初创公司尽调
尽调报告 AI therapeutics / AI-enabled drug discovery biotech private, late-preclinical 2026-05-12

insitro

真实合作伙伴验证和有分量的平台野心都在,但价格纪律仍比叙事热情更重要

继续研究:insitro 已拿到真实合作伙伴背书,也有机会跑出 techbio 溢价价值,但如果没有清晰条款和更强临床就绪证据, 公开材料还不足以支撑激进私募估值。

封面要素

已披露资本总额 01
800 USD M [CV002]
最近融资锚点 02
2570 USD M post-money [CV006]
投资建议 03
research-more [CV028]
风险评级 04
High [CV030]
公开可比区间 05
0.54-2.46 USD B [CV016]
成立时间 07
2018 [CO001]
内部验证阶段 08
IND-enabling / FIH prep [CV014]

公司概况

insitro 是一家位于南旧金山的 AI 药物研发公司,由 Daphne Koller 于 2018 年创立。公司已从机器学习药物发现平台,演进为覆盖代谢、神经科学和眼科的私有管线建设者。公开证据显示,借助 BMS、Lilly 以及更早期 Gilead 合作的经济条款,insitro 的外部验证强于许多 TechBio 同行;但更准确的理解仍是:它是一家后期私有、处于临床前到 IND 准备阶段的平台叠加管线型生物科技公司,而不是商业化软件业务。

官网
www.insitro.com
成立时间
2018-01-01
创始人
Daphne Koller
创立地点
South San Francisco, California, USA
总部
South San Francisco, CA
产品
insitro 将多模态人群队列和细胞数据、机器学习,以及内部和合作伙伴发现工作流组合在一起。可见组合覆盖代谢、神经科学和眼科;CTRO-1013 仍是最清晰、接近进入临床的内部资产。
客户
公开客户验证集中在少数大型药企合作方,而不是广泛的软件客户群;BMS 是最强的公开持续性信号,Lilly 是第二个主要合作伙伴,Gilead 则提供历史验证。
商业模式
公司模式把带里程碑付款的药企合作,与更长期的内部管线期权价值揉在一起。公开定价、续约和收入确认细节仍不透明,因此更适合把 insitro 看成经合作伙伴验证的 TechBio 平台,而不是定价透明的产品公司。
阶段
private, late-preclinical
融资情况
最近一次明确披露的融资是 2021 年 $400M Series C。Forge 估计该轮投后估值约 $2.57B;低置信度的公开追踪器仍把 insitro 集中在低 $2B 区间。管理层现在称累计资本约 $800M,累计合作收入约 $150M,但当前融资条款和优先权结构并未公开。
[CO001, CO002, CO003, CO004, CO005, CV002, CV005, CV006]

执行摘要

主要优势

  • BMS 通过首付款、里程碑转化、续约资金和新增靶点提名,给一家私营 techbio 提供了少见的公开背书。
  • 公开材料提到约 $800M 融资和约 $150M 合作收入,说明它拿到的财务与商业验证强于典型临床前平台故事。
  • 如果 CTRO-1013 和后续自研资产进入临床,同时合作继续扩张,平台加管线模式能给 insitro 多条价值创造路径。
  • Amy Abernethy、Joe Hand 等治理和人才补强,说明公司在努力从早期叙事型创业公司走向更成熟的组织。

主要风险

  • 当前价格、股权结构表、清算优先权和最新融资条款都未公开,公开证据无法直接判断入场纪律。
  • 自研资产价值仍停留在临床前:CTRO-1013 处于 IND-enabling / 首次人体试验准备阶段,还没有人体数据。
  • 收入、当前现金和烧钱速度仍过于不透明,无法可靠做收入倍数或现金跑道驱动的估值模型;公开数据源之间分歧很大。
  • 公开 AI-techbio 可比公司大多落在 $0.5B-$2.5B 区间;没有更强私下证据,能支撑的溢价有限。
  • AI 验证、安全和监管就绪材料仍未公开,而这些材料会直接影响合作伙伴信心和后期融资。

未决问题

  • 当前完全摊薄股权结构表和清算优先权瀑布
  • 当前现金余额、月度烧钱速度和下行情景现金跑道计划
  • 2021 年 Series C 之后的任何定价融资、老股标记或 409A 更新
  • BMS 和 Lilly 合同条款、续约机制和收入确认处理
  • CTRO-1013 及相关工作流的 IND 关键路径和 AI 验证包
  • 后期投资人或战略买家预期看到的安全、质量和审计材料

目录

Chapter 01

01公司概况

1.1 身份、使命与当前阶段

insitro 把自己定位为建立在因果生物学上的 AI 药物公司,而不是纯软件供应商或工具提供商。公司由 Daphne Koller 于 2018 年创立,总部位于南旧金山 East Grand Avenue 279 号;公开使命是用多模态人类和细胞数据加机器学习来解码疾病。这个定位很关键,因为 insitro 现在既不是在出售单一治疗资产,也不是在出售横向机器学习平台;它想搭出一台可复用引擎,从靶点发现推进到特定模态药物设计,再推进到内部控制的管线资产。公开网站和使命页面显示,公司仍扎根药物发现和开发,而不是商业化。它没有上市产品,但披露了项目组合,也有不断增加的数据和设计合作;这让公司超出种子前「AI 赋能生物科技」阶段,进入后期私有、支撑临床前到 IND 准备的运营阶段。[CO001, CO002, CO003, CO004, CO005, CO029]

快照 KPI 表
指标数值或状态日期信心缺口或备注
成立时间20182018由官方宗旨页面和 Forbes 资料支持。
总部279 East Grand Avenue,South San Francisco,CA(总部地址)当前官方地址明确。
运营阶段未上市 AI 治疗药物公司;商业化前,无上市产品2026未公开披露临床阶段或商业化产品。
公开披露风险融资截至 2020 年 $243M;2021 年 Series C 轮 $400M2020-2021融资时间线公开,但当前总额取决于是否计入合作款。
公司披露的资本口径$700M 以上至约 $800M(含合作款)2024-2026不能与纯股权融资直接对比。
跟踪网站估值区间$2.2B 至 $2.5B2025-2026公开跟踪网站相互矛盾,且没有公开的 2026 年定价融资。
员工数区间~230(裁员后);跟踪网站显示 ~250 至 3002025-2026第三方跟踪网站的公开规模指标相互矛盾。
已披露药企合作方Gilead、Bristol Myers Squibb 与 Lilly2019-2026公开证据显示合作伙伴集中度高。
已披露数据或研究合作方Genomics England 与 INSIGHT at Moorfields2022-2025关系到数据集获取和模态扩展。
客户数 / 债务 / 老股交易当前已审阅的开放来源均未披露这些指标。

本表混合了官方披露和低置信度跟踪估算;空值表示该指标在已审阅开放来源中仍无支撑。

[CO001, CO002, CO011, CO012, CO021, CO022]
FO002: 公司快照逻辑

insitro 的数据、平台、合作伙伴和治疗项目如何拼成一个价值创造闭环。

这是公开运营模式的结构化呈现,而不是管理层披露的流程图。

[CO003, CO004, CO018, CO019, CO020, CO027]

1.2 领导层、治理与组织匹配

创始人兼 CEO Daphne Koller 仍是公司最核心的公开面孔,也是创始人与市场匹配最清晰的来源。外部报道把她的可信度与深厚机器学习专长、此前创办 Coursera 绑定;公司材料则强调,她在打造跨学科文化,把数据科学、生物学和药物开发揉到一起。2021 年融资周期之后,董事会和高管新增成员让治理更有药物开发和临床证据深度。Paul McCracken 通过 Series C 融资加入董事会;Amy Abernethy 于 2024 年加入,带来 FDA 和真实世界证据经验;Joe Hand 于 2026 年出任首席人力官,随公司扩张而专业化人才战略。这些任命说明公司在为更成熟的组合管理和组织纪律做准备。与此同时,公开来源仍未披露完整董事会控制图、继任规划或优先权结构,因此治理方向可见,经济细节仍不透明。[CO006, CO007, CO008, CO009, CO010, CO033]

领导层与创始人表
人物职务重要性公开依据关键依赖或缺口
Daphne Koller创始人兼 CEO为公司提供机器学习可信度,也是最明确的创始人-市场匹配锚点。Forbes 2019/2020 与官网关键人物依赖高;未发现公开继任计划。
Amy Abernethy董事(自 2024 年起)为治理补上 FDA、真实世界证据和临床开发经验。官方董事会公告董事会经济安排和委员会结构未公开。
Joe Hand首席人力官(自 2026 年起)显示公司正从研究型创业公司转向具备正式人才架构的规模化组织。官方任命公告任命并不能厘清当前确切组织规模或流失趋势。
Paul McCrackenSeries C 轮进入董事会的董事代表 CPP 支持的治理影响力和长期资本协同。Series C 轮融资公告投票权或投资人控制权没有公开细节。

覆盖范围是部分性的,仅限已审阅开放来源中出现的公开具名高管或董事。

[CO006, CO007, CO008, CO009, CO010, CO033]

1.3 资本结构、合作经济与利益相关方图谱

公开来源显示,insitro 的资本栈由股权融资和合作现金共同搭成,这也是公开数字相互打架的原因。Forbes 报道 2020 年 Series B 后累计风投资金达到 $243 million;insitro 2021 年新闻稿记录了由 CPP Investments 领投的 $400 million Series C。但公司后续公告在把合作现金计入后,先后引用超过 $700 million 和约 $800 million 的资本,而第三方追踪器发布的融资和估值数字彼此不一致。更干净的结论是:insitro 吸引了顶级私募财团和有分量的药企现金,但任何单一公开数字都不应被当成权威。合作伙伴图谱在经济上很重要:Gilead 验证了早期 NASH 逻辑;Bristol Myers Squibb 仍是财务影响最大的神经科学合作方;Lilly 通过 2024 和 2025 年协议拓宽了代谢策略。这种集中度同时解释了上行空间和融资风险,因为少数交易对手驱动了大部分公开外部验证、里程碑流入和下游期权价值证据。[CO011, CO012, CO013, CO014, CO015, CO016]

利益相关方或投资人图谱
利益相关方角色经济重要性重要性尽调追问
Andreessen Horowitz早期领投方早期轮次核心股权财团的一部分显示顶级科技投资人支持 ML 逻辑。确认当前持股和董事会权利。
CPP InvestmentsSeries C 轮领投方及董事席位支持方领投 2021 年 $400M 轮,并让 Paul McCracken 进入董事会代表长期机构资本。索取当前持股比例及任何按比例认购权。
Bristol Myers Squibb神经科学合作伙伴首付款、里程碑款和潜在数十亿美元下游经济权益公开披露最清晰、带里程碑款的药企合作关系。按靶点索取当前里程碑时间表和期权结构。
Eli Lilly代谢疾病和 ADMET 数据合作方提供技术、数据,以及未来里程碑款或特许权使用费路径拓宽代谢领域执行和化学能力。逐协议索取经济条款,以及围绕权利保留的义务。
Gilead早期 NASH 验证合作方用数据和里程碑潜力帮助验证原始平台证明大型药企早期愿意为该平台付费。确认该关系是否仍活跃且仍具经济相关性。
Genomics England数据与研究合作方提供接入 NHS 相关多模态基因组和病理数据的通道关系到临床数据规模和发现工作流。厘清期限、排他性和下游 IP 权利。
INSIGHT at Moorfields(合作方)数据与研究合作方提供 35M 张眼科图像数据集,用于神经退行性疾病研究扩展疾病范围和基础模型训练数据。厘清衍生模型和靶点洞察的权利。
CombinAbleAI / AION 生态收购及生物制剂能力增加模态广度,而非直接现金强化 insitro 的跨模态设计栈。索取交易后留任和整合计划。

这是公开利益相关方图谱,不是股权结构表。经济重要性只是方向性判断,因为持股比例和合同瀑布条款未公开。

[CO013, CO014, CO015, CO018, CO019, CO020]
FO003: 快照 KPI

判断 insitro 当前成熟度和可见度时,最重要的高阶运营指标。

数值混合了公司披露区间和第三方追踪估计;当数字冲突时,该项展示区间,而不是给出虚假的点估计。

[CO021, CO022, CO023, CO024, CO029, CO032]

1.4 里程碑与管线演进

阅读 insitro 轨迹,最有用的方式是看一串里程碑如何把数据与机器学习逻辑转成更宽的治疗组合。早期验证来自 Gilead 合作和 2020 年 Bristol Myers Squibb ALS 协议,说明大型药企愿意为 insitro 的发现引擎付费。下一阶段加入更大规模股权支持和 Genomics England 等研究合作,随后又通过 Lilly 代谢疾病合作、Bristol Myers Squibb 带里程碑付款的 ALS 靶点工作,更明确地搭建管线。到 2025 和 2026 年,运营模式明显更工业化:公司通过 Moorfields 扩大眼科和神经退行性疾病数据网络,在管线页面披露肥胖和 MASH 项目,并在收购 CombinAbleAI 后推出 TherML。当前披露组合覆盖代谢、神经科学和眼科,共有八个命名项目,但仍没有上市产品,也没有公开临床阶段资产。因此,里程碑时间线比传统商业化时间线更重要。[CO005, CO014, CO015, CO016, CO017, CO018]

里程碑表
日期事件类型金额或状态参与方含义
2018Daphne Koller 在 South San Francisco 创立 insitro创立公司成立Daphne Koller启动 ML 优先的药物发现逻辑。
2019-04外部报道公开描述 Gilead NASH 合作合作首付款 $15M;据 Forbes 潜在最高 $1Binsitro、Gilead平台最早的药企验证。
2020-05Series B 轮融资融资$143M;当时 VC 总额 $243Minsitro、a16z、T. Rowe、BlackRock、Casdin、CPP 等为公司越过概念验证阶段后的扩张补足资本。
2020-10Bristol Myers Squibb 神经科学合作合作首付款 $50M;近期 $20M;潜在里程碑款 >$2Binsitro 与 Bristol Myers Squibb形成公开披露中最大的一笔合作经济权益。
2021-03Series C 轮融资融资$400Minsitro、CPP 及财团引入跨界资本并加深董事会资源。
2022-03Genomics England 合作合作向量嵌入搜索部署到 NHS 相关数据集insitro、Genomics England扩大多模态临床数据获取。
2024-04Amy Abernethy 加入董事会治理新增董事insitro、Amy Abernethy补上临床数据和 FDA 经验。
2024-10Lilly 代谢疾病协议合作保留权利的合作结构insitro、Eli Lilly拓宽代谢管线和模态选项。
2024-12首笔 BMS ALS 靶点里程碑付款规模化$25M 里程碑款insitro 与 Bristol Myers Squibb证明靶点提名取得进展。
2025-03Moorfields INSIGHT 合作合作35M 眼科图像资源insitro 与 INSIGHT at Moorfields增加眼科和神经退行性疾病数据规模。
2025-05裁员反向裁员 22%;裁后约 230 名员工insitro显示进入临床前已承受资本纪律压力。
2025-10BMS ChemML 扩展合作最高 $20M 新资金insitro 与 Bristol Myers Squibb把 ALS 工作从生物学推进到分子设计。
2026-01收购 CombinAbleAI 并推出 TherML产品模态无关的治疗设计平台insitro、CombinAbleAI扩展到生物制剂,并新增以色列研发中心。
2026-02任命 Joe Hand 为首席人力官治理高管任命insitro、Joe Hand显示组织围绕人才体系扩张。
2026-02BAT 研究和肥胖靶点披露产品小鼠体重降低 15%insitro显示较新的代谢管线取得进展。
2026-03BMS 合作扩展至 ALS-2 和 ALS-3合作$10M 里程碑付款insitro 与 Bristol Myers Squibb加深神经科学管线宽度。

本时间线仅涵盖已审阅公开来源中明确可见的里程碑,因此不包含未披露的内部事件或融资事件。

[CO001, CO007, CO008, CO010, CO011, CO012]
FO001: 公司里程碑时间线

展示从创立到 2026 年 3 月的公开融资、合作关系、治理变化和反向事件。

金额和日期仅反映公开披露;私下股权结构事件可能缺失。

[CO001, CO011, CO012, CO014, CO015, CO016]

1.5 反向信号与未决尽调缺口

公开来源中最清晰的反向信号是 2025 年裁员。BioPharma Dive 报道 insitro 裁员 22%,剩余约 230 名员工,并明确把此举与临床准备和更艰难的融资环境相连。这个事实很重要,因为当前追踪器页面仍给出差异很大的员工数、融资总额和估值估计,公司快照没有表面叙事那么精确。收入质量也有同样模糊性:低质量追踪器发布收入估计,但没有经过审阅的公开来源给出经审计的当前收入构成或资产负债表桥接。合起来看,反向图景不是 insitro 缺乏技术野心或合作伙伴验证;问题在于,公开尽调记录在投资人最需要精确度的地方很薄。当前估值、客户集中度、债务或老股交易历史、以及精确当前运营规模都仍未解决,后续章节应把这些视为开放的承销问题,而不是既定事实。[CO022, CO023, CO024, CO034, CO036, CO040]

Chapter 02

02市场分析

2.1 市场边界、邻近领域与现状替代方案

分析 insitro 最容易犯的第一个错误,是简单称其市场为「药企 AI」或「肥胖药物」。公司还不是商业化药物销售方,但也不再是纯工具供应商。它自己的平台和管线页面显示,商业模式是用多模态人类和细胞数据、机器学习以及设计工具,产出内部项目和面向合作伙伴的发现结果。因此,近期可变现市场是药企研发买家的合作与里程碑经济;下游经济上行空间则落在大得多的治疗终端市场,只有内部资产进入更后期开发时才真正重要。正确边界有三层:AI 驱动的药物发现合作、insitro 控制的代谢 / 神经科学 / 眼科治疗项目,以及强化平台的数据密集型发现合作。现状替代方案是药企内部发现团队、传统 CRO 和药物化学工作流,以及 Recursion、Schrödinger、Relay、Evotec 等竞争性 AI 生物制药平台,而不是消费者数字健康工具或医院 IT 软件。[CM001, CM002, CM003, CM004, CM040, CM041]

市场定义表
细分 / 类别纳入的支出或活动排除的支出买方 / 付款方与 insitro 的相关性
AI 赋能药物发现合作与靶点发现、模型构建和项目设计挂钩的首付款、里程碑款、期权费和研究资金商业化药品销售、医院软件、广义消费者健康应用大型药企 R&D 和 BD 预算即时变现路径;insitro 今天签约就在这里
内部代谢疾病治疗药物如果资产推进,来自肥胖、MASLD 及相关代谢项目的未来收入基层慢病管理服务或健康订阅未来处方医生、付款方和专科药企渠道重要长期上行空间,但当前大多仍在商业化前
内部神经科学治疗药物通过合作或内部资产获得的 ALS、FTD 及相关神经靶点经济权益与药物无关的神经科医疗服务支出当前是药企合作伙伴;之后是临床医生和付款方当前 BMS 合作已让神经科学具备商业相关性
眼科与神经退行性疾病数据层改善靶点选择和患者分层的视网膜影像与多模态发现输入通用医院 IT、PACS 软件和常规眼科服务研究合作方和未来药企用户属于战略护城河和发现输入,不是清晰独立收入类别
现状替代栈药企内部发现、传统 CRO 工作流、药物化学和竞争对手 AI 平台消费者数字健康、计费软件或通用医院分析已分配给既有工作流的现有药企预算界定 insitro 要赢得预算必须替代什么

这些行定义的是市场边界逻辑,不是可相加的总可用市场(TAM)桶。纳入和排除的支出刻意把当前买方经济性与未来治疗药物上行空间分开。

[CM001, CM002, CM003, CM004, CM041, CM042]
FM001: 从边界到需求的市场规模测算视角

用示意性市场层级,从宽泛的生物医药预算池一路收窄到 insitro 的合作伙伴和特定疾病需求视角。

这些层级是证据视角,不是可相加的市场桶。公开来源衡量口径不同,因此保留币种和单位差异,而不做标准化。

[CM001, CM002, CM013, CM018, CM022, CM025]

2.2 规模测算视角:AI 市场预测、研发预算与疾病负担需求

公开规模数据存在,但无法整齐对齐成一个可承销 TAM。药企 AI 和药物发现 AI 的分析师页面方向上看多,预测从当前低个位数十亿美元,到 2030 年代初的一百多亿美元中段,甚至二百多亿美元中段。但这些估计混合了不同类别边界:有的纳入广义药企 AI 工作流,有的隔离药物发现,有的包括更宽的精准医疗软件栈。第二个视角是买家背后的预算池:EFPIA 估计 2024 年仅欧洲药企研发支出就有 EUR 55 billion,说明足以资助合作的买家基数有多大。第三个视角是疾病负担带来的需求拉动。WHO 报告 890 million 成年人患有肥胖;insitro 的 Lilly 新闻稿称 MASLD 在美国影响约 100 million 人;WHO 报告 2.2 billion 人存在视力损害;CDC 估计美国有 9.6 million 糖尿病视网膜病变患者。这些数字确认了 insitro 项目周围有巨大的需求域,但不能等同于软件或合作 TAM。最干净的结论是:insitro 的市场必须通过多个受证据约束的视角测算,而不能靠一个通用标题数字。[CM009, CM010, CM011, CM012, CM013, CM014]

TAM/SAM/SOM 或市场规模视角表
发布方 / 视角年份地域数值CAGR方法或单位信心关键限制
McKinsey2025/2030全球2025 年 $4B 至 2030 年 $25.7Bn/a制药行业 AI 市场类别较宽且是未来年份预测,并非 insitro 专属
MarketsandMarkets2024/2029全球2024 年 $1.86B 至 2029 年 $6.89B29.9%AI 药物发现市场类别边界比更宽的制药 AI 估算更窄
Precedence Research2025/2034全球2025 年 $1.94B 至 2034 年 $16.49B27%制药 AI 市场周期长,工作流范围宽
Precedence Research2025/2035全球2025 年 $6.93B 至 2035 年 $17.81B9.9%AI 药物发现市场基准年份和方法与其他分析页面不同
Grand View Research2024/2030全球2024 年 $2.29B 至 2030 年 $14.53B36.23%AI 精准医疗市场精准医疗宽于 insitro 当前商业模式
EFPIA2024欧洲€55B 研发支出n/a药企研发预算池预算池不等于软件或合作业务的 TAM
WHO / Lilly / 肝病来源2022-2025全球 / 美国全球 890M 肥胖成人;美国约 100M MASLD 患者;美国 MASLD 患病率区间 10-46%n/a疾病负担视角疾病负担不等于可触达收入
WHO / CDC2022-2025全球 / 美国全球 2.2B 视力损伤人群;美国糖尿病视网膜病变患者 9.603Mn/a疾病负担视角疾病负担只能确认需求领域,不能说明 insitro 的获取率

本表有意混用预算、软件市场和疾病负担三种视角,因为已审阅公开来源没有拆出干净的 insitro 可服务市场(SAM)。各行只提供信息,不能相加。

[CM009, CM010, CM011, CM012, CM013, CM014]
FM002: 市场估计区间

以统一单位 USD billions 展示公开 AI 医药和 AI 药物发现市场预测的区间视图。

预测期限从 2029 到 2035 不等。低值和高值是包裹点估计的不确定性区间,用于呈现分散度;不应取平均或相加。

[CM009, CM010, CM011, CM012, CM013, CM014]

2.3 买家、用户、付款方与采用路径

insitro 的买家图谱取决于所处价值链位置。在当前合作主导模式中,买家通常是药企研发或业务发展组织;他们判断 insitro 的发现引擎、疾病模型或靶点工作是否值得前期资金和里程碑敞口。职能用户是双方科学团队,预算所有者是药企研发或外部创新预算。因此采用路径高度依赖证据:平台验证、疾病模型可信度、靶点提名,随后才是带里程碑结构的开发权谈判。在更后期的内部药物路径里,买家图谱会改变。只有临床证据和监管批准之后,医生、医疗系统和付款方才相关;对 insitro 来说,这条路径目前大多仍是假设。Genomics England 和 Moorfields 的数据合作并列作为赋能输入,而不是直接清晰的市场输出。这就是买家分析对估值如此重要的原因:公司眼前收入路径由成熟交易对手控制;他们在患者或付款方看到产品很久之前,就会要求证据。[CM005, CM006, CM007, CM008, CM035, CM036]

细分市场 / 买方图谱
细分市场买方用户付款方工作流场景预算负责人采用触发点
大型药企发现合作外部创新负责人、BD 负责人或治疗领域研发负责人合作方科学家、计算生物学家、转化团队药企研发预算平台评估、靶点提名、带里程碑的合作CSO、BD 委员会或治疗领域预算负责人有说服力的数据、靶点或模型证据
现有神经科学联盟已进入合作的大型药企伙伴联盟联合团队和项目负责人合作方研发和里程碑预算借助选择权或开发权推进已提名靶点联盟指导委员会和合作方财务靶点选择和里程碑就绪度
未来内部代谢或神经资产获批后才触达医院系统和专科处方医生医生、护理团队和患者商业保险、Medicare 和其他医疗体系试验后临床采用和报销付款方处方集和医疗服务方预算临床疗效、安全性和标签获批
数据和研究合作方学术机构或公共数据机构联合数据科学和生物学团队资助、研究或内部创新预算数据集访问、搜索工具和模型构建研究项目负责人治理、隐私和科学效用
竞争性 AI 生物医药生态寻求合作或资本的同业买方内部平台和管线团队VC、公开市场或合作方资本买方判断品类定位可信度的参照董事会和融资利益相关方临床深度、数据规模或一体化合作模型

当前合作路径和未来商业化路径下,买方与付款方角色差异很大;本表明确保留这种时间差。

[CM005, CM006, CM007, CM008, CM035, CM040]
FM003: 买方 / 细分市场证明负担图

矩阵展示 insitro 当前和未来市场路径中的关键细分市场,以及每个细分市场背负的证据负担。

取值是定性描述,不是调研测量。目标是按细分市场展示买方关系和证明负担。

[CM005, CM006, CM035, CM040, CM041, CM042]
FM004: 采用漏斗或价值链图

相对漏斗展示 insitro 当前商业路径如何从数据和模型价值逐步收窄到下游医疗支付。

取值是序数漏斗权重,不是经验转化率。它们用来说明市场在哪里收窄、哪里需要更多证据。

[CM031, CM032, CM041, CM046, CM047]

2.4 增长驱动、采用约束与保留的矛盾

insitro 所处市场的增长逻辑很直接。肥胖、MASLD、视力和神经退行性疾病负担巨大;药企研发预算庞大;AI 正在沿药物生命周期更深地渗透到发现、安全和运营环节。同时,该类别的采用约束格外重要。WHO 和 EMA 都强调治理、风险管理和生命周期控制。McKinsey 认为药企还没有看到清晰证据,证明 AI 单独就能显著缩短周期或提高成功率;KPMG 指出,更严峻的融资条件把买家推向风险更低的结构。这些事实意味着,热情不会自动转化为溢价经济。投资人需要保留两组张力,而不是把它们压平。第一,分析师市场预测看多,但彼此不可比。第二,疾病负担确认需求,却不能证明 insitro 能以有吸引力的转化率获得软件收入、合作收入或未来产品收入。因此,本章应继续保留围绕精确 SAM、买家转化和类别层面 AI 生产力证据的明确缺口。[CM015, CM016, CM017, CM019, CM020, CM021]

增长驱动因素与约束表
驱动因素 / 约束方向时间对 insitro 的含义尽调追问
全球肥胖负担和成本巨大增长驱动因素长期支撑代谢项目的相关性,也支撑药企投入大型慢病市场的意愿insitro 会优先瞄准哪些患者子群?
眼科和视力负担巨大增长驱动因素长期支持继续投入视网膜数据合作和神经退行性疾病发现insitro 如何把眼部数据优势转化为靶点或生物标志物差异化?
大型药企研发预算池增长驱动因素近期只要 insitro 能跨过证明门槛,买方就有实质预算哪些合作方预算今天真正可被 insitro 触达?
AI 在发现、安全和运营环节被采用增长驱动因素近期让更多买方工作流可能把 insitro 视为相关方案相对通用 AI 供应商,insitro 哪里最强?
治理和监管控制约束持续买方依赖 AI 输出前,验证和文档负担会变重insitro 今天向合作方开放了什么治理栈?
市场重估后偏好低风险交易约束近期买方会压低首付款,或把权重更多放到里程碑在回款后置的交易里,insitro 的经济性有多抗压?
缺乏 AI 单独改善临床结果的明确证据约束持续按溢价估值把品类叙事变现更难哪些证据显示 insitro 提高了转化率或里程碑产出?
开发和报销周期长约束持续发现阶段承诺转化为下游药物收入会被延迟从平台信号到付款方认可的资产,现实时间表是什么?

时间字段为定性判断。约束强调采用摩擦和变现节奏,而不是科学上不可行。

[CM015, CM017, CM018, CM020, CM021, CM025]
Chapter 03

03竞争格局

3.1 格局:直接同行、邻近公司、替代方案与潜在进入者

insitro 没有一个显而易见的单一对手。竞争集合至少分成三类。第一类是直接 AI 药物发现同行,它们同样试图把自有数据、机器学习和湿实验室工作流转成治疗管线或可合作项目。在已审阅公开来源中,最清晰的名字包括 Recursion/Exscientia、Insilico Medicine、Xaira、Isomorphic Labs、Generate Biomedicines、Valo Health 和 BenevolentAI。第二类是 Schrödinger、Evotec 等邻近替代方案,它们用不同包装解决相似买家问题:基于物理的计算发现、整合型研发服务或灵活合作。第三类是药企内部自建;只要大型买家选择把发现留在内部,而不是为外部平台付费,它仍是默认替代方案。这很重要,因为它意味着 insitro 不只在原始科学上竞争,还要竞争买家是否相信,其多模态人类数据方法足够区别于不断增殖的 AI 叙事和替代工作流。[CP001, CP002, CP003, CP004, CP011, CP014]

FP001: 竞争定位图:平台广度 vs 公开证据

象限图按平台雄心广度和已审阅来源中可见的公开证据或验证数量,对 insitro 和主要同行定位。

坐标轴使用从已审阅公开证据推导的 1-10 定性评分。x 越高,平台雄心越广;y 越高,公开证据、合作关系或临床推进越可见。

[CP002, CP006, CP009, CP012, CP017, CP018]

3.2 竞争者画像:规模、融资与公开验证信号

直接同行中,Recursion/Exscientia 是公开市场中最清晰的规模基准:大型生物数据护城河、>10 个内部项目、>10 个合作项目,以及数亿美元已实现合作现金。Insilico 在生成式 AI 同行中公开验证节奏最强,拥有 40 多个总项目、多个 IND 获批资产,以及披露的具体对外授权经济条款。Xaira 和 Isomorphic Labs 在私有同行中呈现最强的公开平台野心;Xaira 推动虚拟细胞逻辑,Isomorphic 则把源自 AlphaFold 的设计延展到大型药企合作和大额融资。Generate Biomedicines 和 Valo Health 的差异化更多来自模态或人类因果生物学角度,而不是清晰公开商业验证。BenevolentAI 仍是有意义的类别参照,因为其知识图谱传统让它成为早期 AI 药物发现领导者;但其战略重组和拟退市也显示,一旦执行或市场胃口转弱,类别叙事会迅速压缩。合起来看,这组画像表明 insitro 有差异化,但不是该类别的规模或融资领导者。[CP005, CP006, CP007, CP008, CP009, CP010]

竞品画像表
竞品类别公开规模 / 融资信号产品范围 / 目标客户关键差异化相对 insitro 的关键局限
insitro参照公司未上市;BMS 和 Lilly 提供公开合作方验证;管线覆盖三类疾病面向药企买方的 AI 治疗药物平台,以及内部 / 合作项目多模态人类和细胞数据,贯穿平台到管线不是公开资本规模领先者;公开定价透明度有限
Recursion / Exscientia直接 AI 平台同业>50 petabytes;>10 个内部项目和 >10 个合作项目;已实现合作方现金约 $450M;约 800 名员工面向药企合作方和内部管线的最宽 AI 发现平台工业化表型组学、化学能力和合作方现金验证覆盖范围太宽,可能稀释疾病聚焦;已审阅来源未披露上市药物
Insilico Medicine直接生成式 AI 同业40+ 个项目;13 条 IND 获批管线;2021-2024 年产出 20 个 PCC;多笔披露交易金额面向内部管线和对外授权的生成式 AI 药物发现产出节奏快,交易经济性罕见地可见相较 insitro,多模态疾病数据护城河没那么明显
Xaira直接前沿平台同业融资近 $1B;最大公开全基因组 Perturb-seq 数据集;推进虚拟细胞AI 优先,覆盖药物发现与开发全栈虚拟细胞雄心和异常雄厚的起步资本公开临床验证仍更早期
Isomorphic Labs直接前沿模型同业$600M 外部轮;与 Novartis、Lilly 和 J&J 合作基于 AlphaFold 继续扩展的 AI 药物设计,面向药企合作和内部发现AlphaFold 邻近品牌,预测式和生成式设计引擎公开管线细节薄于平台雄心
Generate Biomedicines相邻生物药生成器已生成、构建并测试 42,000 个蛋白;设施面积 140k+ sq ft生成式生物学和蛋白治疗蛋白设计和生物药角度强更偏特定模态,与 insitro 当前公开叙事不完全可比
Valo Health相邻人类数据同业人类因果生物学和闭环化学定位;生态系统驱动模型借 AI 指导的靶点和分子发现,并通过合作落地人类数据叙事与 insitro 自身论点共振已审阅来源中的公开验证点较薄
BenevolentAI相邻知识图谱同业本体和知识图谱投入十年;战略重整并拟退市面向生命科学研发和靶点发现的 AI 决策支持老牌 AI 药企品牌和知识图谱底子不利的重组证据削弱可信度
Schrödinger计算型替代方基于物理平台的合作项目和自有项目面向药企和内部管线的计算发现基于物理的模拟和已建立的平台身份没那么明显围绕多模态人类疾病数据
Evotec一体化服务替代方一体化研发价值链和灵活合作模型合作式药物发现和开发服务服务包装商业上更易读,模态覆盖也宽商业模式不同,纯 AI 护城河较弱

规模信号仅限已审阅公开来源披露内容。除非明确可见,私营公司现金和估值字段仍不完整。

[CP004, CP005, CP006, CP007, CP009, CP011]

3.3 能力、包装、GTM 与切换成本对比

能力比较的重点不是谁「使用 AI」,而是每家公司能为这个主张挂上哪类证据。insitro 的公开角度是多模态人类和细胞数据,连接到疾病理解和内部管线形成。Recursion 的公开优势是数据集和工业化规模。Insilico 的公开优势是速度和重复项目输出。Xaira 出售虚拟细胞野心,Isomorphic Labs 出售 AlphaFold 邻近的预测设计。Schrödinger 是最清晰的计算替代方案,Evotec 是最清晰的整合服务替代方案。公开定价比较远弱于能力比较。在已审阅来源中,经济条款通过合作结构、对外授权、里程碑组合、服务或合作姿态披露,而不是通过透明席位定价或菜单式包装披露。这让 GTM 和切换成本分析格外重要。在深度整合之前,买家很可能同时使用多个平台;但一旦数据、模型工作流或项目权利嵌入合作,切换成本会上升。与此同时,分销权力仍偏向大型药企和后期商业化组织,而不是 insitro 和多数私有同行。[CP023, CP024, CP025, CP026, CP027, CP028]

功能 / 能力矩阵
能力insitroRecursion / ExscientiaInsilicoXaira / Isomorphic相邻公司(Generate / Schrödinger / Evotec)
主要数据护城河多模态人类和细胞疾病数据大规模表型组学,加上化学和患者数据生成式模型,加上内部和合作项目数据虚拟细胞 / AlphaFold 风格预测模型蛋白生成、基于物理的模拟或一体化服务数据
湿实验室整合是,偏重细胞模型是,工业化湿实验室和自动化是,项目产出加化学引擎暗示有全栈雄心,公开细节不一因玩家而异;Evotec 强,Schrödinger 较轻
内部管线所有权是或正在发展因公司而异;Generate 最强,服务模型较弱
公开合作方验证披露了 BMS 和 Lilly 结构披露多个合作项目和已实现合作方现金数十项合作和披露交易金额Isomorphic 披露大型药企合作;Xaira 仍更早期合作或服务姿态可见,但经济性常不透明
模态宽度小分子,加上跨疾病领域的更宽设计雄心化学能力和平台覆盖较宽生成式化学,疾病范围宽前沿模型宽度;生物学引擎角度生物药、物理模型或一体化发现服务
商业包装可见度里程碑 / 权利结构,无席位定价合作项目和里程碑经济性对外授权、共同开发、里程碑偏融资和合作,定价不透明服务或软件包装在概念上可见,详细定价不透明

单元格只汇总公开证据。公开验证未知或较薄,不应被解读为能力缺失。

[CP004, CP011, CP014, CP016, CP019, CP020]
定价 / 包装对比
竞品公开包装 / 经济模型已披露验证点定价可见度可能买方含义
insitro发现合作,包含选择权、里程碑、版税和保留权利BMS 和 Lilly 交易结构已高层级披露大型药企研发和 BD卖的是战略项目准入价值,不是软件席位
Recursion / Exscientia合作项目加内部管线和里程碑现金Nasdaq 合并公告披露 >10 个合作项目和约 $450M 已实现合作方现金低-中大型药企和公开市场投资者尽管单项目定价不透明,经济性验证仍强于大多数同业
Insilico对外授权、共同开发、里程碑和内部管线推进ISM8969 共同开发最高 $66M;三项关键对外授权合计最高 $2.1B药企合作方和投资者在已审阅私营生成式同业中,公开经济性最透明
Isomorphic Labs战略性药企合作加大额融资轮与 Lilly、Novartis 和 J&J 合作;$600M 外部轮大型药企和 Alphabet 支持的战略生态包装偏战略合作,而不是透明交易
Schrödinger平台加合作项目和自有项目模型主页可见合作项目和自有项目药企研发以及材料 / 生命科学客户靠软件加项目叙事竞争,但公开详细定价很薄
Evotec灵活合作和一体化研发服务主页强调灵活合作模型生物医药研发买方服务式包装比 AI 平台叙事更容易让买方看懂
Xaira / Generate / Valo / Benevolent融资和平台主导的叙事,披露交易细节有限公开验证集中在数据集、平台主张或战略变化很低合作方、投资者和未来合作者公开定价对比仍高度不完整

本表比较经济包装,不比较标价。多数私营同业披露战略和合作,而不是可逐项对比的商业价格点。

[CP009, CP010, CP017, CP018, CP028, CP030]
FP002: 功能广度 / 能力与信任图

压缩矩阵展示哪些平台在广度、信任状态和商业可读性上看起来最强。

条目是定性判断,只基于已审阅公开证据。未知或较弱的公开证据,并不证明能力不存在。

[CP026, CP027, CP028, CP029, CP031, CP032]

3.4 护城河耐久性、商品化风险与反向竞争证据

看多 insitro 的最强理由,是公司不依赖单一算法叙事。它的护城河主张把多模态疾病数据、细胞模型、项目所有权和已展示的合作伙伴验证组合在一起。但这条护城河并非免疫商品化。如果买家在交易前越来越把 AI 药物发现供应商视为可互换,定价权就会被压缩。如果药企内部自建能力改善,外部平台胃口会变弱。如果 Recursion、Insilico、Xaira 或 Isomorphic Labs 等同行继续在资本、数据集广度或公开验证节奏上放大规模,它们可能塑造整个类别的买家预期。这里的反向证据很重要。BenevolentAI 的战略重组和拟退市说明,早期 AI 药物发现领导地位并不保证持久经济性;Exscientia 被 Recursion 吸收则说明整合已经开始。因此,竞争结论更有层次:insitro 有差异化,但只有后续证据显示其靶点质量更好、里程碑转化更强,或买家锁定比不断扩大的 AI 生物制药替代阵营更可防守,护城河才会显得持久。[CP036, CP037, CP038, CP039, CP040, CP041]

护城河耐久性 / 竞争风险登记表
护城河或风险当前方向公开证据对 insitro 的含义尽调追问
多模态疾病数据护城河潜在强项insitro 平台和管线叙事可能支撑差异化靶点和患者选择洞察相对同业和合作方,数据集有多独特?
合作方验证有保留的强项披露 BMS 和 Lilly 结构有助于提升买方可信度,但还不能证明品类领先结果相比同业,insitro 实际实现了哪些里程碑转化?
资本规模差距风险Xaira 和 Isomorphic 的公开融资信号;Recursion 的公开规模更大的同业可能塑造品类预期,并吸收更多失败成本相对前沿同业,insitro 的实际现金跑道是多少?
虚拟细胞和 AlphaFold 替代风险Xaira 和 Isomorphic 的雄心可能压缩 insitro 自身模型中心化主张的独特性insitro 在哪里明确优于虚拟细胞或结构 AI 路线?
生成式化学速度风险Insilico 产出节奏和已披露 IND 进展如果 insitro 显得更慢,产出更快的同业可能赢走买方注意力insitro 在靶点到候选物速度上如何对标?
多平台并用与内部自建风险KPMG / McKinsey 及替代方案集买方可能把预算分散到多个平台,或把工作留在内部哪些买方只用 insitro,哪些会与其他平台并用?
品类整合 / 脆弱性风险Exscientia 被 Recursion 吸收;BenevolentAI 重组并计划退市投资人与买方信心可能迅速收缩整合会提升买方对规模化龙头的信任,还是给专科平台打开空白空间?
分发不对称持续风险大型药企或既有服务商仍控制许多终端渠道insitro 仍依赖伙伴触达下游仍现实的分发或商业化路径有哪些?

本表区分两类内容:真正的护城河强项,以及公开证据仍太薄、尚不足以支撑耐久性的领域。

[CP021, CP022, CP027, CP028, CP029, CP033]
FP003: 护城河 / 就绪度 KPI

从公开证据看,insitro 当前竞争姿态有多稳,用紧凑指标呈现。

取值是有证据支撑的定性指标,不是财务 KPI。差值表示护城河耐久性的方向性压力。

[CP030, CP036, CP037, CP038, CP040, CP041]
Chapter 04

04财务情况

4.1 收入模式与定价:定制化合作经济,不是标价表

公开证据只支持一个清晰的变现结论:insitro 不是靠产品收入或公开席位制软件定价出售价值。相反,价值被打包进定制化药企合作和带里程碑付款的延展中。2020 年 Bristol Myers Squibb 交易披露 $50 million 前付款、$20 million 近期运营里程碑、超过 $2 billion 下游里程碑以及特许权使用费。到 2024 年末,insitro 又宣布一个 $25 million 的 BMS 里程碑;2025 年 ChemML 延展最高可再增加 $20 million 资金;2026 年靶点扩展公告又增加一个 $10 million 里程碑。Lilly 的 2024 年结构显示了另一种经济设计:insitro 保留完整全球权利,Lilly 则有资格获得里程碑和特许权使用费;这意味着合作可以加速资产,但不能充当清晰的年度收入流。官方 2026 年公司信息称,insitro 已在 BMS、Lilly 和 Gilead 上产生约 $150 million 合作收入,但该数字是累计且未经审计的。结果是,一个真实但难以标准化的收入基础:有意义的合作伙伴现金、薄弱的公开定价透明度,以及没有经审计的产品销售引擎。[CI001, CI002, CI004, CI005, CI006, CI007]

收入来源表
收入来源公开经济信号当前状态收入质量主要不确定性关键来源
BMS 2020 年合作$50M 首付款、$20M 近期运营里程碑、>$2B 下游里程碑、特许权使用费活跃,且持续扩展当前中低;若里程碑持续兑现,则改善已确认收入与递延收入各有多少BMS 2020 年公告
BMS 里程碑 / 扩展现金(2024-2026)2024 年 $25M 里程碑、2025 年最高 $20M 延期资助、2026 年 $10M 里程碑已实现且近期可见的非稀释性现金中;现金真实,但到账节奏和会计处理不透明延期资助是按期摊销、逐项里程碑确认,还是与成本轧差insitro 2024/2025/2026 年 BMS 更新
Gilead 历史合作Forbes 报道的 $15M 首付款,以及最高 $1B 潜在款项早期历史验证中低;可信,但信息偏旧,公开更新少当前状态和任何已实现里程碑未公开Forbes 2020
Lilly 2024 年战略协议insitro 保留完整全球权利;Lilly 有资格获得里程碑款和特许权使用费战略赋能,不是简单披露的年度收入来源当前收入贡献低,但长期期权价值更高近期现金经济条款披露不透明Lilly 2024 年公告
未来自有药物 / 特许权使用费收入来自内部代谢或神经资产,以及合作伙伴挂钩特许权使用费的潜力仍取决于临床就绪度和下游成功当前低;完全前瞻时间线、所有权分成和商业化姿态仍未公开insitro 官方材料

本表区分现金可见度与收入质量。真实现金事件本身,不能证明年度收入具备经常性或高质量。

[CI001, CI002, CI004, CI005, CI006, CI011]
定价 / 变现表
套餐 / 模式公开价格可见度已披露经济条款买方 / 交易对手收入确认含义解读
BMS 2020 年平台合作首付款 + 里程碑款 + 特许权使用费大型药企神经科学买方多要素安排;现金与 GAAP 收入确认节奏可能错位典型企业级药企合作,不是软件定价
BMS 2025-2026 年扩展工作最高 $20M 延期资助,加上 2026 年 $10M 里程碑款现有标杆客户扩展现金很可能绑定特定阶段和里程碑支撑先落地、再扩张的打法
Lilly 2024 年战略协议权利保留,加未来里程碑款 / 特许权使用费大型药企代谢领域伙伴经济条款可能更像资产期权,而非订阅合作能积累期权价值,但年度收入不清晰
Gilead 早期合作Forbes 报道的 $15M 首付款和长尾里程碑上行空间大型药企肝病领域伙伴历史现金对当前收入运行率解释有限可作为早期验证,但作为当前年度收入锚点偏弱
内部项目未来变现很低潜在产品收入、特许权使用费或后续授权未来支付方、伙伴或收购方当前没有收入确认依据主要上行情景仍是前瞻

insitro 公开可见的变现更像战略客户经济条款,而不是透明 SKU 定价。因此,价格可见度在结构上偏低。

[CI006, CI007, CI008, CI009, CI010, CI037]
FI001: 收入模型桥

流程图展示 insitro 如何把已融资本和平台投入转化为当下的合作现金,以及之后期权式的下游经济收益。

该图是结构图,不是预测。它映射公开披露所暗示的价值路径,而不是预测财务结果。

[CI001, CI002, CI003, CI006, CI007, CI008]

4.2 GTM 与销售效率代理指标:企业级药企账户和先落地再扩张动作

insitro 的市场进入动作更像面向大型药企的战略企业业务发展,而不是广泛软件分销。每份披露协议都是定制的、权利很重,并围绕治疗项目设计,而不是围绕订阅设计。BMS 关系尤其有启发性:2020 年发现合作、2024 年里程碑转化、2025 年 ChemML 延展,以及 2026 年新增靶点。这一串动作像先落地再扩张销售:先靠科学进展赚到信任,再通过后续工作变现。KPMG 和 McKinsey 都把药企 AI 采用描述为重证据、重合作,这与 insitro 的公开姿态一致:它是战略发现能力,而不是通用 AI 工具。由于没有公开来源披露获客成本、胜率或回本周期,销售效率只能用少数标杆账户随时间加深的能力来代理。公开证据说明 BMS 已经扩展;但还不能证明 insitro 拥有广泛、可复制、多客户商业引擎,且漏斗效率可衡量。[CI007, CI008, CI009, CI010, CI037]

4.3 成本结构、毛利率路径与公开牵引力

成本结构比验证更容易推断。insitro 运营湿实验室生物学、人类细胞数据生成、化学和大量算力,而不是轻量 SaaS 栈。2025 年 ChemML 延展特别强调了 192-H100 GPU 集群;公司更广泛的平台叙事也依赖大规模数据生成和内部管线工作。因此,即便长期里程碑或特许权使用费经济最终可能有吸引力,当前交付经济也不太可能像纯软件毛利。公开牵引力同样不均衡。官方材料现在引用约 $150 million 累计合作收入,对私有生物科技公司有意义;但第三方追踪器对年化收入和员工数分歧很大。GetLatka 报告 2024 年收入 $69 million、员工 262 人;Usearch 列出收入 $7.5 million、员工 267 人;Awaira 报告约 300 名员工和一个很宽的估计 ARR 区间;WorxForm 则把融资压缩成单一 $400 million 数字。最安全的读法是:合作伙伴现金存在且重要,但没有干净公开来源能对齐年度收入、毛利率或运营规模。[CI011, CI012, CI013, CI014, CI015, CI016]

单位经济表
指标公开信号锚点 / 可比对象为什么重要尽调要求
收入集中度已具名变现伙伴集中在 BMS、Lilly 和早期 Gilead尚未分散到数十个客户少数交易对手可能贡献大部分经济价值,也带来战略风险要求按伙伴拆分收入和递延收入
当前员工规模BioPharma Dive 称 2025 年裁员后为 ~230 人;追踪器仍显示 ~250-300追踪器分歧仍然重大人员很可能是最大成本池之一,但公开口径连分母都不清楚要求提供当前组织架构图和全负担薪酬基数
算力强度2025 年 ChemML 更新强调一个 192-H100 GPU 集群更接近前沿模型 R&D,而不是轻量企业软件支撑一个判断:算力和实验不是小成本中心要求年度云 / 算力和折旧支出
同行 R&D 费用基准(2023)Recursion 为 $241.2M;Schrödinger 为 $181.8M;Relay 为 $330.0M公开 AI 生物制药可比公司给规模化“平台 + 管线”运营年度成本划出边界对照这些同行,展示 insitro 平台与项目 R&D 分配
同行经营现金消耗基准(2023)Schrödinger 约 $136.7M;Relay 约 $300.3M公开现金消耗锚点显示即便有一些收入,现金也可能很快被吃掉要求 insitro 过去 24 个月月度烧钱趋势
毛利率路径insitro 没有公开数字;经济模型可能介于软件与湿实验室服务之间公开可比公司有收入,仍在消化大量 R&D没有毛利率披露,合作收入质量仍难支撑投资判断要求按合作或项目族拆分毛利率

本表有意把 insitro 直接信号与同行锚点放在一起,因为 insitro 自身单位经济披露不完整。

[CI016, CI017, CI018, CI024, CI026, CI028]
FI002: 单位经济性桥

结构图展示可能决定 insitro 当前单位经济性的主要成本项和证明点。

节点标识成本和价值驱动因素;由于 insitro 不公开披露,它们不分配经审计的美元金额。

[CI009, CI010, CI017, CI018, CI019, CI033]

4.4 资本充足性与同业资本强度基准

资本充足性必须从融资、合作现金和成本重置中重建。公开报道描述,2019 年前已融资超过 $100 million,2020 年中累计风投资金 $243 million;随后 insitro 在 2021 年披露进一步 $400 million Series C。到 2026 年 2 月,公司称自己有约 $800 million 资本支持;这个更大数字可能混合了股权、合作现金或其他资本来源。因此,2025 年裁员在财务上很重要。BioPharma Dive 报道 insitro 裁掉 22% 员工,剩余约 230 人,并把运营目标指向 2027 年。行业背景让这种行为并不意外:EY 2025 年生物科技报告称,后续融资是 2016 年以来最差,IPO 窗口仍低迷,而裁员持续广泛到 2026 年。公开 AI 生物制药可比公司进一步强化烧钱风险。Recursion 2023 年在 $44.6 million 收入上亏损约 $328 million;Relay 没有产品收入却亏损 $342 million;Schrödinger 即使收入基数大得多,仍消耗约 $136.7 million 经营现金。insitro 很可能比这些同行更有资本纪律,但公开数据不足以证明。[CI003, CI020, CI021, CI022, CI023, CI024]

资本充足性表
指标公开解读证据含义下一步尽调
历史披露股权融资截至 2019 年 >100M;到 2020 年累计 VC $243M;2021 年 +$400M Series C 轮Forbes 2019/2020 年报道和 Series C 公告资金基数很大,但作为私营公司仍有限要求提供逐轮融资台账,包括 2021 年后的任何资本
当前资本口径insitro 2026 年称资本为 ~$800MJoe Hand 公告公开口径资本高于已披露 VC 轮次总额要求调节股权、伙伴现金和其他资本来源
2025 年重组信号员工减少 22%;运营资金可撑到 2027 年BioPharma Dive暗示公司在下一个验证点前保全资本要求重组前后的内部现金跑道模型
行业融资背景后续融资为 2016 年以来最差;IPO 仍低迷;裁员普遍EY + Fierce追加融资更取决于验证进展和伙伴支持要求在基准 / 下行市场情景下分析融资选项
公开同行现金头寸(2023)Recursion 为 $391.6M;Schrödinger 为 $468.8M;Relay 为 $750.1M公开申报文件显示规模化同行每年大量消耗现金的同时,仍持有多少现金对标同行烧钱速度和管线范围,测算 insitro 现金需求
公开同行股权价值(2026 年 5 月)Recursion 为 $1.73B;Schrödinger 为 $0.95B;Relay 为 $2.46BCompaniesMarketCap框定公开市场当前给予 AI 生物制药可比公司的估值区间用作私有估值讨论的合理性校验

insitro 尚未发布当前资产负债表,也没有明确披露烧钱速度;资本充足性只能做方向性判断。

[CI003, CI020, CI021, CI022, CI023, CI024]
FI003: 上市同业资本强度区间

选取上市可比公司的财务指标做区间参照,用来界定 AI 生物制药平台可能需要多高资本强度。

这些数值来自有来源支撑的上市可比公司数据,不是 insitro 预测;这里只作为品类资本强度的外部边界。

[CI024, CI026, CI028, CI029, CI030, CI031]
FI004: 资本强度 / 现金流图

用矩阵比较 insitro 的公开财务透明度和变现状态,以及上市 AI 生物制药可比公司。

这些数值只是基于公开证据的定性描述。「低」可能反映披露缺失,而不是底层经济性弱。

[CI019, CI024, CI026, CI028, CI029, CI032]

4.5 公开财务缺口与无法调和的估计

从承销角度看,最大问题不是缺少信号,而是缺少可调和的信号。公开来源未披露当前现金、月度烧钱速度、递延收入、债务、租赁承诺或项目层面支出。追踪器在融资、收入、估值和员工数上相互矛盾;公司自己的资本和合作收入说法也是累计口径,而不是审计式期间披露。即使存在公开基准,也回答不了 insitro 特定问题:有多少合作现金被确认为收入,活跃合作工作赚到多少毛利率,或已有多少资本投入内部临床项目。公开可比公司文件适合给类别资本强度划边界,不能替代公司数据。结果是,几乎每个真正的财务判断——现金跑道、扣除合作现金后的烧钱速度、下一轮规模或稀释风险——仍依赖私下披露,而不是完整公开记录。[CI016, CI019, CI034, CI035, CI036, CI040]

公开财务缺口表
缺失指标缺失原因为什么重要当前公开代理指标具体尽调要求
当前现金和非受限现金私营公司;未发布资产负债表无法独立建模现金跑道和稀释风险只有一条关于资金可运营至 2027 年的媒体引述要求最新月度现金报告和董事会现金跑道材料
按类别拆分的月度烧钱审计或管理口径 P&L 明细均未公开总烧钱与伙伴现金抵扣之间的关系,是核心融资问题同行文件只给出类别层面的外部边界要求 24 个月烧钱历史,拆成 R&D、G&A、算力和设施
按伙伴拆分的已确认收入和递延收入公开来源披露现金事件,但不披露会计处理仅靠里程碑标题无法判断收入质量Recursion 文件显示现金与 GAAP 节奏可能错位要求按交易对手列示合作会计处理表
义务:债务、租赁、算力合同、设备融资合同排期或义务表均未公开隐性义务可能显著压缩有效现金跑道只有同行文件显示这些承诺可能有多重大要求完整义务表,包含到期日和契约条款
项目级支出和预期资金用途内部管线预算未公开需要它来判断平台支出是否高效转化为资产公司新闻稿暗示平台和管线扩张,但不披露成本分配要求按项目拆分预算,以及未来 24 个月资金用途计划
董事会批准的融资计划 / 下一轮触发条件投资者信或融资备忘录均未公开决定公司离下一轮融资有多近,以及哪些里程碑最关键公开代理信号只有 2025 年重组和 2026 年伙伴表述要求融资策略备忘录和基于里程碑的资本计划

本表每一行都对估值、稀释分析,或任何收入质量论证的可信度至关重要。

[CI016, CI034, CI035, CI040, CI041]

4.6 财务结论:真实合作伙伴验证仍伴随高融资依赖

insitro 不是一个零验证科学项目。大型药企已经向它支付真实现金,至少反复延展一个核心关系,并给公司一条从平台工作走向里程碑、特许权使用费或内部保留资产价值的可信路径。这比没有任何变现的纯临床前生物科技经济性更强。但从公开证据看,公司仍依赖融资。没有经审计现金头寸、没有干净年度收入桥接、没有披露义务时间表,也没有可靠公开方法来承销当前毛利率或烧钱速度。2025 年重组说明,管理层仍在为到达证据节点而优化,而不是为了自我维持的经营杠杆。相对公开 AI 生物制药同行,insitro 很可能有财务上可信的平台故事,但还不能用传统模型精度承销。关键尽调阻碍仍是当前现金、真实年度合作收入、义务负担,以及按资产和阶段拆分的项目支出。[CI031, CI032, CI033, CI038, CI040, CI041]

Chapter 05

05产品与技术

5.1 工作流语境下的产品定义:先是发现系统,其次才是产品 SKU

公开材料一贯把 insitro 描述得不像软件供应商,更像工业化发现系统。公司从多模态人群队列和临床数据出发,加入内部生成的细胞和扰动数据,再用机器学习推导关于疾病生物学的因果假设。产出随后被推入第二层治疗设计,而不是停在靶点排序。这很重要,因为它澄清了客户视角下「产品」到底是什么:对药企合作伙伴而言,它是一套能从生物学发现推进到优化干预设计的工作流;对数据合作伙伴而言,它可以是向量嵌入搜索或共同开发的基础模型等特定部署工具;对 insitro 内部而言,它是产出保留管线资产的操作系统。没有公开定价、自助 API 界面或打包软件文档,进一步说明其公开产品表面以工作流为中心、按合作定制,而不是通用企业 SaaS 平台。[CE001, CE002, CE003, CE019, CE020, CE029]

工作流 / 使用场景表
用户任务当前工作流痛点insitro 方案公开可见收益关键限制
发现 ALS 疾病驱动因素已知假设带来的进展有限Virtual Human + CellML/POSH,与 BMS 合作已提出多个 ALS 靶点,并推进到治疗模态工作尚无公开人体疗效证明
为难成药靶点设计小分子依赖实验优化 ADMET 和 PK,速度慢、成本高ChemML / TherML 主动学习循环伙伴扩展和 Lilly ADMET 合作显示伙伴需求牵引没有公开候选物质量基准或成功率披露
推进代谢 RNA 和抗体项目模态选择可能受内部工具偏差限制TherML 加 Lilly 递送和抗体协议insitro 保留权利,同时拓宽模态选项具名资产进入临床的时间仍未公开
探索多模态 NHS 病例基于标签的搜索没有充分利用高维组织病理和基因组数据面向 Genomics England 的向量嵌入搜索引擎安全研究环境内的语义检索没有公开使用或结果指标
为神经退行性疾病构建眼部生物标志物模型OCT 数据丰富,但难以规模化用于靶点发现INSIGHT / Moorfields 基础模型合作在安全环境中访问数百万张关联 OCT 图像仍处合作阶段;没有公开部署指标
从规模化人类表型生成肥胖靶点BAT 难以在人群规模测量ClinML 表型加 CellML 筛选和体内跟进具名临床前资产 BAT-01,已有动物疗效数据仅有临床前证明

收益只限于 insitro 或具名伙伴公开描述的内容;商业价值捕获仍大多不透明。

[CE001, CE005, CE006, CE015, CE017, CE018]
FE002: 客户工作流 / 运营流程

insitro 的发现平台如何从数据获取推进到靶点提名、干预方案设计,再进入合作伙伴或内部项目。

该流程把多个合作伙伴版本抽象成一个标准运营闭环。

[CE001, CE005, CE006, CE013, CE015, CE017]

5.2 模块图谱与运营架构:从因果生物学到模态感知设计

insitro 的架构在公开资料中可读成一个分层系统。在发现端,Virtual Human 被描述为因果生物学引擎;ClinML 和 CellML 创建表型与细胞证据,让这张因果图可执行。POSH 位于细胞扰动层,是一种高内涵筛选机制;它在规模化时保留表型深度,而不是把一切压缩成低维读数。下游,TherML 和 ChemML 把生物学洞察转成干预设计,主动学习循环直接连接自动化实验室。结果不是一堆模型,而是管线形状的运营模式:发现疾病结构,识别杠杆点,选择模态,生成化合物或寡核苷酸候选,并用实验反馈迭代。公开 GitHub 资产和 cp-posh 资源让这套架构的切片变得具体;但完整栈仍部分不透明,因为只有面向研究的片段开放,最有价值的数据集、内部模型和运营指标仍是私有的。[CE004, CE005, CE006, CE007, CE008, CE009]

产品模块 / 资产矩阵
模块 / 资产主要用户状态 / 成熟度差异化主要尽调缺口
Virtual Humaninsitro 疾病团队;药企伙伴活跃的核心发现层以人类和细胞数据为锚的因果生物学引擎没有公开基准可比较靶点命中率与替代方案
ClinML人类数据与转化团队活跃研究模块从队列和影像数据规模化生成表型数据权利和表型验证条款仍未公开
CellML / POSH靶点发现与验证团队活跃研究模块具备表型深度的规模化高内涵扰动筛选没有公开吞吐量或单次筛选成本指标
ChemML / 小分子设计化学团队;BMS;Lilly 挂钩工作流活跃,并获伙伴验证QALs、ADMET 模型、主动学习设计循环、大规模算力没有公开先导物到候选物转化数据
寡核苷酸设计栈代谢和 ALS 项目团队活跃 / 临床前AI 引导的 siRNA 设计,加递送技术集成首次人体试验时间和 CMC 就绪度未公开
TherML 生物制剂模块生物制剂 / 抗体设计团队2026 年推出CombinAbleAI 用物理信息优化抗体和其他生物制剂收购后整合成熟度尚无外部基准
外部部署工具Genomics England;INSIGHT/Moorfields 研究人员已部署 / 联合开发中安全环境中的向量嵌入搜索和 OCT 基础模型工作流商业条款、使用指标和可复制性未公开

insitro 对外讲的是发现系统,而不是 SKU 目录;因此,本表根据公开项目和平台披露重建模块。

[CE001, CE004, CE005, CE006, CE007, CE011]
技术 / 运营架构表
层 / 组件作用关键依赖主要风险
人类队列和伙伴数据提供多模态临床和表型输入生物样本库和伙伴数据权利权利碎片化或复用受限
表型 / 表征学习构建向量嵌入、基础模型和规模化人类表型经过整理的高质量关联数据集源队列之外的漂移、偏差或迁移弱
细胞模型生成创建贴近疾病的人类细胞系统,用于发现自动化实验室和分化方案生物学可重复性风险
扰动筛选深入到表型层面,测量基因或扰动带来的影响POSH 成像、条形码和测序栈吞吐量和成本未对外披露
治疗设计引擎选择治疗模态,并围绕药效和可开发性优化干预方案算力、合作方数据集、QAL、分子模拟算力经济性和收购后整合风险
项目转化层将输出推进到合作方或内部资产项目BMS、Lilly 和内部管线执行临床转化和 CMC 风险
治理层用隐私和验证控制约束高风险 AI 使用合作方安全环境;监管框架控制落地情况无法公开审计

详细系统图未公开;表格根据公司自身模块说明和合作方部署还原运营模型。

[CE003, CE004, CE005, CE006, CE007, CE011]
FE001: 产品架构图

从数据获取、因果发现,到疗法设计和项目交付,概览 insitro 的产品栈。

这些层按公开平台和合作伙伴材料重构;未披露内部子服务。

[CE001, CE003, CE004, CE005, CE006, CE011]

5.3 部署、成熟度与路线图:真实合作伙伴使用,但证据仍多停留在临床前

最强成熟度证据不是广泛客户铺开,而是在具体合作和命名项目中反复使用。Bristol Myers Squibb 最初让 insitro 建立 ALS 和 FTD 疾病模型,随后把关系转成已提名靶点、基于 ChemML 的分子设计阶段,最后扩展到覆盖寡核苷酸和小分子的多模态合作。Lilly 同样从围绕 siRNA 递送和抗体发现的代谢协议,扩展到小分子 ADMET 模型开发。药企之外,Genomics England 和 INSIGHT/Moorfields 描述了 insitro 技术在安全环境中的合作伙伴专属部署,说明至少部分平台组件可以交付给外部组织。从 2025 年末到 2026 年的路线图也重要:POSH 从概念推进到论文发表和公开代码发布;TherML 作为带品牌的治疗设计层推出;BAT-01 作为端到端栈产出的命名临床前资产浮出。即便如此,成熟度天花板仍很清楚。公开证据大多止于临床前或合作阶段里程碑,而不是获批产品或公开人体疗效结果。[CE015, CE016, CE017, CE018, CE019, CE020]

路线图 / 发布 / 开发阶段表
日期 / 阶段功能 / 里程碑状态含义来源
2020-10BMS ALS 发现合作启动已完成疾病模型和靶点发现工作流获得外部验证BMS 2020
2024-10Lilly 代谢疾病协议(siRNA 递送 + 抗体发现)进行中确认能力从靶点提名扩展到模态选择Lilly 2024
2024-12BMS 首个 ALS 靶点里程碑已完成表明靶点提名转化为付费后续工作BMS 里程碑 2024
2025-09Lilly 小分子 ADMET 合作进行中增加合作方训练的 ADMET 层,并通过 TuneLab 获得曝光Lilly 小分子 2025
2025-10BMS ChemML 扩展进行中将 ALS 工作从生物学发现推进到分子设计BMS ChemML 扩展
2025-12POSH 论文和公开 cp-posh 资产已完成外部技术证明加开放研究产物POSH 2025 + GitHub
2026-01TherML 发布和 CombinAbleAI 收购已完成补全了包含生物药在内、不限模态的设计叙事TherML / CombinAbleAI
2026-02BAT-01 肥胖症临床前数据已完成具名资产展示从靶点到动物实验的端到端工作流BAT 研究 2026
2026-03BMS ALS 靶点扩展和多模态计划进行中表明平台现在能围绕同一生物学机制支持寡核苷酸和小分子分支BMS 扩展 2026

行顺序展示能力如何从发现合作拓宽到按模态设计,再到具名临床前资产。

[CE015, CE016, CE017, CE018, CE021, CE030]

5.4 差异化、数据护城河与 IP:生物学、数据密度和模态广度交叉处最强

insitro 的差异化似乎并不建立在单一模型架构或单一治疗模态上。更强的公开故事,是把大规模人类数据资产、高密度细胞扰动系统,以及能跨模态选择和工程化干预的整合设计循环组合起来。POSH 给公司提供了一个具体技术证据点,因为它展示了 insitro 如何试图打破表型筛选中规模与深度的取舍;cp-posh 仓库也给出真实开发者信号,说明部分工作可复现。TherML 通过把靶点选择连接到模态选择,延展了护城河主张;这比仅仅围绕一个靶点优化小分子更有野心。Justia 专利记录也表明,insitro 正在建立真实的图像和组学 IP 资产。尽管如此,护城河可见度仍不完整。栈中最有价值的部分很可能是私有数据集、内部反馈循环和合同数据权利,而公开记录无法清晰暴露这些。[CE007, CE008, CE009, CE010, CE011, CE012]

FE004: 产品成熟度 / 能力图

从科学验证、外部验证、模态宽度和审计可见度四个维度,对 insitro 主要模块的强度和可见度作定性评估。

分数是只基于公开证据的定性分析师判断,不能替代内部 KPI 复盘。

[CE007, CE011, CE012, CE019, CE020, CE027]

5.5 信任、安全、隐私、合规与质量控制:原则可见,审计不可见

公开信任证据方向为正,但不完整。Genomics England 和 INSIGHT/Moorfields 都强调安全研究环境和受限数据访问;这很有意义,因为 insitro 平台触及敏感临床和影像数据集。外部框架也清楚展示「好」应该是什么样:FDA 强调用于药物决策的 AI 模型需要基于风险的可信度评估;EMA 强调数据治理和生命周期管理;WHO 把信任和伦理视为基础,因为技术部署跑在法律前面。但 insitro 自身公开披露在运营细节上仍很薄。已发布隐私政策聚焦网站,不是合作数据控制的窗口;没有来源材料披露 SOC 2、ISO 27001、GxP、HIPAA 或 HITRUST 认证姿态。这留下了熟悉的尽调结论:公司似乎理解正确的治理语言,但投资人仍需要私下证据,验证这些控制是否以工业级深度落实到真实发现工作流中。[CE019, CE020, CE032, CE033, CE034, CE035]

信任 / 质量 / 合规表
控制 / 框架状态范围主要缺口
安全研究环境见于合作方材料Genomics England 和 INSIGHT/Moorfields 数据访问合作方特定控制不能证明全公司安全态势
FDA AI 可信度指南适用的外部框架用于药品监管决策的 AI 输出未公开 insitro 控制与 FDA 框架的映射
EMA 良好 AI 实践原则适用的外部框架药物开发 AI 生命周期、验证、治理insitro 未公开实施细节
WHO 伦理与治理指南高层级外部框架医疗健康 AI 的信任、公平与治理不是公司特定控制集
网站隐私政策已公开披露网站 cookie、设备数据和用户咨询不覆盖支撑平台的受治理合作数据集
公开认证和审计已审阅材料中未发现安全、隐私和实验室质量态势未发现 SOC2 / ISO / GxP / HIPAA / HITRUST 披露
开放研究产物可在 GitHub 看到cp-posh 数据集、脚本和模型权重只覆盖研究子集,不覆盖完整内部平台

状态区分了可见的合作方特定控制、适用外部框架,以及仍未披露的真正公司特定审计证据。

[CE009, CE019, CE020, CE032, CE033, CE034]
FE003: 关键依赖图

公开记录显示,这些重大内外部依赖支撑着 insitro 的产品和技术栈。

依赖项根据公开合作伙伴、治理和平台材料重构,而非来自内部架构文件。

[CE015, CE016, CE018, CE019, CE020, CE024]
Chapter 06

06客户情况

6.1 客户群细分:少数药企付款方、少数外部用户,没有广泛安装基础

截至 2026 年 5 月,insitro 的客户图谱很窄。最清晰的直接付款方是 Bristol Myers Squibb、Lilly,以及历史上的 Gilead:这些大型药企交易对手用 insitro 平台发现靶点、设计模态,并建立机器学习驱动的化学或 ADMET 能力。与这些付款方分开的,是 Genomics England 和 INSIGHT/Moorfields 等研究环境合作伙伴;它们很重要,因为它们说明 insitro 工具可以进入带严格治理的真实外部数据基础设施。第三个表面,是通过 Lilly TuneLab 和 Catalyze360 获得的间接生态触达;insitro 构建的模型可能触达生物科技用户,但这些用户不一定成为 insitro 直接客户。同样重要的是缺席项:已审阅来源没有出现自助产品、公开定价、市场平台招商动作或长尾命名企业账户。主页、平台、使命和管线材料仍描述一家在内部项目和合作项目之间平衡的公司,因此当前客户经济依赖少数高价值关系,而不是广泛安装基础。[CU001, CU002, CU003, CU004, CU005, CU006]

客户分层表
分层买方 / 用户 / 付款方使用场景规模收入 / 战略价值主要缺口
大型药企共同发现交易对手买方 / 付款方:BMS、Lilly、Gilead;用户:药企研发和转化团队靶点发现、模态选择、化学、ADMET、疾病建模3 个具名药企账户可见现金价值最高;已披露合作收入大部分可能集中在这里未披露账户级收入拆分、续约条款或客户数分母
研究环境部署合作方买方:合作方管理层;用户:获批的 Genomics England 和 INSIGHT/Moorfields 研究人员;付款方:不明安全环境内的向量嵌入搜索和基础模型工作流2 家具名英国机构有力证明工具能在外部敏感数据上运行商业条款、用户数和重复使用指标未披露
Lilly TuneLab 生态间接用户买方 / 付款方:Lilly 和合作生物科技公司;用户:药物化学和数据科学团队在 Lilly 联邦式基础设施中访问 insitro 构建的 ADMET 模型间接;合作方数量未披露可能成为超越直接双边交易、触达更广用户的路径insitro 可能不是面向这些用户的直接商业供应商
insitro 内部管线项目买方 / 付款方:insitro 自身;用户:内部疾病和平台团队用平台推进全资拥有资产和合作项目多个项目,但不是客户账户战略上重要,但不能分散外部收入消耗资本,却不能证明第三方付费意愿
广泛自助式或长尾企业客户公开材料未见证据未发现公开的产品驱动或市场化工作流披露为 0公开无是否存在长尾客户仍未验证

表格将直接现金交易对手、外部用户和内部平台消耗分开,避免夸大客户质量。

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

insitro 的具名客户关系通常如何从未满足科学需求推进到部署、里程碑验证和扩展。

该旅程图根据公开合作和安全环境材料重构,而非来自公司发布的 GTM 图。

[CU001, CU004, CU006, CU018, CU021, CU026]

6.2 命名采用证据:交易对手描述真实部署或付费里程碑推进之处最强

命名客户证据真实存在,但由合作驱动,而不是订阅驱动。Gilead 2019 年 NASH 交易披露了三年期限、前付款、里程碑潜力,以及推进最多五个靶点的权利;这远强于被动标识背书。Bristol Myers Squibb 是最佳公开采用案例:2020 年五年合作,2024 年 $25 million 里程碑和第一个已提名 ALS 靶点,2025 年 ChemML 延展进入小分子设计,2026 年又新增两个靶点和另一个里程碑。Lilly 是第二个主要先落地再扩张案例:2024 年三个代谢协议,在 2025 年拓宽为用 Lilly 数据训练、并通过 Lilly TuneLab 暴露的 ADMET 模型。药企之外,Genomics England 称 insitro 的向量嵌入搜索将在其安全 Research Environment 中可用;INSIGHT/Moorfields 称双方正基于数百万张关联图像共同开发 OCT 基础模型。这些都是有意义的外部部署,但公开结果仍被表述为里程碑或技术能力,而不是使用量或 ROI。[CU008, CU009, CU010, CU011, CU012, CU013]

客户增长 / 采用轨迹表
指标 / 里程碑日期来源置信度含义缺失分母
Gilead 发现合作启动3 年 NASH 合作;最多 5 个靶点;$15M 首付款加里程碑组合2019-04-16Gilead 官方 + Forbes最早明确证明大型药企愿意为 insitro 平台工作付费已实现里程碑或续约情况未公开
BMS 合作启动5 年 ALS/FTD 发现合作,包含首付款和后续里程碑潜力2020-10-28insitro BMS 2020确立 BMS 是最早可见的持续性药企账户未披露年度收入贡献
Genomics England 部署宣布向 Genomics England 研究合作方开放安全 Research Environment 内的向量嵌入搜索2022-03-09Genomics England 官方合作方侧有力证明外部工具可部署在敏感数据上未公开用户数或查询量
BMS 首次现金转化$25M 里程碑付款,并选出首个新型 ALS 靶点2024-12-18insitro BMS 里程碑 2024让 BMS 从已宣布合作推进到已兑现经济进展截至目前确认的合同总价值未公开
Lilly 关系启动3 项战略协议,覆盖代谢疾病中的 siRNA 递送和抗体发现2024-10-09insitro Lilly 2024在 BMS 之外创造第二个大型药企账户未披露近期现金经济性
Lilly 关系扩展用 Lilly 数据训练 ADMET 模型;模型向 Lilly TuneLab 合作方开放2025-09-09insitro Lilly 2025展示早期落地扩张和间接生态触达未公开使用这些模型的合作生物科技公司数量
INSIGHT / Moorfields 合作宣布在安全环境中基于数百万张 OCT 图像构建基础模型2025-05-05INSIGHT 合作页面展示英国第二个由合作方控制的外部部署界面未公开部署使用或商业化数据
BMS 再次扩展提名 2 个额外靶点,并支付 $10M 里程碑款2026-03-23insitro BMS 2026使 BMS 成为公开信息中最清晰的重复扩展账户未披露剩余里程碑时间表或期限
披露累计合作收入来自 BMS、Lilly 和 Gilead 的约 $150M2026-02-26Joe Hand 2026确认客户在经济上重要未按账户披露收入拆分

轨迹表聚焦可直接观察的外部里程碑和部署。客户数和留存指标未披露时保持 null。

[CU008, CU009, CU010, CU011, CU012, CU013]
具名客户证明表
客户 / 合作方分层部署 / 使用场景生产级 / 试点结果 / 证明主要限制
Bristol Myers Squibb大型药企付款方ALS/FTD 靶点发现,已扩展到 ChemML 小分子设计和更多靶点进行中的战略项目2020 启动,2024 $25M 里程碑,2025 扩展,2026 新增靶点 + $10M 里程碑未公开收入拆分、剩余合同期限或续约机制
Eli Lilly大型药企付款方代谢疾病靶点和模态协议,加上小分子 ADMET / PK 模型开发进行中的战略项目2024 三项战略协议,2025 扩展到 TuneLab 关联 ADMET 模型公开持续时间仍短,经济性仍未披露
Gilead Sciences大型药企付款方NASH 疾病建模和靶点发现,并有权推进最多五个靶点历史上的类生产合作Gilead 官方条款披露了首付款、里程碑、版税和明确的三年期限初始期限之后的当前状态未公开
Genomics England研究数据合作方 / 外部用户界面向研究合作方开放安全 Research Environment 内的向量嵌入搜索在合作方环境中部署的外部工具合作方页面和当前 RE 文档确认了真实受治理的外部环境未公开使用量、满意度或现金价值数据
INSIGHT / Moorfields研究数据合作方 / 外部用户界面在安全环境中共同开发用于神经退行性疾病相关发现的 OCT 基础模型在合作方控制环境中进行的联合开发合作方材料描述了数百万张 OCT 图像、安全访问控制和获批研究人员基础设施未公开重复使用、变现或续约指标

行仅限外部具名、且有具体部署或经济证明的交易对手。证据明显强于单纯 logo,但在使用率和续约上仍不完整。

[CU008, CU009, CU010, CU011, CU012, CU013]
FU002: 采用 / 部署漏斗

insitro 具名客户关系从机会识别到安全部署和扩展的通用流程。

该流程只使用公开公告和文档可见步骤,横跨药企和研究环境关系做抽象。

[CU004, CU008, CU010, CU014, CU018, CU024]

6.3 留存与耐久性:BMS 是最佳信号,Lilly 仍早,Gilead 不透明,使用指标缺席

留存和耐久性正是公开记录明显变薄的地方。公司没有披露 NRR、GRR、流失率、NPS、活跃客户数或账户级收入留存指标。最强耐久性代理指标是观察到的连续性:BMS 自 2020 年到 2026 年保持公开并跨四个不同证据点扩展,使其成为 insitro 在赢得科学信任后可以加深标杆关系的最佳证据。Lilly 令人鼓舞,但更新,目前只有 2024 至 2025 年的公开序列。Gilead 证明大型药企早期愿意付费,但原始期限之后的当前状态在已审阅来源中不可见。研究环境合作加入了另一类粘性。Genomics England 和 INSIGHT 都描述了安全环境:数据留在受控基础设施内,导出受门控,工具或虚拟机为获批研究者配置。这种运营模式意味着上手摩擦和一定切换成本,但仍未揭示续约经济、用户满意度或重复使用质量。[CU022, CU023, CU025, CU026, CU027, CU030]

留存 / 重复使用 / 满意度表
指标值 / null分层置信度尽调问题
具名直接付款方3 个公开具名药企交易对手(BMS、Lilly、Gilead)药企关系要求提供完整活跃账户名单,以及任何休眠或到期账户
公开落地扩张证据BMS 2020→2026 和 Lilly 2024→2025药企关系要求提供逐账户里程碑时间线、续约日期和扩展管线
公开合同期限可见性BMS 原始期限 5 年;Gilead 原始期限 3 年;Lilly 当前期限未披露药企关系获取当前剩余期限、自动续约条款和终止权
NRR / GRR / 流失 / NPS所有外部客户要求提供收入留存队列、流失账户历史和满意度调查数据
部署使用率Genomics England、INSIGHT/Moorfields、Lilly TuneLab 用户要求提供活跃用户数、查询量、模型调用量和重复使用频率
采购 / 切换摩擦安全环境、Airlock 控制、获批导出和预置 VM 暗示中等粘性研究环境合作方量化受治理部署的上线时间、重新验证负担和续约工作量

公司未披露留存或满意度指标处,null 是刻意保留。公开连续性信号以 BMS 最强,Gilead 当前状态可见性最弱。

[CU022, CU023, CU025, CU026, CU030, CU031]
FU003: 客户验证矩阵

定性比较 insitro 具名交易对手和外部部署伙伴的验证强度。

分数是只基于公开证据的定性分析师判断。低分常常反映披露缺失,而非已知关系薄弱。

[CU028, CU029, CU032, CU033, CU034, CU035]
FU004: 留存 / 重复合作队列

insitro 式关系类型的示例性连续性代理情景;公司未披露留存指标,因此仅作参考。

这些百分比是分析师启发式假设,不是公司披露的留存。它们把 BMS、Lilly、Gilead 和合作伙伴部署中可见的公开连续性,转化为仅供尽调使用的可比耐久性框架。

[CU032, CU033, CU034, CU035, CU043]

6.4 扩张上行存在,但集中度风险主导可投资客户故事

扩张上行和集中度风险并存。上行面上,insitro 已证明客户关系可以从靶点发现推进到化学、模态扩展或生态暴露:BMS 跨靶点和模态拓宽;Lilly 拓宽到 ADMET 和 TuneLab;Genomics England 可以把 insitro 搜索暴露给更广的研究网络。下行面上,Joe Hand 2026 年公司声明把约 $150 million 合作收入归入仅三个名字——BMS、Lilly 和 Gilead——但没有披露拆分。这足以表明集中度,却不足以承销收入质量。BioPharma Dive 2025 年裁员报道,包括 22% 裁员和现金跑道至 2027 年的说法,强化了这样一种看法:insitro 仍像资本密集型生物科技公司运作,依赖少数交易对手和内部证据点。KPMG 和 McKinsey 把这置于 AI 生物制药交易的常态中:合作定制化、重证据、重整合。结果是,客户基础在战略上可信,但仍过于不透明,不能当作多元化经常性收入引擎看待。[CU036, CU037, CU038, CU039, CU040, CU042]

扩展和集中度风险表
扩展驱动集中度 / 执行风险影响尽调路径
BMS 跨靶点和模态落地扩张质量最高的公开客户证明集中在一个标杆账户BMS 若流失或放缓,会削弱最强的持久性信号,并影响合作价值中的重要份额要求提供 BMS 收入占比、剩余里程碑时间表和按项目配置的人员支持
Lilly 向 ADMET 和 TuneLab 生态扩展间接生态触达未必能分散直接付款方能扩大平台曝光,但经济性仍可能绑定一个药企赞助方要求提供直接与间接经济性、用户数,以及更多 Lilly 关联项目路线图
Gilead 式发现经济性历史证明可能不能反映当前收入质量可作为早期验证,但当前状态不透明削弱了持久性分析要求提供已实现里程碑、当前义务,以及任何仍存续的版税或共同开发权
Genomics England 和 INSIGHT 安全环境部署战略证明有力,但变现不清晰验证了敏感数据上的外部可用性,但未必转化为有意义的现金收入要求提供每项部署的商业条款、托管义务和使用指标
合作收入整体集中度仅三个交易对手披露约 $150M,未拆分任一合作方暂停都可能实质影响现金规划和外界感知的动能要求提供头部客户贡献占比和未来收入集中度情景
2025 重组和现金跑道表述在资本纪律下,支持能力和账户聚焦可能收紧可能提高账户和内部项目之间的优先级排序风险要求提供重组后账户覆盖计划、BD 管线和支持 SLA 承诺

表格有意把上行空间和集中度风险并列,因为 insitro 最好的客户故事也是最清晰的风险暴露。

[CU036, CU037, CU038, CU039, CU040, CU042]
Chapter 07

07风险

7.1 按严重性排序的风险态势:临床转化和 AI 治理风险居前

insitro 的风险栈由两个紧密相连的问题领头:把 AI 驱动的发现推进临床,以及在监管环境下完成这件事;该环境正更明确要求使用场景、数据治理、生命周期管理和可信度证据。公开证据仍停留在临床前和合作阶段。MASLD 更新把 CTRO-1013 描述为正推进 IND 准备工作并走向首次人体试验;裁员和现金跑道措辞则显示,时间安排在财务上和科学上一样重要。这形成了复合结构:如果临床转化放慢,里程碑时点和融资灵活性可能一起恶化。客户侧无法抵消这一风险,因为合作验证集中在 BMS、Lilly 和 Gilead,而不是分散到广泛组合。因此,本章把临床就绪风险、AI 治理风险、集中度融资风险视为三类最重要风险;数据权利、质量认证不透明和人员执行风险则是放大器,而不是独立干扰项。[CR001, CR002, CR003, CR011, CR019, CR025]

FR001: 风险热力图

insitro 主要风险桶的相对可能性、影响、缓释成熟度和剩余严重性。

分数是只基于公开证据的定性分析师判断。披露缺失常常拉低缓释成熟度分数。

[CR001, CR011, CR019, CR025, CR039, CR040]

7.2 监管、法律与数据治理风险:原则可见,公司特定证据有限

FDA 和 EMA 现在已经相当清楚地展示了 AI 用于药物开发时「好」是什么样。FDA 指出 AI 相关提交显著增加、基于风险的监管框架,以及使用场景的重要性。EMA 明确把其反思文件从药物发现延伸到上市后,并强调偏见、患者安全、隐私和数据治理。对面向美国的项目,这些预期还叠加在围绕 IND 提交和电子记录的正式法律规则旁。相比之下,insitro 的公开记录没有展示针对这些预期的验证包映射。法律图景同样混合。网站隐私政策很窄,且针对网站;它不能替代合作伙伴数据治理文件。Genomics England 和 INSIGHT 展示了严肃的安全环境控制,但这些控制存在于合作伙伴环境中,也可能限制数据可移植性和复用。专利显示真实 IP 表面,但自由实施、授权负担、诉讼历史和赔偿条款仍是私有信息。结果不是已知不合规的红旗,而是在监管方和成熟合作伙伴最关心的领域存在重大可见度缺口。[CR005, CR006, CR007, CR008, CR009, CR010]

监管 / 法律风险登记表
风险 / 规则 / 法律问题司法辖区当前公开状态可能性严重性缓释措施剩余风险暴露尽调路径
AI 模型可信度与监管证据包美国 / 欧盟FDA 与 EMA 的预期已经可见;针对 insitro 的映射证据包未公开FDA/EMA 通用原则,以及内部专业能力补强要求提供面向监管方的 AI 验证包和使用场景映射
领先资产的 IND / 进临床准备度美国只有方向性证明:公司提到支持 IND 的研究和临床准备度,但没有公开申报或给药里程碑项目进展、合作伙伴验证和董事会专业能力要求提供 IND/CTA 状态、关键路径计划和剩余卡点
合作伙伴数据隐私与受控访问约束英国 / 欧盟 / 美国安全环境和导出控制已经可见中高Genomics England 与 INSIGHT 的治理环境降低原始数据泄露风险中高要求提供关键合作的数据权利、复用和导出治理条款
质量 / 安全认证不透明全球未看到公开 SOC 2 / ISO / GxP / HIPAA / HITRUST 证据中高合理预防措辞和合作伙伴控制要求提供审计材料、事件历史和质量体系文档
IP / FTO / 诉讼不透明全球专利表面信息可见;FTO、权利负担、诉讼历史和赔偿安排未公开中高专利组合在增长中高要求提供 FTO 审查、重大许可清单和法律风险摘要

行按剩余严重性排序,而不是按公开描述的难易度排序。

[CR005, CR007, CR008, CR009, CR010, CR012]
FR002: 风险传导图

insitro 的主要风险如何层层传导到里程碑、客户、融资和估值。

这张 DAG 只突出公开信息里最重要的传导路径,不穷尽所有可能的运营互动。

[CR011, CR015, CR019, CR025, CR035, CR040]

7.3 运营、合作伙伴与融资依赖风险:模型强大,但耦合很紧

insitro 的运营模式不是轻量软件业务。它把多模态数据生成、湿实验室工作流、重算力、自有合作伙伴数据集,以及与外部合作争夺资本的内部管线组合在一起。公开来源分片展示了这一点:BMS ChemML 延展提到 QALs、ADMET 模型和 192-H100 集群;Lilly 小分子工作依赖 Lilly 自有数据;Genomics England 和 INSIGHT 部署运行在受治理的合作伙伴环境中;Joe Hand 的公司声明仍把合作收入集中在三个名字上。这意味着多个依赖项可能一起失效。合作伙伴延迟可以同时影响数据访问、里程碑时点和战略验证。临床准备放慢会同时加重融资需求并削弱客户杠杆。BioPharma Dive 和 Fierce 添加了重要背景:裁员并非 insitro 独有,但行业背景强化了资本和执行余量仍然有限。这里的运营风险不是一次戏剧性失败,而是一个耦合很紧、容错空间有限的系统。[CR004, CR014, CR015, CR019, CR020, CR021]

运营 / 质量 / 安全风险登记表
失效模式可能性严重性缓释成熟度剩余风险暴露未解决缺口
临床前发现未能转化为可申报 IND 或有临床价值的项目还没有公开 IND 或人体给药里程碑
计算、数据生成和湿实验室成本跑得比里程碑更快中高中高未公开单位经济、烧钱速度或吞吐量指标
AI 模型验证、文档或性能证据不足以说服监管方或合作伙伴低中未公开映射验证包或审计轨迹
安全环境部署摩擦拖慢合作伙伴执行或数据复用中高中高导出控制和合作伙伴侧治理会限制敏捷性
公开记录中,安全和质量状态缺少足够审计中高中高未公开认证、事件或质量体系披露

运营风险不只关乎科学;科学、文档、基础设施和时点叠在一起,才构成真正风险。

[CR001, CR003, CR010, CR014, CR015, CR016]
合作伙伴 / 依赖风险登记表
依赖项交易对手角色集中度失效场景严重性缓释措施剩余风险暴露
战略神经科学合作伙伴Bristol Myers Squibb验证、里程碑现金、靶点扩展、模态分支靶点没有进一步进展,里程碑也未转化多年深度关系,且反复取得进展
代谢和化学合作伙伴Lilly数据、模态支持、TuneLab 生态触达Lilly 降低项目优先级,或限制生态访问多份协议,范围扩大中高
历史大型药企验证方Gilead外部早期证据显示,客户愿意为平台工作付费关系实质上已经结束,或未来贡献很小中高只有历史验证中高
受治理的数据与部署表面Genomics England / INSIGHT敏感数据访问、安全外部部署、合作伙伴证明数据权利限制或导出控制阻碍复用和规模化中高安全环境降低泄露风险中高
外部融资和里程碑资金私人投资者与合作伙伴现金为临床转型、计算、实验室和招聘供血临床证据改善之前,资金先变贵历史资本基础和合作关系
监管认可FDA / EMA设定 AI 可信度和证据预期文档不足拖慢或阻断申报进展已有公开指南和沟通路径

依赖项按失效可能造成的损害大小列出,而不只看合同状态。

[CR019, CR020, CR021, CR022, CR023, CR026]

7.4 人员、执行与终止标准:缓释措施存在,但举证责任仍在前方

公开记录里确有缓释因素。Amy Abernethy 进入董事会,带来临床开发和前 FDA 经验。Joe Hand 的任命说明,公司进入运营要求更高的阶段后,已经把人才战略摆到台面上。合作机构的安全环境也显示,insitro 对外协作时已经把部分高风险治理控制纳入工作方式。但这些缓释因素都没有消掉核心举证责任。公司仍需证明,它能把项目推进到临床,产出成熟合作方和监管机构都愿意接受的验证文件,并且不再进一步削弱执行能力或加深合作方集中。尽调最有用的可监测触发点不是模糊的进展印象,而是具体事件:申报状态、首次人体试验准备度、进一步裁员、缺失验证包、缺乏审计证据,以及合作方不续约或降低优先级。在这些问题厘清之前,正确姿态是把领导层补强视为有意义但不完整的缓释,残余风险仍然偏高。[CR007, CR008, CR026, CR027, CR028, CR029]

人才 / 执行风险登记表
角色 / 职能依赖或缺口可能性严重性缓释措施尽调路径
临床开发与监管执行公司从偏发现的组织转向临床就绪运营模型,这一点尚未被公开证明中高Amy Abernethy 加入董事会,且公司明确聚焦临床准备度要求提供组织架构图、外部 CRO/监管顾问和 IND 工作流
创始人主导的战略集中度Daphne Koller 仍是科学和战略叙事的核心中高董事会扩容和高管招聘要求提供授权决策权和继任规划
跨职能 AI / 生物 / 化学人才留存裁员削弱了人才稀缺运营模型里的缓冲中高CPO 到任,以及公开的人才战略表述要求提供流失数据、关键空缺和裁员后招聘计划
治理和政策深度AI、临床证据和卫生政策接口需要更成熟的监督Abernethy 带来前 FDA 和临床研究领导经验要求提供董事会委员会结构和外部顾问支持
组织规模化纪律平台、内部资产和合作关系都在争夺管理注意力和资本中高中高透明文化表述和领导层补强要求提供组合优先级框架和资源配置模型

人才风险被放大,因为 insitro 的差异化取决于整合多个职能,而这些职能在生物科技和软件公司通常各自为营。

[CR027, CR028, CR029, CR030, CR031, CR039]
缓释与否决标准表
风险可监测触发项阈值 / 事件行动含义
临床转化延迟IND / 首次人体试验准备度2027 年底前没有申报或人体给药里程碑转入高警戒 / 避免承销平台溢价
AI 可信度缺口验证包可得性管理层无法提供按 FDA / EMA 原则梳理好的控制措施将 AI 治理风险视为未解决阻断项
质量 / 安全不透明审计证据尽调资料室没有有意义的安全、质量或事件文档假设企业级就绪度尚未证明
合作伙伴集中度BMS / Lilly 里程碑和续约节奏头部合作伙伴没有进一步里程碑或扩展进展提高集中度折价和融资谨慎度
执行能力侵蚀员工规模和招聘趋势再次发生重大裁员,或明显无法填补关键临床岗位下调对临床准备时间表的信心
数据权利约束合同数据使用条款限制阻止有意义的复用、模型更新或部署规模化下调护城河和可扩展性假设

触发项按可观察性选择:尽调、未来披露或合作伙伴更新都能核验,而不是只能从叙事语气里推断。

[CR034, CR035, CR036, CR037, CR038, CR040]
FR003: 依赖关系图

insitro 的平台、管线和临床准备路径,背后依赖哪些关键对手方和基础设施。

对手方按尽调最关心的层级归并:具名合作伙伴、监管机构,以及核心赋能型基础设施类别。

[CR014, CR015, CR019, CR020, CR021, CR023]
Chapter 08

08估值

8.1 投资建议与估值框架:先守定价纪律,再谈热情

首要问题不是 insitro 是否令人印象深刻。答案显然是肯定的。真正的问题是,公开证据是否足够给它定价。公开来源显示不少正面因素:头部合作伙伴、BMS 多次扩展合作、累计合作收入约 $150 million,以及已融资或被引用的资本约 $800 million。但它们没有给出当前入场价格、优先股堆叠、当前现金余额、可靠年收入,或内部资产的人体数据验证点。因此,直接给出坚决买入或坚决回避都显得牵强。基于公开证据,诚实姿态是继续研究、价格敏感。正确框架不是单点目标回报,而是一组参考区间,由公开可比公司、最后已知融资锚点和证据状态假设拼出。如果公司报价接近公开可比公司集群并加上合理溢价,尽调值得做;如果价格已经按一个去风险的前沿平台来给,公开记录提示应放慢。[CV001, CV004, CV005, CV006, CV007, CV008]

建议摘要表
维度评估信心决策含义
总体建议继续研究 / 观察只有价格和条款能用公开证据区间做尽调校验时,才继续推进。
风险评级转化、监管、集中度和融资风险仍在相加,而不是互相抵消。
估值立场对价格敏感;基准参考区间约 $1.5B-$2.5B不要仅凭叙事强度承销溢价估值。
进入纪律如果条款干净,有效投后估值约 $2.0B 或以下值得关注低于该区间时,可以支持更深入尽调;约 $2.57B 只有在私有证据更强时才成立。
对公开证据的信心合作伙伴证明中等,经济性和条款较弱足以设定护栏,但不足以给实际轮次定价。

所有估值参考都是从公开证据推导的股权价值启发式判断,不能替代投资条款清单或完整股权结构模型。

[CV001, CV022, CV025, CV026, CV027, CV028]
FV001: 建议逻辑——从证据到继续研究

这张流程图把 insitro 最强的公开正面证据、缺失的估值输入,以及最终对价格敏感的建议串起来。

这条流程刻意服务于决策,而非穷尽所有因素。它展示投资人只靠公开证据为一轮私募融资定价时,最核心的判断逻辑。

[CV001, CV002, CV003, CV004, CV025, CV026]

8.2 投资逻辑与反向逻辑:合作方验证真实,经济性证据仍不完整

insitro 的正向逻辑不是假设。BMS 提供了本文件中最强的公开验证序列:首付款经济权益、里程碑转化、平台扩展资金和新靶点提名。Lilly 增加第二个重大关系,并保留通向可观下游期权价值的路径。公司还受益于治理和人才信号,包括 Amy Abernethy 的董事会角色和 Joe Hand 的人才职责。这些资产并不轻。反向逻辑在于,最重要的价值驱动因素仍处于临床前或私人信息状态。CTRO-1013 仍在推进 IND 支持工作,还没有进入人体验证。收入可见度很差,公开跟踪网站之间差异巨大。合同条款、数据权利、优先权悬置,以及当前现金和烧钱情况都未公开。换句话说,insitro 可能是家好公司,但公开记录仍更容易为愿景定价,而不是为已验证经济性定价。这个差别应主导投资姿态。[CV002, CV003, CV010, CV011, CV012, CV013]

投资逻辑 / 反向逻辑表
维度投资逻辑反向逻辑什么会改变判断
合作伙伴证明BMS 和 Lilly 的关系显示出真实外部验证和付费战略需求。只有 BMS 反复公开扩展;Lilly 经济性仍不够可见,Gilead 基本只是历史验证。更多公开的合作伙伴转化、续约,或新交易对手证明。
内部管线CTRO-1013 给 insitro 留出一条通往自有资产上行的路径,不止是类服务合作。CTRO-1013 仍处于临床前 / IND 支持研究阶段,因此内部资产价值仍需大幅按风险折价。清晰 IND 时间、FIH 状态或早期人体证据。
经济性可见度累计合作收入和大额融资是真实加分项。当前现金、烧钱速度、年收入和合同会计太不透明,无法干净承销。当前现金和烧钱模型,加上合同经济性。
可比公司语境相比更弱或验证更少的上市科技生物公司,insitro 或许应享有溢价。上市可比公司群的估值仍明显低于前沿 AI 叙事。证明 insitro 应高于 Relay 这类上市参照。
融资韧性庞大的历史资本基础和重组纪律,可能为公司争取时间抵达证据点。如果现金跑道仍先于临床证据耗尽,投资者可能面对价格压力和偏重优先权的融资。更新后的现金跑道模型和证据点日历。
治理 / 领导层董事会和领导层补强提升执行力和监管判断。治理无法替代价格、条款或临床证据。可见执行里程碑,加上干净融资条款。

该表把公司质量和特定私募价格下的可投资性分开。

[CV002, CV003, CV010, CV011, CV012, CV013]

8.3 融资背景与入场纪律:旧锚点、嘈杂跟踪网站、容错空间有限

最干净的融资锚点仍是 2021 年 Series C 轮。insitro 官方披露融资 $400 million,Forge 估算投后约 $2.57 billion。除此之外,公开估值可见度明显变弱。GetLatka 和 Awaira 都把 insitro 放在低 $2 billion 区间,但它们在收入、融资总额,甚至基本轮次细节上相互冲突。Usearch 将收入压到 $7.5 million,而 GetLatka 声称 $69 million。这种分歧不是小会计问题;它意味着公开跟踪网站只能当粗略路标。实际结果不是目标价,而是一条入场纪律规则。有效投后估值低于约 $2.0 billion 且条款干净时,机会才开始值得深入研究。在旧的 $2.57 billion 锚点附近,公开记录只支持有限上行。高于约 $3 billion,投资人就在为公开文件没有包含的私人证据付费。[CV003, CV005, CV006, CV007, CV008, CV009]

8.4 可比公司背景与情景区间:公开技术生物企业估值给出护栏

防止付高价的最强公开保护是可比公司组。截至 2026 年 5 月,Eikon 交易市值约 $0.54 billion,Absci 约 $0.90 billion,Schrödinger 约 $0.95 billion,Recursion 约 $1.73 billion,Relay 约 $2.46 billion。这些同行并不完美,但每一个都给出有用约束。Eikon 显示,公开市场可以多快重置早期证据故事。Absci 显示,AI 生物学平台可以长期低于 $1 billion。Schrödinger 显示,即使计算平台商业化程度更高,也拿不到软件式溢价。Recursion 显示,规模、资本和合作伙伴并不自动支撑很高倍数。Relay 显示,更清晰的临床资产可以把公司推到公开区间上沿,但仍不会进入无约束地带。在这个背景下,insitro 的公开证据基准情景应落在约 $1.5-2.5 billion,悲观情景约 $0.8-1.3 billion;只有临床进入和合作方转化明显改善时,乐观情景才约 $3.0-4.5 billion。[CV016, CV017, CV018, CV019, CV020, CV021]

乐观 / 基准 / 悲观情景表
情景关键假设估值区间($B)概率信号什么会证实 / 打破该情景
乐观CTRO-1013 按计划进入临床;BMS 或 Lilly 进一步转化;不再发生重大员工重置;条款干净。$3.0-$4.520-25%IND/FIH 进展和合作伙伴扩展会证实;延迟或惩罚性条款会打破。
基准合作伙伴经济性守住;临床时间仍是方向性、尚未证明;价格落在 $2B 出头;优先权可控。$1.5-$2.540-45%稳定现金跑道且合作伙伴证明没有恶化,会证实该情景;时间停滞或隐藏优先权负担会打破。
悲观2027 年底前没有临床证据;合作伙伴势头平台化;融资条款暴露重度稀释或强优先权。$0.8-$1.330-35%延迟、裁员或苛刻条款会证实;只有证据改善快于预期时才会打破。

概率是分析师信号,不是统计输出。区间用于框定公开证据下的合理结果,并不暗示市场精确性。

[CV022, CV023, CV024, CV025, CV026, CV027]
可比估值表
可比公司当前公开价值状态 / 证据状态与 insitro 的相关性局限
Recursion~$1.73B广义 AI 药物发现平台,有公开文件、合作关系和多个项目可作为中上区间的上市平台锚点,披露比 insitro 更多条款、模态组合或当前证据状态仍不能直接对比
Schrödinger~$0.95B计算平台,既有可观软件 / 服务收入,也有管线可选性说明即使已经变现的平台模型,也可能只按温和的上市估值交易软件收入更多,业务组合也不同于 insitro
Relay Therapeutics~$2.46B临床肿瘤平台,人体数据和资产特定估值逻辑更清晰资产证据更清晰时,可作为公开市场上限锚点比 insitro 当前公开记录更偏临床、也更聚焦肿瘤
Absci~$0.90B有公开市场交易历史、估值仍不高的 AI 生物学平台可作为技术驱动生物平台的下限锚点模态重点和商业化路径不同
Eikon Therapeutics~$0.54B近期 IPO 的疗法平台,公开市场已大幅下修定价可作为早期证据公司公开市场胃纳的警示性可比公司更直接偏临床、上市时间也短,因此不是干净的平台类比对象

该组可比公司刻意选择公开且可定价的对象。它并不主张 insitro 应与其中任何一家公司完全匹配。

[CV016, CV017, CV018, CV019, CV020, CV021]
FV002: 估值对验证状态与公开可比锚点的敏感度

这张条形图对比当前公开 techbio 公司的市值、insitro 悲观 / 基准 / 乐观三种情景中点,以及最后已知融资锚点。

数值为大致股权价值,单位 $M。insitro 的条形是参考情景,不是可观察的市场估值标记。

[CV016, CV017, CV018, CV019, CV020, CV021]
FV003: 估值 / 回报区间

这张区间图展示公开可比公司簇、最后已知融资参考,以及 insitro 悲观 / 基准 / 乐观三种股权价值区间。

数值为股权价值,单位 $M。融资锚点这一行噪声较大,因为各追踪器方法不同,当前条款也未公开。

[CV007, CV016, CV022, CV023, CV024, CV025]

8.5 退出准备度、投资逻辑破裂触发点与最终结论

基于公开证据,insitro 还没到 IPO 准备好状态。私人股份没有公开价格历史,没有经审计的公开财务包,没有干净的当前现金和烧钱视图,也没有具备可见人体数据的内部资产。因此,较近的价值兑现路径更可能来自持续扩大合作伙伴、战略兴趣或后续私募融资,而不是一条直接的公开市场承销故事。这个结论让最终尽调议程更清晰。投资人需要当前股权结构表和优先股堆叠、当前烧钱与现金跑道模型、重大合作的合同条款和收入确认逻辑、CTRO-1013 的 IND 关键路径,以及后期交易对手会要求的 AI 验证和安全材料。关键的投资逻辑破裂触发点同样具体:到 2027 年底仍无 IND 或首次人体试验进展,再次发生重大人员重置,合作方不续约变得可见,或条款清单在溢价估值之上嵌入重度稀释。最终结论:继续研究,高风险、中等信心,且只在定价有纪律时推进。[CV028, CV029, CV030, CV031, CV032, CV036]

投资逻辑破裂与终止触发因素表
触发因素阈值 / 事件对投资逻辑的传导操作含义
临床时间表明显后移到 2027 年底仍无 IND 或首次人体试验进展压低自有资产期权价值,并推高融资压力转向悲观情景 / 放弃,除非价格大幅重置
再次重大裁员关键职能再次出现实质性裁员或明显冻结招聘说明现金跑道或执行能力弱于假设提高下行情景权重,并质疑能否跑到下一证明点
合作伙伴动能停滞BMS / Lilly 没有新增扩展,合作伙伴明确降级优先级,或不续约削弱最强的外部验证支柱提高集中度折扣,并下调基准情景区间
AI / 安全尽调不过关管理层拿不出验证、治理或安全材料监管和企业级就绪风险仍悬而未决视为溢价定价的硬性阻断项
价格或条款变得惩罚性轮次估值高于约 $3B,或附带沉重的参与分配优先权 / 反稀释保护安全边际被吃掉放弃,除非私下证据显著强于公开证据
现金跑道短于预期现有现金无法稳妥撑到下一证明点抬高被迫融资和稀释风险立即按悲观情景重切估值

这些触发因素之所以入选,是因为尽调、管理层更新或后续披露可以核验它们,而不是只能从叙事语气推断。

[CV030, CV035, CV036, CV043, CV044]
最终尽调清单表
主题缺失证据重要性责任方 / 尽调路径
股权结构表与优先权当前完全稀释股本和优先股权利决定名义上行空间是否真的归新投资人所有索取董事会材料、股权结构表和瀑布分析
现金、烧钱速度和现金跑道当前现金余额、月度烧钱速度和下行情景现金跑道决定下一轮融资是被迫还是可选索取财务模型和证明点预算
最近定价轮或老股信号2021 年后任何定价交易、老股标记或 409A 更新用真实入场价取代过时轮次叙事索取融资历史和估值备忘录
合作伙伴合同经济性续约、排他性、数据权利、成果所有权和收入确认细节区分可持续经济性与里程碑表象审阅 BMS、Lilly 以及当前任何涉及 Gilead 的协议
IND 与 AI 验证路径监管沟通、IND 时间表和 AI 验证包决定平台价值能否转化为疗法价值上行的核心因素索取开发与监管尽调材料包
安全与质量状态审计证据、认证、事件历史和企业级控制影响合作伙伴信任、投资人尽调转化和退出就绪度索取安全 / 质量尽调资料室材料

这些尽调问题刻意按决策杠杆筛选:每一项都可能实质性移动价格、风险评级或推荐结论。

[CV003, CV009, CV036, CV037, CV043, CV044]
FV004: 投资 KPI 评分卡

这张 IC 风格评分卡从市场、合作伙伴验证、内部验证、经济性透明度、风险、估值支撑和退出准备度评估 insitro。

评分采用 0-10 分制,反映在私募价格未知时的可投性,而不是公司的绝对质量。

[CV002, CV003, CV011, CV014, CV022, CV028]

8.6 图表证据

免责声明

本报告基于截至 2026 年 5 月 12 日的公开信息生成,用于尽调研究目的,不构成投资建议。关于私营公司估值、融资和合同的结论,应以一手尽调材料核验。

证据索引

结论
编号陈述可信度来源
CO001 insitro was founded in 2018 by Daphne Koller to apply machine learning and multimodal data to drug discovery. SO002, SO016
CO002 insitro’s public headquarters is 279 East Grand Avenue in South San Francisco, California. SO001, SO004
CO003 insitro’s core operating model combines human-cohort data and cellular data with machine learning to identify causal targets and therapeutic hypotheses. SO001, SO002
CO004 insitro describes itself as a pipeline-through-platform company that both advances internal programs and partners with large pharmaceutical companies. SO001, SO003, SO009
CO005 The current disclosed pipeline spans metabolism, neuroscience, and ophthalmology rather than a single-disease portfolio. SO003
CO006 Daphne Koller’s founder-market fit rests on a Stanford machine-learning background, the prior creation of Coursera, and a long-standing focus on applying AI to high-dimensional problems. SO016, SO017
CO007 Amy Abernethy joined insitro’s board in 2024, adding FDA, real-world evidence, and clinical-development experience to governance. SO006
CO008 Joe Hand joined insitro as Chief People Officer in February 2026 to lead global people strategy, organizational development, and culture. SO005
CO009 Joe Hand previously held senior leadership roles at Celgene and Phathom Pharmaceuticals, giving him large-scale biotech HR and transaction experience. SO005
CO010 CPP Investments executive Paul McCracken joined insitro’s board as part of the 2021 Series C financing. SO007
CO011 Forbes reported that insitro raised a $143 million Series B in 2020, bringing total venture funding at that time to $243 million. SO017
CO012 insitro’s 2021 Series C raised $400 million and was led by CPP Investments with participation from both existing investors and new crossover backers. SO007
CO013 Public sources identify a16z, ARCH, GV, Third Rock, CPP, BlackRock, T. Rowe, Casdin, Temasek, SoftBank, and Foresite among insitro’s named equity backers by 2021. SO007, SO017
CO014 By 2020, Forbes described insitro’s Gilead NASH collaboration as including a $15 million upfront payment and up to $1 billion in milestone potential. SO016, SO017
CO015 The 2020 Bristol Myers Squibb collaboration added $50 million upfront, $20 million in near-term operational milestones, and more than $2 billion in downstream milestones plus royalties. SO008
CO016 insitro disclosed a $25 million milestone payment from Bristol Myers Squibb in 2024 after selecting the first novel ALS target from the collaboration. SO010
CO017 The 2025 Bristol Myers Squibb extension could provide up to $20 million in new funding for ChemML-enabled ALS small-molecule work. SO013
CO018 The 2024 Lilly agreements targeted metabolic diseases including MASLD while letting insitro retain global rights and leaving Lilly eligible for milestones and royalties. SO009
CO019 The Moorfields INSIGHT collaboration added access to a 35 million-image ophthalmic dataset to support neurodegeneration and ocular target discovery. SO012
CO020 The 2022 Genomics England partnership gave insitro access to an NHS-linked resource of almost 150,000 whole genomes and associated multimodal phenotypic data. SO011
CO021 insitro’s own 2024 to 2026 communications frame the company as having more than $700 million to approximately $800 million in capital when partnership cash is included. SO005, SO006, SO014, SO015
CO022 BioPharma Dive reported that insitro cut 22 percent of its workforce in 2025, leaving about 230 employees and targeting runway into 2027. SO018
CO023 Open tracker pages still show inconsistent headcount estimates of roughly 250, 262, 267, or 300 employees, so current scale remains only partially verified in public sources. SO022, SO023, SO024, SO025
CO024 Open tracker pages conflict on funding, valuation, and revenue, with GetLatka reporting $643 million raised and $69 million of 2024 revenue while Awaira reports $743 million raised and a $2.2 billion March 2026 valuation. SO022, SO023
CO025 Public company pages show insitro now references colleagues in Israel, Poland, and Malaysia in addition to its South San Francisco base. SO005, SO014
CO026 The company still publicly anchors its office footprint at South San Francisco rather than disclosing a broad network of U.S. office sites. SO001, SO004
CO027 The 2026 CombinAbleAI acquisition launched TherML and extended insitro’s design stack from small molecules and oligonucleotides into biologics and other complex modalities. SO014
CO028 The 2026 BAT study said BAT-01 knockdown drove a 15 percent body-weight reduction in obese mice while preserving lean mass, signaling preclinical obesity momentum. SO015
CO029 insitro’s public pipeline page discloses eight named programs ranging from target credentialing to IND-enabling, but no marketed product or public clinical-stage asset. SO003
CO030 The metabolism pipeline currently includes CTRO-1013, CTRO-1029, CTRO-1035, OBS-1, and OBS-2. SO003
CO031 The neuroscience pipeline currently includes ALS-1 small molecule, CTRO-2018 oligonucleotide, ALS-2, and ALS-3. SO003, SO013
CO032 Publicly disclosed revenue-bearing or milestone-bearing pharma relationships are concentrated in Gilead, Bristol Myers Squibb, and Lilly. SO008, SO009, SO010, SO016, SO017
CO033 Public governance clues are limited to named investors and board additions, with no open-source disclosure of control rights, preference stack, or special voting arrangements. SO006, SO007, SO017
CO034 Reviewed open sources did not disclose customer count, debt facilities, secondaries, or a current audited revenue mix for insitro. SO001, SO005, SO022, SO023
CO035 insitro’s GitHub organization exposes public research artifacts such as cp-posh, kindel, and other 2022 to 2025 repositories, indicating an externally visible technical output stream. SO019
CO036 KPMG and McKinsey both describe AI-biopharma partnerships as a large and still-expanding market, but also note valuation pressure, regulatory scrutiny, and more conservative dealmaking after 2022. SO020, SO021
CO037 insitro’s public narrative evolved by 2026 from a machine-learning-driven drug discovery company to an AI therapeutics or physical AI company built on causal biology. SO007, SO014, SO015
CO038 The company’s current one-line product logic is to use Virtual Human and ClinML for target discovery and TherML or ChemML for modality-specific therapeutic design. SO003, SO013, SO014, SO015
CO039 The current operating architecture combines internal data generation with external datasets, partner technologies, and targeted acquisitions to industrialize discovery. SO011, SO012, SO014
CO040 The main overview-level adverse signals are a 2025 workforce reduction, opaque current valuation and scale metrics, and dependence on a small number of pharma partners for data, milestones, or funding. SO018, SO022, SO023, SO024, SO025
CM001 insitro’s current commercial posture is still a platform-backed drug R&D company rather than a company already selling marketed therapeutics. SM001, SM002
CM002 The relevant market boundary for insitro includes AI drug-discovery partnerships, internally controlled therapeutic programs, and data-driven discovery collaborations. SM001, SM002, SM005, SM006
CM003 insitro says its platform is built to derisk and accelerate multiple steps across the research and development value chain. SM001
CM004 insitro’s disclosed pipeline currently spans metabolism, neuroscience, and ophthalmology. SM002
CM005 The 2024 Lilly agreements support metabolic-disease programs while leaving insitro with full global rights and Lilly eligible for milestones and royalties. SM003
CM006 The 2020 Bristol Myers Squibb agreement lets BMS opt selected targets into downstream development and commercialization. SM004
CM007 The Moorfields collaboration gives insitro access to a 35 million-image ophthalmic dataset for neurodegeneration research. SM005
CM008 The Genomics England collaboration gives insitro multimodal search capability across NHS-linked genomics and pathology data. SM006
CM009 MarketsandMarkets estimates the AI in drug discovery market will grow from USD 1.86 billion in 2024 to USD 6.89 billion by 2029 at a 29.9% CAGR. SM021
CM010 Precedence Research estimates the AI in pharmaceutical market at USD 1.94 billion in 2025 and USD 16.49 billion by 2034. SM022
CM011 Precedence Research separately estimates the AI in drug discovery market at USD 6.93 billion in 2025 and USD 17.81 billion by 2035. SM023
CM012 Grand View Research estimates the AI in precision medicine market at USD 2.29 billion in 2024 and USD 14.53 billion by 2030. SM029
CM013 McKinsey says the pharma AI market is projected to grow from more than USD 4 billion in 2025 to USD 25.7 billion by 2030. SM008
CM014 The reviewed analyst forecasts are directionally bullish but not directly comparable because they measure different boundaries such as AI in pharma, AI in drug discovery, and AI in precision medicine across different forecast horizons. SM008, SM021, SM022, SM023, SM029
CM015 KPMG reports that AI-focused M&A and partnership deals in biopharma grew at a 27.3% compound annual rate from 2013 to 2022. SM009
CM016 KPMG says AI-related biopharma deals during the period concentrated in R&D-only agreements and development or commercialization licensing. SM009
CM017 KPMG says the 2022 downturn pushed pharma companies toward lower-risk partnerships and asset acquisitions. SM009
CM018 EFPIA estimates that research-based pharmaceutical companies invested about EUR 55 billion in Europe in R&D during 2024. SM010
CM019 EFPIA says North America accounted for 54.8% of world pharmaceutical sales in 2024 versus 22.7% for Europe. SM010
CM020 WHO says 890 million adults worldwide were living with obesity in 2022. SM014
CM021 WHO says the global costs of overweight and obesity could reach about USD 3 trillion per year by 2030. SM014
CM022 insitro’s Lilly announcement says MASLD affects about 100 million people in the United States and lacked approved interventional treatments in 2024. SM003
CM023 The American Liver Foundation says MASLD affects up to 25% of people in the United States. SM019
CM024 AASLD says MASLD prevalence estimates run from 10% to 46% in the United States and 6% to 35% worldwide. SM020
CM025 WHO says at least 2.2 billion people globally have near or distance vision impairment. SM013
CM026 WHO says at least 1 billion vision-impairment cases are preventable or unaddressed and the associated annual productivity loss is about USD 411 billion. SM013
CM027 CDC estimates about 9.603 million people in the United States have any diabetic retinopathy. SM016
CM028 CDC’s National ALS Registry publishes national prevalence estimates through 2018 and projections out to 2030. SM017
CM029 CDC’s 50-state ALS prevalence paper reports state prevalence ranging from 2.5 to 7.8 per 100,000. SM017
CM030 NINDS says there is currently no known treatment that stops or reverses ALS progression. SM018
CM031 WHO says AI for health faces a pacing gap because technology is advancing faster than legal frameworks. SM007
CM032 EMA’s 2026 AI principles require human-centric design, a risk-based approach, fit-for-use data, lifecycle monitoring, and clear context of use. SM012
CM033 EFPIA’s 2025 AI lifecycle report says AI is spreading from discovery into clinical development, manufacturing, and post-approval safety under trust and transparency expectations. SM011
CM034 IQVIA says pharma companies are extending AI into pharmacovigilance, medical information, and lifecycle management while relying on experienced technology partners. SM030
CM035 The combined Recursion-Exscientia entity disclosed more than 10 internal programs, more than 10 partnered programs, and about USD 450 million of upfront and milestone cash realized from partners. SM024
CM036 Recursion says its platform is built on more than 50 petabytes of proprietary data and millions of experiments per week. SM025
CM037 Schrödinger says it deploys its computational platform across both collaborative and proprietary drug-discovery programs. SM026
CM038 Relay Therapeutics says it uses integrated experimental and computational discovery against previously intractable biology in precision oncology and genetic disease. SM027
CM039 Evotec says it offers a fully integrated R&D value chain and flexible partnering models across small molecules and biotherapeutics. SM028
CM040 Taken together, public peers suggest the AI-drug-discovery market increasingly rewards platform-plus-pipeline or platform-plus-partnering capability rather than tools-only positioning. SM024, SM025, SM026, SM027, SM028
CM041 insitro’s immediate buyers are large-pharma R&D and business-development teams, while physicians, payers, and patients matter mainly in a later downstream drug path. SM003, SM004, SM009
CM042 The Genomics England and Moorfields relationships function primarily as strategic discovery inputs and moats, not as a clean standalone monetizable market category. SM005, SM006, SM011
CM043 A defensible market view for insitro needs multiple lenses—AI market forecasts, pharma R&D budget pools, and disease-burden demand—rather than one generic TAM number. SM008, SM010, SM013, SM014, SM020
CM044 No reviewed public source cleanly isolates insitro’s exact SAM or SOM across its metabolic, neuro, and ophthalmic strategy. SM008, SM009, SM010, SM021, SM022, SM023, SM029
CM045 Growth drivers include high disease burden, large R&D budgets, multimodal data availability, and ongoing AI adoption across the medicines lifecycle. SM010, SM013, SM014, SM030
CM046 Adoption constraints include governance demands, long drug timelines, demand for proof beyond hype, and buyer preference for lower-risk deal structures. SM008, SM009, SM011, SM012
CM047 The partnership adoption path runs from data and model validation to target nomination and then to negotiated development rights and milestone-bearing execution. SM003, SM004, SM005, SM006
CM048 The main status-quo substitutes for insitro are in-house pharma discovery teams, conventional CRO-style workflows, and competing AI-platform companies rather than consumer health or hospital IT vendors. SM009, SM024, SM025, SM026, SM027, SM028
CP001 insitro’s competitive landscape spans direct AI-drug-discovery peers, adjacent integrated-service substitutes, and internal pharma build rather than one like-for-like rival. SP001, SP002, SP016, SP018, SP020, SP021, SP022
CP002 The most visible direct AI-drug-discovery peers in reviewed public sources are Recursion/Exscientia, Insilico Medicine, Xaira, Isomorphic Labs, Generate Biomedicines, Valo Health, and BenevolentAI. SP005, SP008, SP010, SP011, SP013, SP014, SP016
CP003 Adjacent substitutes include Schrödinger’s physics-based platform, Evotec’s integrated R&D services, and in-house pharma discovery teams. SP018, SP020, SP021, SP022
CP004 insitro’s own differentiation claim rests on multimodal human and cellular data, machine learning, and a pipeline spanning metabolism, neuroscience, and ophthalmology. SP001, SP002
CP005 Recursion says its platform is built on more than 50 petabytes of data and millions of cell experiments per week. SP016
CP006 The Recursion-Exscientia combination disclosed more than 10 internal programs, more than 10 partnered programs, roughly USD 450 million of realized partner cash, and about 800 employees. SP017
CP007 Insilico’s pipeline page lists more than 40 total programs, 30 preclinical candidates nominated since 2021, and 13 pipelines that received IND approval. SP006
CP008 Insilico says it nominated 20 preclinical candidates from 2021 to 2024 with an average 12- to 18-month timeline and only 60 to 200 molecules synthesized and tested per program. SP007
CP009 Insilico’s ISM8969 received FDA IND clearance in 2026 for Parkinson’s disease, with a co-development arrangement that includes 50/50 global rights and up to USD 66 million in upfront and milestones. SP007
CP010 Insilico says its three key license-out deals add up to as much as USD 2.1 billion in total contract value. SP007
CP011 Xaira says it is building predictive and agentic AI models across the complete spectrum of drug discovery and development. SP008
CP012 Xaira’s news page says X-Cell launched in 2026 as a virtual-cell model trained on the largest-ever genome-wide perturbation dataset, X-Atlas/Pisces. SP009
CP013 Xaira’s news page also says the company raised nearly USD 1 billion and released the largest publicly available genome-wide Perturb-seq dataset in 2025. SP009
CP014 Generate Biomedicines positions itself as a generative-biology company that uses machine learning to create medicines on demand across therapeutic modalities. SP010
CP015 Generate says it has generated, built, and tested 42,000 proteins and operates with more than 140,000 square feet of space, making it more biologics-oriented than insitro’s current public narrative. SP010
CP016 Isomorphic Labs says it is building predictive and generative AI beyond AlphaFold to design novel medicines. SP011
CP017 Isomorphic Labs’ news page lists a USD 600 million external investment round announced in 2025. SP012
CP018 Isomorphic Labs’ news page also lists collaborations or collaboration expansions with Lilly, Novartis, and Johnson & Johnson between 2024 and 2026. SP012
CP019 Valo Health emphasizes AI-enabled human causal biology and closed-loop chemistry through an ecosystem-led innovation model. SP013
CP020 BenevolentAI emphasizes a decade of investment in a knowledge graph and proprietary ontologies supporting life-science decision making. SP014
CP021 BenevolentAI’s news page shows both a strategic overhaul and a proposed delisting via merger, which is adverse evidence for category durability. SP015
CP022 The Exscientia news URL resolving to Recursion and the Nasdaq merger announcement together show that Exscientia has effectively been absorbed into the Recursion platform. SP017, SP026
CP023 Schrödinger says it deploys a physics-based computational platform across both collaborative and proprietary drug-discovery programs. SP018
CP024 Relay Therapeutics says it uses integrated experimental and computational discovery against challenging targets in precision oncology and genetic disease. SP019
CP025 Evotec competes through a fully integrated R&D value chain and flexible partnering models rather than a pure AI-only value proposition. SP020
CP026 MarketsandMarkets says pharmaceutical and biotechnology companies are the largest and fastest-growing end users of AI in drug discovery. SP027
CP027 WHO, EMA, and EFPIA all frame trust, transparency, risk-based governance, and lifecycle controls as critical to AI adoption in medicines, making trust posture a competitive variable rather than only a compliance issue. SP023, SP024, SP025
CP028 KPMG says the post-2022 market pushed buyers toward lower-risk partnership structures. SP021
CP029 McKinsey says pharma still has not seen clear evidence that AI alone materially shortens development timelines or improves success rates. SP022
CP030 insitro’s own BMS and Lilly deals show that it packages value through collaboration rights, milestones, royalties, and option-style economics rather than through public seat-based software pricing. SP003, SP004
CP031 Across the reviewed peer set, public pricing is seldom list-priced; economics appear through milestone deals, license-outs, co-development, or integrated service bundles. SP007, SP017, SP020, SP021
CP032 Recursion and Insilico provide clearer public economic proof points than most private peers in the reviewed source set. SP007, SP017
CP033 Pharma buyers can likely multi-home across several AI platforms because many pre-deal value propositions remain partially substitutable. SP021, SP022, SP027
CP034 Switching costs rise only after shared data pipelines, model tuning, or program rights become embedded inside an active collaboration. SP003, SP004, SP021
CP035 Distribution power still sits with large pharma or later-stage commercial organizations rather than insitro or most private AI-biopharma peers. SP003, SP004, SP018, SP020
CP036 insitro’s moat is not raw capital scale but the combination of multimodal disease data, platform-to-pipeline continuity, and partner validation. SP001, SP002, SP003, SP004
CP037 Recursion, Xaira, and Isomorphic appear stronger than insitro on capital or platform ambition in current public evidence. SP009, SP012, SP017
CP038 Insilico and Recursion show stronger public industrialization or proof cadence than insitro in the reviewed sources. SP006, SP007, SP017
CP039 Generate, Valo, and Benevolent are more adjacent or differently shaped substitutes than direct one-for-one peers for insitro’s current pipeline. SP010, SP013, SP014
CP040 insitro’s moat could commoditize if platform claims become substitutable across many AI vendors and internal pharma build teams. SP021, SP022, SP023, SP024, SP025
CP041 BenevolentAI’s restructuring and Exscientia’s absorption provide concrete adverse evidence that the category can consolidate or compress before durable pricing power appears. SP015, SP017, SP026
CP042 The net picture is a crowded but still fragmented field in which insitro is differentiated, yet not clearly dominant on public scale, proof, or pricing power. SP021, SP022, SP023, SP024, SP025
CI001 insitro’s public financial model is partnership-driven and still pre-product, with monetization disclosed through collaboration economics rather than product sales or list-priced software. SI001, SI003, SI004, SI005, SI006, SI007
CI002 Official sources document a $50M BMS upfront, a $25M BMS milestone in 2024, up to $20M of new funding in the 2025 ChemML extension, and a $10M milestone in the 2026 target expansion; 2026 company messaging also cites about $150M of partnership revenue across BMS, Lilly, and Gilead. SI001, SI003, SI004, SI005, SI006, SI008
CI003 Public funding history supports more than $100M raised by 2019, $243M total venture funding by mid-2020, and a further $400M Series C in 2021; later company messaging describes insitro as backed by roughly $800M in capital. SI001, SI002, SI009, SI010
CI004 Publicly disclosed revenue-bearing or milestone-bearing pharma relationships are concentrated in Gilead, Bristol Myers Squibb, and Lilly. SI001, SI003, SI007, SI010
CI005 Forbes reported that insitro’s Gilead collaboration included a $15M upfront payment and up to $1B in milestone potential. SI010
CI006 The 2024 Lilly structure is not a simple up-front monetization event: insitro keeps full global rights to its programs while Lilly becomes eligible for milestones and royalties. SI007
CI007 Public pricing is bespoke and milestone-heavy rather than list-priced, with economics disclosed as upfronts, milestones, royalties, retained rights, and option-style collaboration terms. SI003, SI004, SI005, SI006, SI007, SI025, SI026
CI008 The Bristol Myers Squibb relationship shows a land-and-expand GTM pattern: discovery collaboration in 2020, milestone conversion in 2024, ChemML extension in 2025, and additional target expansion in 2026. SI003, SI004, SI005, SI006, SI008
CI009 insitro’s apparent buyer is large-pharma R&D and business development, implying a low-customer-count, high-contract-value enterprise selling motion rather than broad-distribution software sales. SI003, SI007, SI025, SI026
CI010 Public CAC or payback metrics are unavailable; the best public proxy for sales efficiency is whether insitro can deepen a small number of strategic accounts over multiple years. SI003, SI006, SI025, SI026
CI011 In February 2026, insitro itself said it was backed by roughly $800M in capital and about $150M in partnership revenue from BMS, Lilly, and Gilead. SI001
CI012 GetLatka reports $69M of 2024 revenue, $643M total funding, a $2.4B 2021 valuation, and 262 employees. SI014
CI013 Awaira reports $743M of disclosed funding, an approximately $2.2B valuation as of March 2026, around 300 employees, and only a very broad estimated ARR range. SI015
CI014 Usearch lists $7.5M of revenue and 267 employees for insitro. SI016
CI015 WorxForm summarizes insitro as having raised $400M with roughly 250 employees, reinforcing that public trackers disagree even on basic company scale. SI017
CI016 Public third-party trackers conflict materially on insitro’s revenue, funding, valuation, and headcount, so none should be treated as canonical without management backup. SI001, SI014, SI015, SI016, SI017
CI017 insitro’s cost structure likely skews toward wet-lab biology, data generation, compute, chemistry, and senior scientific talent rather than software-only delivery costs. SI001, SI005, SI006, SI007
CI018 The 2025 ChemML extension highlighted a 192-H100 GPU compute cluster, reinforcing that high-end compute is a real cost center inside the discovery model. SI005
CI019 Public traction is visible mainly through collaboration cash events and cumulative partnership-revenue claims, not through audited annual revenue, bookings, or backlog disclosure. SI001, SI004, SI005, SI006, SI007
CI020 BioPharma Dive reported that insitro cut 22% of its workforce in 2025, leaving about 230 workers and aiming to operate into 2027. SI011
CI021 The 2025 restructuring suggests management prioritized cash preservation and clinic readiness over continued hiring velocity. SI011, SI012
CI022 EY said 2024 follow-on financings were the worst since 2016, IPO activity remained muted, and biotechs were being pushed toward cost reductions, alternate funding, and pharma partnerships. SI012
CI023 Fierce’s 2026 layoff tracker shows workforce reductions remained widespread across biotech after 2025, reinforcing that cost resets were a sector-wide feature rather than an insitro-only anomaly. SI012, SI013
CI024 Recursion’s 2023 annual report shows $44.6M of total revenue, a $328.1M net loss, $241.2M of R&D expense, and $391.6M of year-end cash. SI018
CI025 Recursion’s filing also says revenue recognition is not directly correlated to cash receipts and includes milestone-based variable consideration, illustrating why collaboration cash does not map cleanly to GAAP revenue. SI018
CI026 Schrödinger’s 2023 annual report shows $216.7M of total revenue, $181.8M of R&D expense, $468.8M of cash, cash equivalents, restricted cash, and marketable securities, and about $136.7M of operating cash use. SI019
CI027 Macrotrends shows Schrödinger revenue at roughly $208M in 2024 after $216.7M in 2023, suggesting the most revenue-visible public comp still does not scale like hypergrowth software. SI019, SI024
CI028 Relay’s 2023 annual report shows no product revenue, a $342.0M net loss, $330.0M of R&D expense, $750.1M of cash/investments, and $300.3M of operating cash use. SI020
CI029 Taken together, Recursion, Schrödinger, and Relay show that AI-biopharma platforms can remain highly cash consumptive even with collaboration or software revenue. SI018, SI019, SI020
CI030 CompaniesMarketCap listed public 2026 market caps of about $1.73B for Recursion, $0.95B for Schrödinger, and $2.46B for Relay. SI021, SI022, SI023
CI031 Tracker-based private valuation signals for insitro of roughly $2.2B to $2.4B sit inside the public-equity band of these comps despite far lower disclosure quality. SI014, SI015, SI021, SI022, SI023
CI032 Because insitro has public evidence of realized collaboration cash, revenue quality is better than that of a zero-monetization preclinical biotech, but still materially weaker than a disclosed commercial or audited recurring-revenue model. SI001, SI004, SI007, SI018, SI019, SI020
CI033 Gross margin path is hard to underwrite publicly; partnership economics carry milestone and royalty upside, but current delivery almost certainly absorbs significant lab, compute, and personnel expense. SI005, SI018, SI019, SI020
CI034 Working-capital, capex, lease, and debt burdens are essentially opaque for insitro because no audited statements or contract schedules are public. SI001, SI002, SI011
CI035 No public source reviewed disclosed insitro’s current cash balance, monthly burn, deferred revenue, lease commitments, or debt facilities. SI001, SI002, SI011
CI036 The trackers themselves acknowledge that at least some of their figures are estimated or based on public data, limiting their weight for real underwriting. SI014, SI015
CI037 KPMG and McKinsey both describe pharma AI adoption as proof-sensitive and partnership-oriented, consistent with insitro selling strategic program access rather than generalized software subscriptions. SI025, SI026
CI038 The most plausible next value-inflection triggers are additional partnership expansion, milestone conversion, or successful clinic-readiness progression rather than near-term public-market liquidity. SI005, SI006, SI011, SI012
CI039 The 2021 Series C proceeds were earmarked for platform expansion, enabling datasets, complementary technologies, and in-licensed assets, implying capital deployment is platform- and pipeline-building rather than near-term profitability. SI002
CI040 Financial model quality is currently blocked by missing current cash, burn, revenue-recognition, and obligation data. SI001, SI002, SI011, SI014
CI041 Overall financial verdict: insitro is partnership-validated but still opaque and financing-dependent until internal programs or milestone conversion create clearer, auditable proof of revenue quality and capital efficiency. SI001, SI011, SI012, SI018, SI019, SI020
CE001 insitro’s public product is not a general-purpose software application but an end-to-end discovery workflow that turns multimodal human and cellular data into targets, biomarkers, and therapeutic designs. SE001, SE002, SE004
CE002 insitro publicly describes the platform as powering both wholly owned and partnered programs across metabolism, neuroscience, and oncology. SE002, SE004
CE003 The company’s platform narrative emphasizes a modular, reusable, automated stack that generates or acquires data and then converts it into decision support for target and drug discovery. SE001, SE004
CE004 Virtual Human is publicly framed as a genetically anchored causal AI engine that explains disease mechanisms and helps decide what biology to target. SE011, SE012, SE015
CE005 ClinML creates scalable human-derived phenotypes from cohort data, illustrated by a brown-adipose-tissue phenotype derived from 69,598 UK Biobank MRI scans. SE012
CE006 CellML is used to screen genetically supported targets in relevant human cells using high-content imaging, transcriptomics, and functional assays. SE012, SE015
CE007 POSH integrates pooled CRISPR screening, Cell Painting, and self-supervised deep learning to infer gene function at scale. SE010, SE018
CE008 In the public POSH study, self-supervised models recovered 2.5 times more functional gene relationships than conventional expert-designed analysis and yielded validated target insights. SE010, SE018
CE009 The cp-posh repository publicly exposes datasets, training and inference scripts, notebooks, and pretrained model weights, showing that part of insitro’s research tooling is reproducible outside the company. SE017, SE018
CE010 insitro’s GitHub organization shows a maintained public engineering and research surface that includes redun and multiple public-research repositories. SE016, SE017
CE011 TherML is publicly described as an integrated design layer spanning small molecules, oligonucleotides, and complex biologics within one platform. SE011, SE014
CE012 The CombinAbleAI acquisition extends TherML into antibodies and other complex biologics using a physics-informed optimization engine pre-trained on more than 100,000 molecular-dynamics surrogates. SE011, SE014
CE013 TherML operates as a closed-loop active-learning system connected to automated labs so that experimental results iteratively refine design predictions. SE011, SE014
CE014 TherML and ChemML are positioned to optimize both target activity and developability rather than deferring manufacturability and ADMET concerns until late in the workflow. SE011, SE014
CE015 The BMS ChemML extension and the Lilly small-molecule collaboration together show that insitro’s small-molecule stack includes QAL-driven data generation, active-learning medicinal chemistry, ADMET models, and substantial compute infrastructure including a 192-H100 GPU cluster. SE008, SE009
CE016 The 2025 Lilly collaboration shows insitro building ADMET models on Lilly preclinical data and making those models available to Lilly TuneLab partners on a federated infrastructure. SE009
CE017 The 2024 Lilly metabolic agreements show the platform pairing internal target discovery with both siRNA-delivery technology and antibody discovery while insitro retains global rights. SE006, SE009
CE018 The 2026 BMS expansion shows insitro pursuing ALS-1 across both oligonucleotide and small-molecule modalities and using TherML to select the optimal intervention for each target. SE013, SE015
CE019 Genomics England’s own partner page confirms that insitro’s embedding search engine is deployed into the secure Genomics England Research Environment for research partners. SE019
CE020 INSIGHT’s own collaboration page says only INSIGHT researchers access the OCT data while the foundation model is built inside a secure research environment. SE020
CE021 The BAT program demonstrates a public workflow in which a ClinML-derived human phenotype feeds a CellML screen and then in vivo validation, culminating in a named preclinical asset, BAT-01. SE002, SE012
CE022 BAT-01 knockdown produced a 15 percent body-weight reduction and a 25 percent fat-mass reduction in obese mice without reducing caloric intake. SE012
CE023 The 2020 BMS collaboration used insitro’s Human platform to build iPSC-derived ALS/FTD disease models and gave BMS opt-in rights for targets that insitro identified. SE005, SE007
CE024 The 2024 BMS milestone update described more than 200 engineered and patient ALS cell lines, ML-enabled motor-neuron differentiation, high-content imaging, and POSH as core proprietary elements of the discovery engine. SE007, SE015
CE025 The ALS playbook says POSH can measure thousands of cellular features across millions of perturbed cells, underscoring data density as a core design principle of Virtual Human. SE015
CE026 insitro’s differentiation is publicly anchored in combining large-scale human cohort data with internally generated cellular perturbation data rather than in molecule-generation models alone. SE001, SE011, SE012, SE015
CE027 Justia’s assignee page lists multiple 2025-2026 insitro patents on biological image transformation, autonomous cell imaging and modeling, and machine-learning-based spatial omics imputation. SE025
CE028 Public patent activity suggests a real imaging and omics IP estate, but the scope, field-of-use limits, and licensing terms are not auditable from open sources alone. SE025
CE029 External proof exists for discrete platform components, but there is no public evidence of a broad self-serve product surface with pricing, API documentation, or support SLAs for general customers. SE001, SE002, SE016, SE017, SE018, SE019, SE020
CE030 The public roadmap from 2024-2026 shows insitro moving from discovery-engine branding toward modality-agnostic therapeutic design and nearer-clinic asset language. SE006, SE008, SE011, SE013, SE014
CE031 Despite richer platform messaging, public proof still centers on preclinical, discovery, or partner-delivery milestones rather than approved products or public human efficacy readouts. SE002, SE006, SE012, SE013
CE032 Trust posture in external data partnerships is framed around secure partner environments, restricted data access, and use-specific model development rather than consumer-style product controls. SE019, SE020, SE021
CE033 FDA’s January 2025 draft guidance says AI used to support drug regulatory decisions should undergo risk-based credibility assessment for a specific context of use. SE024
CE034 EMA’s January 2026 principles require human-centric design, data governance, validation, lifecycle management, and a clear context of use for AI in drug development. SE022
CE035 WHO’s AI-for-health guidance emphasizes ethics, governance, and trust because legal frameworks lag technical deployment. SE023
CE036 insitro’s public privacy policy applies to website and cookie or inquiry data and does not publicly describe enterprise-grade controls for collaboration datasets. SE019, SE020, SE021
CE037 No public SOC 2, ISO 27001, GxP, HIPAA, or HITRUST certification disclosure was found across the sourced platform, partner, and privacy materials. SE001, SE019, SE020, SE021
CE038 No public source reviewed discloses platform throughput metrics such as screens per quarter, targets advanced per year, average design-make-test cycle time, or support uptime. SE001, SE002, SE010, SE011
CE039 The public IP and open-source surface is directionally supportive of moat, but externally auditable evidence still covers only slices of the platform and not the proprietary datasets or internal model weights that likely matter most. SE016, SE017, SE018, SE025
CE040 Overall product-tech verdict: insitro appears to have a genuinely differentiated, modular, modality-agnostic discovery stack with real external deployments and partner pull, but public evidence still leaves quality-system, data-rights, throughput, and clinical-translation validation gaps that matter for underwriting. SE011, SE014, SE019, SE020, SE021, SE024, SE025
CU001 insitro’s visible external customer base is best described as a small set of high-value pharma counterparties plus a small number of research-data partners, not a broad software install base. SU001, SU002, SU005, SU007, SU011, SU012, SU013
CU002 insitro’s home and pipeline materials show that the platform supports both partnered and wholly owned programs, so some platform output is consumed internally rather than sold to external customers. SU001, SU002, SU003, SU004
CU003 The public record does not show a self-serve product, public pricing page, marketplace channel, or broad long-tail customer roster. SU001, SU002, SU003, SU004
CU004 The visible channel motion is direct, science-led business development around bespoke collaborations rather than transactional software sales. SU005, SU007, SU011, SU023, SU024
CU005 Named external customer proof is geographically concentrated in U.S. big pharma and U.K. health-data institutions. SU005, SU007, SU011, SU012, SU013
CU006 The visible buyer-user-payer map has three layers: direct pharma payers, partner-controlled research-environment users, and indirect biotech users reached through Lilly TuneLab. SU007, SU009, SU012, SU013, SU015, SU020
CU007 There is no public evidence that insitro has diversified into a large number of mid-market biotech or self-serve customers. SU001, SU002, SU003, SU004, SU023, SU024
CU008 Gilead’s 2019 agreement was a three-year NASH collaboration in which insitro’s platform was used to create disease models and identify targets that Gilead could advance. SU011, SU027
CU009 The Gilead agreement disclosed enterprise-scale economics: $15 million upfront, up to $35 million in near-term operational milestones, up to $200 million per target plus royalties, and opt-in co-development rights. SU011, SU025
CU010 Bristol Myers Squibb entered a five-year discovery collaboration with insitro in 2020 around ALS and FTD, including upfront cash and downstream milestone potential. SU005
CU011 In December 2024 insitro disclosed a $25 million BMS milestone and the selection of the first novel ALS target, proving follow-on economics beyond the original launch announcement. SU006
CU012 In October 2025 the BMS relationship expanded into ChemML-enabled small-molecule design for a novel ALS target. SU008
CU013 In March 2026 the BMS relationship expanded again with two additional targets and a $10 million milestone payment, making BMS the clearest public land-and-expand account. SU010, SU026
CU014 Lilly and insitro announced three strategic agreements in 2024 spanning metabolic disease target programs, siRNA delivery, and antibody discovery. SU007
CU015 The 2025 Lilly small-molecule collaboration broadened the relationship into ADMET and pharmacokinetic model building on Lilly’s proprietary preclinical data. SU009
CU016 The 2025 Lilly collaboration explicitly says insitro-built models will be available to Lilly, insitro, and Lilly TuneLab partners, giving indirect reach into a broader biotech partner ecosystem. SU009
CU017 Lilly ExploR&D describes itself as supporting emerging biotech companies from early stage through clinical proof of concept and cites 50+ external programs over 15 years, which contextualizes insitro within an established long-duration external-innovation channel. SU020
CU018 Genomics England’s official partner page says insitro will make its embedding search engine available to Genomics England’s network of research partners within the secure Research Environment. SU012
CU019 Genomics England also says insitro became a broader research partner, implying the relationship extends beyond a one-off technical demo. SU012
CU020 Genomics England’s current Research Environment page describes a multi-user platform for academia and industry built around one of the largest genomic datasets linked to clinical records. SU014
CU021 Genomics England’s industry-researcher page shows three visible commercial engagement modes—self-service data access, bioinformatics consulting, and R&D collaboration—which clarifies how insitro’s tool sits inside a broader buyer and user workflow. SU015
CU022 Genomics England’s documentation shows the Research Environment is an AWS-accessed virtual desktop in which data stay inside the environment and only results are exported. SU016
CU023 Genomics England security guidance emphasizes Airlock-mediated import/export and explicitly forbids screenshots that bypass controls, highlighting the operational friction of partner-environment deployments. SU017
CU024 INSIGHT and Moorfields say they are collaborating with insitro to build an OCT foundation model on millions of linked retinal images and clinical records for neurodegeneration-related discovery. SU013
CU025 INSIGHT/Moorfields also states that only INSIGHT researchers access the underlying data while building the model in a secure research environment, so insitro benefits from the model without direct raw-data possession. SU013, SU018
CU026 INSIGHT’s Secure Research Environment provisions approved researchers with anonymised data, requested software tools, and virtual machines, with export approval controlled by the data custodian. SU018
CU027 INSIGHT’s Data Use Register shows multiple approved external research projects with completed contracts and secure access granted, evidencing that the surrounding infrastructure supports real external research use rather than a marketing pilot. SU018, SU019
CU028 The strongest named customer proof combines either paid milestone progression or partner-side descriptions of concrete tool and model usage; it is materially better than a passive logo wall. SU006, SU010, SU011, SU012, SU013, SU019
CU029 Public outcome evidence exists mainly as disclosed milestones, target nominations, and deployment descriptions, not as user counts, query volumes, ROI statistics, or partner satisfaction metrics. SU006, SU009, SU010, SU012, SU013, SU014, SU015, SU019
CU030 No public source reviewed disclosed NRR, GRR, churn, NPS, active-customer counts, or account-level revenue retention metrics. SU001, SU002, SU003, SU004, SU021, SU022
CU031 There is no public evidence of broad recurring software revenue; the disclosed economics are collaboration-, milestone-, royalty-, and rights-structured. SU005, SU007, SU009, SU011, SU021, SU023, SU024
CU032 BMS is the strongest public durability signal because the relationship progressed from 2020 launch to 2024 milestone conversion, 2025 chemistry extension, and 2026 target expansion. SU005, SU006, SU008, SU010
CU033 Lilly provides positive but still early durability evidence because the public relationship broadened from 2024 multi-agreement collaboration into 2025 small-molecule model work. SU007, SU009, SU020
CU034 Gilead provides strong early customer proof but weak current durability visibility because reviewed public materials do not describe post-term renewal, later milestones, or the collaboration’s present status. SU011, SU025
CU035 Research-environment collaborations validate external adoption and create some operational stickiness, but they do not prove large cash revenue because commercial terms and utilization metrics are undisclosed. SU012, SU013, SU016, SU017, SU018, SU019
CU036 A 2026 company statement says insitro has generated about $150 million in partnership revenue from BMS, Lilly, and Gilead, implying meaningful concentration in just three names. SU021
CU037 Because the company does not disclose the split of partnership revenue across BMS, Lilly, and Gilead, exact top-customer concentration cannot be quantified even though concentration is likely high. SU021
CU038 BioPharma Dive’s 2025 layoff report and insitro’s runway-to-2027 framing suggest the company still depends on concentrated partnerships and internal pipeline proof points rather than diversified customer cash flows. SU022
CU039 KPMG and McKinsey both describe AI-biopharma commercialization as partnership-heavy, proof-sensitive, and integration-intensive, which is consistent with insitro’s public customer motion. SU023, SU024
CU040 Expansion upside exists through deeper BMS and Lilly scopes, broader use inside the Genomics England research network, and indirect reach to biotech users through Lilly TuneLab. SU009, SU010, SU012, SU015, SU020
CU041 Customer satisfaction and procurement quality remain largely private because public materials surfaced no marketplace reviews, independent user testimonials, or disclosed renewal outcomes. SU001, SU002, SU012, SU013, SU014, SU015, SU019
CU042 Overall, insitro’s customer base looks strategically credible and scientifically integrated, but concentrated and under-disclosed relative to what investors would want for a repeatable revenue engine. SU021, SU022, SU023, SU024
CU043 Illustrative continuity proxies suggest BMS-style integrated partnerships should retain better than single-indication discovery deals, but these are analyst heuristics rather than company-reported retention figures. SU005, SU007, SU011, SU022, SU023, SU024
CR001 The highest residual risk in the current insitro case is clinical translation, because public proof remains preclinical or collaboration-led rather than human-outcome-based. SR008, SR009, SR013, SR016, SR017, SR018, SR025
CR002 Public company messaging frames clinic readiness as an upcoming milestone rather than an achieved state. SR009, SR025
CR003 The MASLD program update says CTRO-1013 is still in IND-enabling work and preparing for first-in-human trials, which confirms preclinical progress but not regulatory clearance. SR008
CR004 BMS and Lilly milestones validate discovery and partner appetite, but they do not prove that insitro can convert AI-driven discovery outputs into successful human clinical programs. SR013, SR014, SR015, SR016, SR017, SR018
CR005 FDA says it is seeing a significant increase in AI-related submissions across nonclinical, clinical, postmarketing, and manufacturing phases of the drug lifecycle. SR001
CR006 FDA says its 2025 draft guidance was informed by over 500 submissions with AI components from 2016 to 2023 plus more than 800 external comments on the discussion paper. SR001
CR007 FDA’s guiding-principles page and EMA’s reflection paper both emphasize human-centric design, risk-based assessment, data governance, performance evaluation, and lifecycle management for AI in medicines. SR002, SR004, SR005
CR008 FDA’s discussion paper says careful assessments tied to the specific context of use and a risk-based approach are needed for AI/ML in drug development. SR002, SR003
CR009 EMA’s reflection paper explicitly extends AI considerations from drug discovery through post-authorisation, expanding the compliance surface beyond a narrow clinical-model question. SR004, SR005
CR010 Public insitro materials reviewed do not disclose a mapped control framework showing how the company satisfies FDA and EMA expectations on context of use, data governance, validation, or lifecycle management. SR001, SR002, SR004, SR005, SR028, SR029, SR030
CR011 AI regulatory credibility risk is therefore material before any filing, pivotal partner expansion, or public re-rating tied to clinic entry. SR001, SR002, SR003, SR004, SR005
CR012 insitro’s public privacy policy governs the website only and says no security measures can guarantee security. SR010
CR013 The website privacy policy does not provide assurance about governance for the multimodal collaboration datasets that matter most to the business model. SR010, SR020, SR023, SR024
CR014 Genomics England and INSIGHT partner materials show that sensitive data are kept inside partner-controlled secure environments with gated export processes. SR020, SR022, SR023, SR024
CR015 Those controls mitigate direct leakage risk but also create friction, limit data portability, and may constrain how broadly insitro can compound partner data into reusable platform assets. SR020, SR021, SR022, SR023, SR024
CR016 No public SOC 2, ISO 27001, GxP, HIPAA, or HITRUST disclosure surfaced in the reviewed materials. SR010, SR022, SR024, SR028, SR030
CR017 The public patent record shows a growing IP surface across imaging, biomarker discovery, and platform workflows, but it does not expose freedom-to-operate analyses, inbound license encumbrances, or indemnity terms. SR011
CR018 No public litigation or enforcement action surfaced in the reviewed materials, but that remains a visibility gap rather than a strong clean-bill-of-health signal because insitro is private and lightly disclosing. SR010, SR011, SR025
CR019 Customer and partner concentration risk is high because the company groups about $150 million of partnership revenue into just three names: BMS, Lilly, and Gilead. SR009, SR013, SR017, SR019
CR020 BMS is the strongest de-risking partner relationship and also the clearest single dependency because it has expanded across multiple milestones, targets, and modalities. SR013, SR014, SR015, SR016
CR021 Lilly adds a second major partner but also creates dependency on Lilly-controlled data, TuneLab distribution, and Catalyze360 ecosystem choices. SR017, SR018
CR022 Gilead proves large-pharma willingness to pay, but the current status of that relationship is opaque, reducing visibility into true partner durability. SR019
CR023 Genomics England and INSIGHT relationships show partner-controlled data and deployment dependencies that could constrain reuse, scaling, and external portability. SR020, SR021, SR022, SR023, SR024
CR024 The BMS ChemML extension publicly references QALs, ADMET models, and a 192-H100 GPU cluster, implying material compute dependence and fixed-cost exposure. SR015
CR025 insitro’s operating model is capital intensive because it combines multimodal data generation, wet-lab experimentation, compute infrastructure, and multiple internal pipeline programs. SR008, SR015, SR028, SR029, SR030
CR026 Sector layoff reporting and insitro’s own restructuring show that biotech funding and execution conditions remain difficult. SR006, SR025, SR026, SR027
CR027 BioPharma Dive reports that insitro cut 22% of its workforce and sought to keep operating into 2027, which is a direct signal of both discipline and pressure. SR025
CR028 The 2025 layoff raises execution-capacity risk even if it improves runway, because insitro still needs to move multiple programs and partnerships toward more operationally complex milestones. SR006, SR008, SR025
CR029 Amy Abernethy’s board appointment is a real mitigation for clinical and regulatory-evidence risk because she brings former FDA leadership and clinical-development expertise. SR007
CR030 Joe Hand’s appointment is a real mitigation for talent-strategy risk, but it is not proof that insitro can retain or cheaply hire scarce cross-functional talent after layoffs. SR007, SR009, SR025
CR031 KPMG and McKinsey both describe AI-biopharma success as proof-sensitive, data-dependent, and multidisciplinary, reinforcing that execution risk is structural rather than incidental. SR026, SR027
CR032 FDA and EMA’s joint work on AI principles suggests regulatory expectations are converging internationally, which raises the compliance burden for globally ambitious AI-drug platforms. SR001, SR002, SR004, SR005
CR033 No public IND, CTA, or human-dosing milestone surfaced in the reviewed materials. SR008, SR009, SR028, SR029
CR034 Public clinic-readiness evidence is directional rather than audited: approaching clinic, IND-enabling, and next-year clinic readiness are all visible, but none prove filing success. SR008, SR009, SR025
CR035 If model credibility or regulatory documentation is insufficient, partner milestone timing and financing dependence are likely to worsen because discovery validation and clinic progression are intertwined. SR001, SR009, SR013, SR014, SR015, SR016, SR017, SR018, SR025
CR036 A further material workforce reduction or failure to reach IND or first-in-human readiness by 2027 would be a strong thesis-break signal. SR006, SR008, SR025
CR037 Inability to produce a private AI validation package or mapped controls against FDA and EMA principles would keep regulatory risk high. SR001, SR002, SR003, SR004, SR005
CR038 Inability to produce security audits, quality documentation, or collaboration data-rights summaries would keep legal and operational risk high. SR010, SR020, SR022, SR023, SR024
CR039 Overall residual exposure remains high even though board additions, leadership hires, and secure-environment controls are meaningful mitigants. SR007, SR009, SR014, SR029, SR030
CR040 Risk transmission is reinforcing: AI-governance gaps can slow clinic progress; slower clinic progress can delay partner milestones; delayed milestones can tighten capital and talent flexibility. SR001, SR002, SR003, SR004, SR005, SR009, SR025
CR041 Data-rights and privacy risk is partially mitigated, not eliminated, because the strongest public controls sit in partner environments rather than in an auditable insitro-wide security disclosure set. SR010, SR020, SR021, SR022, SR023, SR024
CR042 IP and legal risk is medium: patents exist, but freedom-to-operate, contract terms, and litigation visibility remain thin. SR011, SR019, SR025
CR043 People and execution risk is medium-high after the layoffs, partially offset by visible investment in leadership and governance. SR007, SR009, SR025
CR044 The overall risk rating should remain high until clinic entry, validation documentation, and concentration transparency improve materially. SR008, SR009, SR019, SR025, SR026, SR027
CR045 For US-facing programs, the documentation burden is also legal rather than merely best-practice: IND workflow expectations sit alongside formal rules such as 21 CFR Part 312 and 21 CFR Part 11. SR012, SR031, SR032
CV001 Public evidence supports a research-more, price-sensitive recommendation rather than a clean buy or pass today. SV001, SV002, SV011, SV017, SV019, SV020, SV024, SV025, SV026, SV027, SV029
CV002 insitro has unusually strong positive inputs for a private techbio: about $800 million of capital and roughly $150 million of cumulative partnership revenue are both cited publicly. SV001, SV002, SV003, SV004, SV005, SV006, SV007, SV009
CV003 The public record still lacks the inputs needed for hard valuation underwriting: current share price, cap table, preference stack, audited current cash, and reliable annual revenue. SV001, SV011, SV013, SV014, SV015, SV016, SV017
CV004 Because entry price and term quality are not public, scenario ranges and public comparables are more honest than a precise IRR or target-return claim. SV017, SV019, SV020, SV024, SV025, SV026, SV027, SV029
CV005 The last clearly disclosed financing benchmark is insitro’s $400 million Series C in 2021. SV002
CV006 Forge estimates the 2021 Series C post-money valuation at about $2.57 billion and lists a $18.29 price per share for that round. SV017
CV007 Low-confidence public trackers cluster insitro around roughly $2.2 billion to $2.4 billion, but they disagree on revenue, funding, and timing details. SV013, SV014, SV017
CV008 That low-$2 billions tracker cluster suggests public secondary references still anchor near the old round rather than proving a later step-up. SV013, SV014, SV017
CV009 No reviewed public source disclosed a later clearly priced financing round, current preference stack, or current liquidation waterfall after the 2021 Series C. SV002, SV013, SV014, SV017
CV010 Joe Hand’s 2026 announcement and BioPharma Dive’s 2025 layoff coverage show management emphasizing clinic readiness and runway into 2027 rather than near-term IPO readiness. SV001, SV011
CV011 BMS is the strongest value-supporting proof asset because the relationship shows upfront cash, later milestones, extension funding, and further target nominations. SV003, SV004, SV005, SV006
CV012 Lilly adds real upside optionality, but public near-term economics are less visible than the BMS relationship. SV007, SV008
CV013 Gilead is useful as historical proof of willingness to pay, but it is not a strong current pillar for valuation support. SV009
CV014 CTRO-1013 remains in IND-enabling and first-in-human preparation, so internal-asset value should still be heavily risk-adjusted. SV010
CV015 FDA and EMA guidance plus insitro’s still-preclinical internal asset state argue against paying frontier-AI style multiples today. SV010, SV034, SV035, SV036
CV016 The most relevant public comp cluster spans roughly $0.5 billion to $2.46 billion across Eikon, Absci, Schrödinger, Recursion, and Relay. SV024, SV025, SV026, SV027, SV029, SV031, SV032, SV033
CV017 Eikon’s public reset to about $539 million and its 44.6% decline since its 2026 IPO show how quickly public markets can discount precommercial therapeutics stories. SV029, SV030
CV018 Recursion at about $1.73 billion despite broader public disclosure, multiple programs, and a large capital base shows that scale and financing alone do not command a frontier premium. SV021, SV024, SV031
CV019 Schrödinger at about $0.95 billion despite real software and services revenue shows that even the most monetized AI-drug-discovery comp is not valued like hypergrowth SaaS. SV022, SV025, SV033
CV020 Relay at about $2.46 billion shows public markets can support a higher mark when clinical programs are clearer, but that upper end is still only low single-digit billions. SV023, SV026, SV032
CV021 Absci at about $0.90 billion shows AI-biology platforms can remain sub-$1 billion in public markets even after multiple years of operating history. SV027, SV028
CV022 A base public-evidence valuation range of about $1.5 billion to $2.5 billion is defensible if partner durability and clinic timing remain intact but no human data or cleaner term disclosure emerges. SV006, SV007, SV011, SV012, SV016, SV018, SV019, SV020, SV021
CV023 A bear range of about $0.8 billion to $1.3 billion is appropriate if clinic timing slips, partner validation stalls, or financing terms prove investor-unfriendly. SV009, SV010, SV014, SV015, SV017, SV021
CV024 A bull range of about $3.0 billion to $4.5 billion requires actual clinic entry, further BMS or Lilly conversion, and private-market willingness to look through current public comp marks. SV011, SV012, SV014, SV017, SV018, SV020
CV025 At or below roughly $2.0 billion effective post-money with clean terms, the asymmetry becomes interesting enough to pursue diligence aggressively. SV007, SV022, SV023, SV024
CV026 Around the last-known ~$2.57 billion reference, upside looks modest unless private diligence reveals evidence materially stronger than the public record. SV006, SV022, SV024
CV027 Above roughly $3 billion, the current public record does not support underwriting the round. SV017, SV018, SV019, SV020, SV024
CV028 The correct public-evidence stance is therefore research-more / track rather than buy or pass outright. SV001, SV004, SV025, SV026, SV027
CV029 Recommendation confidence should be medium: strong enough to set price guardrails, not strong enough to bless an undisclosed premium valuation. SV002, SV003, SV007, SV028
CV030 Overall risk rating should remain high, because valuation uncertainty compounds regulatory, translation, concentration, and financing risk rather than offsetting them. SV010, SV011, SV014, SV015, SV034, SV035, SV036
CV031 insitro is not IPO-ready on public evidence because no current price history, audited public financials, or clinic-stage product proof are visible. SV003, SV010, SV014, SV030
CV032 The most plausible nearer-term value-realization path is further partner expansion or strategic interest rather than a clean near-term public-market underwriting case. SV011, SV012, SV031
CV033 The bull case requires CTRO-1013 to enter clinic on schedule, continued BMS or Lilly conversion, no further workforce reset, and clean terms. SV010, SV011, SV012, SV018
CV034 The base case assumes partner economics hold, clinic timing remains directional but unproven, and price lands around the low-$2 billions without harsh preferences. SV007, SV010, SV022
CV035 The bear case assumes no clinic proof by end-2027, partner plateau, or term-sheet features that reveal heavy dilution or preference overhang. SV009, SV010, SV018, SV030
CV036 Thesis-break triggers include another material layoff, no IND or first-in-human progress by end-2027, inability to produce AI validation or security materials, and visible partner non-renewal. SV010, SV018, SV034, SV035, SV036, SV037
CV037 The top diligence blockers are current cash and burn, current cap table and preferences, AI validation documentation, partner contract terms, and the IND path. SV003, SV009, SV034, SV035, SV036, SV037
CV038 Tracker revenue estimates ranging from roughly $7.5 million to $69 million are too noisy to support a revenue-multiple valuation. SV013, SV015
CV039 Because annual revenue is unreliable and internal asset value is preclinical, valuation should anchor to proof state and public comp cluster rather than SaaS-like multiples. SV014, SV016, SV019, SV020, SV038
CV040 KPMG and McKinsey both describe AI-biopharma adoption as proof-sensitive, supporting only a disciplined premium until insitro converts promise into durable external or clinical outcomes. SV019, SV020
CV041 Public peer filings show severe cash consumption even after public listing: Recursion lost $328 million, Schrödinger used about $137 million of operating cash, and Relay lost $342 million in 2023. SV021, SV022, SV023
CV042 Public comp one-year moves are volatile: Recursion is down about 24.9%, Schrödinger about 51.0%, Eikon about 44.6%, while Relay and Absci moved sharply in the opposite direction. SV028, SV030, SV031, SV032, SV033
CV043 If private diligence shows price below the public-comp-plus-premium range and terms are clean, the recommendation could upgrade to selective pursue. SV025, SV029, SV037
CV044 If private diligence shows price above $3 billion or harsh preferences without stronger proof, the recommendation should downgrade to pass. SV027, SV035, SV037
CV045 The central valuation stance is not that insitro lacks quality; it is that price discipline has to dominate because public proof still lags private narrative potential. SV002, SV028, SV029, SV030
CV046 Amy Abernethy’s board addition is a meaningful governance mitigant, but governance strength alone cannot carry valuation without price, term, and clinic proof. SV010, SV011, SV038
来源
编号出版方标题引文
SO001 insitro Making Medicines Differently - insitro
SO002 insitro Purpose - insitro
SO003 insitro Our Pipeline Focused On Insights & Patient Value - insitro
SO004 insitro Join Us - insitro
SO005 insitro insitro Appoints Joe Hand as Chief People Officer to Advance Talent Strategy for Next Stage of Development
SO006 insitro Leading Clinical Research Innovator, Amy Abernethy, M.D., Ph.D., Joins insitro Board of Directors
SO007 insitro insitro Raises $400 Million in Series C Financing
SO008 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for ALS and FTD
SO009 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases
SO010 insitro insitro Receives $25 Million in Milestone Payments from Bristol Myers Squibb for ALS Discovery Milestones
SO011 insitro insitro and Genomics England Announce Partnership to Provide Multimodal Search Capabilities
SO012 insitro insitro and INSIGHT at Moorfields Eye Hospital Announce Collaboration to Expand Research in Neurodegeneration
SO013 insitro insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets
SO014 insitro insitro to Acquire CombinAbleAI to Complete its Full Stack, Modality-Agnostic AI Platform for Drug Discovery and Design
SO015 insitro insitro Completes First AI-Enabled Human Genetics Study of Brown Adipose Tissue, Shares Differentiated Targets with Anti-Obesity Effects
SO016 Forbes Coursera Cofounder Daphne Koller Melds AI And Biology In Drug Startup insitro
SO017 Forbes Exclusive: Machine Learning Company insitro Raises $143 Million to Bridge Biology and AI
SO018 BioPharma Dive 4 more biotechs cut staff amid market tumult
SO019 GitHub insitro · GitHub
SO020 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SO021 McKinsey How pharma is rewriting the AI playbook: Perspectives from industry leaders
SO022 GetLatka insitro revenue, valuation, funding and headcount profile
SO023 Awaira Insitro company profile and valuation tracker
SO024 Usearch Insitro overview, layoffs and company signals
SO025 WorxForm Insitro careers, culture and funding overview
SO026 Business Wire insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets
SM001 insitro Platform - insitro
SM002 insitro Our Pipeline Focused On Insights & Patient Value - insitro
SM003 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases
SM004 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for ALS and FTD
SM005 insitro insitro and UK’s INSIGHT at Moorfields Eye Hospital Announce Collaboration to Expand Research Efforts in Neurodegeneration and Related Conditions
SM006 insitro insitro and Genomics England Announce Partnership to Provide Multimodal Search Capabilities
SM007 World Health Organization Harnessing Artificial Intelligence for Health
SM008 McKinsey & Company How pharma is rewriting the AI playbook: Perspectives from industry leaders In the pharma industry alone, the AI market is projected to grow from more than $4 billion this year to a whopping $25.7 billion by 2030.
SM009 KPMG Artificial intelligence and its expanding role across the biopharma landscape AI-focused M&A and partnership deals showed a compounded annual growth rate of 27.3 percent from 2013 to 2022.
SM010 EFPIA The Pharmaceutical Industry in Figures: Key Data 2025
SM011 EFPIA AI Across the Medicines Lifecycle: Insights from Preliminary Case Studies and Considerations for Policy
SM012 European Medicines Agency Guiding principles of good AI practice in drug development
SM013 World Health Organization Vision impairment and blindness
SM014 World Health Organization Obesity and overweight
SM015 National Eye Institute Eye Health Data and Statistics
SM016 Centers for Disease Control and Prevention VEHSS Modeled Estimates: Prevalence of Diabetic Retinopathy (DR)
SM017 Centers for Disease Control and Prevention National ALS Disease Estimates
SM018 National Institute of Neurological Disorders and Stroke Amyotrophic Lateral Sclerosis (ALS)
SM019 American Liver Foundation The Facts About Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD)
SM020 American Association for the Study of Liver Diseases Steatotic-What? Changes in Fatty Liver Nomenclature
SM021 MarketsandMarkets Artificial Intelligence in Drug Discovery Market worth $6.89 billion by 2029
SM022 Precedence Research AI in Pharmaceutical Market Size to Hit USD 16.49 Billion by 2034
SM023 Precedence Research Artificial Intelligence In Drug Discovery Market Size to Surpass USD 16.52 Bn by 2035
SM024 Nasdaq / GlobeNewswire Recursion and Exscientia, Two Leaders in the AI Drug Discovery Space, Have Officially Combined
SM025 Recursion Recursion
SM026 Schrödinger Schrödinger
SM027 Relay Therapeutics Relay Therapeutics
SM028 Evotec Evotec
SM029 Grand View Research Artificial Intelligence In Precision Medicine Market Report, 2030
SM030 IQVIA AI Trends in Pharma: Enhancing Drug Safety and Regulatory Compliance for 2025
SP001 insitro Platform - insitro
SP002 insitro Our Pipeline Focused On Insights & Patient Value - insitro
SP003 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for ALS and FTD
SP004 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases
SP005 Insilico Medicine Main | Insilico Medicine
SP006 Insilico Medicine Pipeline | Insilico Medicine
SP007 Insilico Medicine Insilico Medicine Receives IND Approval from FDA for ISM8969, an AI-empowered Potential Best-in-class NLRP3 Inhibitor
SP008 Xaira Therapeutics Xaira Therapeutics
SP009 Xaira Therapeutics News & Content | Xaira Therapeutics
SP010 Generate Biomedicines Home
SP011 Isomorphic Labs Reimagining Drug Discovery Process with AI - Isomorphic Labs
SP012 Isomorphic Labs News - Isomorphic Labs
SP013 Valo Health This is Intelligent Health
SP014 BenevolentAI BenevolentAI | AI Drug Discovery | AI Pharma
SP015 BenevolentAI News and Media | BenevolentAI
SP016 Recursion Pioneering AI Drug Discovery | Recursion
SP017 Nasdaq / GlobeNewswire Recursion and Exscientia, Two Leaders in the AI Drug Discovery Space, Have Officially Combined
SP018 Schrödinger Schrödinger
SP019 Relay Therapeutics Relay Therapeutics
SP020 Evotec Evotec
SP021 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SP022 McKinsey & Company How pharma is rewriting the AI playbook: Perspectives from industry leaders
SP023 World Health Organization Harnessing Artificial Intelligence for Health
SP024 European Medicines Agency Guiding principles of good AI practice in drug development
SP025 EFPIA AI Across the Medicines Lifecycle: Insights from Preliminary Case Studies and Considerations for Policy
SP026 Exscientia / Recursion Exscientia News (redirects to Recursion)
SP027 MarketsandMarkets Artificial Intelligence in Drug Discovery Market worth $6.89 billion by 2029
SI001 insitro insitro Appoints Joe Hand as Chief People Officer to Advance Talent Strategy for Next Stage of Development
SI002 insitro insitro Raises $400 Million in Series C Financing
SI003 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for ALS and FTD
SI004 insitro insitro Receives $25 Million in Milestone Payments from Bristol Myers Squibb for ALS Discovery Milestones
SI005 insitro insitro and Bristol Myers Squibb Discover New ALS Medicines in ChemML Collaboration Extension
SI006 insitro insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets
SI007 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases
SI008 Business Wire insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets
SI009 Forbes Coursera Cofounder Daphne Koller Melds AI And Biology In Drug Startup insitro
SI010 Forbes Exclusive: Machine Learning Company insitro Raises $143 Million to Bridge Biology and AI
SI011 BioPharma Dive 4 more biotechs cut staff amid market tumult
SI012 EY EY 2025 Biotech Beyond Borders Report: Biopharma focus on fundamentals to bounce back
SI013 Fierce Biotech Fierce Biotech Layoff Tracker 2026
SI014 GetLatka insitro revenue, valuation, funding and headcount profile
SI015 Awaira Insitro company profile and valuation tracker
SI016 Usearch Insitro overview, layoffs and company signals
SI017 WorxForm Insitro careers, culture and funding overview
SI018 AnnualReports.com / Recursion Pharmaceuticals Recursion Pharmaceuticals 2023 annual report (Form 10-K)
SI019 AnnualReports.com / Schrödinger Schrödinger 2023 annual report (Form 10-K)
SI020 AnnualReports.com / Relay Therapeutics Relay Therapeutics 2023 annual report (Form 10-K)
SI021 CompaniesMarketCap Recursion Pharmaceuticals market capitalization
SI022 CompaniesMarketCap Schrödinger market capitalization
SI023 CompaniesMarketCap Relay Therapeutics market capitalization
SI024 Macrotrends Schrodinger revenue 2019-2025
SI025 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SI026 McKinsey How pharma is rewriting the AI playbook: Perspectives from industry leaders
SE001 insitro AI/ML-driven Discovery
SE002 insitro Systematically Advancing AI Therapeutics Across Diseases
SE003 insitro Purpose
SE004 insitro insitro homepage
SE005 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for Amyotrophic Lateral Sclerosis and Frontotemporal Dementia
SE006 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases
SE007 insitro insitro Receives $25 Million in Milestone Payments from Bristol Myers Squibb for ALS Discovery Milestones
SE008 insitro insitro and Bristol Myers Squibb Discover New ALS Medicines in ChemML Collaboration Extension
SE009 insitro insitro partners with Lilly to build first-in-kind machine learning models to advance small molecule drug discovery
SE010 insitro insitro Validates AI-Enabled POSH Platform in Nature Communications, Bridging Critical Gap in Drug Discovery
SE011 insitro Introducing insitro’s TherML: Rapidly engineering the right therapeutic for the right target
SE012 insitro insitro Completes First AI-Enabled Human Genetics Study of Brown Adipose Tissue, Shares Differentiated Targets with Anti-Obesity Effects
SE013 insitro insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets
SE014 insitro insitro to Acquire CombinAbleAI to Complete its Full Stack, Modality-Agnostic AI Platform for Drug Discovery and Design
SE015 insitro Rewriting the Playbook for ALS Drug Development
SE016 GitHub insitro · GitHub
SE017 GitHub insitro/insitro-research
SE018 GitHub insitro/cp-posh
SE019 Genomics England Insitro and Genomics England announce partnership to provide multimodal search capabilities and derivation of novel insights
SE020 INSIGHT insitro collaboration | INSIGHT
SE021 insitro Privacy Policy
SE022 European Medicines Agency Guiding principles of good AI practice in drug development
SE023 World Health Organization Harnessing Artificial Intelligence for Health
SE024 FDA Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products
SE025 Justia Patents Patents Assigned to Insitro, Inc.
SE026 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SE027 McKinsey How pharma is rewriting the AI playbook: Perspectives from industry leaders
SU001 insitro insitro homepage
SU002 insitro Systematically Advancing AI Therapeutics Across Diseases
SU003 insitro AI/ML-driven Discovery
SU004 insitro Purpose
SU005 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for Amyotrophic Lateral Sclerosis and Frontotemporal Dementia insitro has entered into a five-year, discovery collaboration with Bristol Myers Squibb focused on the discovery and development of novel therapies for ALS and FTD.
SU006 insitro insitro Receives $25 Million in Milestone Payments from Bristol Myers Squibb for ALS Discovery Milestones insitro today announced it has received $25 million from Bristol Myers Squibb representing both the achievement of discovery milestones and the selection of the first novel target for ALS.
SU007 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases insitro today announced the execution of three strategic agreements with Eli Lilly and Company focused on advancing potential new medicines for metabolic diseases.
SU008 insitro insitro and Bristol Myers Squibb Discover New ALS Medicines in ChemML Collaboration Extension The collaboration extension will leverage insitro’s AI-enabled ChemML platform to design new medicines for a novel ALS target that was identified in the first biological evaluation phase.
SU009 insitro insitro partners with Lilly to build first-in-kind machine learning models to advance small molecule drug discovery The machine learning models developed by insitro will be available to insitro and Lilly, as well as their partners, including biotech companies that partner with Lilly TuneLab.
SU010 insitro insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets BMS has nominated two additional targets, ALS-2 and ALS-3 ... insitro received a $10 million milestone payment in connection with the selection of the two additional targets.
SU011 Gilead Sciences Gilead and insitro announce strategic collaboration to discover and develop novel therapies for nonalcoholic steatohepatitis Under the terms of the three-year collaboration ... Gilead can advance up to five targets identified through this collaboration.
SU012 Genomics England Insitro and Genomics England announce partnership to provide multimodal search capabilities and derivation of novel insights insitro will make its embedding search engine available to Genomics England’s network of research partners within the secure Genomics England Research Environment.
SU013 INSIGHT insitro collaboration | INSIGHT only INSIGHT researchers at Moorfields will access the data while building the foundation model in a secure research environment.
SU014 Genomics England The Research Environment | Genomics England Genomics England Research Environment has one of the largest genomic data sets enriched with clinical data. We enable scientists from academia and industry to make discoveries.
SU015 Genomics England Join as an industry researcher | Genomics England Get direct access to data ... within our Research Environment ... Bioinformatics Consulting ... R&D Collaboration.
SU016 Genomics England Research Environment User Guide Welcome pack - Genomics England Research Environment User Guide The RE is a virtual computer that you access through AWS ... Data cannot be exported from the RE, you must carry out all your analyses then only export the results.
SU017 Genomics England Research Environment User Guide Data security and you - Genomics England Research Environment User Guide Do not "screenshot" the Research Environment or otherwise shortcut the Airlock.
SU018 INSIGHT Secure Research Environment | INSIGHT INSIGHT’s Secure Research Environment provides a dedicated, secure platform to provision approved researchers with access to anonymised data and supporting software.
SU019 INSIGHT Data Use Register | INSIGHT successful Data Use Applications ... where a contract between the NHS Data Controller and the lead research applicant has been fully completed and secure access to anonymised data has been granted.
SU020 Lilly ExploR&D Enabling Bold Innovation with Lilly’s R&D Expertise Lilly ExploR&D is a provider of customized R&D solutions ... supporting 50+ external programs in over 15 years of biotech collaboration.
SU021 insitro insitro Appoints Joe Hand as Chief People Officer to Advance Talent Strategy for Next Stage of Development Backed by ~$800M in capital ... including ~$150M in revenue from partnerships with BMS, Lilly, and Gilead.
SU022 BioPharma Dive 4 more biotechs cut staff amid market tumult Insitro ... is laying off 22% of its workforce ... ensure "clinic readiness" next year, and keep running into 2027.
SU023 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SU024 McKinsey How pharma is rewriting the AI playbook: Perspectives from industry leaders
SU025 Forbes Coursera Cofounder Daphne Koller Melds AI And Biology In Drug Startup insitro last year it inked a deal with Gilead to work on a drug for liver disease with $15 million up front and the potential for $1 billion down the road.
SU026 Business Wire insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets
SU027 Business Wire Gilead and insitro Announce Strategic Collaboration to Discover and Develop Novel Therapies for Nonalcoholic Steatohepatitis
SR001 FDA Artificial Intelligence for Drug Development | FDA CDER has seen a significant increase in the number of drug application submissions using AI components over the past few years ... over 500 submissions with AI components from 2016 to 2023.
SR002 FDA Guiding Principles of Good AI Practice in Drug Development | FDA The 10 principles are tailored to the drug development cycle and emphasize ... human-centric by design ... risk-based approach ... data governance and documentation ... life cycle management.
SR003 FDA Using Artificial Intelligence & Machine Learning in the Development of Drug & Biological Products — Discussion Paper and Request for Feedback diverse, and careful assessments that consider the specific context of use are needed. Taking a risk-based approach to evaluate and manage the use of AI/ML can help facilitate innovations and protect public health.
SR004 European Medicines Agency Use of Artificial Intelligence (AI) in the medicinal product lifecycle - Scientific guideline | European Medicines Agency (EMA) This paper reflects on principles relevant to the application of AI and machine learning at any step of a medicines’ lifecycle, from drug discovery to the post-authorisation setting.
SR005 European Medicines Agency Reflection paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle new risks are introduced that need to be mitigated to ensure the safety of patients and integrity of clinical study results ... active measures must be taken to minimise the integration of bias into AI/ML applications.
SR006 Fierce Biotech Fierce Biotech Layoff Tracker 2026 In 2025, industry layoffs continued to rise year over year, prompting the need for another edition of this article.
SR007 insitro Leading Clinical Research Innovator, Amy Abernethy, M.D., Ph.D, Joins insitro Board of Directors Dr. Abernethy ... held the position of Principal Deputy Commissioner of Food and Drugs for the U.S. Food and Drug Administration and ... will lead initiatives to inform health policy.
SR008 insitro Revealing MASLD’s Genetic Architecture, Machine Learning Discovery, and the Path to IRS1 CTRO-1013 is advancing toward First-in-Human studies ... We are progressing CTRO-1013 through IND-enabling studies and preparing for First-in-Human clinical trials.
SR009 insitro insitro Appoints Joe Hand as Chief People Officer to Advance Talent Strategy for Next Stage of Development Backed by ~$800M in capital ... including ~$150M in revenue from partnerships with BMS, Lilly, and Gilead.
SR010 insitro Privacy Policy This Privacy Policy applies only to the Site ... We have implemented reasonable precautions ... Please be aware that despite our best efforts, no data security measures can guarantee security.
SR011 Justia Patents Patents Assigned to Insitro, Inc. Patents Assigned to Insitro, Inc. ... Biological image transformation using machine-learning models ... Discovery platform.
SR012 FDA Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products
SR013 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for Amyotrophic Lateral Sclerosis and Frontotemporal Dementia insitro has entered into a five-year, discovery collaboration with Bristol Myers Squibb focused on ALS and FTD.
SR014 insitro insitro Receives $25 Million in Milestone Payments from Bristol Myers Squibb for ALS Discovery Milestones
SR015 insitro insitro and Bristol Myers Squibb Discover New ALS Medicines in ChemML Collaboration Extension Robust compute infrastructure: A large compute cluster of 192 H100 GPUs ... Advanced ML models for absorption, distribution, metabolism, excretion, and toxicity.
SR016 insitro insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets BMS has nominated two additional targets ... insitro received a $10 million milestone payment in connection with the selection of the two additional targets.
SR017 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases insitro today announced the execution of three strategic agreements with Eli Lilly and Company focused on advancing potential new medicines for metabolic diseases.
SR018 insitro insitro partners with Lilly to build first-in-kind machine learning models to advance small molecule drug discovery The models being developed are designed to improve the efficiency of hit-to-lead and lead optimization efforts ... Lilly TuneLab is part of the Lilly Catalyze360 model.
SR019 Gilead Sciences Gilead and insitro announce strategic collaboration to discover and develop novel therapies for nonalcoholic steatohepatitis Under the terms of the three-year collaboration ... Gilead can advance up to five targets identified through this collaboration.
SR020 Genomics England Insitro and Genomics England announce partnership to provide multimodal search capabilities and derivation of novel insights insitro will make its embedding search engine available to Genomics England’s network of research partners within the secure Genomics England Research Environment.
SR021 Genomics England The Research Environment | Genomics England Genomics England Research Environment has one of the largest genomic data sets enriched with clinical data. We enable scientists from academia and industry to make discoveries.
SR022 Genomics England Research Environment User Guide Data security and you - Genomics England Research Environment User Guide Do not "screenshot" the Research Environment or otherwise shortcut the Airlock.
SR023 INSIGHT insitro collaboration | INSIGHT only INSIGHT researchers at Moorfields will access the data while building the foundation model in a secure research environment.
SR024 INSIGHT Secure Research Environment | INSIGHT INSIGHT’s Secure Research Environment provides a dedicated, secure platform to provision approved researchers with access to anonymised data and supporting software.
SR025 BioPharma Dive 4 more biotechs cut staff amid market tumult Insitro ... is laying off 22% of its workforce ... ensure “clinic readiness” next year, and keep running into 2027.
SR026 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SR027 McKinsey How pharma is rewriting the AI playbook: Perspectives from industry leaders
SR028 insitro insitro homepage
SR029 insitro Systematically Advancing AI Therapeutics Across Diseases
SR030 insitro AI/ML-driven Discovery
SR031 Cornell Legal Information Institute 21 CFR Part 312 - Investigational New Drug Application
SR032 Cornell Legal Information Institute 21 CFR Part 11 - Electronic Records; Electronic Signatures
SV001 insitro insitro Appoints Joe Hand as Chief People Officer to Advance Talent Strategy for Next Stage of Development Backed by ~$800M in capital ... including ~$150M in revenue from partnerships with BMS, Lilly, and Gilead.
SV002 insitro insitro Raises $400 Million in Series C Financing insitro ... announced the closing of a $400 million Series C financing.
SV003 insitro insitro Announces Five-Year Discovery Collaboration with Bristol Myers Squibb to Discover and Develop Novel Treatments for Amyotrophic Lateral Sclerosis and Frontotemporal Dementia insitro has entered into a five-year discovery collaboration with Bristol Myers Squibb.
SV004 insitro insitro Receives $25 Million in Milestone Payments from Bristol Myers Squibb for ALS Discovery Milestones
SV005 insitro insitro and Bristol Myers Squibb Discover New ALS Medicines in ChemML Collaboration Extension The extension could provide up to $20 million in additional funding for expanded platform discovery work.
SV006 insitro insitro and Bristol Myers Squibb Collaboration Expanded with Nomination of New Targets BMS has nominated two additional targets ... insitro received a $10 million milestone payment.
SV007 insitro insitro and Lilly Enter Strategic Agreements to Advance Novel Treatments for Metabolic Diseases insitro today announced the execution of three strategic agreements with Eli Lilly and Company.
SV008 insitro insitro partners with Lilly to build first-in-kind machine learning models to advance small molecule drug discovery The models being developed are designed to improve the efficiency of hit-to-lead and lead optimization efforts.
SV009 Gilead Sciences Gilead and insitro announce strategic collaboration to discover and develop novel therapies for nonalcoholic steatohepatitis Under the terms of the three-year collaboration ... Gilead can advance up to five targets identified through this collaboration.
SV010 insitro Revealing MASLD’s Genetic Architecture, Machine Learning Discovery, and the Path to IRS1 CTRO-1013 is advancing toward First-in-Human studies ... progressing through IND-enabling studies.
SV011 BioPharma Dive 4 more biotechs cut staff amid market tumult Insitro ... is laying off 22% of its workforce ... ensure “clinic readiness” next year, and keep running into 2027.
SV012 EY EY 2025 Biotech Beyond Borders Report: Biopharma focus on fundamentals to bounce back
SV013 GetLatka insitro revenue, valuation, funding and headcount profile insitro reached a $2.4B valuation in 2021 ... In 2024, insitro's revenue reached $69M.
SV014 Awaira Insitro company profile and valuation tracker The current market valuation is approximately $2.2B ... Capital was most recently raised through a Series C of $200M in October 2021.
SV015 Usearch Insitro overview, layoffs and company signals Revenue: $7.5 Million ... Number of Employees: 267.
SV016 WorxForm Insitro careers, culture and funding overview $400M (Series C) ... at around 250 employees.
SV017 Forge Insitro IPO: Investment Opportunities & Pre-IPO Valuations - Forge $2.57B ... 03/15/2021 ... $400MM ... $18.29 price per share.
SV018 Fierce Biotech Fierce Biotech Layoff Tracker 2026 In 2025, industry layoffs continued to rise year over year.
SV019 KPMG Artificial intelligence and its expanding role across the biopharma landscape
SV020 McKinsey How pharma is rewriting the AI playbook: Perspectives from industry leaders
SV021 AnnualReports.com / Recursion Pharmaceuticals Recursion Pharmaceuticals 2023 annual report (Form 10-K)
SV022 AnnualReports.com / Schrödinger Schrödinger 2023 annual report (Form 10-K)
SV023 AnnualReports.com / Relay Therapeutics Relay Therapeutics 2023 annual report (Form 10-K)
SV024 CompaniesMarketCap Recursion Pharmaceuticals market capitalization As of May 2026 Recursion Pharmaceuticals has a market cap of $1.73 Billion USD.
SV025 CompaniesMarketCap Schrödinger market capitalization As of May 2026 Schrödinger has a market cap of $0.95 Billion USD.
SV026 CompaniesMarketCap Relay Therapeutics market capitalization As of May 2026 Relay Therapeutics has a market cap of $2.46 Billion USD.
SV027 CompaniesMarketCap Absci market capitalization As of May 2026 Absci has a market cap of $0.90 Billion USD.
SV028 StockAnalysis Absci market cap Absci has a market cap or net worth of $901.13 million as of May 12, 2026.
SV029 CompaniesMarketCap Eikon Therapeutics market capitalization As of May 2026 Eikon Therapeutics has a market cap of $0.53 Billion USD.
SV030 StockAnalysis Eikon Therapeutics market cap Eikon Therapeutics has a market cap or net worth of $538.68 million as of May 12, 2026 ... down 44.56% since the IPO.
SV031 StockAnalysis Recursion Pharmaceuticals market cap Recursion Pharmaceuticals has a market cap or net worth of $1.73 billion as of May 12, 2026. Its market cap has decreased by -24.94% in one year.
SV032 StockAnalysis Relay Therapeutics market cap Relay Therapeutics has a market cap or net worth of $2.46 billion as of May 12, 2026.
SV033 StockAnalysis Schrödinger market cap Schrödinger has a market cap or net worth of $950.45 million as of May 12, 2026. Its market cap has decreased by -51.03% in one year.
SV034 FDA Artificial Intelligence for Drug Development | FDA CDER has seen a significant increase in the number of drug application submissions using AI components over the past few years.
SV035 FDA Guiding Principles of Good AI Practice in Drug Development | FDA The 10 principles ... emphasize human-centric design, risk-based approach, data governance and documentation, and life cycle management.
SV036 European Medicines Agency Reflection paper on the use of Artificial Intelligence (AI) in the medicinal product lifecycle new risks are introduced that need to be mitigated to ensure the safety of patients and integrity of clinical study results.
SV037 insitro Privacy Policy This Privacy Policy applies only to the Site ... no data security measures can guarantee security.
SV038 insitro Leading Clinical Research Innovator, Amy Abernethy, M.D., Ph.D, Joins insitro Board of Directors Dr. Abernethy ... held the position of Principal Deputy Commissioner of Food and Drugs for the U.S. Food and Drug Administration.