI data center operators. SP010
CP007 AWS P5 instances offer 8×H100 SXM GPUs with 3,200 Gbps EFA networking and SageMaker HyperPod managed ML cluster support. SP011, SP012
CP008 Azure NDv5 GPU VMs provide H100 SXM for distributed AI training, with Azure Machine Learning compute cluster integration. SP013
CP009 Google Cloud offers A100 and H100 GPU instances with committed use discount programs as part of GCP AI infrastructure. SP014
CP010 Oracle OCI SuperCluster delivers 400 H100 GPUs in a single bare-metal cluster with RDMA networking for large-scale AI training. SP009
CP011 CoreWeave holds NVIDIA preferred partner status with priority allocation of H100, H200, and Blackwell (B200/GB200) GPUs. SP003, SP018
CP012 Nebius AI raised $700M in September 2024 from investors including NVIDIA, targeting the EU AI cloud market. SP006, SP018
CP013 Lambda Labs does not own data center power generation or build campus-scale infrastructure, relying on third-party colocation. SP005, SP015
CP014 Lancium does not offer a general-purpose GPU cloud product; its focus is on HPC batch compute with grid-responsive energy management. SP010
CP015 Crusoe's NVIDIA strategic collaboration provides early access to NVIDIA Nemotron models and integration with NVIDIA Dynamo inference framework. SP021, SP020
CP016 Crusoe achieved ISO 27001 (information security management) and ISO 42001 (AI management) certifications as of 2026. SP022, SP001
CP017 Crusoe Cloud operates four cloud regions: us-northcentral1-a, us-east1-a, us-southcentral1-a, and eu-iceland1-a. SP001, SP024
CP018 Crusoe's Spark modular AI factory product manufactures prefab data center units at its Brighton, Colorado facility with first commercial deliveries expected Q3 2026. SP023, SP032
CP019 Nebius AI claims 112% better TCO for inference compared to AWS as of its 2026 pricing page. SP006, SP007
CP020 AWS P5 on-demand pricing is substantially higher per GPU-hour than neocloud operators, reflecting managed service premium and broader ecosystem integration. SP012, SP015
CP021 CoreWeave Kubernetes Service control plane costs $1,500 per month, reflecting a managed platform premium over raw GPU rental. SP004
CP022 Azure Machine Learning compute clusters support NDv5 GPU VMs with H100 for distributed training; pricing is per-VM-hour including Azure ML service overhead. SP013
CP023 Google Cloud offers 1-year and 3-year flexible committed use discounts (CUDs) on GPU instances, reducing effective pricing below on-demand rates. SP014
CP024 Oracle OCI Stargate-related infrastructure is partly deployed by Crusoe under a 15-year capacity agreement, making Oracle simultaneously a hyperscaler competitor and a Crusoe customer. SP019, SP026
CP025 Microsoft committed over $80 billion in AI infrastructure capex for 2025, a major driver of hyperscaler demand that Crusoe is partly fulfilling through the Abilene 900 MW deal. SP019
CP026 Crusoe's vertical integration—spanning power sourcing, AI factory construction, and GPU cloud—is a structural differentiator that pure-play neoclouds cannot replicate without years of capital investment. SP015, SP025
CP027 Crusoe Spark units (approximately 1 MW prefab capacity each) can be deployed at customer sites or near energy sources, a modular deployment capability no neocloud competitor currently replicates. SP023, SP032
CP028 H100 spot GPU rental prices declined from approximately $8/GPU-hour in 2023 to $2–3/GPU-hour by late 2024, compressing neocloud gross margin on spot-priced capacity. SP015, SP027
CP029 Switching from Crusoe Cloud to a hyperscaler requires workload replatforming and renegotiation of long-term capacity agreements, creating meaningful but not prohibitive switching costs. SP024, SP015
CP030 CoreWeave's key structural limitation relative to Crusoe is that it relies on third-party colocation for power and data center infrastructure, exposing it to supply-chain constraints as GPU demand tightens. SP003, SP015
CP031 A documented connectivity outage impacted Crusoe's compute instances in the us-east1-a cloud region, providing competitors with concrete adverse reliability evidence. SP028
CP032 Heatmap News reported that Crusoe uses 360 MW of natural gas turbines at its Abilene campus and is developing 4.5 GW of new gas capacity via an Engine No. 1 joint venture, contradicting its clean-energy branding. SP017
CP033 Crusoe's dual role as hyperscaler capacity partner (Microsoft, Oracle) and GPU cloud competitor for enterprise buyers creates competitive complexity not seen at other neoclouds. SP019, SP026
CP034 The neocloud sector faces long-term commoditization risk as providers that remain pure hardware resellers will be squeezed by falling GPU rental prices and rising power costs, per analyst forecasts. SP015, SP029
CP035 Crusoe was named to Fast Company's Most Innovative Companies of 2026 list and won North American Data Center Project of the Year at the 2025 DCD Global Awards, signaling brand differentiation. SP031, SP018
CP036 Crusoe Cloud's 100% CSAT score over more than 12 months is a company-reported customer satisfaction metric that has not been independently verified. SP001
CP037 The BYO Power approach enables deploying AI compute adjacent to energy sources, bypassing the interconnection queue that delays traditional data center builds by 3–5 years. SP023, SP025
CP038 Crusoe Cloud's four-region footprint (2 US, 2 EU/Nordic) gives it geographic diversity that single-region neoclouds like Lambda Labs currently lack. SP001, SP018
CP039 Neocloud providers that move up the stack into software and orchestration services are forecast by analysts to capture higher margins, validating Crusoe's MemoryAlloy inference and Command Center investments. SP015, SP016
CP040 Crusoe's total equity raised of approximately $3.4B as of Series E compares to CoreWeave's approximately $1.1B equity raised at IPO, while Nebius raised $700M; Crusoe has materially more equity capital. SP018, SP027
CI001 Crusoe generated $276M in total revenue in 2024, split approximately 55% Bitcoin/DFM mining ($152M) and 45% AI cloud ($124M). SI007, SI008
CI002 Crusoe's AI cloud revenue grew 460% year-over-year to $124M in 2024, making it the fastest-growing segment. SI007, SI015
CI003 Crusoe disclosed 150% cloud ARR growth and 17x total contract value growth in 2025 without publishing the absolute base figures. SI005, SI006
CI004 Crusoe sold its Bitcoin mining and digital flare mining operations—425+ modular data centers, 250+ MW, ~135 employees—to NYDIG in March 2025. SI008, SI007
CI005 Following the NYDIG sale, Crusoe's revenue composition shifted to pure AI cloud infrastructure, eliminating commodity Bitcoin mining revenue. SI008, SI005
CI006 Crusoe's Spark modular AI factory product creates a potential product-sale or managed-service revenue line beyond GPU cloud rental, with Redwood Materials expanding from 4 to 24 Spark units. SI021, SI022
CI007 Analyst estimates for Crusoe's full-year 2025 revenue range between $500M and $1B, but no company-disclosed absolute 2025 revenue figure is publicly available. SI018, SI025
CI008 Crusoe Cloud lists H100 SXM GPU instances at $4.29/GPU-hour as of 2026, positioning it at the premium end of neocloud on-demand pricing. SI001, SI013
CI009 Crusoe Cloud lists A100 SXM GPU compute at $1.95/GPU-hour and CPU compute at $0.04/vCPU-hour. SI001
CI010 H100 spot GPU rental prices declined from approximately $8/GPU-hour in 2023 to $2–3/GPU-hour by late 2024, compressing neocloud margins on spot-priced capacity. SI013, SI028
CI011 Crusoe's managed inference product (MemoryAlloy) uses a pay-per-token model billed per 1 million tokens, with cached tokens billed at a lower rate. SI001
CI012 Crusoe's PUE of 1.2–1.3 versus the industry average of 1.8 implies approximately 30–40% lower power overhead per GPU-hour than typical data centers. SI013, SI026
CI013 Estimated power cost per H100 GPU-hour at Crusoe is approximately $0.02–$0.04, based on 700W TDP, 1.2 PUE, and $0.03–$0.06/kWh blended energy cost. SI013, SI030
CI014 Estimated GPU capex depreciation cost per H100 GPU-hour is approximately $0.60–$1.20, based on a $25K–$30K unit purchase price, 3–5 year life, and 80% utilization. SI013
CI015 Neocloud peer companies have disclosed gross margins in the 20–40% range; Crusoe's energy efficiency advantage suggests it could be at the high end, but no Crusoe gross margin figure has been disclosed. SI013, SI025
CI016 Crusoe's Series E raised exactly $1,374,999,988 from 71 investors, filed with the SEC on October 23, 2025, per Form D (CIK 0001924674). SI003, SI005
CI017 Crusoe's Series D raised $817,744,542 from 70 investors, filed with the SEC on November 21, 2024, per Form D (CIK 0001924674). SI004, SI011
CI018 Crusoe's total equity raised across six rounds is approximately $3.4B, including all tranches from 2019 seed through the 2025 Series E. SI005, SI007
CI019 Blue Owl Capital committed $3.4B in a joint venture for the Abilene data center construction, reducing Crusoe's equity capital requirements for the campus. SI007, SI029
CI020 JPMorgan is arranging approximately $9.6B in project financing for the Abilene campus, secured by campus assets and long-term hyperscaler contracts. SI007, SI029
CI021 The Microsoft 900 MW Abilene capacity agreement and Oracle Stargate 15-year contract represent multi-year take-or-pay revenue floor underpinning Crusoe's project finance covenants. SI010, SI029
CI022 Total committed capital for the Abilene campus exceeds $15B including Blue Owl JV ($3.4B) and JPMorgan project financing (~$9.6B), a scale unusual for a private company. SI007, SI019
CI023 Michael Gordon joined Crusoe as COO and CFO in December 2025; he previously led MongoDB's 2017 IPO and grew MongoDB revenue approximately 50x during his tenure. SI007, SI012
CI024 AI hyperscale data center development requires approximately $10–$15M per MW in civil engineering, power infrastructure, and cooling before GPU racks are installed. SI013, SI029
CI025 Crusoe has not disclosed gross margin, EBITDA, cash position, burn rate, net revenue retention, or audited financial statements for any period. SI025, SI018
CI026 The absence of absolute 2025 revenue figures makes the 150% ARR growth claim uninterpretable without the base; the implied range of $310M–$1B+ is too wide for underwriting. SI003, SI025
CI027 GPU spot pricing compression from $8/hr to $2–3/hr for H100 represents a structural pricing ceiling that constrains upside on Crusoe's on-demand cloud revenue. SI010, SI013
CI028 At Crusoe's $10B+ Series E valuation against an estimated $600M–$1B 2025 revenue base, the implied revenue multiple is approximately 10x–17x, consistent with premium neocloud multiples. SI003, SI007
CI029 Heatmap News reported that Crusoe uses 360 MW of natural gas turbines at Abilene and is developing 4.5 GW of new gas capacity via an Engine No. 1 joint venture, creating ESG premium pricing risk. SI009
CI030 Customer concentration in Microsoft and Oracle creates revenue dependency risk; a contract modification by either hyperscaler anchor could materially impact Crusoe's revenue. SI010, SI029
CI031 No public covenant terms, loan-to-value ratios, or debt service coverage requirements are available for Crusoe's JPMorgan project finance; this is the largest undisclosed financial risk. SI025, SI019
CI032 Crusoe's high-capex model with $9.6B in project finance creates refinancing risk if construction delays or hyperscaler contract modifications occur before revenue ramps to service covenants. SI020, SI007
CI033 The implied revenue multiple at $10B valuation is not underwritable without private financial disclosure including gross margin, NRR, burn rate, and project finance covenant terms. SI025, SI018
CI034 Crusoe Spark modular AI factory revenue model—whether product sale, lease, or managed service—is company-claimed rather than independently verified, and the margin profile is unknown. SI021, SI022
CI035 Codeium CEO reported 99.98% GPU cluster uptime on Crusoe, cited in the Series D announcement as customer validation of Crusoe Cloud reliability. SI011, SI002
CI036 Crusoe's 2024 Impact Report describes clean-energy milestones and operational commitments, which underpin its ESG-linked premium pricing claim to enterprise customers. SI026, SI009
CI037 Series E investors include Valor Equity Partners (co-lead) and Mubadala Capital (co-lead), with over 71 total investors, indicating broad institutional confidence. SI003, SI005
CE001 Crusoe Cloud is a full-stack, NVIDIA-native AI cloud platform offering GPU IaaS, Managed Inference, Managed Kubernetes, Managed Slurm, and AutoClusters orchestration for AI/ML workloads. SE001, SE002
CE002 Crusoe Cloud's on-demand GPU pricing is $3.90/hr for H100 SXM (80 GB), $4.29/hr for H200 SXM (141 GB), and $3.45/hr for AMD MI300X (192 GB) as of June 2026. SE003, SE013
CE003 Crusoe Cloud supports NVIDIA A100, H100, H200, B200, L40S GPUs and AMD MI300X accelerators across US East, US South-Central, and European (Iceland, Norway) regions. SE003, SE013
CE004 Crusoe Cloud uses cloud-hypervisor as its VM virtualization layer for all GPU VM types, enabling multi-tenant GPU isolation; single-GPU to 8-GPU node configurations are available. SE003, SE010
CE005 Crusoe Cloud provides Managed Kubernetes (CMK), Managed Slurm, and AutoClusters as orchestration services, with AutoClusters offering automated fault-tolerant training via proactive node health checks. SE002, SE004
CE006 Crusoe Cloud provides VPC networking with RDMA InfiniBand at up to 3200 Gbps, topology-aware cluster placement, and block, object, and shared storage integrated with the GPU compute layer. SE002, SE003
CE007 Crusoe Cloud exposes a REST API at api.cloud.crusoe.ai/v1, a CLI, Terraform provider, and SDKs for Go, Python, and TypeScript; authentication uses signed requests with access and secret keys. SE009, SE004
CE008 Crusoe Intelligence Foundry is a unified model hub supporting leading open-source LLMs including Llama 3.3 70B, DeepSeek V3 0324, DeepSeek-R1 0528, Qwen3 235B, Kimi-K2, Gemma 3 12B, GPT-OSS-120B, Nemotron 3 Super, and Nemotron 3 VoiceChat as of June 2026. SE012, SE004
CE009 MemoryAlloy is a proprietary cluster-native memory fabric that implements a cluster-wide KV cache, enabling GPUs to fetch prefix caches from both local and remote nodes to eliminate duplicate prefill computation and enable persistent inference sessions. SE012
CE010 Crusoe Managed Inference achieves up to 9.9× faster time-to-first-token and 5× higher token throughput compared to vLLM for Llama 3.3 70B, per Crusoe's own benchmark. SE012, SE011
CE011 Crusoe contributed a high-performance Rust BPE tokenizer (fastokens) to the NVIDIA Dynamo open-source inference framework, claiming ~9× speedup over HuggingFace tokenizers and up to 31× on long prompts. SE005, SE008
CE012 Crusoe is an early adopter of NVIDIA Nemotron 3 Super (hybrid MoE, 50%+ higher token generation, 1M token context) and NVIDIA Nemotron 3 VoiceChat; both are available on the Intelligence Foundry following GTC 2026 announcement. SE005, SE012
CE013 Crusoe Spark is a turnkey, prefabricated modular AI factory delivering approximately 1 MW of high-density compute per unit with integrated power, cooling, fire suppression, and GPU racks; units can be delivered within three months of order. SE006, SE018
CE014 Crusoe manufactures Spark modules at a 352,000-sq-ft Spark Factory in Brighton, Colorado representing more than $200 million of investment; production of first factory-built modules is targeted for Q3 2026. SE007, SE024
CE015 Crusoe's flagship 1.2 GW Abilene, Texas campus on Lancium Clean Campus land is now operational in its first two buildings (980,000 sq ft, 200 MW+), serving Oracle Cloud Infrastructure for OpenAI's Stargate compute needs. SE025, SE020
CE016 Crusoe broke ground on a second 900 MW Abilene campus for Microsoft in 2026, expanding the Lancium Clean Campus footprint with a sixth additional data center building. SE020, SE025
CE017 Crusoe's contracted AI infrastructure capacity reached 4.9 GW across its data center and cloud operations as of June 9, 2026, with a total development pipeline exceeding 40 GW. SE020, SE019
CE018 Crusoe partners with Redwood Materials to power Spark modular units using second-life EV battery microgrids; the first deployment at Redwood's Tahoe Campus, Nevada used a 63 MWh system and demonstrated 99.2% uptime. SE015, SE006
CE019 Crusoe secured 12 GWh of iron-air battery storage from Form Energy to power AI data centers, with deployment beginning in 2027. SE016, SE015
CE020 Crusoe has a multi-year strategic framework agreement with Energy Vault for Spark unit deployment at Energy Vault's Snyder, Texas site up to 25 MW, with deployment beginning in 2026. SE017
CE021 Crusoe's Digital Flare Mitigation (DFM) technology achieves approximately 99% combustion efficiency by capturing flared natural gas at oil extraction sites and converting it to electricity for on-site modular data centers; over 120 DFM units have been deployed. SE023, SE024
CE022 Crusoe will be an early adopter of NVIDIA Vera CPU—supporting over 22,500 concurrent agentic environments per rack—alongside Rubin GPU and Vera Rubin NLV72 systems targeted for late 2026 and throughout 2027. SE005
CE023 Crusoe is adopting NVIDIA Omniverse DSX Blueprint and Vera Rubin DSX Reference Design to inform the design and operation of its next-generation gigawatt-scale AI factories, using digital twins and AI-driven power and cooling optimization. SE005
CE024 Crusoe's public GitHub organization (github.com/crusoecloud) contains repositories including the Terraform provider, CLI tool, Go client library (client-go), Slurm configurations, CSI driver Helm charts, and fastokens Rust BPE tokenizer with commits active through June 2026. SE008, SE004
CE025 Crusoe's developer hub showcases six live demos including multi-node Llama 3.1 8B pre-training on CMK via TorchTitan, PyTorch distributed training on Slurm (FSDP/DDP), and Ray inference with vLLM and KubeRay on CMK. SE004
CE026 Crusoe Cloud achieved SOC 2 Type I attestation in December 2023 and SOC 2 Type II attestation in July 2024; the SOC 2 Type II report is available to customers through Crusoe's Trust Center. SE007, SE001
CE027 Crusoe Cloud guarantees 99.98% cluster uptime via AutoClusters fault tolerance, complies with GDPR, and offers 24/7 enterprise support with an average first reply time under six minutes. SE002, SE014
CE028 Crusoe Cloud achieved SemiAnalysis "Gold" status in the GPU Cloud Cluster Max Rating System, recognizing reliability, enterprise-grade support, and a user-friendly console interface. SE011, SE010
CE029 SemiAnalysis's ClusterMAX review found Crusoe's Slurm-on-Kubernetes offering unusable out-of-the-box: login pods lacked vim, nano, git, python, and sudo; CMK clusters lacked a default ReadWriteMany StorageClass; and NVML driver mismatches were reproducible in containerized workloads. SE010, SE013
CE030 Crusoe's Iceland data center experienced widespread fiber end contamination (described as volcanic ash or debris), requiring physical cleaning of 20,000 fiber ends; the issue caused link flaps and random filesystem unmounts for customers who tested clusters at that facility. SE010
CE031 Crusoe Cloud's ICAT region experienced a load balancer outage on June 17, 2026 that rendered load balancer endpoints unreachable for approximately two hours before resolution. SE014
CE032 SemiAnalysis noted that Crusoe is at risk of a Gold rating downgrade due to high engineering turnover in the cloud division, citing too many middle managers and slowing feature releases. SE010
CE033 Crusoe Cloud bookings grew 5× year-over-year in 2025, reflecting rapid enterprise AI adoption of the platform. SE019, SE021
CE034 Crusoe co-develops power infrastructure and construction from project inception, manufactures long-lead electrical components at facilities in Colorado, Oklahoma, and Louisiana, and ships prefabricated equipment ready for installation—compressing timelines vs. conventional sequential development. SE020, SE025
CE035 Crusoe Cloud VMs can boot in under 90 seconds, per SemiAnalysis ClusterMAX benchmark findings, which the report describes as setting the "gold standard" for GPU scaling speed. SE010, SE011
CE036 Crusoe Cloud operates a global private backbone providing high-bandwidth connectivity from hyperscalers and data centers across North America and Europe. SE002
CE037 Crusoe Cloud has maintained a 100% CSAT score from customers for more than twelve consecutive months as of June 2026, per company disclosure. SE002
CU001 Crusoe's cloud segment serves AI-native startups and enterprise machine learning teams as its primary buyer cohorts, with secondary segments including research institutions seeking cost-effective burst GPU capacity. SU015, SU016
CU002 Crusoe's data-center campus segment serves hyperscale cloud operators and large AI labs through decade-long power purchase agreements, representing a structurally distinct market from the cloud tier with different procurement, risk, and contract dynamics. SU009, SU011
CU003 Crusoe Cloud's customer base is predominantly US-based, with European deployments in Iceland and Norway serving customers with data-residency requirements or seeking renewable-energy-sourced compute. SU015, SU009, SU026
CU004 Crusoe Cloud's primary sales channel is direct self-service via crusoe.ai/cloud, supplemented by enterprise direct sales for large reservation contracts and NVIDIA Cloud Partner channel referrals. SU015, SU020
CU005 Crusoe targets GPU workloads where hyperscaler pricing and reservation scarcity create buyer friction, including LLM pre-training, fine-tuning, and real-time inference for AI startup teams unable to secure consistent H100/H200 availability on AWS or Azure. SU011, SU015, SU027
CU006 Crusoe's sustainability narrative — energy-first positioning, renewable sourcing, flare-gas elimination — is a secondary purchase driver for ESG-committed buyers, particularly AI startups and consumer-facing tech companies with public carbon goals. SU017, SU009
CU007 Windsurf (formerly Codeium), an AI coding assistant platform serving over 800,000 developers, migrated its GPU workloads to Crusoe Cloud, achieving a 50% reduction in infrastructure costs versus AWS and Azure and sustaining 99.98% cluster uptime over a twelve-month production period. SU001, SU016
CU008 Decart AI, an AI startup specializing in real-time LLM inference for human simulation applications, runs production workloads on Crusoe Cloud H100 clusters, citing sub-200ms response latency for Llama-class models in customer-facing applications. SU015, SU011
CU009 BosonAI is a named Crusoe Cloud customer using the platform for multilingual LLM pre-training at scale, citing cost-effectiveness relative to hyperscaler alternatives as the primary adoption driver. SU015, SU011
CU010 Oracle and OpenAI are the Phase 1 anchor tenants of Crusoe's 1.2GW Abilene, Texas data center campus, with Phase 1 (200MW+) confirmed live in May 2026. Oracle CEO Safra Catz publicly attributed the 12-month construction timeline to Crusoe's "unmatched execution." SU013, SU009
CU011 Microsoft is the anchor tenant for Abilene Phase 2 (900MW), currently under construction as of June 2026, with an expected completion date in 2027. SU013, SU012
CU012 Meta has reportedly signed a 1.6GW data-center capacity agreement with Crusoe spanning Childress, Texas and Warrenton, Missouri, as reported by Data Center Dynamics, but neither Meta nor Crusoe has officially confirmed the contract. SU003, SU022
CU013 xAI (Elon Musk's AI lab) is reported by multiple media outlets to have contracted over 1GW of Crusoe-built AI infrastructure for the Colossus 2 expansion, positioning Crusoe as a key infrastructure partner for xAI's 2026 Grok model training buildout. SU004, SU011
CU014 Databricks, Together AI, Luma AI, Codeium (now Windsurf), Wonderful.ai, Yutori, Oaklet, and Sony are named as Crusoe Cloud customers across training and inference workloads, cited in Crusoe marketing materials and analyst coverage. SU015, SU011
CU015 SemiAnalysis awarded Crusoe Cloud a ClusterMAX Gold rating based on evaluation across eight criteria including hardware quality, networking, storage, developer experience, support responsiveness, and documentation — though the same review identified usability gaps in the Slurm-on-Kubernetes product representing downgrade risk. SU014, SU018
CU016 Crusoe Cloud's ARR grew 150% year-over-year from 2024 to 2025, reflecting a combination of new customer acquisition and expansion within existing accounts, according to Crusoe's 2025 Impact Report. SU017, SU010
CU017 Crusoe's total contracted AI infrastructure capacity reached 4.9GW as of June 2026, with a development pipeline exceeding 40GW, according to reporting by Business Insider and SiliconAngle citing Crusoe statements. SU012, SU009
CU018 Crusoe reported 70% new customer growth in 2025 and a 17× increase in average cloud contract value year-over-year, signaling both volume expansion and maturation toward larger enterprise deal sizes. SU017, SU011
CU019 Sacra Research projected Crusoe's total revenue at approximately $2 billion for FY 2026, up from $276 million in FY 2024, representing approximately 7× compound growth driven primarily by data-center campus contracts with Oracle, OpenAI, and Microsoft. SU010, SU012
CU020 Cheddarflow analysis estimated Crusoe Cloud's standalone ARR at $150–200 million as of early 2026, with the cloud segment representing a growing but still minority share of total company revenue compared to campus and DFM segments. SU011, SU010
CU021 Crusoe reported 100% CSAT over the trailing twelve months and average first support response time under six minutes, based on internal customer feedback in the 2025 Impact Report and corroborated by SemiAnalysis's reviewer test experience. SU017, SU014
CU022 Crusoe launched Managed Inference in April 2026, adding an API-first consumption tier serving inference buyers without dedicated GPU cluster reservations, expanding Crusoe's addressable customer base to include token-economy buyers. SU021, SU009
CU023 Crusoe Cloud's standard IaaS offering includes a contractually guaranteed 99.98% cluster uptime SLA and a sub-90-second VM boot time across all GPU VM types, embedded in standard enterprise reservation contracts. SU015, SU017
CU024 Crusoe Cloud's ICAT region experienced a load-balancer outage on June 17, 2026 that rendered load-balancer endpoints unreachable for approximately two hours, documented on the public Crusoe status page and corroborated by DeployBase review coverage. SU019, SU014
CU025 Crusoe Cloud's Net Revenue Retention rate has not been publicly disclosed; Cheddarflow estimated NRR in the 110–130% range for pre-2025 cohorts based on qualitative expansion signals, but this estimate cannot be independently verified. SU011, SU008
CU026 Hyperscale data-center tenant contracts with Crusoe span 10-to-20-year power purchase agreements, providing structurally durable revenue that contrasts with the monthly-to-annual reservation terms used in the cloud tier. SU009, SU011
CU027 Windsurf's twelve-month production tenure without a reported SLA breach, combined with its expansion from GPU IaaS to Managed Inference, demonstrates both retention and organic upsell within Crusoe Cloud's most thoroughly documented customer relationship. SU001, SU021
CU028 Crusoe's energy-sector partnerships (Form Energy 12GWh iron-air batteries, Energy Vault Snyder TX site, Redwood Materials EV battery microgrids) provide infrastructure-layer retention anchors for the DFM legacy business but do not directly reduce cloud customer churn risk. SU016, SU017
CU029 Crusoe's hyperscale campus procurement follows a land-and-expand pattern: an anchor tenant commits to an initial phase of 200–400MW, triggering construction, with subsequent phases of 600–1200MW following upon Phase 1 delivery and tenant satisfaction, as demonstrated at Abilene with Oracle Phase 1 and Microsoft Phase 2. SU013, SU009
CU030 Cloud customer procurement occurs through direct self-service sign-up with API key and payment card, or through enterprise sales with negotiated annual GPU reservation contracts; Managed Inference uses an API-key token consumption model without cluster reservations. SU015, SU021
CU031 Crusoe paused work on its 1.8GW Cheyenne, Wyoming data center at the request of an unnamed customer, according to a statement from Crusoe reported by Data Center Dynamics, establishing evidence that hyperscale pipeline customers can and do withdraw before project completion. SU002, SU011
CU032 Crusoe's data-center campus business exhibits high customer concentration: the top three reported tenants (Oracle, Microsoft, Meta) account for the majority of announced campus capacity, and the failure of any one to execute their reported agreement would materially reduce contracted capacity and construction-finance capacity. SU013, SU003
CU033 xAI's reported 1GW+ infrastructure partnership with Crusoe — spanning Colossus 2 buildout and ongoing Grok inference GPU provisioning — represents both a revenue anchor and a brand-validation signal, but its unconfirmed status introduces concentration and disclosure risk. SU004, SU011
CU034 Crusoe and Upstream Data Inc. resolved a patent-infringement lawsuit in 2025 via a licensing agreement covering DFM technology intellectual property, according to Crusoe's newsroom statement; the district court had found infringement on at least one claim before damages were contested. SU005, SU006
CU035 The Upstream Data patent litigation was resolved before Crusoe's main revenue center shifted fully to AI cloud; the licensed DFM technology remains the basis of Crusoe's legacy modular gas-capture business but is not the core of the cloud or campus product lines. SU005, SU007
CU036 No independent customer satisfaction survey, Net Promoter Score disclosure, or third-party reference program data is publicly available for Crusoe Cloud; all CSAT and support-quality data is self-reported by Crusoe. SU008, SU011
CU037 Crusoe's 4.9GW contracted figure mixes signed leases with reported agreements; if two major hyperscale tenants each at 1GW+ fail to execute final contracts, the contracted capacity metric and associated construction-financing plans would be materially reduced. SU012, SU009
CR001 Crusoe describes itself as an 'AI factory company' with a mission to accelerate the abundance of energy and intelligence. SR016
CR002 Crusoe's power pipeline exceeded 45 gigawatts as of October 2025, a more than 4x increase in the prior year. SR016, SR015
CR003 CEO Chase Lochmiller stated publicly that many pipeline commitments exist where he 'can't vouch for the reality that they will happen.' SR015
CR004 Crusoe contracted approximately 4.9 gigawatts of data center infrastructure for Crusoe Cloud as of June 2026. SR013, SR020, SR031
CR005 Crusoe's total data center development pipeline including contracted projects, sites under active negotiation, and advanced development exceeds 40 GW as of June 2026. SR013
CR006 EPA's 2023/2024 OOOOb/OOOOc methane rules ban routine flaring at new oil-and-gas wells after May 7, 2026, with limited temporary-service-interruption exemptions. SR005, SR006, SR002
CR007 The BLM 2024 Waste Prevention Rule enforcement was delayed for two flare-measurement and LDAR provisions until December 10, 2026, due to proposed rule revisions. SR003, SR004
CR008 North Dakota, Texas, Montana, Wyoming, and Utah operate under a preliminary injunction blocking BLM enforcement of the 2024 Waste Prevention Rule. SR003, SR004
CR009 The Department of the Interior published proposed revisions to the Biden methane waste prevention rule on June 22, 2026, including rollbacks to waste-minimization plan requirements. SR004
CR010 Texas Senate Bill 6, signed by Governor Abbott in June 2025, requires data centers ≥75 MW to accept curtailment during firm load-shed events and mandates remote disconnect capability for new connections after December 31, 2025. SR010, SR012
CR011 ERCOT forecasts 138 GW of large loads on its Texas grid by 2030, up from approximately 87 GW in 2025, with 87% of the queued load from data centers. SR010
CR012 Crusoe divested its Digital Flare Mitigation business in 2025 to focus exclusively on vertically integrated AI infrastructure. SR011, SR016
CR013 Heatmap News reported that Crusoe plans to operate 360 MW of on-site natural gas turbines at the Abilene facility, representing 30% of the 1.2 GW campus capacity. SR001
CR014 Crusoe acquired 4.5 GW of natural gas capacity via a joint venture with Engine No. 1 and Chevron, with plans to include post-combustion carbon capture systems. SR001
CR015 The carbon capture technology planned for Crusoe/Chevron/Engine No. 1 gas turbines has no confirmed design partner and is expected to deliver power by end 2027, per Chevron's announcement. SR001
CR016 Crusoe committed $400 million to AMD MI355X accelerators (~13,000 units) in June 2025, representing one of AMD's largest GPU orders to date. SR018
CR017 NVIDIA is an equity holder in Crusoe, having participated in both the Series D and Series E rounds. SR016, SR018
CR018 Crusoe expanded its NVIDIA collaboration in March 2026 to span Vera CPU, Rubin GPU NVL72, Nemotron 3 Super, and Dynamo open-source inference integration. SR019
CR019 US semiconductor tariffs averaged 18.2% by mid-2025, increasing GPU procurement costs for AI infrastructure providers including Crusoe. SR028
CR020 Crusoe Cloud achieved 99.98% uptime and holds ISO 27001 and ISO 42001 certifications as of February 2026, making it one of the first AI cloud providers with both certifications simultaneously. SR011, SR016
CR021 Crusoe delivered Phase 1 of the 1.2 GW Abilene Stargate campus within 12 months of breaking ground, setting a record for greenfield hyperscale development. SR011, SR014
CR022 Crusoe's Abilene Phase 1 was built to an annualized design PUE of 1.2 to 1.4, well below the industry average of 1.54. SR011
CR023 EDF reported that approximately $4.2 billion in natural gas was wasted in the year after the Trump administration stopped enforcing oil-and-gas methane regulations. SR002
CR024 Crusoe paused development of Project Jade, a 1.8 GW campus in Cheyenne, Wyoming, at a customer's request in early June 2026; neither the customer identity nor the reason was disclosed. SR013
CR025 Oracle and OpenAI exited the later stages of the Abilene data center project due to delays; Microsoft subsequently took on the remaining capacity. SR013
CR026 Data Center Dynamics reported that Meta signed a 1.6 GW capacity agreement with Crusoe. SR024
CR027 CoreWeave went public March 27, 2025, at a $23 billion market cap and $30 billion enterprise value, with 2024 revenue of $1.9 billion and Microsoft representing 62% of revenue. SR032
CR028 Crusoe and Blue Owl Capital announced a $3.4 billion joint venture in October 2024 to build and operate modular data centers capable of hosting up to 100,000 GPUs. SR018
CR029 Crusoe has a $750 million credit facility from Brookfield Asset Management arranged in mid-2025 to accelerate AI factory development. SR028
CR030 Crusoe's announced strategic collaborations with NVIDIA and AMD, plus investor equity holdings by NVIDIA, create both supply-chain alignment and a potential conflict of interest risk. SR018, SR019
CR031 Crusoe grew to 1,217 employees in 2025, more than doubling headcount for the second consecutive year. SR011
CR032 Crusoe's 2025 debt facilities totalled approximately $975 million based on company and investor disclosures. SR028
CR033 Crusoe has raised approximately $3.9 billion total across all rounds since its 2018 founding. SR017, SR026
CR034 Press reports estimate the full Abilene campus will cost approximately $12 billion, a figure that exceeds Crusoe's total raised capital by more than 3x. SR015, SR028
CR035 Crusoe divested its Bitcoin mining business to NYDIG in 2025, removing a cash-generative but cyclical revenue stream. SR016
CR036 Crusoe's 2025 average training hours per employee grew to 53, more than double 2024 levels, with employee engagement at 85% favorable. SR011
CR037 Crusoe's cloud electricity usage is 100% matched with renewable energy through VPPAs, energy attribute certificates, and direct supply at Iceland and Norway sites. SR011
CR038 The Heatmap article notes Crusoe's climate-tech credentials depend on 'unprovable counterfactuals': first that flared gas would otherwise be vented, and now that its data centers are inherently cleaner than competitor-built alternatives. SR001
CR039 Crusoe holds both ISO 27001 (information security) and ISO 42001 (responsible AI) certifications, achieved simultaneously — among the first AI cloud providers to do so. SR011
CR040 Crusoe has patents US10862307, US10862309, US11437821, and US11451059 covering systems and processes for oil-field natural gas computing, as cited in its July 2025 complaint against Upstream Data. SR009
CR041 The November 2025 Crusoe-Upstream Data settlement includes a confidential license covering all pending suits but does not address Upstream Data founder Steve Barbour's 13 remaining patents. SR007, SR008
CR042 Crusoe's built-in.com profile notes that its large campuses are tied to a small number of marquee anchor tenants where 'evolving plans' and concentration create volatility risk. SR022
CV001 Crusoe raised $600M in a Series D funding round at a $2.8B post-money valuation, closing December 12 2024. SV001, SV019
CV002 Crusoe raised $1.375B in a Series E funding round at a post-money valuation exceeding $10B, announced October 24 2025. SV002, SV003, SV035
CV003 Crusoe has raised approximately $3.9B in total equity financing across Series A through Series E rounds as of October 2025. SV004, SV005
CV004 CoreWeave completed its IPO on March 26 2025 at approximately $23B enterprise value, trading at roughly 12x its FY2024 trailing revenue of $1.92B. SV008, SV009, SV036
CV005 CoreWeave's FY2024 revenue of $1.92B represented 737% year-over-year growth, per the company's SEC S-1 registration statement filed February 28 2025. SV008, SV009
CV006 Sacra Research estimated Crusoe's FY2024 revenue at approximately $276M; this is an analyst estimate, not audited or management-confirmed revenue. SV006, SV007
CV007 Sacra Research and Cheddar Flow projected Crusoe's FY2025 revenue at approximately $900M–$1B; these are analyst estimates and have not been independently confirmed. SV006, SV007
CV008 At a $10B post-money Series E valuation and $276M FY2024 revenue (Sacra estimate), Crusoe's implied trailing EV/revenue multiple is approximately 36x — roughly 3x CoreWeave's 2024 trailing multiple. SV006, SV008
CV009 Meta Platforms signed a 1.6 GW capacity agreement with Crusoe, per industry reports in April 2026; the binding nature and term of this agreement has not been independently confirmed. SV013, SV015
CV010 Crusoe Cloud had approximately 4.9 GW of contracted AI data center capacity as of June 2026 per company disclosure. SV014, SV015
CV011 Crusoe secured a $500M credit facility from Brookfield Asset Management to fund power infrastructure development. SV005, SV012
CV012 Blue Owl Capital and Crusoe formed a $450M joint venture for the Abilene, Texas campus Phase 2 buildout. SV014, SV016
CV013 Crusoe committed approximately $400M to AMD MI355X GPUs (~13,000 units) in a strategic supply agreement announced in June 2025. SV027, SV016
CV014 Crusoe's total power pipeline exceeded 45 GW as of October 2025 per its Series E announcement, representing more than a 4x increase in 12 months. SV003, SV015
CV015 Oracle and OpenAI reportedly exited the Abilene campus later phases; Microsoft replaced them as anchor customer for those phases, per industry reporting. SV010, SV014
CV016 xAI (Elon Musk's AI company) became a significant anchor customer for Crusoe in 2025-2026, with dedicated capacity commitments at the Abilene campus. SV025, SV016
CV017 Crusoe introduced the Crusoe Spark modular AI data center product line in March 2026, designed to reduce capital deployment risk by scaling in 10–100 MW units. SV028, SV031
CV018 Crusoe paused construction on its 1.8 GW Cheyenne, Wyoming Project Jade campus at a customer's request as of June 2026, with no public disclosure of the customer identity or resume timeline. SV024, SV010
CV019 Combined equity and structured debt financing for Crusoe totals approximately $5–5.5B when including the Brookfield credit facility and Blue Owl JV, based on public disclosures. SV003, SV004
CV020 Form Energy and Crusoe announced a 12 GWh iron-air battery storage agreement for AI data center backup power in April 2026. SV021, SV022
CV021 NVIDIA holds an equity stake in Crusoe and has extended its strategic collaboration to encompass GPU deployment, AI development tooling, and inference infrastructure alignment. SV030, SV025
CV022 Crusoe Cloud achieved SOC 2 Type II certification, demonstrating minimum operational security and compliance standards for enterprise customers. SV032, SV016
CV023 Third-party benchmark reviews (DeployBase, ClusterMax) rated Crusoe Cloud positively on price-performance for bare-metal H100 GPU instances as of 2026, with pricing below AWS P5 instances. SV017, SV018
CV024 CoreWeave reported $8.7B in committed revenue backlog at the time of its S-1 IPO filing, largely concentrated in a Microsoft hyperscale anchor customer relationship. SV008, SV009
CV025 Applying CoreWeave's 12x trailing IPO multiple to Crusoe's projected FY2025 revenue of ~$998M implies an enterprise value of approximately $12B — modestly above the current $10B+ mark. SV008, SV006
CV026 The bull scenario for Crusoe assumes FY2026 revenue of $1.5–2.0B, EV/revenue of 13–15x forward, and implies enterprise value of $22–26B on the basis of Abilene full ramp and multi-campus activation. SV007, SV004
CV027 The bear scenario for Crusoe assumes revenue of $400–600M in FY2026 with 5–7x forward EV/revenue multiple, implying $2.4–4.8B enterprise value — below the Series E post-money — if major anchor customers reduce commitments. SV010, SV024
CV028 The base scenario for Crusoe implies FY2026 revenue of $900M–$1.2B at 8–12x forward EV/revenue, valuing the company at $8–13B — slightly above Series E entry for exiting investors at the high end. SV007, SV006
CV029 The 360 MW natural gas turbine buildout at Abilene (30% of 1.2 GW capacity) and the Engine No. 1/Chevron gas JV create a material ESG contradiction that could reduce Crusoe's access to sustainability-linked capital at premium terms. SV010, SV016
CV030 A Series D investor entering at $2.8B achieves an estimated 2.9–9.3x MOIC under base to bull scenarios if Crusoe exits via IPO or strategic acquisition in 2027–2028. SV001, SV006
CV031 Crusoe's clean-power ESG positioning attracted sustainability-linked financing from Brookfield; loss of that positioning could raise the company's weighted average cost of capital by an estimated 50–150 basis points. SV022, SV016
CV032 xAI's capacity commitments at Crusoe are described as large-scale dedicated GPU cluster deployments; the exact contractual terms and cancellation provisions are not publicly disclosed. SV025, SV016
CV033 The Crusoe Spark modular product line was announced in March 2026, targeting enterprise customers with 10–100 MW deployments and aiming to reduce the capex concentration risk of gigawatt-scale campus builds. SV028, SV031
CV034 Crusoe Cloud offers bare-metal H100 GPU instances at pricing 30–81% below comparable AWS, GCP, and Azure on-demand instances per third-party benchmarking and company pricing pages. SV026, SV017
CV035 Crusoe Energy Systems was founded in 2018 by Chase Lochmiller and Cully Cavness; the company employs approximately 500 people as of 2025–2026. SV016, SV020
CV036 CoreWeave's market capitalisation declined below $30B in secondary trading post-IPO, suggesting investor uncertainty about AI infrastructure multiples even for the most comparable public company. SV008, SV025
CV037 Series E investors in the October 2025 round include Valor Equity Partners and Mubadala Investment Company, per Data Center Dynamics reporting. SV002, SV003
CV038 Reuters reported in June 2025 that NVIDIA-backed Crusoe was targeting a $750M fundraise including Brookfield; this likely refers to a combination of Series D tranches and structured debt rather than a separate priced equity round. SV012, SV011
CV039 Crusoe's strategic infrastructure model targets 10–100 MW modular deployments via Crusoe Spark for faster time-to-revenue relative to gigawatt-scale campuses. SV028, SV016
CV040 Lambda Labs and Voltage Park, as private neo-cloud comparables, are estimated to trade at 4–7x ARR in secondary market transactions, well below Crusoe's $10B+ at estimated $276M FY2024 revenue. SV017, SV018
CV041 Energy Vault and Crusoe announced a strategic framework agreement in February 2026 for grid-scale energy storage, supporting Crusoe's long-term power sustainability and operational resilience. SV022, SV021
CV042 Crusoe's total compute infrastructure footprint across all active campuses exceeds 1.2 GW as of mid-2026, with Abilene Phase 1 representing the largest single site at ~900 MW. SV016, SV015
CV043 Crusoe Cloud achieved ClusterMax Gold status benchmark in 2026, representing one of the highest third-party quality ratings available for GPU cloud providers. SV029, SV018
CV044 Crusoe's 45 GW pipeline is largely without fully binding signed contracts; industry analysts estimate only 10–15% of announced pipeline may convert to final executed agreements within 24 months. SV007, SV005
CV045 Crusoe's estimated debt stack of approximately $975M (Brookfield facility + Blue Owl JV) represents roughly 100% of projected FY2025 revenue, creating significant free-cash-flow constraints and limiting financial flexibility. SV006, SV020
CV046 Crusoe Cloud's GPU infrastructure pricing delivers measurable total cost of ownership savings for AI application companies, as demonstrated by customer case studies including Windsurf (AI code editor) achieving reduced compute spend relative to hyperscaler alternatives. SV033, SV017
CV047 SEC EDGAR filings confirm CoreWeave Inc. is a publicly registered company with an S-1 on file; the CoreWeave public disclosure record provides the primary comparable-company data anchor for Crusoe's valuation analysis. SV034, SV009
来源
编号出版方标题引文
SO001 Crusoe Energy Systems Crusoe Closes $600M Series D Funding Round and Announces General Availability of Crusoe Cloud Crusoe is reimagining AI infrastructure from the ground up to meet and exceed organizations' demands, powering the next wave of innovation in a sustainable way.
SO002 Crusoe Energy Systems Crusoe Expands NVIDIA Collaboration Across the Full AI Factory Stack, Delivering the Complete Infrastructure for the Agentic AI Era We're building every layer of Crusoe's stack in lock step with NVIDIA's roadmap in order to manufacture intelligence with unprecedented speed and scale.
SO003 Crusoe Energy Systems Understanding the Problem Crusoe Solves
SO004 ESG News Crusoe Secures $600M to Scale Clean Energy-Powered AI Infrastructure Platform
SO005 Carbon Credits Crusoe Energy's $600M Raise Fuels AI Revolution with Clean Energy Data Centers
SO006 TechStartups Crusoe raises $600 million in Series D funding to power the AI cloud, now valued at $2.8 billion The biggest risk with AI is that we don't go big enough. Crusoe is here to liberate us from the island of limited ambition.
SO007 Bain Capital Ventures Crusoe's Climb: Betting on Power Before AI Was Cool
SO008 Tech Funding News Clean energy data centres fuel AI revolution as Crusoe raises $600M led by Peter Thiel's Founders Fund
SO009 Data Center Dynamics Oracle to lease Texas data center from cryptomining and AI firm Crusoe — report The first potential client interested in the Abilene site was reportedly Elon Musk's xAI.
SO010 Forbes Meet The Tiny Startup Building Stargate, OpenAI's $500 Billion Data Center Moonshot Some customers worry Crusoe's cloud system is unreliable. In early March one of its centers experienced an outage that lasted 45 hours.
SO011 IPO Club Crusoe (2025) – AI & Clean-Energy Cloud Infrastructure Pioneer
SO012 IsDown (Crusoe Status Tracker) Crusoe Energy Compute instance availability degraded in us-east1-a — May 2026 We are investigating a networking issue in the us-east1-a region affecting connectivity to a subset of compute hosts.
SO013 Forbes (Christopher Helman) Customer quote on Crusoe outage: 'They're not as reliable' This was a big deal. We were pretty unhappy about it. They're not as reliable.
SO014 Data Center Dynamics Crusoe raises $1.375bn in latest funding round The oversubscribed round was co-led by technology investors Valor Equity Partners and Mubadala Capital, with participation from more than 20 firms.
SO015 Tech Funding News AI infrastructure startup Crusoe hits $10B valuation to power AI data centres with cleaner energy
SO016 Crusoe Energy Systems Crusoe Unveils Crusoe Edge Zones to Deliver High-Performance AI Infrastructure
SO017 Crusoe Energy Systems Crusoe Achieves ISO 27001 and ISO 42001 Certifications
SO018 Cheddar Flow Crusoe Energy: Deep Dive Analysis of the AI Infrastructure Pioneer
SO019 World Metrics Crusoe Energy Statistics: 2026 Market Report
SO020 Crusoe Energy Systems Crusoe homepage and product overview
SO021 Bain Capital Ventures (via Keep Cool) Crusoe Energy DFM deep dive — climate impact, scale potential, and founding narrative Each of these digital flare mitigation systems we deploy eliminates 8,000 tons of CO2 equivalent from being emitted to the atmosphere on an annual basis.
SO022 IDC AI Infrastructure Spending Caps Historic Year at ~$90 Billion in Q4 2025; 2029 Spending to Eclipse $1 Trillion
SO023 S&P Global Market Intelligence AI infrastructure results in 2025 top expectations, forecast upgraded
SO024 Harvard Belfer Center AI, Data Centers, and the U.S. Electric Grid: A Watershed Moment
SO025 Network World Neoclouds roll in, challenge hyperscalers for AI workloads
SO026 We The Flywheel AI Compute and Neocloud Providers 2026: Vendor Comparison Energy strategy is the real moat in this market, which is why Crusoe (stranded gas, behind-the-meter renewables) and CoreWeave (long-dated PPAs) score so highly on it.
SO027 Flexential 2025 State of AI Infrastructure Report
SM001 ARK Investment Management The State of AI Infrastructure: Demand, Costs, and Custom Silicon ARK's forecast of nearly $1.5 trillion in annual AI infrastructure spending by 2030 reflects a market driven by genuine, accelerating demand from both consumers and enterprises.
SM002 Data Center Frontier The Gigawatt Bottleneck: Power Constraints Define AI Data Center Growth Power strategy is increasingly moving to the front of the development process. Operators that secure electricity early will be able to deploy capacity faster and at larger scale.
SM003 EnkiAI Data Center Power Crisis 2026: The Grid Bottleneck Goldman Sachs projects a 15% compound annual growth rate in U.S. data center power demand through 2030. By that year, data centers are forecast to consume 8% of all electricity in the United States.
SM004 World Economic Forum Is power grid connectivity the strategic bottleneck for AI? Aligning AI growth with grid connectivity is now a central leadership responsibility.
SM005 Fortune Business Insights GPU as a Service (GPUaaS) Market Size, Share & Industry Analysis
SM006 GM Insights AI Data Center Market Size, Share & Growth Forecast 2025–2034
SM007 Turion AI GPU Clouds Compared 2026: CoreWeave, Lambda, RunPod and More CoreWeave went public on March 28, 2025 at $40 per share with a $35 billion valuation. The company reported $1.9 billion in revenue for 2024 (737% year-over-year growth) and $5 billion for full-year 2025.
SM008 RCR Wireless News Neoclouds revenue surge 205% y-o-y in Q2, research finds Currently, the [neocloud] market is heavily oriented towards the United States, which accounts for around 70% of all neocloud revenues.
SM009 Artificial Intelligence Companies Best AI Cloud Infrastructure Providers 2026 — GPU Cloud for AI For most AI companies, neocloud providers are more cost-effective than owned hardware until sustained monthly spend exceeds $2–5 million.
SM010 COR Advisors Energy Grid, Data Center Capacity and AI Bottlenecks 2026: The Real Constraints
SM011 Emergen Research AI Infrastructure Market Report 2025
SM012 AI Tool Discovery Cloud GPU Providers Compared: H100 Pricing and Buyer Guide For $50k+/month committed: Multi-cloud strategy. Reserved at CoreWeave/Crusoe/Lambda; hyperscaler for compliance-sensitive workloads; hosted APIs for spiky.
SM013 CoreWeave CoreWeave — About Page
SM014 MarketsandMarkets AI Data Center Market — Global Forecast to 2029
SM015 Statista Global Artificial Intelligence Market Revenue 2018–2025
SM016 IndexBox AI Infrastructure Growth 2026: CoreWeave and Nebius Revenue and Projections
SM017 Channel Dive Neoclouds helped drive Q3 cloud market surge Forrester Principal Analyst Lee Sustar expects CoreWeave, Lambda, Nebius and other GPU service providers to take in $20 billion in revenue next year.
SM018 Hanwha Data Centers Data Center Grid Limitations: The Power Bottleneck Power availability should be a primary screening criterion, evaluated before detailed analysis of other site characteristics.
SM019 Gartner Gartner Says AI-Optimized IaaS Is Poised to Become the Next Growth Engine for AI Infrastructure
SM020 S&P Global Market Intelligence AI infrastructure results in 2025 top expectations, forecast upgraded
SM021 IDC AI Infrastructure Spending Caps Historic Year at ~$90 Billion in Q4 2025; 2029 Spending to Eclipse $1 Trillion
SM022 Flexential 2025 State of AI Infrastructure Report
SM023 Network World Neoclouds roll in, challenge hyperscalers for AI workloads
SM024 We The Flywheel AI Compute and Neocloud Providers 2026: Vendor Comparison Energy strategy is the real moat in this market, which is why Crusoe (stranded gas, behind-the-meter renewables) and CoreWeave (long-dated PPAs) score so highly on it.
SM025 Harvard Belfer Center AI, Data Centers, and the U.S. Electric Grid: A Watershed Moment
SP001 Crusoe Crusoe Cloud — AI Cloud Platform For more than 12 months our customers have rated Crusoe support with a 100% CSAT score.
SP002 Crusoe Crusoe Cloud Pricing Our infrastructure is purpose-built and engineered for AI.
SP003 CoreWeave CoreWeave — AI Cloud Platform
SP004 CoreWeave CoreWeave Pricing $1,500 — CoreWeave Kubernetes Service Control Plane
SP005 Lambda Labs Lambda Labs — GPU Cloud The engines of superintelligence
SP006 Nebius AI Nebius AI — AI Cloud Platform 112% better TCO for inference vs. AWS.
SP007 Nebius AI Nebius AI Pricing
SP008 Vast.ai Vast.ai — GPU Marketplace Start with $5. Scale to 20,000 GPUs.
SP009 Oracle Oracle Cloud Infrastructure SuperCluster
SP010 Lancium Lancium Clean Campuses Lancium Clean Campuses enable the scale and speed demanded by hyperscale data center operators.
SP011 Amazon Web Services AWS P5 Instances — H100 GPU Compute
SP012 Amazon Web Services AWS EC2 On-Demand Pricing
SP013 Microsoft Azure Azure Machine Learning Compute Pricing
SP014 Google Cloud Google Cloud GPU Pricing
SP015 Compute Forecast The Neocloud GPU Cloud Service: A Long Read Those who remain mere hardware resellers risk being squeezed by falling GPU rental prices and rising power costs.
SP016 Signisys The Neocloud Revolution: How $20 Billion in GPU-Focused Providers Are Reshaping the Cloud Market
SP017 Heatmap News Crusoe Energy and the Stargate Bet: The Climate Tech Company That Went All-In on Gas Presumably, these plans will also incorporate either some way to utilize the CO2 in industry or to permanently sequester it underground, though the company hasn't mentioned anything to this effect.
SP018 Data Center Dynamics Crusoe Raises $1.375bn in Latest Funding Round Crusoe's cloud operates four cloud regions across the US and Europe.
SP019 Forbes Meet the Tiny Startup Building Stargate: OpenAI's $500 Billion Data Center Moonshot After all, Crusoe financed Project Ludicrous on the back of a 15-year contract with Oracle and Microsoft.
SP020 NVIDIA NVIDIA Announces Strategic Collaboration with Crusoe AI
SP021 Crusoe Crusoe Expands NVIDIA Collaboration Open models, open infrastructure, and performance that scales — that's the architecture of immediacy in practice.
SP022 Crusoe Crusoe Achieves ISO 27001 and ISO 42001 Certifications
SP023 Crusoe Welcome to the Era of BYO Power
SP024 Crusoe Crusoe Cloud Documentation
SP025 Bain Capital Ventures Crusoe's Climb: Betting on Power Before AI Was Cool Identify stranded or wasted energy, deploy modular data centers at the power source, monetize via high-density GPU clusters.
SP026 Data Center Dynamics Crusoe Exits Crypto Operations to Focus on AI; Sells Business to NYDIG
SP027 Sacra Crusoe Company Profile
SP028 IsDown Crusoe Energy — Connectivity Issues Impacting Compute Instances in us-east1-a Connectivity issues impacting compute instances in us-east1-a
SP029 WeTheFlywheel AI Compute Neocloud Providers Guide 2026
SP030 Network World Neoclouds Roll In, Challenge Hyperscalers for AI Workloads
SP031 Crusoe Crusoe Named to Fast Company's Most Innovative Companies of 2026
SP032 Forbes From Gigawatts to Grab-and-Go: Crusoe Leans Into Modular AI Data Centers
SI001 Crusoe Crusoe Cloud Pricing Crusoe Managed Inference uses a usage-based, pay-as-you-go model, billed per 1 million tokens.
SI002 Crusoe Crusoe Cloud — AI Cloud Platform For more than 12 months our customers have rated Crusoe support with a 100% CSAT score.
SI003 SEC EDGAR Crusoe Inc. Form D — Series E Offering (CIK 0001924674, Filed 2025-10-23) 1374999988 — total offering amount; 71 investors
SI004 SEC EDGAR Crusoe Inc. Form D — Series D Offering (CIK 0001924674, Filed 2024-11-21) 817744542 — total offering amount; 70 investors
SI005 Crusoe Crusoe Announces Series E Funding Today's Series E enables Crusoe to rapidly expand our vertically integrated approach to delivering AI factories.
SI006 GlobeNewswire Crusoe Raises $1.375 Billion, Reaches $10 Billion Valuation
SI007 Data Center Dynamics Crusoe Raises $1.375bn in Latest Funding Round Crusoe's cloud operates four cloud regions across the US and Europe.
SI008 Data Center Dynamics Crusoe Exits Crypto Operations to Focus on AI; Sells Business to NYDIG Crusoe is developing a large data center campus in Texas alongside Oracle for OpenAI's Stargate project.
SI009 Heatmap News Crusoe Energy and the Stargate Bet: The Climate Tech Company That Went All-In on Gas Whether or not Crusoe is a boon for the climate ultimately depends upon the degree to which that unquantifiable claim ends up being true.
SI010 Crusoe Crusoe Announces New 900 MW AI Factory Campus in Abilene, Texas to Support Microsoft AI Infrastructure
SI011 Crusoe Crusoe Closes Series D Funding Codeium's NVIDIA H100 Tensor Core GPUs on Crusoe have been incredibly reliable with a cluster uptime of 99.98%.
SI012 Forbes Crusoe's Wind and Solar Powered AI Data Centers Could Lead to IPO
SI013 Compute Forecast The Neocloud GPU Cloud Service: A Long Read Those who remain mere hardware resellers risk being squeezed by falling GPU rental prices and rising power costs.
SI014 WeTheFlywheel AI Compute Neocloud Providers Guide 2026
SI015 TechFunding News AI Infrastructure Startup Crusoe Hits $10B Valuation to Power AI Data Centres with Cleaner Energy
SI016 TechFunding News Crusoe Clean Energy AI Data Centres
SI017 CarbonCredits.com Crusoe Energy's $600M Raise Fuels AI Revolution with Clean Energy Data Centers
SI018 Contrary Research Crusoe Company Research Profile
SI019 IPO Club Crusoe Energy — Initiation of Coverage
SI020 CheddarFlow Crusoe Energy Deep Dive: Analysis of the AI Infrastructure Pioneer
SI021 Crusoe Crusoe and Redwood Materials Expand Strategic Partnership, Scaling to 7x the Original AI Infrastructure Density
SI022 Crusoe Crusoe Announces New Manufacturing Facility to Produce Modular AI Factories
SI023 SEC EDGAR Crusoe Energy Form D EDGAR Filing Index (CIK 0001924674)
SI024 Bain Capital Ventures Crusoe's Climb: Betting on Power Before AI Was Cool
SI025 Sacra Crusoe Company Profile
SI026 Crusoe Crusoe 2024 Impact Report
SI027 Crusoe The New Equation: What AI Leaders Need to Know About Infrastructure in 2026
SI028 Signisys The Neocloud Revolution: How $20 Billion in GPU-Focused Providers Are Reshaping the Cloud Market
SI029 Forbes Meet the Tiny Startup Building Stargate: OpenAI's $500 Billion Data Center Moonshot After all, Crusoe financed Project Ludicrous on the back of a 15-year contract with Oracle and Microsoft.
SI030 Crusoe Welcome to the Era of BYO Power
SE001 Crusoe Crusoe | The energy-first AI factory company Now with Crusoe Managed Inference for breakthrough speed and scale.
SE002 Crusoe Crusoe Cloud | AI Platform & Services Train and deploy up to 20x faster, with 81% less cost. Crusoe's AI-optimized infrastructure cuts waste and unlocks performance, so you get more done with less.
SE003 Crusoe Overview - Crusoe Cloud VM Types
SE004 Crusoe Crusoe for Developers | GPU Cloud for AI Developers
SE005 Crusoe Crusoe Expands NVIDIA Collaboration Across the Full AI Factory Stack, Delivering the Complete Infrastructure for the Agentic AI Era We're building every layer of Crusoe's stack in lock step with NVIDIA's roadmap in order to manufacture intelligence with unprecedented speed and scale.
SE006 Crusoe Crusoe introduces Crusoe Spark: Modular AI data centers for scalable edge computing With over 400 modular units already deployed globally, operating in some of the harshest conditions, Crusoe brings proven reliability to the edge.
SE007 Crusoe Crusoe Cloud achieves SOC 2 Type II attestation
SE008 Crusoe (GitHub organization) CrusoeCloud — GitHub
SE009 Crusoe API Reference - Crusoe Cloud
SE010 SemiAnalysis (ClusterMAX) Crusoe Review 2026: Gold Tier GPU Cloud | ClusterMAX by SemiAnalysis The new Slurm-on-Kubernetes experience is in its early days and is not usable out of the box. The login pod was missing vim, nano, git, python, and sudo permissions.
SE011 EIN Presswire Crusoe Cloud Achieves "Gold" Status in The GPU Cloud Cluster Max Rating System from SemiAnalysis
SE012 Crusoe Crusoe Managed Inference: Crusoe delivers 9.9x faster AI production Breakthrough speed: Achieve up to 9.9x faster TTFT with our inference engine featuring MemoryAlloy, a cluster-wide KV cache with intelligent routing that eliminates duplicate prefills.
SE013 DeployBase Crusoe Review 2026: Pricing, Performance, Pros & Cons
SE014 Crusoe Crusoe Cloud Status
SE015 Latitude Media Crusoe and Redwood Materials are powering a data center with old EV batteries
SE016 Form Energy Form Energy & Crusoe Announce Agreement for 12 Gigawatt-Hours of Iron-Air Batteries for AI Data Centers
SE017 BusinessWire Energy Vault and Crusoe Announce Strategic Framework Agreement for Deployment of Crusoe Spark Modular AI Factory Units
SE018 Forbes From Gigawatts To Grab-And-Go: Crusoe Leans Into Modular AI Data Centers
SE019 SiliconAngle Crusoe AI puts energy first to power sustainable AI infrastructure at scale — PureAccelerate
SE020 Markets Insider / Business Insider Crusoe's Contracted AI Infrastructure Capacity Approaches 5 Gigawatts Across Data Centers and Cloud Crusoe co-develops them from the start, manufactures long-lead electrical components at its own facilities in Colorado, Oklahoma, and Louisiana, and ships prefabbed equipment ready for installation.
SE021 Sacra Crusoe revenue, valuation & funding | Sacra
SE022 CheddarFlow Crusoe Energy: Deep Dive Analysis of the AI Infrastructure Pioneer
SE023 Crusoe Energy Systems About Crusoe Energy
SE024 Crusoe Crusoe's 2025 Impact Report
SE025 Data Center Dynamics Crusoe's Abilene data center campus officially live, serving Oracle and OpenAI's Stargate Crusoe has proven its leadership in AI data center design, engineering and construction and has become a trusted partner of Oracle.
SU001 Crusoe Windsurf Saves 50% on Infrastructure Costs Using Crusoe Cloud GPU Infrastructure
SU002 Data Center Dynamics Crusoe Pauses Work on 1.8GW Cheyenne Wyoming Data Center at the Request of Our Customer
SU003 Data Center Dynamics Meta Signs 1.6GW Capacity Agreement with Crusoe (Report)
SU004 Heatmap News Crusoe Energy and the Stargate AI Infrastructure Race
SU005 Crusoe Crusoe and Upstream Data to Resolve Lawsuit
SU006 Blockspace Media Crusoe Energy — Upstream Data Patent Dispute Ends with Mixed Ruling
SU007 CourtListener Crusoe Energy Systems LLC v. Upstream Data Inc. — Federal Docket
SU008 Worldmetrics Crusoe Energy Statistics and Key Facts 2026
SU009 SiliconAngle Crusoe AI Puts Energy First to Power Sustainable AI Infrastructure at Scale — PureAccelerate
SU010 Sacra Crusoe Energy — Company Research Profile
SU011 Cheddarflow Crusoe Energy Deep Dive — Analysis of the AI Infrastructure Pioneer
SU012 Business Insider (Markets) Crusoe's Contracted AI Infrastructure Capacity Approaches 5 Gigawatts
SU013 Data Center Dynamics Crusoe's Abilene Data Center Officially Live, Serving Oracle and OpenAI's Stargate
SU014 SemiAnalysis / ClusterMAX Crusoe Cloud — ClusterMAX GPU Cloud Review
SU015 Crusoe Crusoe Cloud — GPU Cloud Platform
SU016 Crusoe Crusoe | The Energy-First AI Factory Company
SU017 Crusoe Crusoe 2025 Impact Report
SU018 EIN Presswire Crusoe Cloud Achieves Gold Status in the GPU Cloud ClusterMAX Rating System from SemiAnalysis
SU019 DeployBase Crusoe Cloud Review 2026 — Pricing, Performance, Pros and Cons
SU020 Crusoe Crusoe Expands NVIDIA Collaboration for AI Infrastructure
SU021 Crusoe Crusoe Launches Managed Inference Delivering Breakthrough Speed for Production AI
SU022 Forbes From Gigawatts to Grab-and-Go: Crusoe Leans into Modular AI Data Centers
SU023 Crusoe Energy Systems Crusoe Energy — About
SU024 Crusoe Crusoe Cloud Achieves SOC 2 Type II
SU025 Crusoe (GitHub) Crusoe Cloud — GitHub Organization
SU026 Built In Crusoe Energy Systems — Company Stability and Growth Profile
SU027 DataCenters.com Crusoe's $400M AMD Deal — A Game Changer for AI Data Centers
SR001 Heatmap News Crusoe Is Pushing the Definition of Climate Tech According to Business Insider, public filings with the Texas Commission on Environmental Quality show that so far, Crusoe plans to operate on-site natural gas turbines at the Abilene facility totaling 360 megawatts of power.
SR002 Environmental Defense Fund EPA Issues Guidance on Oil & Gas Methane Pollution Flaring Standards Nearly $5 billion in methane has been wasted since the Trump administration's EPA stopped enforcing oil and gas methane regulations.
SR003 Bureau of Land Management 2024 Waste Prevention Rule | Bureau of Land Management BLM will not enforce these December 10, 2025, deadlines for an additional year until December 10, 2026.
SR004 Harvard Environmental and Energy Law Program BLM Methane Waste Prevention Rule – Environmental and Energy Law Program June 22, 2026 The Department of the Interior announced proposed revisions to the Biden administration's methane waste prevention rule.
SR005 US Environmental Protection Agency Controlling Air Pollution from Oil and Natural Gas Operations
SR006 US Environmental Protection Agency EPA's Final Rule to Reduce Methane and Other Harmful Pollution from Oil and Natural Gas Operations December 2, 2023 -- EPA has issued a final rule that will sharply reduce emissions of methane and other harmful air pollution from oil and natural gas operations.
SR007 Crusoe (Official) Crusoe, Upstream Data to Resolve Lawsuit Both parties have agreed to dismiss the lawsuits and enter into a license agreement on confidential terms.
SR008 Blockspace Media Crusoe Energy, Upstream Data patent dispute ends with mixed ruling The USPTO ruled that Crusoe Energy proved that the concept of powering a bitcoin mine with a natural gas generator is unpatentable, but it failed to prove that specific design aspects of Upstream Data's oil and gas bitcoin mining systems are also unpatentable.
SR009 CourtListener (Free Law Project) Crusoe Energy Systems LLC v. Upstream Data Inc., 2:25-cv-00743 – CourtListener.com ORDER DISMISSING CASE WITH PREJUDICE. Signed by District Judge Rodney Gilstrap on 10/28/2025.
SR010 Utility Dive Texas law gives grid operator power to disconnect data centers during crisis Data centers and other large, non-critical power consumers connected to the Electric Reliability Council of Texas transmission grid must accept curtailment during firm load shed events under a landmark law Republican Gov. Greg Abbott signed Friday.
SR011 Crusoe (Official) Crusoe's 2025 Impact Report: Building sustainable intelligence Divested our Digital Flare Mitigation® business to focus on vertically integrated AI infrastructure.
SR012 DataCenterKnowledge Texas Gets Tough on Data Center Power – Who's Next? SB6 applies to large-load customer interconnecting within ERCOT… targets loads of 75 MW or more.
SR013 Data Center Dynamics Crusoe pauses work on 1.8GW Cheyenne, Wyoming, data center "at the request of our customer" At the request of our customer, Crusoe has paused its development activities.
SR014 Data Center Dynamics Crusoe's Abilene data center officially live, serving Oracle and OpenAI's Stargate
SR015 DCPulse Crusoe Raises $1.38B to Supercharge AI Data Centers Power is very scarce right now… there have been a lot of commitments made in the space where I can't vouch for the reality that they will happen.
SR016 Crusoe (Official) Crusoe, the AI factory company, raising $1.375 billion at a valuation above $10 billion Crusoe's power pipeline grew over 4x and is now over 45 gigawatts.
SR017 Data Center Dynamics Crusoe raises $1.375bn in latest funding round
SR018 DCPulse Crusoe Backs AMD with $400M: Can Startups Take on NVIDIA in AI? On June 12, 2025, CEO of Crusoe, Chase Lochmiller, announced a USD400 million investment plan in AMD's MI355X chips. The deal includes approximately 13,000 units.
SR019 Crusoe (Official) Crusoe Expands NVIDIA Collaboration Across the Full AI Factory Stack
SR020 Crusoe (Official) Crusoe Newsroom | Company news & AI announcements Crusoe's Contracted AI Infrastructure Capacity Approaches 5 Gigawatts Across Data Centers and Cloud.
SR021 ESG News Crusoe Secures $600M to Scale Clean Energy-Powered AI Infrastructure Platform Customer count increased 7x in the past year.
SR022 Built In Crusoe Energy Systems Company Growth, Stability & Outlook 2026 Concentrated Customer Base: Major milestones and visibility are closely tied to marquee partners and anchor campuses (e.g., Abilene for OpenAI/Oracle), where reports have noted evolving plans.
SR023 World Metrics Crusoe Energy Statistics: 2026 Market Report
SR024 Data Center Dynamics Meta signs 1.6GW capacity agreement with Crusoe — report
SR025 Crusoe (Official) Crusoe Unveils Crusoe Edge Zones and Crusoe Spark Modular AI Data Centers
SR026 AI Curator Crusoe Energy Scores Massive $1.38B AI Computing Funding
SR027 Economic Times AI data centre startup Crusoe raising $1.38 billion in latest funding round
SR028 Cheddar Flow Crusoe Energy: Deep Dive Analysis of the AI Infrastructure Pioneer Crusoe Energy demonstrated exceptional financial performance in 2024, with revenue reaching an estimated $276 million, representing an 82% year-over-year increase.
SR029 Crusoe (Official) Crusoe Cloud Achieves SOC 2 Type II
SR030 Crusoe (Official) Crusoe Launches Managed Inference
SR031 Business Insider via Markets Insider Crusoe's Contracted AI Infrastructure Capacity Approaches 5 Gigawatts
SR032 Multiples.vc CoreWeave IPO Valuation Deep Dive CoreWeave (trading under ticker CRWV) went public on Friday 27th March 2025, valued at $23B (enterprise value of $30B).
SV001 SiliconAngle AI-focused data center startup Crusoe raises $600M at $2.8B valuation
SV002 Data Center Dynamics Crusoe raises $1.375bn in latest funding round
SV003 Crusoe Energy Systems Crusoe Announces Series E Funding
SV004 AInvest $10 Billion Valuation — Crusoe's AI Infrastructure Gold Rush
SV005 DCPulse Crusoe Raises $1.38B in AI Data Centers Funding
SV006 Sacra Research Crusoe Company Profile and Financial Estimates
SV007 Cheddar Flow Crusoe Energy Deep Dive — Analysis of the AI Infrastructure Pioneer
SV008 Multiples.vc CoreWeave IPO Valuation Deep Dive
SV009 U.S. Securities and Exchange Commission CoreWeave S-1 Registration Statement Index (Filed 2025-02-28)
SV010 Heatmap News Crusoe Energy Joins Stargate — but the AI Giant Plans to Burn Natural Gas
SV011 AInvest Crusoe Energy $750M Inflection Point — AI Infrastructure Play
SV012 Reuters Crusoe Energy, backed by Nvidia, to raise $750 million — sources
SV013 Data Center Dynamics Meta signs 1.6GW capacity agreement with Crusoe — report
SV014 Data Center Dynamics Crusoe's Abilene data center officially live, serving Oracle AI
SV015 Markets Insider / Business Insider Crusoe's Contracted AI Infrastructure Capacity Approaches 5 GW
SV016 Crusoe Energy Systems Crusoe 2025 Impact Report
SV017 DeployBase AI Crusoe Review 2026 — Pricing, Performance, Pros and Cons
SV018 ClusterMax AI Crusoe Cloud Review — ClusterMax GPU Benchmark
SV019 ESG News Crusoe Secures $600M to Scale Clean Energy-Powered AI Infrastructure
SV020 WorldMetrics Crusoe Energy Statistics and Key Metrics
SV021 Form Energy Form Energy and Crusoe Announce Agreement for 12 GWh of Iron-Air Batteries for AI Data Centers
SV022 Business Wire Energy Vault and Crusoe Announce Strategic Framework Agreement
SV023 Latitude Media Crusoe and Redwood Materials Are Powering a Data Center with Old EV Batteries
SV024 Data Center Dynamics Crusoe pauses work on 1.8GW Cheyenne, Wyoming data center at customer's request
SV025 SiliconAngle Crusoe AI Puts Energy First to Power Sustainable AI Infrastructure at Scale
SV026 Crusoe Energy Systems Crusoe Cloud Pricing Page
SV027 DataCenters.com Crusoe's $400M AMD Deal — A Game-Changer for AI Data Centers
SV028 Crusoe Energy Systems Crusoe Introduces Crusoe Spark Modular AI Data Centers
SV029 EIN Presswire Crusoe Cloud Achieves Gold Status in the GPU Cloud ClusterMax Rating System
SV030 Crusoe Energy Systems Crusoe Expands NVIDIA Collaboration
SV031 Forbes From Gigawatts to Grab-and-Go — Crusoe Leans Into Modular Data Centers
SV032 Crusoe Energy Systems Crusoe Cloud SOC 2 Type II Certification
SV033 Crusoe Energy Systems Windsurf Cost Savings Using Crusoe Cloud GPU Infrastructure
SV034 U.S. Securities and Exchange Commission CoreWeave Inc. SEC EDGAR Filing Index
SV035 Silicon Republic Crusoe valued at more than $10bn with closure of $1.37bn Series E funding
SV036 CoreWeave Investor Relations CoreWeave Announces Pricing of Initial Public Offering