初创公司尽调
尽调报告 Semiconductor Equipment / Advanced Manufacturing Strategic financing / pre-commercial stealth 2026-08-11

Source Foundry

战略信号和资本都很真实,但公开证据仍太少,尚不足以稳妥支撑据称约 $5B 的估值。

Source Foundry 切入真实且战略重要的半导体瓶颈,但公开记录远不足以干净支撑传闻约 $5B 估值。

封面要素

总部 02
San Francisco, California [CO002]
已披露融资总额 04
500 USD M [CO013, CV002]
公开客户 06
None disclosed [CU010]
披露画像 07
Stealth / private-undisclosed [CO032, CO034]

公司概况

Source Foundry 是一家位于旧金山的隐身半导体设备初创公司,由 Stanford 研究人员 Abdulmalik Obaid 和 Joe Burg 于 2025 年创立。公开记录和商标申请显示,公司野心很宽:覆盖光刻机、晶圆和芯片加工软件、掩模和光罩、技术支持,甚至设施工程或定制制造服务。投资逻辑押在一个真实瓶颈上:支持者认为,AI 的约束正从芯片本身上移到制造芯片的工具。但公开证据仍追不上叙事。留存资料没有披露收入、具名客户、精确产品架构或股权结构条款;在据称约 $5B 估值下,公司战略上值得关注,运营上仍高度不透明。

成立时间
2025-07-18
创始人
Abdulmalik Obaid, Joe Burg
创立地点
San Francisco, California
总部
San Francisco, California
产品
公开备案显示,Source Foundry 追求的不只是单台机器:商标覆盖范围包括光刻机、半导体制造机器、先进光刻软件、晶圆加工软件、掩模和光罩,以及售后维修、维护和技术支持服务。因此,公司可能做的是一套设备 + 软件 + 服务栈,瞄准芯片制造上游瓶颈,而不是传统无晶圆厂芯片公司。
客户
具备深厚制程能力和资本预算的企业级半导体买家:领先制程晶圆代工厂、IDM、试验线或研究基础设施运营方,以及能帮助制造工具完成导入验证的生态伙伴。
商业模式
可能是资本设备、软件和技术服务模式,收入可能覆盖工具销售、原型或集成项目、制程软件和装机后支持;公开资料尚未确认哪条收入流先启动。
阶段
Strategic financing / pre-commercial stealth
融资情况
公开报道称,Situational Awareness 于 2026 年 8 月追加 $400M,使已披露承诺总额达到约 $500M,对应约 $5B 估值;Sequoia 此前也支持了公司,但公开条款、售出股权和优先权结构仍未披露。
[CO001, CO002, CO003, CO004, CO013, CO014, CO015, CO016]

执行摘要

主要优势

  • 公司瞄准 AI 时代的真实瓶颈:半导体制造工具,尤其是 lithography;现有产能和交付周期都已成战略约束。
  • 创始人拥有可信的 Stanford 相关材料科学和微制造背景,而不是单纯金融或宣传型履历。
  • 对一家隐身硬科技初创公司来说,公开披露资本异常大:既有 Sequoia 支持,也有 Situational Awareness 报道中的 $500M 总承诺。
  • 商标范围暗示公司可能要做更宽的设备、软件和服务平台,而不是单一狭窄点产品。

主要风险

  • 留存公开来源均未披露收入、具名客户、装机基础或边界清晰的首款产品。
  • 传闻约 $5B 估值已接近后期硬科技平台,而后者披露的经营证明实质上更多。
  • 赞助方集中度高,而且 Situational Awareness 在最近一次跟投前,不得不清掉大部分公开股票组合。
  • 与 ASML 及相邻既有玩家竞争,意味着技术、认证、可靠性和支持要求都极高。
  • 现实买方池很小,因此任何最终商业路径都可能高度集中。

未决问题

  • Cap table 条款、出让股权比例、清算优先权,以及 2026 年 8 月融资是否全为 primary capital 仍未披露。
  • 公开来源没有说明 Source Foundry 的首款产品边界、原型指标或认证里程碑。
  • 尚无公开客户、试点、设计伙伴或生态验证路径被点名。
  • 收入模型、毛利率目标、现金消耗、runway 和按里程碑拆分的资本计划均未公开。
  • 留存来源仍看不清 Situational Awareness 和 Sequoia 之外的投资财团深度。

目录

Chapter 01

01公司概览

1.1 身份、设立与公开足迹

相比据称估值,Source Foundry 的公开身份异常稀疏,因此最可靠的底层事实来自记录,而不是营销。California 商业备案显示,Source Foundry Inc. 是一家存续公司,2025-07-18 注册,成立地为 Delaware,并把 San Francisco 的 674 Harrison Street 作为主要地址和邮寄地址。2026 年 8 月的报道在这个法律壳上叠加了更高层叙事:公司由 Stanford 研究人员 Abdulmalik Obaid 和 Joe Burg 于 2025 年创立,试图让 AI 时代的半导体制造更快、更便宜、更可扩展。这个组合很关键。法律记录证明公司存在,也说明它在哪里;报道解释投资人为什么在意。审阅过的证据仍没有给出一个普通初创公司的公开入口。留存资料没有公开半导体产品网站、工具规格表、客户案例或发布路线图,外部人无法一步从法律存在走到商业理解。[CO001, CO002, CO003, CO004, CO020, CO031]

KPI 快照表
指标数值 / 状态日期置信度缺口
法律实体Source Foundry Inc.,加州有效存续股份公司;成立地为 Delaware2026-04-01仅有私营公司法律壳;备案未嵌入公开经营指标
创立日期2025-07-18 备案日期;媒体称 2025 年创立2026-08-10未找到更早的公开起源日期
总部674 Harrison St., San Francisco, California(美国加州旧金山)2026-04-01除通信地址外,没有公开园区、晶圆厂或实验室场地
创始人Abdulmalik Obaid 和 Joe Burg2026-08-08公开记录未披露创始人股权或董事会角色
现任高管Obaid(CEO/CFO);Burg(公司秘书)2026-04-01留存来源中没有公开具名的 COO、CTO 或董事会名单
业务重点半导体制造机器、光刻工具、晶圆软件和设施工程服务2026-03-04商标范围传递意图,不等于已交付产品
最新披露新增资本$400M 来自 Situational Awareness2026-08-08融资工具和交割机制未披露
Situational Awareness 披露总承诺额$500M2026-08-08公司累计融资总额仍未精确披露
最新报道估值~$5,000M2026-08-08来自第三方报道,而非公司公告
阶段隐身、收入前硬件 / 工具公司,增长轮融资规模异常大2026-08-11收入、客户和产品发布日期均未公开
收入 / ARR未公开披露2026-08-11需要管理层披露或一手财务资料
员工数 / 客户未公开披露2026-08-11留存来源未披露员工、客户或装机

这张表有意把硬事实和仅有状态的字段放在一起,因为 Source Foundry 没有发布一家非隐身独角兽通常应披露的标准经营指标。

[CO001, CO002, CO003, CO004, CO008, CO013]
FO002: 公司快照逻辑

公开证据把公司串成三层逻辑:源自 Stanford 的创始团队、宽泛的光刻 / 工具范围,以及围绕芯片制造瓶颈搭起的投资论点。

这是分析逻辑图,不是字面组织架构或供应链图。它概括公开证据如何把创始人履历、商标范围和投资人叙事连起来。

[CO004, CO005, CO008, CO009, CO010, CO011]

1.2 创始人、领导层与治理表面

创始人画像真实,但厚度仍低于后期硬件估值通常需要的水平。公开记录目前只识别出三个高管角色:Obaid 任 CEO 和 CFO,Burg 任秘书。Stanford 来源让两人的学术背景可读。Obaid 的 Bio-X fellowship 以及后续 Stanford 神经科学报道显示,他具备强材料科学、微制造和硅电子能力,并反复强调可扩展工艺开发。Burg 的公开 Stanford 页面不同:重点是有机硅酸盐玻璃和力学性能相关的计算材料工作,而不是光刻工具公司的公开运营经验。这不排除团队资格,但会改变尽调重点。公开记录支持一个技术严肃、扎根 Stanford 的创始二人组;它还没有展示更宽的高管梯队、董事会委员会、运营型招聘或治理机制,无法为数十亿美元级半导体设备扩张降风险。即便最宽松地解读,公开治理披露也仍处于机构化之前。[CO003, CO004, CO005, CO006, CO007, CO032]

领导层与创始人表
人员公开角色已记录背景职能匹配关键人物依赖
Abdulmalik ObaidCEO 兼 CFO;联合创始人Stanford 材料科学 / 微加工研究员;Bio-X 研究员;脑机接口硅工艺研究技术雄心和可规模化工艺框架很强;公开财务 / 运营履历仍未披露非常高,因为公开备案中一人同时担任 CEO 和 CFO
Joe Burg公司秘书;联合创始人Stanford 计算材料与有机硅玻璃力学 PhD 研究员增加了深厚的材料 / 计算可信度;公开光刻工具商业化记录未披露高,因为公开披露的高管班底极薄
更广管理团队留存来源中未公开披露留存来源未具名额外高管、董事或董事会委员会仅凭公开证据无法评估组织制度化程度在更完整的高管和董事会名单公开前,依赖度高

各行只反映留存公开来源可见的角色和背景。治理深度缺失本身就是尽调信号,不是这张表的遗漏。

[CO003, CO004, CO005, CO006, CO007, CO032]

1.3 融资基础、估值与投资人解读

资本故事是 Source Foundry 眼下最重要的理由。审阅到的 2026 年 8 月报道口径一致:Situational Awareness 本周新增 $400M 投资,使该基金总承诺达到 $500M,并支撑约 $5B 估值。公开资料也称 Sequoia 更早支持了公司,但距离完整股权结构描述还很远。这个缺口重要,因为交易条款基本空白:没有审阅来源披露工具是优先股、债务还是某种承诺额度,也没有给出持股、清算优先权或董事会权利细节。因此,这笔融资必须分两层理解。第一,这是对上游 AI 制造命题的一次巨大信任票。第二,单靠公开证据,它仍不是一轮可以完整承做的融资。领投方背景也复杂。Situational Awareness 刚经历公开市场仓位的惨烈出清,上市 AI 基础设施押注失手后,Source Foundry 这张支票既显得大胆,也有几分被迫转向长期私募押注的意味。[CO013, CO014, CO015, CO016, CO017, CO029]

利益相关方或投资者地图
利益相关方角色控制或经济重要性当前证据尽调要求
Situational Awareness披露的主要资本提供方可见经济重要性最高,因为据报已承诺 $500M2026 年 8 月新增 $400M,总承诺 $500M,条款未披露获取融资工具、所有权、董事会权利、拨款计划和清算优先权
Sequoia Capital更早的风险投资支持者重要的信号型投资者,但当前持股未披露公开报道称 Sequoia 支持了 Source Foundry,Stephanie Zhan 阐述了该论点澄清融资轮历史、初始支票规模、按比例跟投行为和治理权利
Abdulmalik Obaid创始人运营者运营控制看起来集中,因为他在公开文件中列为 CEO 和 CFO法律备案列出双重执行和财务角色核验管理团队、授权财务控制和招聘计划
Joe Burg创始人高管可能对技术方向和公司搭建很重要法律备案列出公司秘书角色;公开来源确认其联合创始人身份澄清日常运营角色和技术领域归属
ASML现有龙头竞争参照不是投资者,但其市场主导地位定义了 Source Foundry 的野心报道把 Source Foundry 描述为光刻挑战者理解 Source Foundry 攻击的是光源、扫描仪、工艺流程还是服务栈
其他未披露支持者潜在其他股东可能实质影响稀释和治理,但目前不可见已审阅公开来源未披露完整股权结构表要求完整资本结构表和过往融资轮时间线

这张图把资本提供方、创始人和定义性现有龙头放在一起,因为公开证据仍太稀疏,无法做正常的仅股权结构视图。 因此,经济重要性是方向性的,并非穷尽。

[CO003, CO013, CO014, CO015, CO016, CO017]
FO003: 快照 KPI

少数已披露信号都是资本和范围标记;标准运营 KPI 仍然缺席。

这些是目前最强可见指标,不是均衡运营仪表盘。客户、收入和员工数缺席,本身就是概览的一部分。

[CO003, CO013, CO014, CO022, CO023, CO032]

1.4 公司似乎在造什么

Source Foundry 仍处于隐身状态,商标申请眼下就是最有信息量的准产品文件。它们暗示的不是一台秘密机器,而是更宽的东西。2026 年 3 月的申请覆盖光刻机、半导体制造设备、晶圆加工软件、先进光刻开发工具、技术支持、维护、设施工程、培训,甚至定制制造。换句话说,公开纸面痕迹描述的是横跨机器、软件和制造服务的平台野心。这与 Sequoia 相关内容和 8 月融资报道反复讲述的投资人故事一致:Source Foundry 不是又一家 AI 芯片设计公司,而是一家试图攻击芯片设计上游制造瓶颈的公司。战略背景解释了这套叙事为什么当下有共鸣。ASML 仍把持光刻前沿,Intel 已经把 High-NA EUV 带入生产,Canon 和 Nikon 继续占据有意义的相邻地带,xLight 也拿到联邦支持,试图做替代 EUV 光源。Source Foundry 进入的是一个真实且拥挤的问题集,不是空白市场。[CO008, CO009, CO010, CO011, CO012, CO018]

FO001: 公司里程碑时间线

公开里程碑虽少,但都关键:创始人前期工作、公司注册、商标正式化,以及 2026 年 8 月融资浮出水面。

这条时间线刻意保持稀疏,因为 Source Foundry 仍处隐身状态。时间线只追踪留存记录和报道明确给出的日期,不推断未披露的产品或客户里程碑。

[CO001, CO005, CO006, CO008, CO010, CO013]

1.5 里程碑、证据缺口与未知项

公司概览里的核心分析张力在于,范围野心和融资规模都已公开,运营证据却没有。今天能梳理出的带日期时间线,主要由公司设立、商标申请和 2026 年 8 月媒体曝光构成,而不是产品发布、具名客户、工具装机或量化生产里程碑。对隐身公司来说这并不致命,但会影响承做。没有审阅来源披露 Source Foundry 的制程节点、晶圆尺寸、吞吐量、工具架构、客户名单、收入、员工数或发布日期。即便据称 $5B 估值,也来自第三方报道,而非公司公告。结果是,本章呈现异常高的信息不对称:投资人释放了巨大信心,但外部人仍无法检验性能、经济性或采用情况等核心主张。对后续章节而言,纪律很重要。任何关于产品成熟度、客户牵引或财务准备度的表述,都必须框定为已证明、从记录推断,或仍然未知。[CO016, CO017, CO031, CO032, CO033, CO034]

里程碑表
日期事件类型金额 / 状态参与方含义
2018-01-01Abdulmalik Obaid 获 Stanford Bio-X SIGF 研究员创立创始人背景里程碑Abdulmalik Obaid、Stanford Bio-XObaid 微加工和材料科学履历的最早公开证明
2019-04-17Stanford 公开 Obaid 牵头的硅神经接口研究产品公司成立前的工艺规模化证明Abdulmalik Obaid、Stanford 研究团队显示创始人即使在光刻之外,也有构建可规模化硅基系统的历史
2025-07-18Source Foundry Inc. 在加州备案,成立于 Delaware创立有效存续公司Source Foundry Inc.创建正式法律壳,后续融资和 IP 活动都挂靠其上
2026-03-04首批 SOURCE FOUNDRY 商标申请提交产品机器、晶圆、软件、服务、R&DSource Foundry Inc.、USPTO / Justia 记录首次公开描述产品和服务边界
2026-03-12追加 SOURCE FOUNDRY 商标申请提交治理同一商标家族下的第二波申请Source Foundry Inc.暗示公开亮相前有意规范品牌和范围
2026-06-02美国政府最终敲定 xLight CHIPS 奖励监管$150M 给另一家光刻挑战者NIST、美国商务部、xLight显示政策层面对替代光刻基础设施有兴趣
2026-08-08Situational Awareness 总承诺据报达 $500M,并新增 $400M 注资融资新增 $400M;该基金总额 $500M;报道估值 $5BSituational Awareness、Source Foundry,背景中有 SequoiaSource Foundry 从隐秘的隐身创业公司变成资本充足的登月式项目
2026-08-09TechCrunch 在 Situational Awareness 公开市场减仓后解释这笔投资的语境不利领投方承压TechCrunch、Situational Awareness在未否定核心技术论点的情况下,把财务支持方风险带入公司叙事

时间线受限于公众真正能确定日期的事项。因此,它混合了成立、IP 规范化、生态语境和融资浮出水面, 而不是假装运营历史已经完全可见。

[CO001, CO005, CO006, CO008, CO010, CO012]

1.6 图表

Chapter 02

02市场分析

2.1 市场边界:大于单纯光刻,小于全部半导体资本开支

分析 Source Foundry 时,第一个陷阱是把市场定义得太宽。若只说公司在攻击半导体瓶颈,标题 TAM 就会变成半导体制造设备上的每一美元,甚至每一美元芯片需求。这在方向上有用,但对决策没用。更好的起点,是把公司商标定义出的范围与现有厂商对自身技术栈的描述放在一起。Source Foundry 的公开备案覆盖机器、光刻、晶圆加工软件、工具控制和设施工程。ASML 同样把自身价值主张放在系统、量测、检测和软件上,而不是一台裸扫描仪。因此,市场边界应该包括光刻工具,以及与其紧密绑定的制程控制层。但默认不应包括商品化后道搬运、通用 EDA 或全部半导体资本开支。因此,可触达支出池很大,却仍只是晶圆厂和封装投资总额中的一部分。[CM001, CM002, CM006, CM016, CM017, CM018]

市场定义表
细分 / 类别纳入支出排除支出买方 / 付款方与 Source Foundry 的相关性
先进制程光刻工具EUV、High-NA、先进 DUV、替代图形化系统没有光刻控制差异化的大宗成熟制程工具晶圆代工厂、IDM、先进节点项目核心市场,因为 Source Foundry 明确提到光刻和制造机器
光刻相邻软件与控制晶圆加工软件、工具监控、校准、优化通用企业软件或独立 EDA晶圆厂和设备所有者纳入,因为 Source Foundry 商标提到运行和优化制造工具
晶圆厂工程与部署服务半导体设施定制设计和机器集成通用施工或无关设施工作先进制程晶圆厂和政府支持的建设项目如果 Source Foundry 销售系统级基础设施,这会成为潜在相邻市场
先进封装图形化与 AI 封装复杂性相关的面板、中介层和重布线层曝光大宗封装搬运或组装劳动力OSAT、晶圆代工厂、封装专家值得关注的相邻市场,因为封装已经吸收大量光刻支出
全部半导体设备WFE、测试、封装、设施,作为自上而下语境宏观口径下无排除项全球半导体制造商只适合作为 TAM 上限,不是 Source Foundry 的直接市场
终端市场芯片销售AI 加速器、存储、移动、汽车需求与半导体无关的软件或云收入超大规模云厂商和设备 OEM只是需求驱动,不是直接市场边界

纳入 / 排除线是分析边界,不是公司表述。目的在于避免把 Source Foundry 的市场高估成全部半导体资本开支或全部 AI 收入。

[CM003, CM006, CM010, CM016, CM017, CM018]
FM001: 规模层级视角

自上而下的设备支出口径会大幅高估 Source Foundry 的市场,除非先收窄到光刻,再因产品不清晰进一步打折。

底层仍比真正的 Source Foundry SAM 更宽,因为公司没有披露首个商业产品层。这也是目前公开资料里最接近产品的口径。

[CM001, CM002, CM006, CM018, CM020, CM022]

2.2 规模口径与买家分层

公开证据支持多个市场口径,而不是一个整齐数字。最宽口径是 SEMI 对 2026 年设备市场 $165.9B 的预测。最相关的公开光刻口径,是 Mordor 对 2026 年光刻设备 $30.44B 的估算。在其内部,Source Foundry 真正可服务市场还会继续收窄,指向那些为领先逻辑、存储、封装和制程控制改进买单的买家子集。买家集中度极高。Mordor 称纯晶圆代工厂占光刻设备支出近一半,而 TSMC 披露的 534 个客户显示,一家领先制程制造商如何把广泛终端需求聚合到相对少数的制程开发项目里。Samsung 和 Intel 进一步说明,下一代节点需求集中在少数拥有资本、人员和意愿去验证新工具的组织手中。超大规模云厂商在经济上重要,但它们往往隔着一层,是 AI 需求推动晶圆代工资本开支的间接付款方,而不是直接购买扫描仪。[CM006, CM007, CM009, CM011, CM012, CM013]

TAM / SAM / SOM 或规模测算口径表
口径来源 / 年份地理范围数值方法局限
总半导体设备 TAMSEMI 2026全球2026 年 $165.9B全部半导体制造设备销售预测远宽于 Source Foundry 可能的首个产品范围
晶圆厂设备口径SEMI 2026全球2026 年 $143.9B晶圆加工、掩模 / 光罩和晶圆厂设施设备仍然过宽,因为包含许多非光刻类别
晶圆代工 / 逻辑 WFE 口径SEMI 2026全球2026 年 $78.0B与晶圆代工和逻辑应用相关的子集更接近先进制程买方,但仍不是纯光刻数字
光刻设备口径Mordor 2026全球2026 年 $30.44B具体光刻设备市场估计第三方市场模型,不是经审计行业账本
先进封装光刻相邻口径Mordor 2025 份额全球2025 年光刻设备收入的 36.74%光刻市场内部的应用份额相邻份额,不是独立的 Source Foundry 市场
Source Foundry 近产品 SAM作者综合先进制程逻辑、存储、封装地区公开证据无法解析取决于首个产品层和买方类别公司披露的产品具体性不足,无法精确隔离

这张表有意保留多个口径,因为公开证据不足以为 Source Foundry 给出一个干净的单一市场规模数字。

[CM001, CM002, CM006, CM007, CM010, CM019]
细分 / 买方地图
细分直接买方用户付款方 / 预算所有者工作流采用触发因素
先进制程晶圆代工厂TSMC 级晶圆厂的制程技术和资本设备团队晶圆厂工程师和制程整合团队受 AI 需求影响的公司资本开支委员会节点迁移和良率爬坡需要缩短图形化步骤或提高前沿节点吞吐
集成器件制造商类 Intel 或 Samsung 的器件和代工集团内部产品和制造团队公司资本开支加战略节点路线图将新 CPU、GPU、存储或代工产品推向生产每瓦性能和密度提升足以支撑重新认证
先进封装专家 / OSAT封装厂工程和采购团队RDL、中介层、TSV 和面板线操作方封装和 AI 平台预算为 AI 系统规模化异构集成封装图形化需要更好的套刻、吞吐或成本
政府支持的研究和试点晶圆厂联盟、国家实验室或 CHIPS 支持站点R&D 和原型团队公共补贴和联盟预算原型验证替代光刻或本土工艺能力大规模投产承诺前,需要测试非现有龙头架构
受超大规模云厂商影响的平台项目通过晶圆代工厂、OEM 或系统伙伴间接影响系统架构师间接来自云 / AI 基础设施财务把 AI 需求转化为节点和封装要求需要在每瓦、每机架或每个数据中心建设中产出更多芯片

超大规模云厂商往往是经济驱动方,但直接采购权通常在晶圆代工厂、IDM、封装厂或公共资助的试点项目手中。

[CM009, CM011, CM012, CM013, CM023, CM024]
FM002: 市场估算区间

公开规模估算因口径不同差异很大:全部设备、晶圆厂设备,或仅光刻支出,结果完全不同。

低位是仅光刻口径,基准是代工 / 逻辑晶圆厂设备支出,高位是全部半导体设备。这些口径范围不同,不是统计置信区间。

[CM001, CM002, CM006, CM019, CM020]
FM003: 直接买方与间接付款方地图

购买权在晶圆厂和封装厂手里,但经济拉力常来自超大规模云厂商 AI 预算和先进封装需求。

条目综合公开市场结构,而不是公司指引。图的重点是谁必须先被说服,不是谁最终在经济上受益。

[CM009, CM010, CM011, CM012, CM013, CM023]

2.3 增长驱动与采用约束

市场的多头逻辑很直接。SIA、IDC 和 SEMI 都显示,2026 年的环境里,AI 基础设施正把多年半导体需求增长压缩进更短的投资周期。SEMI 特别把更高支出与先进逻辑、HBM、封装和测试相连;Mordor 又补充了补贴、节点缩小和异构集成。但空头一面同样真实,而且对初创公司更关键。这个市场受残酷切换成本塑形:跨季度导入验证、深度嵌入制程配方、巨大营运资本需求。Mordor 称 High-NA 工具单台价格可超过 $400M,且可能要求交付前一年或更早支付定金。出口管制、ESG 约束和供应商集中又增加摩擦。即便需求强劲,买家也不会轻易塞进一种新光刻架构。他们需要对良率、正常运行时间、服务和晶圆厂兼容性有信心;这也是 Canon、xLight、Nikon、Intel、Samsung 和 ASML 都把产品放在系统层面讲,而不是讲成一次性发明的原因。[CM001, CM003, CM004, CM005, CM014, CM015]

增长驱动与约束表
驱动 / 约束方向时点含义证据
AI 加速器与 HBM 需求正向近期推高先进逻辑、存储、封装和测试资本开支SEMI、SIA、IDC 都指向 AI 拉动的需求加速
制程节点缩小和 High-NA 导入正向短期至中期先进晶圆厂对新图形化能力的需求上升Intel 和 ASML 已在讨论 High-NA 层的生产使用
政府晶圆厂补贴正向中期帮助资助新的本土晶圆厂和试产线,它们可能考虑挑战者SEMI 和 NIST 显示政策支持上游制造工具
替代架构关注升温正向中期为 NIL、FEL 或其他非既有路线留出空间Canon 和 xLight 显示买方与政府愿意资助替代路线
极高资本开支和长周期预付款负向即时让买方认证更保守,也抬高创业公司销售阻力Mordor 引述 High-NA 定价 >$400M,订金周期 12-18 个月
出口管制和地缘政治碎片化负向即时至中期限制可触达地区,并让供应链更复杂Mordor 和 ASML 都描述了出口管制压力
切换成本和稼动率要求负向即时新工具不只要证明物理方案新,还要证明良率、服务和集成既有厂商和买方材料都强调生产就绪度
Source Foundry 产品细节不清负向当前尽调阻断项阻碍清晰测算 SAM/SOM 和采用时间线商标覆盖很宽,但首款产品披露缺失

各行把结构性需求驱动与商业摩擦放在一起,这些摩擦对试图卖进先进晶圆厂的创业公司最相关。

[CM001, CM003, CM005, CM013, CM014, CM025]
FM004: 采用漏斗 / 价值链地图

最陡降幅出现在发现瓶颈之后、让新生产工具通过前沿晶圆厂认证之前。

数值是序数型就绪度评分,不是实际赢单率。评分代表设备资本开支、正常运行时间要求和漫长认证周期带来的商业化摩擦。

[CM013, CM014, CM024, CM028, CM029, CM035]

2.4 对 Source Foundry 的含义

放在一起看,市场毫无疑问足够大,值得关注,但它的大并不会让商业化变容易。相关买家群体很小、技术要求极高,并且已经与现有厂商和可信替代架构供应商深度绑定。这会影响投资判断。Source Foundry 不需要赢得消费级客户数量;它需要说服少数世界级晶圆厂、IDM 或先进封装项目相信,公司能在前沿瓶颈上改善吞吐量、良率、成本或物理极限。公开证据足以说明奖池以十亿美元计,也足以说明 AI 需求正把投资向上游拉。公开证据不足以为 Source Foundry 单独推导干净的 SOM 或时间曲线,因为公司仍未披露首款产品形态到底攻击光源、扫描仪、软件还是晶圆厂服务层。因此,市场章节以一个克制结论收束:机会真实、集中且具有战略重要性,但公开 TAM 叙事仍跑在公开公司细节前面。[CM018, CM019, CM020, CM021, CM022, CM033]

2.5 图表

Chapter 03

03竞争对手

3.1 格局概览:先看现有体系,再看替代方案

最核心的竞争事实是,Source Foundry 试图进入一个默认答案已经可用的市场。ASML 称,其 EUV 系统能打印 7 nm、5 nm 和 3 nm 节点最精细的层,而 DUV 仍需并行用于许多其他层。这意味着现有厂商不是一个孤立盒子,而是一套已安装工作流,横跨扫描仪、软件、服务和生产率调优。ASML 自己的 2025 年年报显示净销售额 €32.7B、售出 535 套系统、供应商基础达 5,100 家,凸显挑战者必须跨过多大的工业规模。公开证据因此提示 Source Foundry 面对三类对手:第一,ASML 装机基础形成的现状;第二,Nikon、Canon、xLight 等替代架构挑战者;第三,大型晶圆厂在更广的代工和封装生态内自己解决瓶颈,而不是引入新的独立供应商。[CP001, CP002, CP003, CP004, CP005, CP026]

FP001: 竞争定位图

商业就绪度偏向在位者,而架构新颖性集中在 Canon、xLight 和 Source Foundry 这类挑战者身上。

X 轴是基于留存证据的序数型商业就绪度;Y 轴是相对已装机 FEOL 现状的序数型架构新颖性。该图是解释性图,不是市场份额图。

[CP004, CP008, CP009, CP012, CP014, CP023]

3.2 直接、相邻与替代竞争者画像

Nikon 和 Canon 是最清晰的公开案例,说明挑战者可以切入更窄楔形市场,而不是直接替代 ASML。Nikon 仍销售半导体光刻、对准和量测系统,其 DSP-100 封装平台展示了进入先进封装的一条路径:无掩模曝光和大面板支持。Canon 走的是另一条路,即纳米压印光刻:其 FPA-1200NZ2C 宣称 14 nm 最小线宽、通向 2 nm 级图形化的路径,以及相比 EUV 更低成本、更低功耗的方法,但公开部署仍集中在 TIE 等研究和原型场景。xLight 攻击的又是另一层,试图用可置于晶圆厂外、为多台扫描仪供能的自由电子激光公用设施替代 EUV 光源本身。EV Group 对前道 EUV 更偏相邻而非直接竞争,但它仍重要,因为其掩模对准机、无掩模工具和封装专长显示,买家注意力可能转向后道和异构集成瓶颈,而不只是停留在前道扫描仪分辨率。[CP006, CP007, CP008, CP009, CP010, CP011]

竞争者画像表
竞争者 / 路线类别规模 / 融资目标细分差异化局限
ASML既有龙头路线2025 年净销售额 €32.7bn;售出 535 台系统;员工 >44,000 人先进逻辑和存储光刻唯一商业化 EUV 供应商,同时有 DUV、软件和服务组合成本高、暴露在出口管制下;若只是稼动率或经济性更好,仍难以取代它
Nikon既有替代 / 邻近路线全球已安装 >8,000 台光刻系统DUV、对准、计量和先进封装半导体光刻栈较宽,并新增 DSP-100 封装路线公开证据看不出真正替代 EUV 的定位
Canon替代架构挑战者上市公司工业级规模;NIL 已向 TIE 出货部分先进节点层、研发和专用应用纳米压印声称在部分层上成本和功耗低于 EUV商业验证仍早,套刻 / 缺陷控制仍卡采用
xLight光源层挑战者$150M CHIPS 激励在 2026 年最终敲定先进晶圆厂的 EUV 光源瓶颈晶圆厂外 FEL 公用设施,可驱动最多 16 台扫描仪仍在原型阶段,不是已在现场验证的扫描仪部署
EV Group邻近封装竞争者私营设备商;此处公开规模未披露先进封装、MEMS、化合物半导体、光子掩模对准机、无掩模曝光、300 mm 就绪封装流程不是最先进 FEOL 层的前道 EUV 替代品
Intel / Samsung / TSMC内部自建 / 替代路径具备 HVM 规模的大型代工和封装生态客户在可信生态内选择节点、封装或协同设计可沿制程和封装路线图解决瓶颈,无需采用创业公司设备商并非每种情形都是直接商用设备替代,但替代力量很强
Source Foundry隐形新进入者据报估值 $5B,最新融资 $400M;Sequoia 支持先进制造中未披露的光刻瓶颈潜在范围较宽,覆盖设备、软件和支持尚无公开产品、客户、定价或就绪度披露

该表把直接竞争者和替代路线放在一起,因为买方可用三种方式处理同一个瓶颈:继续使用既有厂商、采用替代架构,或在代工和封装生态内部解决。

[CP004, CP006, CP008, CP009, CP010, CP012]
定价 / 商业打包比较
供应商 / 路线商业模式公开价格信号包含能力对 Source Foundry 的含义
ASML资本设备销售,加软件、升级和服务此处未保留公开标价;第三方报道称先进工具可能超过 $400MEUV、DUV、软件、支持Source Foundry 必须胜出的不只是物理方案,还包括总拥有成本和服务可信度
Nikon设备销售和支持留存来源未披露价格DUV 扫描仪、封装系统、对准、计量在买方偏好更窄、更低风险导入而非完整架构切换的场景里竞争力较强
Canon设备销售IEEE 引述 EUV 级工具约 $150M,Canon 将 NIL 定位为低成本NIL 系统、工艺创新、原型用途如果 Source Foundry 的卖点是更便宜的先进图形化,而不是完整替代既有厂商,Canon 会构成压力
xLight原型平台和未来公用设施式光源服务定价未披露晶圆厂外 FEL 光源;每套系统最多支持 16 台扫描仪表明子系统或基础设施商业模式无需交付完整扫描仪,也能参与竞争
EV Group设备销售和工艺集成定价未披露封装光刻、无掩模曝光、晶圆轨道集成提高瓶颈预算转向封装而非 FEOL 替代的可能性
Source FoundryUnknownUnknown潜在是设备加软件加支持定价缺失,无法公开推断胜率或毛利率

该表主要比较合同模式,因为留存公开来源很少披露实际设备定价。未知项是证据缺口,不是作者遗漏。

[CP008, CP009, CP011, CP013, CP014, CP024]

3.3 能力对比与替代路径

最重要的竞争对比不只是供应商对供应商,而是路径对路径。Intel、Samsung 和 TSMC 都宣传系统级代工或封装能力,让客户在可信生态内部靠共同开发、封装和制程节点迁移吸收新瓶颈。Intel 18A 已被描述为可用于客户项目,并在美国进入大批量生产;Samsung 则把 EUV、3 nm GAA 和 2.5D/3D 封装服务结合起来。TSMC 的先进封装栈走得更远,围绕 SoIC、CoWoS 和 InFO 提供集成式交钥匙服务,并由 2025 年超过 1,700 万片 12 英寸等效晶圆的制造规模支撑。对 Source Foundry 来说,这意味着真正替代者往往根本不是另一家初创公司,而是客户决定继续把支出留在现有厂商或代工厂对齐的路线图里;那里服务、封装、配方知识和量产制造都已就位。[CP015, CP016, CP017, CP018, CP019, CP024]

功能 / 能力矩阵
供应商 / 路线先进 FEOL 光刻先进封装图形化软件 / 控制 / 系统集成现场服务 / 支持证据公开生产就绪度
Source Foundry公开未知公开未知基于商标范围看可能存在无公开装机基础证据隐形 / 未披露
ASML是:EUV 和 High-NA,加 DUV间接 / 有部分邻近软件和检测是:硬件、软件、服务是:大型全球供应商和服务基础大批量生产
NikonDUV 和浸没式;此处无公开 EUV 主张是:DSP-100 和对准是:光刻、对准、计量是:全球支持网络商业产品线
Canon经由 NIL 的选择性替代路径可能用于专用和原型工作有一定工艺诀窍,但平台比 ASML 窄留存来源中的公开服务细节有限NIL 已商业出货,但部署仍早期
xLight为 EUV 提供光源层支持,不是完整扫描仪无留存证据公用设施式光源架构此处未披露公开现场服务基础原型 / 演示阶段
EV Group无留存证据显示先进 FEOL 领先是:强聚焦封装、MEMS、光子集成式晶圆轨道和计量邻接能力声称有全球支持,但此处未量化邻近领域已商业化
Intel / Samsung / TSMC 路线是,经由制程节点路线图是,经由先进封装产品是,经由代工、封装和共同开发栈是,在现有客户生态内大批量生产或已部署晶圆厂规模

未知单元格是有意保留。公开证据不足以支撑对 Source Foundry 本身做精确的产品级比较,这也是本章主要尽调发现之一。

[CP005, CP006, CP009, CP013, CP014, CP015]
FP002: 功能宽度 / 能力地图

真正重要的比较是 FEOL、封装、控制和服务的覆盖宽度,不只是看谁声称有新的物理突破。

定性标签反映留存公开证据。「未知」表示未找到证据,不代表没有能力。

[CP005, CP008, CP011, CP013, CP014, CP017]

3.4 切换成本、分销权力与监管摩擦

光刻切换成本不只是技术问题,也是组织和地缘政治问题。ASML 和 Nikon 都推广包含图形化工具、软件、量测、对准和服务网络的组合,因此导入验证既关乎正常运行时间和晶圆厂兼容性,也关乎第一性原理物理。Canon 的 NIL 漫长开发周期,以及 IEEE Spectrum 对套刻和缺陷挑战的报道,说明从有趣物理走到可靠生产有多难。出口管制不确定性又叠上一层。Bloomberg 报道称,美国官员警告 ASML,可能有一台先进工具流入中国;CISES 则认为真正瓶颈不是一家公司,而是荷兰、德国、日本和美国供应商组成的联盟。对初创公司来说,这种碎片化会提高可信供应链的价值,也可能让销售周期比标题里的 AI 需求故事更慢、更窄、更受政治条件约束。[CP011, CP020, CP021, CP022, CP030, CP031]

3.5 护城河耐久性与公开证据仍无法证明的事项

公开证据足以说明 Source Foundry 获得资金是为了攻击真实瓶颈,但不足以支撑一条公开可承做的护城河。据报融资和商标申请暗示公司在光刻相关工具、软件、维护和支持上有宽广野心。如果公司打造的是系统级产品,而不是一台孤立机器,这种宽度可能变成优势。但它也意味着竞争集合仍无法清晰锁定:如果 Source Foundry 攻击光源,xLight 更相关;如果攻击完整扫描仪,ASML 和 Nikon 占主导;如果攻击封装或制程控制相邻环节,Canon、EVG、TSMC、Samsung 和 Intel 更重要。因此,投资结论偏保守。Source Foundry 可能具备战略时点和资本,但公开证据尚未显示客户证明、定价杠杆、装机服务能力或生产准备度指标,无法对抗根深蒂固的现有厂商和资金充足的替代路径,提出强护城河主张。[CP023, CP024, CP025, CP034, CP035]

护城河耐久性 / 竞争风险台账
护城河主张威胁严重性为何重要缓释措施 / 尽调问题
新型光刻架构Canon NIL 或 xLight FEL 先解决更窄切口替代架构可能在 Source Foundry 证明更广平台前,先拿下有限生产或试点预算要求说明首款产品定义、目标层和试点晶圆厂路线图
系统级产品宽度ASML 装机基础已把硬件、软件和服务打包在一起关键固化的买方工作流会压低对单点挑战者的兴趣要求提供稼动率、服务模式,以及接入现有晶圆厂工艺配方的证据
隐形状态和新资金带来速度优势Intel、Samsung、TSMC 可沿既有节点和封装路线图吸收需求客户可能在可信生态内花钱,而不是试用新供应商要求具名客户接触、认证阶段和经基准验证的 ROI
宽商标范围暗示软件 + 机器 + 支持护城河范围宽可能只说明雄心,不代表产品已出货未披露首个切口时,竞争类别仍不稳定要求 SKU 架构、BOM 边界,以及 MVP 实际包含什么
地缘政治紧迫性催生替代需求出口管制和供应链联盟会拖慢认证和地域扩张政治支持可帮助试点,但也会让采购和销售更复杂要求供应商地图、出口管制依赖和可信伙伴名单
封装邻接扩大 TAMEVG、Nikon、TSMC、Samsung 已掌握封装关系如果瓶颈转向异构集成,既有厂商可能已经控制预算负责人要求封装专项主张、合作伙伴策略,以及相对 CoWoS / 3DFabric 生态的差异化

严重性仅用公开证据评估竞争承保风险。每行都落到具体尽调问题,因为 Source Foundry 自身公开数据仍很稀疏。

[CP018, CP020, CP021, CP025, CP026, CP029]
FP003: 护城河就绪度 KPI

公开 KPI 凸显 Source Foundry 仍需跨越多大的在位者规模和生态成熟度。

最后一项 KPI 是基于缺少公开证据的尽调发现,而不是公司披露指标。

[CP004, CP008, CP019, CP023, CP025, CP034]
Chapter 04

04财务

4.1 收入模式与核心披露缺口

公开报道给出了 Source Foundry 的估值和融资标题,但几乎没有商业细节。Benzinga 称,公司从 Situational Awareness 新募 $400M,使该基金总承诺达到 $500M,对应据称 $5B 估值,并把公司描述为正在开发光刻工具。商标申请进一步拓宽了图景:它们覆盖晶圆加工软件、机器安装、维护、维修和技术支持服务。这意味着可能的收入模式比一次性硬件销售更宽。合理收入流包括工具销售、试点或非经常性工程费、软件或控制许可,以及装机后服务合同。问题在于,这些收入流没有任何一项获得公开量化。留存资料没有披露收入、订单、积压订单、付费试点、平均售价或客户数。因此,财务分析从缺席出发:公开证据足以勾勒变现路径,但不足以确认今天是否已有任何路径启动。[CI001, CI002, CI003, CI004, CI005, CI023]

收入来源表
来源机制当前公开状态质量尽调问题
工具或子系统销售资本设备合同未披露潜在高客单价,但波动大、客户集中度高要求首款产品、ASP、试点到量产转化假设
试点或 NRE 收入付费原型、认证或集成工作未披露是有用验证点,但不是耐久经常性收入要求任何 LOI、试点合同或工程服务发票
软件或控制许可证晶圆加工或优化软件从商标范围看可能存在,但未获商业确认若可与硬件拆分,可能改善利润率结构要求 SKU 边界和软件附加假设
安装和维护现场服务、维修和技术支持从商标范围看可能存在,但未获商业确认可经常化,但依赖装机基础要求服务模式、备件计划和人员配置假设
政府或联盟项目原型或研究合同未保留 Source Foundry 相关证明可在没有完整量产销售前资助验证要求资助管线、试点晶圆厂伙伴和任何非稀释性奖励
当前报道收入已确认客户收入未发现公开披露无法验证要求经审计或管理口径收入桥

该表把可能的变现路径与其已经启动的证据分开。Source Foundry 的公开商标范围暗示多条收入来源,但留存来源没有量化任何当前收入。

[CI003, CI004, CI005, CI023, CI024, CI025]
定价 / 变现表
路线价格 / 单位 / 合同标价与实际来源支持信号含义
Source Foundry 工具销售Unknown无公开标价或实际成交价未保留公开定价披露公开估值目前还无法和合同经济性挂钩
Source Foundry 软件 / 控制Unknown无公开标价或实际成交价商标范围仅暗示可能性可能影响利润率,但尚无法建模
Source Foundry 服务 / 支持Unknown无公开标价或实际成交价商标范围仅暗示可能性只有形成装机基础后,才可能产生经常性收入
Canon NIL 路线设备定价未披露未作为实际成交价公开留存IEEE 将 EUV 既有工具描述为约 $150M,Canon 理论上成本更低说明买方比较的不只是分辨率,也包括资本开支和能耗
xLight 公用设施式光源未披露未留存实际成交价公司称采用晶圆厂外公用设施式供能模式替代变现方式可能是基础设施服务,而非完整设备销售
ASML 既有厂商模式企业合同加软件和服务此处未留存实际成交价季报和年报披露的是规模,不是单台标价如果成功,Source Foundry 可能也进入类似议价型企业合同

留存来源中的定价大多未知。可靠公开证据指向合同结构和资本开支强度,而不是发布的 ASP 或折扣。

[CI006, CI007, CI008, CI021, CI022, CI025]
FI001: 收入模型桥

Source Foundry 可能的收入路径从原型和认证工作走向设备、软件和服务合同,但公开来源尚未显示商业化从哪里开始。

这座桥的证据来自商标范围和同行合同结构,不来自 Source Foundry 已公开的客户披露。

[CI003, CI004, CI023, CI024, CI025]

4.2 成本结构与单位经济代理指标

Source Foundry 没有披露成本数据,因此最好的公开代理来自成熟半导体设备和代工运营商。ASML 2025 年年报显示销售额 €32.7B、毛利率 52.8%、年度 R&D €4.7B;2026 年 Q1 和 Q2 各自约有 €1.2B 至 €1.3B 的 R&D,毛利率为 53% 至 54%。这些数字并不意味着 Source Foundry 很快能接近类似利润率;它们说明,即便主导厂商已经达到规模,持续工程投入仍然巨大。Intel 2025 年 10-K 在风险结构上更直接相关,强调对 R&D 和制造设施进行重大、长期且内生高风险的投资,下一代节点依赖客户赢单,并暴露于供应、贸易和激励政策。对 Source Foundry 来说,近期经济性很可能为负或无法判断:原型开发、供应商验证、安装、现场支持,以及可能的定制晶圆厂集成,可能会在经常性收入可见之前占据主导。[CI006, CI007, CI008, CI009, CI010, CI011]

单位经济性表
指标数值置信度为何重要尽调问题
当前收入没有收入,就无法分析收入质量要求月度和年度收入桥
当前阶段毛利率原型阶段硬件公司在规模化前毛利可能深度为负要求原型 BOM、服务负担和分阶段预期毛利率
成熟供应商毛利率代理ASML:FY2025 52.8%;2026 年 Q1 53.0%;2026 年 Q2 54.0%提供规模阶段基准,而不是创业公司预测要求 Source Foundry 目标稳态毛利率及达成路径
研发强度代理指标ASML FY2025 研发 €4.7bn;2026 Q1 €1.2bn;2026 Q2 €1.3bn即使做到规模化,先进光刻的经济模型仍然重研发投入索取年度工程预算、外部晶圆厂支出,以及按职能拆分的现金消耗
装机基础服务杠杆服务利润率取决于机队规模和正常运行时间承诺索取预测装机量、现场服务人员配置和备件假设
客户集中度少数几家晶圆厂可能贡献几乎全部收入,波动性随之放大索取按客户类别拆分的管线和单一客户敞口上限
营运资本 / 资本开支负担存货、工装、供应商承诺和安装可能在收入确认前吃掉现金索取资本开支计划、供应商预付款和验收里程碑

Source Foundry 未公开披露之处,本表故意保留 null。这里唯一保留的具体经济数字,是 ASML 规模化阶段的代理指标,不应误读为创业公司利润率。

[CI006, CI007, CI008, CI009, CI010, CI026]
FI002: 资本到利润率桥

可持续毛利率路径首先取决于资本能否撑过认证阶段;该图先强调融资依赖,再看规模经济。

这是定性桥,因为 Source Foundry 未披露 BOM、利润率或现金消耗。唯一数字锚点是最后节点中的成熟厂商代理指标。

[CI006, CI007, CI008, CI026, CI027, CI028]

4.3 资本充足性与融资依赖

Source Foundry 据报融资规模可观,但资本是否充足取决于需要资助什么。如果公司要打造一家深度嵌入前沿晶圆厂的资本设备和服务业务,$500M 可能足以支持数年原型、招聘、伙伴验证和早期现场试验,但相对于先进光刻隐含的工业规模,仍可能显得有限。公开可比对象强化了这一点。TSMC 的 2 nm 和先进封装路线图已经处于量产或交钥匙部署阶段,Intel 的文件也把下一代制造描述为依赖大额资本开支、高风险 R&D 和外部客户里程碑。xLight 获得 $150M CHIPS 奖励也显示,相邻光刻挑战者商业化前往往需要公共资本支持。融资集中度又增加一层变量。TechCrunch 和 Reuters 关于 Situational Awareness 的报道描述了剧烈回撤、公开股票出清和大幅缩水的剩余资产基础,尽管该基金保留了私募投资。这并不否定 Source Foundry 已有资金,但如果商业化比计划更久,它会带来一个合理尽调问题:未来跟投能力和赞助方稳定性是否足够。[CI012, CI013, CI014, CI015, CI016, CI017]

资本充足性表
指标公开数值 / 状态置信度重要性尽调要求
最近承诺资本总额来自 Situational Awareness 的总额 $500M,其中包括 $400M 新一轮融资界定当前公开融资基数索取实际到账现金、到账时间,以及任何分期或条件
账上现金没有现金余额,就算不出资金续航索取最新资产负债表和银行层面的现金
月度现金消耗没有月度现金消耗,就无法判断资本是否够用索取当前月度净现金消耗和计划爬坡
资金续航月数估值和下一轮风险取决于资金续航索取含招聘和资本开支假设的基准与下行情景资金续航
计划资金用途原型、招聘、供应商和验证工作可推断;公开信息未细化必须知道当前资金要买下哪些里程碑索取按类别和里程碑拆分的资金用途
下一轮触发条件未披露决定融资依赖度索取可触发下一轮融资或项目融资的里程碑
非稀释性支持保留资料中未见 Source Foundry 获得此类支持;相邻可比公司 xLight 获得 $150M CHIPS 激励在这个行业,公共支持可能显著拉长资金续航索取赠款策略、补贴和债务额度
出资方稳定性领投方在 2026 年 7 月经历公开市场回撤和投资组合出清如果商业化延后,未来跟投资金能力可能变得关键索取出资调用确定性和备选融资方案

公开融资标题很清楚;但支撑资金续航的几乎每个输入项都不清楚。邻近可比来源说明,在前沿制造业务里,这类缺口为什么关键。

[CI001, CI015, CI016, CI017, CI020, CI028]
FI004: 资本强度 / 现金流地图

现有融资可能足以跨过几个技术关口,但商业化时间和资助方稳定性仍是关键外部变量。

该图是定性的,因为 Source Foundry 未披露现金、消耗或募资用途细节。资助方稳定性风险来自 2026 年 7 月关于 Situational Awareness 回撤和退出的报道。

[CI001, CI015, CI016, CI020, CI029, CI030]

4.4 财务结论、收入质量与剩余缺口

财务结论刻意收窄。公开证据不支持围绕收入质量建立承做案例,因为没有公开收入可评估。它支持的,是围绕资本强度的判断。商标范围、竞争者变现模式和公开代工披露都指向一家公司:它更可能像企业设备、平台和服务经济,而不是 SaaS 经济。这意味着长导入验证周期、集中买家、合同不均匀和沉重固定成本基础。因此,据称 $5B 估值主要押在技术杠杆和未来市场重要性上,而不是已披露财务牵引。若要从叙事走向可承做模型,尽调需要现金余额、月度烧钱、原型和现场试验支出、供应商承诺、定价假设、客户管线,以及首个产品的确切边界。在此之前,Source Foundry 应被视为一个收入前、依赖融资的前沿制造押注,而不是已经商业证明的半导体工具公司。[CI022, CI024, CI025, CI026, CI027, CI032]

公开财务缺口表
缺失的未公开指标影响当前公开替代依据具体尽调路径
已确认收入和订单积压卡住任何收入质量或估值倍数分析None索取月度收入、订单积压和已签客户承诺
现金余额和现金消耗卡住资金续航分析只有融资新闻口径索取最新现金余额、过去 12 个月现金消耗历史和董事会预算
原型和验证支出卡住资本开支和里程碑投资判断只有 Intel 和 ASML 的可比风险披露索取按工程、供应商工装、晶圆厂试跑和现场试验拆分的支出
定价和毛利率假设卡住单位经济和 TAM 到收入的转换只有可比结构索取 ASP、服务附加率、支持负担和毛利率瀑布
客户管线和集中度卡住预测和下行情景未保留到公开客户名称索取按阶段拆分的 CRM 管线和集中度上限
融资条款和或有条件卡住稀释和出资方风险分析只有融资轮次新闻索取融资条款清单中的经济安排、分期、权利和备选融资方案

这些不是泛泛的愿望清单。每个缺失指标都直接卡住对一家 $5 billion 前沿制造公司的具体承销步骤。

[CI005, CI024, CI025, CI029, CI032, CI035]
FI003: 财务估算区间

最干净的公开财务区间不是 Source Foundry 预测,而是 ASML 披露的 FY2025、2026 年 Q1 和 2026 年 Q2 毛利率构成的规模阶段代理指标。

该区间不应被解读为 Source Foundry 预测。该区间是公开基准,展示一旦形成规模和装机基础杠杆,成熟先进光刻经济性可能是什么样。

[CI006, CI007, CI008, CI027]
Chapter 05

05产品与技术

5.1 产品定义:范围很宽,已验证细节很窄

最具体的公开产品证据来自商标覆盖范围,而不是公司产品页。USPTO.report 和 Justia 对 Source Foundry 商标的记录显示,公司意图覆盖半导体、芯片、晶圆、晶圆加工软件、先进半导体光刻软件、光刻掩模和光罩,以及半导体制造机器安装或维护服务。这个组合很重要,因为它把 Source Foundry 放进了客户工作流:买家接触公司,可能不只是为了买一台机器,也可能为了制程软件、消耗品或使能工件,以及装机后支持。不过,公开证据到此为止。留存资料没有给出商用 SKU 名称,没有公开架构图,也没有数据表说明公司首个产品到底是光源、扫描仪、软件层还是封装相邻工具。因此,产品定义在范围上部分可知,但具体产品形态不可知。这种缺席应该改变尽调行为。在企业级半导体设备里,即便更深的性能数据保密,买家通常至少能看到一组狭窄公开证明点——产品名称、架构家族、试点引用或应用说明。这里正相反:Source Foundry 展示了异常宽的法律范围,却几乎没有公开产品细节。因此,本章把商标范围视为公司希望获准销售内容的下限,而不是证明每一类都已经工程化、集成或商业就绪。[CE001, CE002, CE003, CE004, CE005, CE021]

产品模块 / 资产矩阵
模块 / 资产可能用户状态 / 成熟度差异化尽调缺口
晶圆加工软件工艺工程师仅见商标范围可能改善工艺控制或优化经济性没有公开产品名、UI、部署模式或指标
先进光刻软件光刻和集成团队仅见商标范围可能瞄准控制层瓶颈,而不只押注硬件没有公开工作流或功能集
半导体 / 芯片 / 晶圆晶圆厂、试点线或合作伙伴仅见商标范围暗示其硬件或工艺资产野心不止于软件未披露具名产品或样品
光刻掩模和光罩工艺和掩模团队仅见商标范围暗示公司会触达赋能型制程载体,不只是主设备没有公开掩模策略或商业模式
安装 / 维护 / 维修服务晶圆厂运营和现场支持负责人仅见商标范围暗示公司有装机基础或部署野心没有公开现场服务覆盖或人员模型
技术支持服务客户工程团队仅见商标范围可能带来安装后的经常性收入和信任没有公开 SLA、支持模式或服务地点

每一行都来自商标范围,而不是保留到的公开产品页。因此,每一行都写明了具体尽调缺口。

[CE001, CE003, CE004, CE021, CE022, CE023]
工作流 / 用例表
用户任务当前工作流Source Foundry 方案可衡量收益限制
开发先进光刻工艺使用既有工具,叠加内部工艺工程能力Source Foundry 可能提供光刻软件和 / 或硬件如果工具切实提高吞吐或图形化能力,瓶颈可能减轻没有公开证据说明产品实际如何运作
优化晶圆加工在晶圆厂工具内调工艺配方可能用于晶圆和芯片加工的软件可能提升良率或缩短周期时间没有公开基准测试或案例研究
验证新的图形化路线在研究线或晶圆厂验证线上试点可能插入原型或子系统可能打开成本更低或架构不同的路径未公开具名试点客户
安装和维护设备依赖既有服务组织Source Foundry 可能提供现场服务和技术支持支持做到位,停机时间可能缩短未披露公开支持网络
管理掩模或光罩相关工作流既有掩模生态和供应商关系可能覆盖掩模 / 光罩相关产品如果与工具紧密配套,集成可能改善未披露公开制程载体策略

本表把 Source Foundry 的公开范围映射到客户工作流,而不是猜测隐藏的机器架构;这种写法更站得住。

[CE003, CE004, CE021, CE023, CE027]
FE001: 产品架构地图

公开产品边界似乎覆盖软件、物理图形化资产和安装后服务,但细节不足以给首款产品精确定名。

这是由公开商标覆盖范围推导的范围栈,不是 Source Foundry 披露的产品 BOM。

[CE001, CE003, CE004, CE022, CE023]
FE002: 客户工作流 / 运营流程

任何 Source Foundry 产品都必须嵌入企业级晶圆厂的认证、安装、优化和支持流程,而不是简单自助式销售动作。

运营流程由商标覆盖的服务范围,以及公开可比公司如何定位安装和工艺集成推断而来。

[CE003, CE012, CE021, CE027, CE034]

5.2 架构与 Source Foundry 要竞争必须面对的技术栈

Source Foundry 的确切设计没有披露,但公开可比证据说明,任何可信产品都必须处理哪些层。ASML 技术页面显示,领先制程光刻是光学、机械、控制软件和服务的混合体,EUV 和 DUV 在不同层中配合使用。Nikon 的技术栈同样横跨光刻、对准、量测和封装,其 DSP-100 又为大尺寸基板加入无掩模投影。Canon 的 NIL 路径展示了另一种架构,用压印替代光学投影,但它经历了多年开发,仍需要通过研究和原型渠道商业化。xLight 攻击的是光源层,而非完整扫描仪;TSMC 的 3DFabric 和先进封装材料则显示,部分关键产品价值正从前道光刻迁移到系统级集成。关键产品结论是,除非某个单一组件能解开更大的系统瓶颈,Source Foundry 不太可能只在一个孤立部件上竞争。这也是为什么产品尽调不能止步于“ASML 替代者”这个标题。Canon 证明,即便方法完全不同,仍需要围绕缺陷控制、对准、试点采用和制造准备度建立完整计划。xLight 说明,初创公司可以选择楔入一个关键子系统,而不是替代整机。TSMC 和 Intel 表明,部分客户价值可能通过集成或封装层被捕获,而不是直接赢下前道扫描仪之战。Source Foundry 合理上可能追求这些形态中的任何一种,但公开证据尚无法区分。[CE006, CE007, CE008, CE009, CE010, CE011]

技术 / 运营架构表
层级 / 组件作用依赖风险
光学或图形化核心生成或转移图形化机制与投影、纳米压印或光源层替代方案竞争公开证据没有说明 Source Foundry 攻的是哪一层
控制软件协调工艺调校、校准和优化需要深度接入晶圆厂,并让工作流接受没有公开技术栈、API 或界面证据
掩模 / 光罩或同等制程载体支撑或塑造图形转移需要工艺兼容性和供应链精度未披露公开制程载体策略
安装和维护从原型走向可靠的晶圆厂运行需要现场工程师、备件和正常运行时间流程没有公开装机基础支持覆盖面
封装 / 系统集成邻近领域如果瓶颈移出 FEOL,仍可捕捉价值依赖先进封装生态和合作伙伴入口可能是替代市场,而不是直接产品范围
供应商和出口管制输入提供组件、设施或关键技术依赖受半导体设备供应链和政策约束监管和本地化压力可能重塑可获得性

由于 Source Foundry 的具体产品仍未披露,本架构按层级拆解。可比公司和政策来源界定相关层级,但不假装知道隐藏设计。

[CE009, CE010, CE012, CE014, CE015, CE018]
FE003: 关键依赖地图

隐藏产品可能不同,但公开依赖集合清楚包括晶圆厂、工艺制品、服务能力和半导体设备供应约束。

依赖按类别层面描述,因为没有可用的 Source Foundry 公开 BOM 或供应商名单。

[CE015, CE018, CE023, CE028, CE029, CE033]

5.3 成熟度、路线图与信任 / 质量控制

公开成熟度信号很弱。商标记录仍处于意图使用状态,公司也没有发布留存官方页面来提供产品文档、发布说明或认证细节。这与现有厂商和相邻挑战者形成鲜明区别,后者至少会发布架构主张、试点去向或技术里程碑。可以公开验证的路线图基于法律和融资:2026 年 3 月商标活动、2026 年 8 月融资报道,以及围绕光刻瓶颈的宽泛创始人或论点叙事。这些都不等同于已验证产品里程碑。信任和质量控制同样不透明。没有留存资料显示 ISO 式制造认证、公开安全声明、正常运行时间或良率指标、状态页、安全控制或正式合规界面。光刻采用高度受可靠性、维护和制程集成制约,这些控制缺席不是文档瑕疵,而是核心产品尽调缺口。创始人可信度信号略微改善了这幅图景。与 Abdulmalik Obaid 相关的 Stanford 和 Science Advances 材料显示,他反复研究复杂硅相关制造和芯片集成问题的扩展,这与半导体工具开发所需思维相关。但这些来源仍只是代理。它们没有展示 Source Foundry 专属专利、制程窗口、套刻数字、吞吐量、正常运行时间、污染控制或已发布的发布节奏。因此,产品案例在技术野心上强于在公开落地证据上。[CE018, CE019, CE020, CE024, CE025, CE026]

信任 / 质量 / 合规表
控制项 / 指标状态范围缺口
公开质量认证保留证据中未找到公司层面的质量和制造主张若卖进晶圆厂,需要 ISO 或同等文件
可靠性 / 正常运行时间指标保留证据中未找到工具或服务表现需要 MTBF、正常运行时间、服务响应或缺陷率数据
安全 / 软件信任面保留证据中未找到任何软件或控制层需要安全架构、访问控制和更新政策
安全 / 合规主张保留证据中未找到设备部署和运营需要晶圆厂安全和监管处理细节
状态 / 事故沟通保留证据中未找到客户运营支持需要事故响应或状态沟通界面
支持组织证明仅有商标意图现场支持、维修和技术支持需要组织覆盖、人员配置和 SLA 证据

这些缺口尤其重要,因为半导体买家看重信任、正常运行时间和工艺兼容性,不亚于理论上的技术优越性。

[CE003, CE024, CE030, CE031, CE034, CE035]
路线图 / 发布 / 开发阶段表
日期 / 阶段里程碑状态含义来源
2014Canon 收购 Molecular Imprints已完成说明替代光刻路线商业化可能拖得很久Canon 2014 年收购新闻稿
2023-10Canon 发布 FPA-1200NZ2C NIL已完成可比案例显示,一条公开替代光刻路线已经进入产品形态Canon 2023 年发布新闻稿
2024-09Canon 向 TIE 出货 NIL已完成可比案例显示,更广泛生产采用之前,会先有原型和研究部署Canon 2024 年出货新闻稿
2026-03Source Foundry 商标已提交并处于有效状态,申请类型为意向使用已完成法律步骤显示范围很宽,但不代表商业可用USPTO.report 和 Justia 记录
2026-08Source Foundry 融资被报道为 $5B 估值已完成融资事件资本先于公开产品披露到位Benzinga 报道
当前公开产品成熟度具名产品、发布说明、公开文档未找到公司对已出货产品成熟度仍然公开不透明保留的公司特定来源中均未出现

Source Foundry 唯一能公开验证的路线图事项,是法律和融资里程碑。本表有意加入可比公司里程碑,用来展示该领域公开产品成熟度通常是什么样。

[CE006, CE007, CE025, CE026, CE031, CE035]
FE004: 产品成熟度 / 能力地图

公开范围横跨软件和支持,但每个模块的公开成熟度证据都很薄。

空白或「否」单元格表示缺少现存公开证据,不等于证明这家私营公司没有该项。

[CE003, CE004, CE005, CE024, CE030, CE035]
Chapter 06

06客户

6.1 买家分层与采购逻辑

Source Foundry 几乎肯定没有广泛的 SMB 或开发者式客户基础。商标记录里可见的产品范围——先进光刻软件、晶圆加工软件、掩模和光罩、半导体制造支持与现场服务——指向的是一小群企业买家,他们有深厚制程工程团队和漫长导入验证周期。最可能的一线买家是领先制程晶圆代工厂和大型 IDM,它们已经管理先进逻辑、封装和制程开发项目。第二类是 TIE 这样的试验线和研究基础设施买家,说明新制造工具如何先通过原型、开放访问或国家能力路径进入市场,再走向主流生产。第三类是生态或渠道买家:晶圆代工联盟、封装供应商、设计服务伙伴或制程使能项目,它们能帮助新工具获得导入验证,而不要求立刻承诺全厂队列生产。每个细分里,用户、买家和付款方都不是同一群人。制程工程师和设备团队可能评估技术,但预算权往往在晶圆厂领导层、平台负责人或政府支持的联盟领导层手中。这形成了缓慢、高摩擦的采购模式,但每个账户的战略价值很大。因此,单个合格账户的商业上行空间可能异常大。[CU001, CU002, CU003, CU004, CU005, CU006]

客户分群表
分群买方 / 用户 / 付费方用例规模收入 / 战略价值缺口
先进制程晶圆代工厂和 IDM晶圆厂管理层 / 工艺工程师 / 资本预算验证新的光刻或晶圆加工能力账户数极少,ACV 极高潜在锚定收入和可信度没有公开具名 Source Foundry 客户
试点线和研究机构联盟管理层 / 研究人员 / 公私资金原型生产、工艺评估、培训数量少,信号价值高可在主流机队采用前验证产品没有公开具名 Source Foundry 机构或试点线
先进封装和异构集成项目封装平台负责人 / 集成工程师 / 项目预算如果瓶颈移出 FEOL,可走替代或邻近路线选择性强,按项目推进即使不替代前道扫描仪,也能打出切入口精确产品边界未知
晶圆代工生态合作伙伴联盟经理 / 设计赋能 / 合作伙伴预算设计、封装、掩模或集成赋能合作伙伴数量中等加快验证和信任建立Source Foundry 未公开披露生态信息
政府支持的半导体项目国家或州级项目经理 / 技术运营方 / 公共资金本土能力建设和原型基础设施账户少,但具战略性可补贴早期安装,并证明本地供应链没有公开具名 Source Foundry 项目
设备和服务渠道合作伙伴OEM 或服务管理层 / 现场团队 / 联合业务预算安装、维护、维修和支持选择性可能带来重复性服务收入和更广支持覆盖没有公开渠道或现场服务覆盖

由于保留的公开来源都没有披露 Source Foundry 的现有客户名单,本分群表强调买方逻辑,而不是当前数量。

[CU001, CU002, CU003, CU005, CU006, CU007]
采购摩擦与渠道依赖表
摩擦点为何重要公开代理证据对 Source Foundry 的含义
漫长设计和认证周期客户需要工艺匹配,也需要首硅信心TSMC OIP 和大学项目强调设计赋能和首次成功Source Foundry 可能也需要类似赋能界面
掩模和制造工作流集成一旦牵涉掩模和工艺窗口,客户买的就不只是一台设备TSMC 掩模服务和 GF 制造服务商业成功可能取决于能否掌控更宽的工作流
先进封装生态协同前道和后道路径越来越重叠Samsung SAFE MDI/OSAT 项目、Intel Foundry 封装、Rapidus 前道加后道卖点买方图谱可能超出传统光刻负责人
IP 安全和可信制造半导体客户要求可信工艺处理SkyWater TaaS 使命和认证创业公司必须证明可信,不只是技术新颖
本地化和政策约束出口管制和本土能力政治会影响买方决策CSIS 出口管制分析和 TIE 州级加联邦资金客户时点可能同样由政策和产品驱动
现场支持和维护覆盖停机风险让服务能力成为产品的一部分Source Foundry 服务商标范围,以及在位厂商支持网络尚无公开 Source Foundry 现场运营证据

额外纳入本表,是因为半导体设备基础设施的客户采用离不开采购摩擦和生态赋能。

[CU004, CU007, CU017, CU019, CU020, CU024]
FU001: 客户旅程图

Source Foundry 最可能的客户旅程,是企业从识别瓶颈出发,先走资质验证路径拿到试点证明,再进入生产和服务扩张。

这张图反映商标范围和可比赋能项目隐含的买方逻辑,不代表 Source Foundry 已披露的销售漏斗。

[CU001, CU003, CU004, CU006, CU022, CU027]

6.2 公开采用证据与具名客户缺口

核心商业事实是负面的:没有留存公开资料识别出 Source Foundry 的具名客户、部署地点、评估群体或生产结果。融资报道把公司描述为隐身且面向未来,但没有补充客户证据。这很关键,因为相邻半导体设备挑战者一旦拥有客户证据,通常会发布某种形式的证明。Canon 公开点名 TIE 为其 NIL 系统去向,并把确切用例描述为先进半导体 R&D 和原型生产。TIE 自身描述了一种聚焦先进微系统的纯晶圆代工厂和培训中心模式,指示了能吸收规模化前制造基础设施的试点或机构买家类型。晶圆代工生态运营者也会发布客户接触界面:Samsung SAFE、TSMC OIP、Intel Foundry Accelerator、GlobalFoundries 制造服务、UMC 技术产品、Rapidus 试验线服务和 SkyWater 的 TaaS 模式都说明,这一类别的客户采用往往始于生态使能,而不是简单硬件转售。Source Foundry 尚未披露任何等价公开证明,因此即便产品野心可信,商业成熟度仍未验证。[CU010, CU011, CU012, CU013, CU014, CU015]

客户增长 / 采用轨迹表
指标或阶段数值日期来源置信度含义缺失分母
Source Foundry 具名量产客户2026-08-11隐身期融资报道加商标语料没有公开量产使用证据量产客户实际数量
Source Foundry 具名试点客户2026-08-11隐身期融资报道加商标语料没有公开试点证据评估或试点账户实际数量
Source Foundry 具名生态伙伴2026-08-11留存公开语料没有公开联盟或伙伴披露伙伴数量和角色
相邻 NIL 试点证据Canon NIL 系统落地对象为 TIE2024-10-01Canon USA / TIE展示真实试点阶段公开证据点应有的样子商业采购条款和复购行为
晶圆代工客户赋能模式公开生态体系和服务门户2026-01-01Intel / Samsung / TSMC / GF / SkyWater / UMC / Rapidus 等晶圆厂该市场客户通常通过漫长赋能流程采用这些项目的实际转化率
公开留存指标2026-08-11留存公开语料公开信息无法支撑耐久性判断续约、GRR、NRR、流失

采用轨迹主要由明确空白项主导,因为 Source Foundry 尚未披露能把商业兴趣转化为牵引力证据的分母。

[CU010, CU011, CU012, CU014, CU015, CU019]
具名客户证据表
客户 / 证据点细分市场部署 / 使用场景量产还是试点结果限制
未具名 — Source Foundry 量产客户终端买方没有识别到公开部署未确认留存公开来源没有证明量产使用公开证据缺失就是核心发现
未具名 — Source Foundry 试点或评估客户终端买方没有识别到公开试点现场未确认留存公开来源没有证明试点使用仍可能在 NDA 下私下存在
Texas Institute for Electronics 研究机构研究 / 试验线买方原型Canon NIL 系统用于先进半导体研发和原型生产试点 / 原型基础设施展示相邻光刻挑战者如何公开给出具名证据这是对照证据,不是 Source Foundry 客户
TSMC University FinFET Program 机构高校或研究设计群体为硅测试芯片提供 MPW 服务和设计资料研究和原型路径展示机构客户如何通过结构化赋能项目进入项目参与方不是 Source Foundry 客户
TIE 生态背书方和产业支持方战略赞助网络为纯晶圆代工能力提供资金和产业支持基础设施建设说明大规模商业出货前,机构买方或赞助方也可能关键支持方未披露为 Source Foundry 买方

本表有意把两行 Source Foundry 空白项与相邻基准行放在一起,方便读者对照:缺失证据是什么样,正常公开试点证据通常包含什么。

[CU010, CU011, CU013, CU014, CU016, CU017]
FU002: 采用 / 部署漏斗

理论上漏斗很宽,但很快收窄,因为只有少数组织有能力评估或购买这类基础设施。

数值是定性漏斗权重,不是 Source Foundry 披露的数量。它们展示的是集中度逻辑,不是实际管线数字。

[CU002, CU005, CU028, CU029, CU030, CU031]
FU003: 客户证明矩阵

相邻市场展示了基准水平的客户证明质量,而 Source Foundry 自身在具名部署证据上仍为空白。

“项目层面”单元格表示公开来源命名的是客户赋能载体,而非某个具体终端客户账户。

[CU010, CU013, CU014, CU016, CU017, CU018]

6.3 留存、扩张与集中风险

没有公开客户名单,传统留存分析就会坍缩为代理和明确空白。Source Foundry 没有公开 NRR、GRR、流失、合同期限、续约或满意度数据。最多只能说,相邻品类动态暗示,一旦工具或工作流完成导入验证,留存会很黏,因为晶圆厂制程变更和现场服务关系在运营上很难撤回。问题是,在采用得到证明前,不能提前给耐久性记功。扩张逻辑比留存逻辑更容易勾勒。一个主要晶圆代工厂、试验线或国家半导体项目完成单一锚点验证,可能带来跨制程模块、封装、支持服务或相邻晶圆厂的后续需求。但同一结构也制造严重集中风险:现实的近期账户宇宙很小,每个账户技术要求都很高,一个导入验证延迟就能让收入预期移动数年。公开可比生态也强调同一点。它们的设计目标是降低设计门槛、缩短上市时间,并帮助客户更快达到首片硅或原型状态。如果 Source Foundry 缺少类似生态渗透,客户扩张被卡住的原因可能不是理论需求,而是导入验证吞吐量和信任。在公司披露哪怕狭窄参考客户集之前,投资人应假设集中度会极端高,耐久性大体不可知。这种不透明尤其重要,因为目前没有公开迹象显示客户引用已按地域、买家类型或采购渠道多元化。[CU021, CU022, CU023, CU024, CU025, CU026]

留存 / 重复使用 / 满意度表
指标数值 / 空值细分市场置信度尽调问题
总留存全部细分市场索取任何试点或量产账户的续约和流失数据
净收入留存全部细分市场索取按同期群或账户拆分的支出扩张
重复采购设备和服务客户索取重复服务或工具订单历史
合同期限晶圆厂和机构索取评估、试点或服务合同期限
客户满意度 / NPS全部细分市场索取客户访谈和支持满意度证据
切换成本代理指标通过认证后可能较高晶圆厂和试验线真实装机基础披露后再验证
现场服务粘性代理指标若维护范围真实,可能有意义服务客户索取 SLA、备件和正常运行时间义务

空白项占主导,因为没有公开披露 Source Foundry 客户基础。两行代理指标只是情景逻辑,不是实际留存证据。

[CU021, CU022, CU023, CU024, CU025, CU026]
扩张和集中度风险表
扩张驱动因素集中度风险影响尽调路径
拿下一个锚定晶圆厂认证头部客户可能主导早期收入重塑可信度和销售漏斗索取按具名项目拆分的账户阶段图和预测
一个试验线或联盟装机公开证据仍可能无法转化为量产可带来战略正当性和技术数据索取下游商业转化计划
参与生态或联盟伙伴依赖可能遮蔽终端客户牵引力可能缩短认证周期索取具体联盟角色和客户交接流程
贴近封装的切入点可降低 ASML 式直接暴露,但会压窄 TAM打出另一条采用路径索取产品边界和买方负责人图
政府背书项目支持政策驱动需求可能阶段性出现可补贴早期装机和本土能力索取非稀释资金和采购可见度
到 2027 年仍未披露锚定客户商业叙事仍完全靠论点支撑会实质性压低 $5B 叙事承销估值前,索取受 NDA 保护的客户证明

本章的核心商业结论是,在 Source Foundry 证明相反情况前,应假设集中度风险极端。

[CU027, CU028, CU029, CU030, CU031, CU032]
Chapter 07

07风险

7.1 最高严重度风险:证据、导入验证与政策暴露

最高风险来自野心和公开证据之间的缺口。商标记录、创始人履历和融资报道支持一个宽广的半导体制造推进,但没有展示具名产品、已发布客户部署或可见质量体系。这很关键,因为半导体制造买家不会只为新颖性买单。他们需要零缺陷目标、供应商控制、支持基础设施和导入验证路径,并且这些要在成熟代工和设备生态里可见。公开可比来源把门槛说得很清楚:TSMC、UMC 和 SkyWater 发布质量与风险管理系统;Samsung、Intel 和 TSMC 发布客户使能项目;Canon 和 xLight 发布具名原型或机构证明点。Source Foundry 尚未发布类似证据。因此,最严重风险不是抽象初创风险,而是具体工业化风险:公司可能必须同时解决产品准备度、客户证明和符合政策安全要求的供应准入。当前问题不是已知这些控制在内部缺失,而是它们尚未在公开层面可见,这迫使外部投资人假设更大的剩余执行风险。[CR001, CR002, CR003, CR004, CR005, CR006]

监管 / 法律风险登记表
风险司法辖区状态可能性严重性缓释措施剩余暴露尽调路径
半导体制造设备的出口管制暴露美国及盟友管制体系政策环境活跃围绕合规路径设计供应链和客户组合仍容易受到规则收紧或许可延误影响将 BOM 和客户地域映射到 BIS 敏感类别
未披露的产品级 IP 地位美国及全球公开信息不清创始人可能掌握专有技术,公司也可能持有未公开申请尚无公开专利语料支撑防御性索取专利清单、权属转让和自由实施审查
意向使用商标范围先于已披露商业化美国公开商标于 2026 年 3 月提交商标广泛预留商业类别不能证明整个产品栈已真实就绪或拥有经营权利索取与商标和产品级申请对应的产品路线图
环境、健康和安全合规可见度缺口多个公开信息不清可通过晶圆厂级合规建设解决尚无公开 EHS 或安全信息索取 EHS 计划、危险材料处理和场地认证
设备装机后的潜在合同和支持责任客户司法辖区前瞻性服务商标范围显示公司预期承担支持义务尚未公开披露 SLA 或保修框架索取服务条款草案、正常运行时间承诺和保修准备金

各行按公开证据下的当前剩余严重性排序,而非按理论可能性排序。

[CR001, CR005, CR006, CR007, CR018, CR019]
FR001: 风险热力图

残余风险最高的部分集中在商业化证明、制造执行和政策敏感依赖。

该矩阵把引用的公开证据转换为相对风险位置;它不是统计模型。

[CR004, CR005, CR006, CR013, CR027, CR028]

7.2 运营、供应链与依赖风险

运营风险异常高,因为 Source Foundry 似乎瞄准的是半导体栈中可靠性、正常运行时间、掩模、制程窗口和现场支持都必须先到位,规模收入才能稳定的环节。公开范围包括维护和技术支持,这意味着服务准备度就是产品承诺的一部分。但公开证据没有显示已安装客户基础支持组织、认证界面或制造防御体系。依赖风险又放大这个问题。BIS 的出口管制行动和 CSIS 的分析显示,半导体制造设备仍是政策敏感瓶颈,尤其是 DUV、EUV 和相关制造项目。商业上依赖少数晶圆代工厂、试点机构或生态项目,也会进一步提高集中风险。即便需求存在,若供应商准入、伙伴导入验证或政策许可变成门槛,初创公司仍可能失败。Intel、Samsung、TSMC、Rapidus 和 GlobalFoundries 的公开生态页面显示,为了把客户带到首片硅、原型或生产状态,通常需要大量协调。实践中,运营风险应该被读成一条链:质量缺口拖慢导入验证,导入验证延迟客户证明,证明延迟又加大融资和估值压力。[CR013, CR014, CR015, CR016, CR017, CR018]

运营 / 质量 / 安全风险登记表
失败模式可能性严重性缓释成熟度剩余暴露未解决缺口
技术达不到晶圆厂级可靠性或良率可见度低没有公开产品认证证据
现场支持和维护能力落后于装机可见度低没有公开服务覆盖或 SLA 信息
质量体系建设低于客户预期可见度低没有 Source Foundry 公开质量认证或零缺陷项目
掩模 / 掩模版 / 工作流集成比计划更难可见度低没有公开运营模型或集成证据
原型到量产转换比预期多耗数年对照证据显示周期很长没有从原型到量产的公开里程碑桥接
安全、业务连续性或事件处理不成熟可见度低没有公开信任或事件信息

运营风险以成熟晶圆代工和设备平台披露的质量、连续性和供应商控制为基准。

[CR002, CR003, CR008, CR009, CR010, CR011]
伙伴 / 依赖风险登记表
依赖交易对手角色集中度失败情景严重性缓释措施剩余暴露
客户赋能生态晶圆厂、设计服务和封装伙伴认证以及通往首硅或试点的路径缺少伙伴支持,客户采用就会停滞锁定早期生态位置和试点赞助方
机构试点入口TIE 式机构或原型线早期证据和工艺验证没有试点场地,证据就停留在理论层面拿下一个具名机构或试验线关系
政策安全供应链BIS 敏感类别下的供应商硬件、制造投入和出口许可规则收紧或许可拖延关键投入设计合规供应链,并保留备选供应商
资金方依赖领投方或融资池过窄后续融资跑道与里程碑压力资本到位,也带来公司兑现不了的证据要求错开里程碑,保留融资弹性
买方池极小先进制程代工厂和 IDM收入集中度一次认证延误就会压垮近期计划在试点、封装和区域项目之间分散布局
支持渠道覆盖OEM 或现场服务伙伴安装与正常运行响应没有可信支持层,客户会拒绝部署搭建直营服务或认证伙伴模式

依赖风险很重,因为买方池小,政策环境又异常活跃。

[CR015, CR016, CR017, CR018, CR019, CR020]
FR002: 风险传导图

关键风险会经由客户、收入时点、毛利率和后续融资传导,而不是停留在技术层面。

传导图展示了本章所引风险证据隐含的因果路径。

[CR006, CR017, CR018, CR021, CR028, CR029]
FR003: 依赖关系图

Source Foundry 似乎需要靠少数买方、试点场地、供应商和生态项目,才能跨过商业化缺口。

由于 Source Foundry 尚未公开合作伙伴名单,依赖项按类别呈现。

[CR015, CR016, CR018, CR019, CR020, CR025]

7.3 财务、团队与论点破裂风险

财务和团队风险位于运营风险下游。据称 $5B 估值和 $400M 融资可以买到时间,但也抬高了证明门槛。如果 Source Foundry 到 2026 年公开层面仍处于商业化前阶段,任何客户证明、伙伴验证或产品可靠性的延迟,都可能迅速转化为估值压缩和后续融资风险。客户宇宙很小,因此一个锚点账户分量极重;这让招聘、现场服务和业务开发执行变得关键。创始人的技术深度是正面输入,但不能替代制造、安全、质量、采购和服务领导力。因此,论点不应因为模糊失望而破裂,而应被可观察触发项打破:在合理里程碑前没有具名客户或生态证明、没有公开质量和支持控制证据、无法展示产品专属 IP 或导入验证进展,或有证据表明政策约束实质阻断供应链或客户准入。正确的尽调反应,是把这些列为可监控风险并给出明确问题,而不是把它们笼统归为普通初创不确定性。这种框定让风险章节绑定可观察里程碑,而不是泛泛担心硬科技初创。投资人还应注意,下行大概率不会一次性出现。它可能先表现为试点证明滑坡,随后是伙伴验证缺失,再之后是融资谈判变难。把这些视为连续预警灯,比等待单次灾难性失败更能形成有用的治理框架。[CR027, CR028, CR029, CR030, CR031, CR032]

人员 / 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
创始人与技术领导层技术信号强,但公开运营证明有限补齐制造、质量和现场服务深度索取组织架构图和关键招聘清单
制造运营需要从概念推进到可重复的工业化执行招聘有晶圆厂和设备规模化经验的负责人索取简历和过往规模化履历
质量与 EHS 领导层需要认证、安全体系和连续性流程及早搭建正式质量与合规职能索取明确负责人和认证路线图
商务拓展 / 客户管理买方池极小,需要顶尖企业销售能力招募拥有代工厂关系的客户团队索取各客户的销售管线归属
现场服务与支持支持义务就是产品承诺的一部分搭建直营服务组织或认证伙伴模式索取安装与支持手册
知识产权 / 法务运营需要自由实施空间和权属转让清洁度建立专利、商标和合同管理索取律师备忘录和权属转让清单

执行风险不只是创始人风险;公司还要围绕一个技术野心很高的产品,搭起工业运营系统。

[CR001, CR002, CR003, CR032, CR033, CR034]
缓释措施与止损标准表
风险可监测触发项阈值 / 事件行动含义
客户证明风险具名试点或生产证明到下一次重大融资或 2027 里程碑前,仍没有具名客户、试点或生态证明从观察上调为投资逻辑破裂候选
质量体系风险公开或 NDA 下的质量证据尽调拿不到认证路线图、正常运行时间模型或服务政策不要把生产规模部署假设纳入投资测算
政策风险供应链或出口管制扰动关键 BOM 项或客户地区落入受阻或延迟的管制路径下调商业化时间和资本开支效率假设
执行风险产品特定里程碑证明没有产品特定认证、基准测试或集成证据将建议转向继续研究 / 观察
融资风险估值与证据之间的缺口拉大公司在产品或客户证据没有实质改善时寻求新资本假设存在降价轮或优先权压顶风险
集中度风险客户分散度销售管线集中在一两个客户,且试点没有分散给上行情景封顶,提高下行概率

这些止损标准强调可观察、可审计,而不是凭印象判断。

[CR027, CR028, CR029, CR037, CR038, CR039]
Chapter 08

08估值

8.1 公开定价信号真实,但估值历史很浅,条款披露缺席

Source Foundry 现在有了真实的估值标记,但公开披露的材料还不够,不能把这个标记当成已经独立验证的公允价值。 Yahoo Finance、AI Weekly、Best AI 和 toflow 给出的 2026 年 8 月核心事实一致:Situational Awareness 新增 $400 million 承诺,披露总持仓升至约 $500 million,公司估值约 $5 billion。这足以说明发生了一次严肃的资本事件。 但还不足以说明这个价格已经被充分背书。留存的公开来源都没有披露公司有多少股权易手、最新资金是否全是一级增资、 是否存在清算优先权或棘轮条款,也没有说明本轮是否改变了治理安排。更早的融资历史也很薄。公开聚合报道显示, 2026 年初可能有一笔 $100 million 初始投资,Sequoia 参与其中;但这一步没有配套公司一手披露、备案文件或董事会权利细节, 外部投资者无法还原一条干净的估值演进线。结果就是,估值历史里有一个可观察的后期标记,底层公开条款却极少。 这一点在这里比在软件公司更重要,因为公司仍处于隐身状态,产品边界未披露,客户或收入的公开证据也仍然缺席。 从估值角度看,8 月轮融资首先应理解为资方信念,其次才是市场出清的价格发现。[CV001, CV002, CV003, CV004, CV005, CV006]

建议摘要表
维度当前判断落在这里的原因决策含义
建议继续研究融资事件真实存在,但公开记录还不足以用常规估值工具支撑以 $5B 入场。不要直接按表面信息认定该轮可投,先做更深尽调。
信心多个独立来源都指向这轮融资,但价格仍主要来自第三方,条款细节缺失。用情景区间,而不是单一公允价值点估。
风险评级在公开收入出现之前,认证风险、客户不透明、赞助方集中和倍数压缩都可能先打过来。假设下行会通过延迟和重定价体现,而不只是彻底失败。
估值立场昂贵当前估值标记已接近那些披露了更多运营证明的后期硬科技平台。接受价格前,要求证据或结构性保护。
入场纪律只在证据驱动条款下入场叙事变好但没有新证据还不够;钱更多但条款更不清晰,反而可能更糟。同时坚持里程碑、客户和股权结构清晰。

这张表总结了据报道的 ~$5B 估值标记下的投资测算姿态;它是判断表,不是已披露的运营 KPI 表。

[CV001, CV003, CV011, CV041, CV042, CV043]
估值历史和公开融资背景
日期 / 期间事件已披露资本公开估值信号变化证据质量
2025-07加州注册记录公开n/a未披露估值相比产品成熟度,法律实体和创始人身份更容易核验。官方公司记录,但不是融资披露。
2026-01(聚合平台报道)与 Situational Awareness 和 Sequoia 挂钩的初始战略投资报道约 $100M估值未公开披露公司首次以半导体制造高确信度押注的形象进入公开视野。仅有第三方聚合平台级别证据;需管理层确认。
2026-08Situational Awareness 后续投资报道新增 $400M报道估值约 $5B公司从隐身叙事跃迁为重大融资事件。多家独立第三方报道。
2026-08 轮后当前公开状态披露总承诺约 $500M只有一个当前估值标记,条款未公开公开历史仍缺少投资条款、收入和客户披露。融资存在性证据较强;完整定价背景较弱。

估值历史很短,且部分由第三方报道重构。它只能作为融资背景,不能视为完整股权结构历史。

[CV001, CV002, CV003, CV005, CV012]
FV001: 建议逻辑

建议从真实融资信号出发;一旦检验条款不透明、出资方集中和商业证明缺失,可定价性就走弱。

该流程用决策关口替代数值公式,因为公开记录披露的输入不足,无法做确定性估值。

[CV007, CV009, CV011, CV026, CV035, CV041]

8.2 上市半导体可比公司显示,$5 billion 通常需要多少已验证收入

检验 Source Foundry 估值标记,最干净的方法不是套用私募市场神话,而是看已经反映规模化半导体运营证明的上市收入倍数。 StockAnalysis 显示,ASML、Applied Materials、Lam Research 和 KLA 的 EV/Sales 大约在 14x to 19x, 规模化晶圆代工和 IDM 锚点 TSMC、Intel 则更接近 9x to 13x。这些并不是完美可比。ASML 是垄断级光刻龙头, Applied 和 Lam 位于相邻设备层,KLA 享有制程控制溢价,TSMC 和 Intel 是制造商而不是工具供应商。 但这个区间仍能说明举证压力。按这些倍数计算,$5 billion 企业价值意味着年收入大致需要 $268 million to $556 million,具体取决于选用哪个上市基准。Source Foundry 没有披露任何收入。这不代表公司不能成长到值 $5 billion;它只说明当前公开证据还没有搭出这座桥。成熟上市同行还披露了让定价成为可能的信息:收入历史、 利润率、现金流、装机基础经济性和投资者材料。Source Foundry 没有。宏观背景确实有利——SEMI、SIA 和 IDC 都显示 2026 年 AI 驱动半导体增长和设备需求——但品类顺风不等于能直接落到某家公司。因此,今天上市半导体设备公司的倍数区间更适合作为纪律工具, 不能直接背书当前私募估值。[CV013, CV014, CV015, CV016, CV017, CV018]

投资逻辑 / 反向逻辑表
视角投资逻辑反向逻辑什么会改变判断
宏观瓶颈AI 需求和设备约束,让本土光刻挑战者具备战略价值。战略重要性不等于一定能商业转化,也不等于定价干净。证明 Source Foundry 正在解决一个具体瓶颈,并有可衡量的客户拉动。
融资信号已披露的 $500M 承诺说明成熟资本认真看待公司。单一集中赞助方也能制造强信号,并不等于市场已广泛验证。增加独立估值标记、战略伙伴,或披露条款的更广泛投资人组合。
公开可比公司半导体资本设备龙头证明,高质量设备生意可以长期维持溢价倍数。这些倍数建立在数百亿美元收入、装机基础、服务经济性和披露之上,而 Source Foundry 尚不具备。披露通往 $268M-$380M+ 收入的可信路径,并证明早期转化。
私营可比公司Hadrian、Helion、Pacific Fusion 以及 2026 年芯片轮次说明,投资人仍会大规模押注前沿瓶颈。这些可比对象要么披露了更多证据,要么采用里程碑分期,要么处在不同终端市场。弥合证据缺口,或围绕里程碑设计新资本,而不是只靠叙事。
赞助方质量Situational Awareness 在公开清盘期间仍保留私人押注,显示出信念。但这场清盘也在新投资人最不希望看到的时候,抛出治理和后续出资能力问题。证明融资韧性不只来自单一赞助方资产负债表。

每行都把最强的支持估值论点与最强反制因素配对,确保建议对价格敏感。

[CV007, CV009, CV019, CV026, CV028, CV031]
可比估值表
可比对象估值 / 交易信号指标 / 倍数可比原因主要限制对 Source Foundry 的启示
Source Foundry2026 年 8 月后续融资据报估值 ~$5B已披露总承诺 ~$500M;无公开收入锚定市场今天隐含愿意为什么付费。价格主要来自第三方,条款披露缺失。当前估值标记真实存在,但没有独立投资测算支撑。
ASML公开上市的垄断级半导体资本设备龙头16.51x EV/Sales;$40.29B TTM 收入最直接的光刻基准,也证明这个品类能有多值钱。成熟、主导、披露充分的平台,且有装机基础经济性。高质量上沿基准,不是干净的阶段可比。
Applied Materials公开上市的全线设备供应商14.25x EV/Sales;$29.02B TTM 收入相邻制程设备可比公司,装机基础多元。产品覆盖更广,收入证明远多于 Source Foundry。说明十几倍中段半导体资本设备倍数通常需要多少收入托底。
Lam Research公开上市的制程设备专家16.44x EV/Sales;$23.23B FY2026 收入能体现设备层战略重要性的公开溢价可比对象。制程层不同,服务版图也已成熟。暗示若按公开可比公司衡量,Source Foundry 需要很大收入才能支撑 $5B。
KLA公开上市的过程控制溢价可比公司18.64x EV/Sales;$13.58B FY2026 收入证明当证据和利润率成立时,公开半导体资本设备倍数的上沿在哪里。溢价来自盈利规模,不是隐身期可选性。即便是公开高端可比,也假设业务证明远强于 Source Foundry。
Hadrian美国私营制造平台融资轮Series D 轮融资 ~$1.37B,对应估值 ~$7.9B可映射主权制造和国家安全资本偏好的可比对象。终端市场不同,公开运营披露明显更强。说明市场愿意资助本土工业平台,但通常要求更多证据。
Helion私营资本密集型登月式项目Series F 轮 $425M,对应投后估值 ~$5.425B有助于比较前沿硬件公司如何靠具名商业对手方达到相近估值。聚变是不同市场,也有不同资本曲线。相比更薄的公开证据,Source Foundry 相近的头部估值显得偏满。
Pacific Fusion采用里程碑结构的前沿硬件融资轮>$900M Series A 已承诺;估值未披露可作为结构可比:投资人如何把资金释放与证据关卡匹配。行业不同,估值未披露。凸显当前沿风险明显时,投资人往往要求里程碑结构。

覆盖范围有意保持有限:这里选的是抓取语料中最影响决策的公开半导体资本设备和资本密集型私营参照,不是所有硬件创业公司的全量清单。

[CV003, CV013, CV014, CV015, CV016, CV017]
FV002: 估值敏感性

按公开基准 EV/Sales 倍数测算,要支撑 $5B 估值所需收入仍然很大,尤其是公司没有公开收入披露。

收入要求只是简单算术:$5,000M 除以各基准 EV/Sales 倍数。它们是敏感性锚点,不是预测。

[CV019, CV020, CV021, CV022, CV023, CV024]

8.3 情景区间可以成立;传统 DCF 不成立

Source Foundry 没有披露收入基线、利润率结构、资本开支曲线、烧钱速度或客户转化路径,传统折现现金流模型只会制造伪精确。 正确方法是把情景分析绑到验证里程碑上。市场背景确实解释了为什么投资者愿意下大注:Crunchbase 称 2026 年半导体创业公司融资仍然很热, 私募市场也支持了 Ayar Labs、Etched、Hadrian、Pacific Fusion 和 Helion 这类围绕瓶颈叙事开出的大额支票。 但这些交易反而让纪律更重要,而不是更宽松。Hadrian 近 $8 billion 的 Series D 伴随了更多关于工厂、软件和客户的公开运营语境。 Helion 的 $5.425 billion 投后估值跟在技术里程碑和具名商业交易对手之后。Pacific Fusion 的 $900 million Series A 按里程碑分批拨付,明确把资金释放绑定到验证闸门。相比之下,Source Foundry 在公开记录中的估值水位接近这些公司, 却仍然几乎没有产品、客户或融资结构细节。这会拉宽区间。悲观情景低于当前估值,适用于公司只证明了子系统切口或原型、 但没跑通多客户认证的世界。基准情景接近当前估值,需要可信的原型进展,以及一条未来通向数亿美元收入的可相信路径。 乐观情景高于当前估值,则要求 Source Foundry 看起来像真正的本土半导体设备平台,而不是只有论点的挑战者。 三种情景里,当前价格都几乎不容执行打滑。[CV029, CV030, CV031, CV032, CV033, CV034]

乐观 / 基准 / 悲观情景表
情景基于情景的估值区间(USD M)假设概率信号相对当前据报估值标记的回报决策含义
悲观$1,500-$3,000Source Foundry 证明了某个技术切入点,但在下一轮融资窗口前,未能展示多客户认证或清晰商业牵引。只要证据继续留在私下、且客户验证到来前资本市场降温,这一情景就有可能。-70% 到 -40%,相对 $5,000M 入场价。不要为了拿额度硬上;等待价格重置或证据实质改善。
基准$3,000-$5,000原型进展变得可信,至少出现一条晶圆厂或生态路径,通往 $200M-$300M 长期收入的桥开始站得住。这是今天最能支撑的公开区间,因为它假设会有进展,但不假装公司已经去风险。相对当前据报估值标记持平至 -40%。密切跟踪,但只有在条款和里程碑可见时才纳入投资测算。
乐观$5,000-$8,000Source Foundry 拿出公开产品边界、客户或试点证明,并形成可支撑未来数亿美元收入的半导体资本设备平台叙事。需要截至 2026-08-11 尚未公开的证据。相对当前据报估值标记 0% 至 +60%。只有证据改善之后才有吸引力,之前不是。

这些是以 USD 百万美元计的情景区间。它们不是已披露的当前估值模型,也刻意避开虚假的 DCF 精度,因为核心运营输入仍未公开。

[CV035, CV036, CV037, CV038, CV039, CV040]
FV003: 估值 / 回报区间

情景区间仍然很宽,因为公开记录缺少可做 DCF 的经营输入,而且未来稀释轻重取决于里程碑时点。

这些是基于情景的估值区间,单位为 USD millions。它们不模拟准确瀑布、分期或 DCF 现金流,因为这些条款并未公开。

[CV037, CV038, CV039, CV040]

8.4 建议取决于新证据,不取决于更热的叙事

本报告的建议必须明确对价格敏感。如果进入价位低得多,Source Foundry 兼具国家安全相关性、真实融资事件和可信的宏观瓶颈, 投资者可以先保持跟踪,等待证据。到了报道中的约 $5 billion 估值,门槛就高得多。公司仍没有公开客户证明、 公开收入、公开产品边界或公开融资条款。资方语境也不是脚注:Situational Awareness 的公开市场减仓,让人有理由追问当前故事中有多少是集中信念、 又有多少是广泛验证。这并不否定公司,但意味着新投资者在接受价格前应要求更多证据。因此,立场落在继续研究(research-more), 信心中等,风险高,估值观点偏贵。能改变立场的是具体证据:具名试点或设计伙伴、披露的产品边界、可信的原型里程碑, 或者证明最新融资结构是在保护而不是拖累未来投资者。会打穿论点的则是相反顺序:公开证明出现前又进行一轮大额融资、 估值跌破当前水平,或公司继续不透明,却要求外部投资者按成熟半导体设备平台给它定价。尽调清单应当毫不客气地务实: 重建股权结构,梳理产品架构,核验原型里程碑,识别首批客户路径,并判断 $400 million 跟投究竟是普通增长资本, 还是更不寻常的融资事件。[CV041, CV042, CV043, CV044, CV045, CV046]

投资逻辑破裂与止损触发项表
触发项阈值 / 事件对投资逻辑的传导行动含义
融资低于当前估值标记下一轮定价融资或可靠老股交易成交价显著低于 ~$5B证实今天的价格跑在证据前面。在新证据或条款披露前,按投资逻辑破裂处理。
无客户或试点证明下一次重大融资事件前,没有具名试点、设计伙伴或晶圆厂验证路径削弱从瓶颈叙事走向商业转化的桥。建议维持继续研究,或直接退出。
无产品边界披露公开记录或尽调材料仍说不清 Source Foundry 销售的具体层级可比分析和里程碑规划过于模糊,无法纳入投资测算。第一个切入点清楚前,不要按平台给公司定价。
赞助方不稳定加剧领投赞助方融资能力或治理可信度进一步走弱公司可能仍需要耐心资本时,后续融资和信号风险反而上升。下调上行空间,提高稀释风险。
认证延误原型或集成里程碑延误,且没有客户证据抵消在当前高估值标记不变时,把收入桥进一步推后。将期望值向悲观情景移动。
政策或供应链堵点出口管制、供应商或设施约束显著拖慢商业化把艰难技术故事变成延迟兑现的经济故事。降低乐观情景概率,扩大下行空间。

这些触发项强调可观察、可转化为投资动作,而不是修辞。

[CV043, CV044, CV045, CV046]
最终尽调问题表
主题缺失证据重要性负责人或尽调路径紧迫性
股权结构表与条款出售股权比例、清算优先权、反稀释、董事会权利,以及 8 月资金是新股还是老股交易这是判断 $5B 是真实经济价值还是叙事头条价值的核心。向公司或领投方索取融资文件、董事会同意文件和投资人摘要。关键
产品边界第一个产品或子系统的精确范围、架构层级,以及相对 ASML 哪些在范围内、哪些不在没有这些,可比分析和里程碑映射仍然太抽象。索取产品路线图、工程简报和客户问题陈述。关键
原型里程碑原型、子系统和认证就绪的日期、指标和闸门标准这是映射情景时间和资本充足性的必要输入。索取里程碑计划和内部评审节奏。关键
客户路径NDA 下的具名试点、设计伙伴、生态赞助方或晶圆厂对话需要把战略重要性转化为实际收入概率。按客户、阶段和技术卡点索取销售管线。关键
经济性桥收入模型、ASP 或合同结构、服务组成和毛利率目标$5B 估值标记需要一条通往数亿美元收入的可信路径。索取定价框架、利润率模型和装机基础假设。
融资韧性Situational Awareness 之后谁能继续供资,以及采用什么结构如果认证周期很长,后续出资能力就很关键。索取投资财团地图、储备策略和替代融资路径。

这些问题优先覆盖那些比泛泛市场尽调更快改变建议的事项。

[CV004, CV026, CV035, CV047, CV048]
FV004: 投资 KPI

公开评分最强的是战略相关性和融资信号;最弱的是估值可见度、商业证明和尽调完整度。

这些是截至 2026-08-11 基于章节证据得出的定性序数评分,不是经审计的经营指标。

[CV042, CV043, CV044, CV045, CV047]

8.5 展示材料

免责声明

本报告是 AI 辅助生成的尽调材料,仅基于截至 2026-08-11 的公开信息,不构成投资建议。私营公司的融资条款、技术性能、客户沟通和内部指标,可能与公开报道存在重大差异;在作出任何投资或合作决策前,请用一手文件核验所有重大事实。

证据索引

结论
编号陈述可信度来源
CO001 Source Foundry Inc. is an active California stock corporation filed on 2025-07-18 and formed in Delaware. SO001, SO005
CO002 The company's principal and mailing address in reviewed public records is 674 Harrison Street, San Francisco, California 94107. SO001
CO003 California business records list Abdulmalik Obaid as chief executive officer and chief financial officer and Joe Burg as secretary. SO001, SO005
CO004 Reviewed August 2026 coverage identifies Abdulmalik Obaid and Joe Burg as Stanford researchers who founded Source Foundry in 2025. SO003, SO005, SO016
CO005 Stanford Bio-X describes Abdulmalik Obaid's training as Materials Science & Engineering and his work as combining materials science and microfabrication with modern microelectronics. SO009, SO010
CO006 Stanford's 2019 neuroscience coverage quotes Obaid saying he had to design new fabrication processes to bring silicon electronics into the third dimension in a scalable way. SO010
CO007 Joe Burg's public Stanford profile centers on computational methods, organosilicate glasses, and structure-property relationships rather than a publicly documented lithography-operating track record. SO011
CO008 Source Foundry's March 4, 2026 trademark applications cover machines for manufacturing semiconductors and lithography machines for manufacturing electronic components. SO012, SO013
CO009 The same trademark package covers semiconductor chips, wafers, lithographic masks, reticles, and downloadable software for processing wafers and developing advanced semiconductor lithography. SO013
CO010 Source Foundry also filed for repair, maintenance, technical support, and information services tied to semiconductor manufacturing machines and lithography systems. SO012, SO014
CO011 A separate service-mark filing covers online software for processing wafers, operating and optimizing manufacturing tools, advanced lithography development, and custom design of semiconductor facilities and machinery. SO015
CO012 The owner page for Source Foundry's trademarks shows additional filings for custom manufacturing, training, and research-and-design services beyond core equipment and software. SO012
CO013 Reviewed August 2026 reporting says Situational Awareness added a new $400 million investment to Source Foundry and brought the fund's total commitment to $500 million. SO002, SO003, SO006, SO016
CO014 The same August 2026 reporting values Source Foundry at approximately $5 billion. SO002, SO003, SO004, SO008
CO015 Public reporting says Sequoia Capital backed Source Foundry before Situational Awareness's disclosed August 2026 investment. SO002, SO003, SO016
CO016 Reviewed public reporting does not disclose whether the new Situational Awareness financing was common equity, preferred equity, debt, or a committed facility. SO005
CO017 Public reporting also does not disclose investor ownership percentages, liquidation preferences, board rights, or whether the full commitment was funded immediately. SO005
CO018 Stephanie Zhan publicly framed Source Foundry as attacking the tightest bottleneck in the AI supply chain: semiconductor-manufacturing tooling starting in lithography. SO002, SO007, SO016
CO019 Zhan also described a chip wall in which AI demand compounds faster than semiconductor manufacturing capacity and said Source Foundry planned to use different underlying physics to increase production speed and scale. SO002
CO020 Multiple August 2026 articles describe Source Foundry's goal as making chip manufacturing simpler, faster, and less expensive rather than designing AI accelerators itself. SO002, SO003, SO006
CO021 The combination of business records and trademark filings points to an equipment, software, and facility-engineering business rather than a conventional fabless chip startup. SO001, SO013, SO015
CO022 ASML reported €32.7 billion of 2025 net sales and remains the benchmark incumbent in advanced lithography that Source Foundry is implicitly trying to displace. SO017, SO016
CO023 ASML shipped 48 EUV systems in 2025 and ended the year with a €38.8 billion backlog, underscoring the scale and entrenchment of the market Source Foundry is entering. SO017
CO024 ASML announced in July 2026 that Intel Foundry had already moved a subset of Panther Lake layers into high-volume logic production using High-NA EUV. SO018, SO019
CO025 Intel's own foundry materials describe 18A as already in high-volume production in the United States and positioned for advanced AI workloads. SO019, SO020
CO026 Canon commercialized a nanoimprint lithography system in 2023 and shipped one to the Texas Institute for Electronics in 2024, showing that alternative lithography approaches are already reaching public prototypes. SO021, SO022
CO027 Nikon continues to market a full portfolio of semiconductor lithography, alignment, and metrology systems, so Source Foundry is also entering a market that still has established non-EUV incumbents besides ASML. SO023
CO028 The U.S. Department of Commerce finalized a $150 million CHIPS award to xLight in June 2026 for a free-electron-laser EUV light source, evidencing broader investor and policy appetite for upstream lithography challengers. SO024, SO025
CO029 TechCrunch reported that Situational Awareness had sold most of its public portfolio and that its assets under management had fallen from roughly $20 billion to $10 billion before the new Source Foundry investment. SO006
CO030 The Source Foundry investment therefore arrived as a contrarian private deep-tech bet during a period of significant turmoil for its lead financial backer. SO003, SO006
CO031 No reviewed public source disclosed Source Foundry's process node, wafer size, throughput, tool architecture, or commercial launch date. SO005, SO016
CO032 No reviewed public source disclosed Source Foundry's revenue, ARR, customers, headcount, or production installations. SO001, SO005
CO033 The 2026 trademark burst suggests the company formalized a broad equipment, software, and services perimeter several months before its funding emerged publicly. SO012, SO013, SO015
CO034 Source Foundry remains a stealth company even after reaching a reported $5 billion valuation and attracting a disclosed half-billion-dollar commitment from Situational Awareness. SO002, SO005, SO006
CO035 Because the strongest public descriptions come from business records, trademark filings, and financing coverage, the current company overview is rich on scope ambition but thin on proof of operating execution. SO001, SO012, SO015, SO005
CM001 SEMI forecasts total semiconductor manufacturing equipment sales will reach $165.9 billion in 2026 and $229.5 billion in 2028. SM001
CM002 SEMI projects wafer-fab-equipment sales to rise to $143.9 billion in 2026, with foundry and logic applications accounting for $78.0 billion. SM001
CM003 SIA reported global semiconductor sales of $403.3 billion in Q2 2026 after April through June demand accelerated sharply year over year. SM002
CM004 SIA said 2025 global semiconductor sales reached $791.7 billion and projected roughly $1 trillion worldwide sales in 2026. SM002
CM005 IDC characterizes the 2026 semiconductor market as crossing the trillion-dollar threshold because AI infrastructure is driving the strongest share of demand. SM003
CM006 Mordor Intelligence estimates the semiconductor lithography equipment market at $30.44 billion in 2026, up from $27.83 billion in 2025. SM009
CM007 The same market report projects the lithography equipment market will reach $47.63 billion by 2031, implying a 9.37% CAGR from 2026 to 2031. SM009
CM008 Mordor says deep-ultraviolet systems still held 67.31% of lithography-equipment revenue in 2025 even while High-NA EUV is projected to grow faster. SM009
CM009 Mordor attributes 46.89% of 2025 lithography-equipment spending to pure-play foundries, indicating that foundries are the largest buyer class for advanced patterning tools. SM009
CM010 Mordor says advanced packaging captured 36.74% of 2025 lithography-equipment revenue, showing that the addressable patterning market extends beyond front-end EUV scanners alone. SM009
CM011 TSMC said it manufactured 12,682 products for 534 customers in 2025 across 305 distinct process technologies, illustrating how concentrated yet broad leading-edge foundry demand can be. SM008
CM012 Samsung Foundry markets itself as offering high-volume manufacturing, EUV from 5nm, and 3nm gate-all-around, while partners now highlight second-generation 2nm for AI infrastructure. SM006, SM007
CM013 Intel says 18A is already in high-volume production and 18A-P has entered risk production, confirming that the frontier buyer set is already deploying new lithography-intensive nodes rather than waiting for entrants. SM011, SM012
CM014 ASML says Intel has entered high-volume manufacturing on selected Panther Lake layers using High-NA EUV, which raises the performance bar any new lithography challenger must beat or complement. SM013
CM015 ASML reported €32.7 billion of 2025 sales and 48 EUV systems shipped, showing that the leading-edge lithography market is already dominated by a very large incumbent base. SM016
CM016 ASML frames its own offering as a portfolio of lithography systems, metrology, inspection, and software, which implies that market demand bundles hardware with process-control layers. SM005
CM017 Source Foundry’s own trademark record points to a similar broad boundary that includes machines, wafer software, and facility-engineering services rather than a single SKU. SM020, SM021, SM023
CM018 A reasonable market boundary for Source Foundry therefore includes lithography equipment, tool-control software, and certain fab-engineering workflows, but excludes commodity back-end handling and generic chip-design tools. SM005, SM020, SM021
CM019 The broadest TAM lens is total semiconductor-manufacturing equipment spend, but Source Foundry’s likely SAM is much narrower because its public scope centers on lithography and adjacent process-control layers. SM001, SM009, SM020, SM021
CM020 The most defensible public lithography lens is roughly $30.44 billion in 2026, because it is specifically scoped to lithography rather than all fab equipment. SM009
CM021 A narrower product-near SAM for Source Foundry would sit inside leading-edge logic, memory, and packaging programs where buyers are actively funding EUV, High-NA, or alternative-patterning improvements. SM001, SM009, SM011, SM012
CM022 No retained public source is sufficient to compute Source Foundry’s SOM because the company has not disclosed whether it targets the source, scanner, mask, control stack, or fab-service layer in production form. SM020, SM021, SM023
CM023 Pure-play foundries, integrated device manufacturers, and advanced-packaging houses are the primary direct buyers for new lithography tooling, while hyperscalers are usually indirect payers through AI-driven demand. SM001, SM009, SM008
CM024 Cadence and Samsung’s 2nm collaboration shows ecosystem buyers increasingly care about AI infrastructure, 3D-IC integration, memory IP, and high-bandwidth interconnect together rather than process nodes in isolation. SM006
CM025 SEMI highlights AI, HBM, advanced memory, logic, test, and packaging as simultaneous capital-formation drivers, which broadens the number of chip categories that can justify new lithography investment. SM001
CM026 SIA and IDC together support the view that AI demand is the central short-term growth driver behind semiconductor-capex acceleration in 2026. SM002, SM003
CM027 Mordor identifies shrinking process nodes, AI and data-center demand, government fab subsidies, and heterogeneous integration as the main positive drivers for lithography spending. SM009
CM028 Export-control restrictions, ultra-high EUV capex, helium volatility, and ESG constraints are material public restraints on lithography-market expansion. SM009
CM029 Mordor says High-NA scanners can list above $400 million each and require deposits 12 to 18 months before delivery, which sharply raises buyer switching and qualification risk. SM009
CM030 IEEE says Canon’s NIL system is pitched as a lower-cost, lower-energy alternative to EUV, while xLight aims to serve up to 16 scanners from outside the fab with a utility-scale free-electron-laser source. SM010, SM014, SM015
CM031 Canon claims 14 nm patterning today with a path toward 10 nm, showing that not every credible alternative to EUV is trying to replicate ASML’s current system architecture. SM017, SM010
CM032 Nikon still markets a full lithography, alignment, and inspection lineup, which makes the broader market more diverse than the ASML-only EUV narrative suggests. SM019
CM033 Asia-Pacific held 63.58% of 2025 lithography-equipment revenue, while North America is projected to grow fastest through 2031, according to Mordor. SM009
CM034 SEMI expects China, Taiwan, and Korea to remain the top three equipment-spending destinations through 2028, while U.S. subsidies support regionalization elsewhere. SM001
CM035 For Source Foundry, the market is large enough to matter but concentrated enough that commercial success depends on winning a very small set of deeply technical buyers with extreme uptime and qualification requirements. SM001, SM009, SM011, SM012, SM013
CP001 ASML says its EUV systems use 13.5 nm light and print the highly complex foundation layers of 7 nm, 5 nm, and 3 nm chips. SP002
CP002 ASML says EUV and DUV technologies will both be required in parallel for many years to come. SP002
CP003 ASML says its latest DUV immersion systems can run in excess of 6,000 wafers per day, highlighting why incumbent productivity is hard for challengers to match. SP003
CP004 ASML reported €32.7 billion of 2025 net sales, 535 system sales, more than 44,000 employees, and 5,100 suppliers in its 2025 annual report. SP001
CP005 ASML describes its offering as a holistic lithography portfolio where hardware, software, and services work together rather than as a single scanner sale. SP001, SP004
CP006 Nikon publicly markets lithography systems, alignment stations, metrology or inspection systems, and a worldwide support network for semiconductor customers. SP006, SP007
CP007 Nikon Precision's public lineup spans KrF scanners, ArF immersion scanners, advanced packaging tools, and alignment stations. SP006
CP008 Nikon's DSP-100 entered orders in July 2025, targets advanced packaging, supports 600 by 600 mm substrates, delivers 1.0 μm line and space resolution, and uses maskless SLM projection. SP008
CP009 Canon launched the FPA-1200NZ2C NIL system in 2023 and says it can pattern 14 nm minimum linewidths equivalent to a 5 nm node, with a path toward 10 nm corresponding to 2 nm-node logic. SP009
CP010 Canon shipped the FPA-1200NZ2C to the Texas Institute for Electronics in September 2024 for advanced-semiconductor R&D and prototype production. SP010
CP011 IEEE Spectrum reported that Canon's NIL could challenge EUV on cost and energy use, but also described a 20-year commercialization effort and significant overlay, contamination, and defect-control hurdles. SP011
CP012 xLight finalized a $150 million CHIPS award in June 2026 for construction and demonstration of a first-of-its-kind free-electron-laser prototype in Albany, New York. SP012, SP013
CP013 xLight says its FEL systems are built outside the fab, can deliver power to up to 16 scanners at once, and aim for 4x more EUV power with 50% lower costs. SP012, SP014
CP014 EV Group positions itself around mask aligners, maskless exposure, 300 mm-ready systems, and advanced-packaging or MEMS workflows, making it an adjacent packaging competitor rather than a direct EUV front-end rival. SP015
CP015 Intel says Intel 18A is ready for customer projects, in high-volume production in the United States, and that Intel 18A-PT is purpose-built for advanced 3DIC integration with up to 9x die-to-die bandwidth density. SP017
CP016 Intel frames its broader manufacturing roadmap as a systems foundry for the AI era spanning multiple process nodes and advanced packaging capabilities. SP016
CP017 Samsung Foundry publicly combines EUV from 5 nm, 3 nm GAA, high-volume manufacturing, and integrated 2.5D or 3D packaging offerings for customers. SP018, SP019
CP018 TSMC says its advanced-packaging services offer integrated turnkey solutions across SoIC, CoWoS, and InFO and collaborate deeply with substrate, memory, and materials suppliers to reduce time-to-volume. SP021
CP019 TSMC says annual capacity across facilities it manages exceeded 17 million 12-inch-equivalent wafers in 2025, showing the scale of substitute foundry ecosystems. SP020
CP020 Bloomberg reported that U.S. officials warned ASML that one of its advanced lithography machines may have reached China, highlighting the live regulatory sensitivity around this market. SP023
CP021 CISES argues that the EUV chokepoint is a supplier coalition across Dutch, German, Japanese, and American firms rather than ASML alone. SP022
CP022 The same CISES analysis says China is pursuing multiple alternatives, including SMEE's domestic DUV work, a Huawei or SiCarrier-linked alternative light-source effort, and a Tsinghua accelerator path. SP022
CP023 Benzinga reported that Source Foundry is a $5 billion-valued lithography startup with a new $400 million infusion, founded by Stanford researchers and previously backed by Sequoia. SP024
CP024 Source Foundry's public trademark filings cover wafer-processing software as well as installation, maintenance, repair, and technical-support services for semiconductor manufacturing and lithography systems. SP025, SP026
CP025 Because Source Foundry has not publicly disclosed a first product, public evidence cannot determine whether its closest comparison set is full scanners, light sources, software, or packaging-adjacent tools. SP024, SP025, SP026
CP026 The default solution to advanced lithography bottlenecks today is to buy or extend incumbent ASML workflows rather than adopt a new startup architecture. SP001, SP002, SP003, SP005
CP027 Nikon and Canon are attacking narrower wedges such as DUV, packaging, or NIL rather than displacing ASML's entire installed base in one step. SP006, SP008, SP009, SP010, SP011
CP028 xLight is best understood as a source-layer or utility competitor, not a complete scanner substitute, which means it can threaten Source Foundry even if Source Foundry is not building a full EUV machine. SP012, SP013, SP014
CP029 EVG, Intel, Samsung, and TSMC show that bottleneck spending can migrate toward advanced packaging, co-design, and systems integration instead of toward a new front-end scanner vendor. SP015, SP016, SP017, SP018, SP019, SP020, SP021
CP030 Switching costs are reinforced by incumbent bundles of hardware, software, metrology, alignment, packaging, and service rather than by tool performance alone. SP004, SP006, SP007, SP018, SP021
CP031 ASML and Nikon both present public evidence of broad field support or ecosystem depth, while Source Foundry has no public installed-base or service-footprint evidence in retained sources. SP001, SP006, SP007, SP024, SP025
CP032 Regulatory fragmentation increases the value of trusted incumbent supply chains and can slow challenger qualification, sourcing, and geography expansion. SP022, SP023
CP033 For the largest fabs, internal build usually means solving the bottleneck within a foundry, process, or packaging roadmap rather than literally building an EUV machine from scratch. SP016, SP017, SP018, SP019, SP020, SP021
CP034 Source Foundry's fresh capital may buy time and credibility, but public evidence still does not show pricing, named customers, yield data, uptime, or qualification evidence versus incumbent and alternative routes. SP023, SP024, SP025, SP026
CP035 The best-supported competitive conclusion is that Source Foundry faces three rival classes—ASML status quo incumbency, alternative-architecture challengers like Canon, Nikon, and xLight, and internal-build or foundry substitute routes—so moat durability cannot yet be underwritten from public evidence alone. SP001, SP009, SP012, SP016, SP018, SP021, SP024, SP026
CI001 Benzinga reported that Situational Awareness invested $500 million total in Source Foundry, including a fresh $400 million infusion, at a reported $5 billion valuation. SI001
CI002 The same report says Source Foundry was founded last year by Stanford researchers and previously received backing from Sequoia Capital. SI001
CI003 Source Foundry's trademark filings cover wafer-processing software as well as installation, maintenance, repair, and technical support for semiconductor manufacturing and lithography systems. SI002, SI003
CI004 That trademark scope implies Source Foundry could monetize through a mix of software, equipment, and recurring service rather than through one isolated revenue stream. SI002, SI003
CI005 No retained public source discloses Source Foundry revenue, backlog, customer count, unit volumes, or pricing, so current commercial traction is not publicly verifiable. SI001, SI002, SI003
CI006 ASML's 2025 annual report shows €32.7 billion of total net sales, 52.8% gross margin, and €4.7 billion of R&D. SI006
CI007 ASML's Q1 2026 results reported €8.8 billion of net sales, 53.0% gross margin, and €1.2 billion of R&D costs. SI008
CI008 ASML's Q2 2026 results reported €9.3 billion of net sales, 54.0% gross margin, and €1.3 billion of R&D costs. SI009
CI009 ASML says it releases complete financial results each quarter and a full annual report each year, highlighting how sparse Source Foundry's public financial disclosure remains by comparison. SI007
CI010 Intel's 2025 10-K says its long-term R&D and manufacturing-facility investments are significant, inherently risky, and may not realize a favorable return. SI010
CI011 Intel's 2025 10-K also identifies risks around future capital expenditures, foundry capacity, supply constraints, customer concentration, government incentives, tariffs, and export controls. SI010
CI012 Intel describes itself as a strategically important U.S. company that conducts leading-edge process R&D, high-volume logic manufacturing, and third-party foundry services. SI010, SI012
CI013 Intel says Intel 18A is ready for customer projects and in high-volume production in the United States. SI010, SI013
CI014 TSMC publishes both annual reports and a 2026 Q2 investor-results package, giving public detail that Source Foundry does not currently provide. SI014, SI015
CI015 TSMC's 2025 Form 20-F says the company manufactured 12,682 products for 534 customers in 2025, illustrating the breadth and concentration advantages of a scaled foundry substitute. SI016
CI016 TSMC says its advanced-packaging services offer integrated turnkey solutions across SoIC, CoWoS, and InFO. SI018
CI017 TSMC says annual capacity across the facilities it manages exceeded 17 million 12-inch-equivalent wafers in 2025. SI019
CI018 TSMC says its 2nm technology started volume production in 4Q25 as planned, showing that the incumbent substitute route is already monetizable at the frontier. SI017
CI019 Samsung Foundry publicly combines design technologies, IP, EUV and GAA process leadership, high-volume manufacturing, and integrated packaging offerings. SI021, SI022
CI020 Samsung maintains an investor-relations earnings-release surface, further underscoring how limited Source Foundry's public financial disclosure is. SI020
CI021 NIST and Business Wire say xLight finalized $150 million of CHIPS incentives for a first-of-its-kind FEL prototype and frames its business as utility-scale light delivery to fabs. SI023, SI024
CI022 Nikon says its DSP-100 reduces production costs with maskless technology and minimizes maintenance costs with solid-state light sources, illustrating how enterprise buyers evaluate cost structure beyond raw performance. SI025
CI023 The most plausible public revenue paths for Source Foundry are equipment or subsystem sales, paid qualification work, software or controls, and recurring support tied to an installed base. SI002, SI003, SI025
CI024 Public evidence supports treating Source Foundry as pre-revenue or at minimum revenue-undisclosed, because no retained source shows signed customer revenue or product shipments. SI001, SI002, SI003
CI025 If Source Foundry commercializes, the public evidence suggests its business model is more likely to resemble negotiated capital-equipment or infrastructure contracts than SaaS pricing. SI003, SI021, SI022, SI024, SI025
CI026 Source Foundry's near-term unit economics are likely dominated by engineering, qualification, supplier, installation, and field-support costs rather than by low-marginal-cost software delivery. SI006, SI010, SI012, SI018, SI025
CI027 ASML's disclosed 52.8% to 54.0% gross margins are useful as a scale-stage benchmark for mature lithography economics, but not as a near-term forecast for Source Foundry. SI006, SI008, SI009
CI028 The combination of Intel's risk disclosures and xLight's CHIPS-backed prototype shows that frontier semiconductor-tool efforts often require sustained outside capital long before revenue scale is proven. SI010, SI023, SI024
CI029 TechCrunch reported that Situational Awareness sold most of its public portfolio after losses, leaving a smaller asset base and raising a reasonable question about future follow-on flexibility. SI004
CI030 Reuters, via Yahoo Finance, reported that Situational Awareness was down around 67% in July 2026 and held a book of roughly $10 billion after the Citadel deal. SI005
CI031 TechCrunch also reported that Situational Awareness retained private investments such as Anthropic, MatX, and Fluidstack even after unwinding most public holdings. SI004
CI032 Because cash balance, burn, and runway are not publicly disclosed, Source Foundry's capital adequacy cannot be directly underwritten from public evidence alone. SI001, SI004, SI005
CI033 A reported $500 million funding base is large for a stealth startup but may still be modest relative to frontier lithography ambitions if commercialization requires manufacturing-scale deployment and field support. SI001, SI006, SI010, SI018, SI019
CI034 The most plausible next-round trigger is not simple user growth but proof of prototype performance, named pilot customers, supplier readiness, and clarity on the first product's contract model. SI001, SI002, SI003, SI010, SI018
CI035 The defensible public financial verdict is that Source Foundry is a pre-revenue, capital-intensive, financing-dependent manufacturing startup whose $5 billion valuation cannot be justified from disclosed revenue quality today. SI001, SI005, SI006, SI010, SI018, SI023
CE001 USPTO.report says Source Foundry's mark is intended to cover semiconductors, semiconductor chips, semiconductor wafers, software for wafer and chip processing, software for advanced lithography, and lithographic masks and reticles. SE001
CE002 The same USPTO.report page says the mark is an intent-to-use application with no current use basis and was awaiting assignment after filing in March 2026. SE001
CE003 USPTO.report says Source Foundry's service mark is intended to cover installation, maintenance, repair, and technical support services for semiconductor manufacturing machines and lithography systems. SE002
CE004 Taken together, the public trademark records support a product envelope spanning software, physical semiconductor or lithography assets, masks or reticles, and post-install service. SE001, SE002, SE003, SE004
CE005 No retained public Source Foundry source names a commercial SKU, product page, release note, or data sheet. SE001, SE002, SE023
CE006 Canon said its FPA-1200NZ2C NIL platform had already been commercialized and was shipped to TIE in September 2024 for advanced-semiconductor R&D and prototype production. SE008, SE009
CE007 Canon said it acquired Molecular Imprints in 2014 to accelerate development and early market entry of next-generation lithography systems. SE007
CE008 IEEE Spectrum reported that Canon's NIL effort took roughly 20 years to commercialize and had to solve defects, contamination, and alignment challenges. SE010
CE009 ASML describes lithography as a hardware-plus-software system and says entire chips require patterns to be printed across many repeated layers. SE011, SE024
CE010 ASML explicitly says software is critical to its lithography systems and that the systems are now hybrids of high-tech hardware and advanced software. SE011
CE011 ASML says EUV and DUV remain used in parallel for different chip layers rather than as one-for-one substitutes. SE012, SE013
CE012 Nikon's public product stack includes lithography systems, alignment stations, metrology or inspection systems, and a worldwide support network. SE014, SE016
CE013 Nikon's DSP-100 uses a spatial light modulator for maskless projection onto large substrates, showing that alternative product architectures can live in packaging rather than only in front-end scanners. SE015
CE014 xLight positions its FEL as an external utility-scale light source that can power multiple scanners rather than as a full scanner product. SE017, SE018
CE015 TSMC says packaging technologies are now critical to product performance, function, and cost, and that 3DFabric enables holistic system-level optimization beyond monolithic die design. SE019, SE020
CE016 Intel 18A-PT is described as purpose-built for advanced 3DIC designs, reinforcing that frontier product value increasingly spans integration and packaging layers. SE026
CE017 Samsung's foundry materials emphasize a progression from HKMG and FinFET to EUV and GAA, which shows that any challenger product must fit into a rapidly evolving process stack. SE025
CE018 CSIS says export-control and localization policies explicitly target DUV and EUV equipment categories, making policy and supply constraints part of the technical dependency map. SE021, SE022
CE019 Stanford Bio-X and Science Advances materials tied to Abdulmalik Obaid show work on microfabrication, CMOS-linked integration, and scaling three-dimensional silicon-linked device architectures. SE027, SE028, SE029
CE020 That Stanford work is evidence of founder technical depth, but it is not itself a published Source Foundry product document. SE006, SE027, SE028, SE029, SE023
CE021 In customer-workflow terms, Source Foundry's public scope most plausibly maps to process development, installation, optimization, and ongoing support rather than to end-user self-service software. SE002, SE004, SE012, SE016
CE022 A credible product-module map can be drawn from mark classes, but the company has not publicly identified which of those modules is actually first to market. SE001, SE002, SE023
CE023 The best-supported public architecture hypothesis is some combination of software, physical patterning assets, masks or reticles, and post-install support, with the exact operating model still undisclosed. SE001, SE002, SE003, SE004, SE012
CE024 No retained public source shows Source Foundry quality certifications, safety metrics, uptime data, security documentation, or status communications. SE001, SE002, SE023
CE025 The only publicly verifiable Source Foundry milestones in retained evidence are March 2026 trademark filings and August 2026 financing reports. SE001, SE002, SE023
CE026 That roadmap pattern suggests the company was still in a pre-commercial public phase when it surfaced in August 2026. SE001, SE002, SE023
CE027 If Source Foundry is software-first, the customer workflow centers on process tuning and support; if it is hardware-first, the workflow centers on qualification, deployment, and uptime management. SE003, SE011, SE012, SE016
CE028 The public dependency map most clearly includes fabs or pilot lines, masks or reticles, control software, service capability, and semiconductor-equipment supply constraints. SE001, SE002, SE018, SE019, SE021, SE022
CE029 Advanced packaging is both an adjacency and a substitute route because some future value creation may migrate away from classical front-end exposure and toward system integration. SE015, SE019, SE020, SE026
CE030 Source Foundry's trust surface is unusually thin for a company attacking fab-critical infrastructure because no public evidence shows quality metrics, compliance, or support processes. SE002, SE023
CE031 The lack of a retained public product-doc surface materially increases diligence burden because even alternate-architecture challengers like Canon publish more concrete milestone evidence. SE008, SE009, SE023
CE032 Public differentiation cannot be credited from performance data today; the defensible public differentiation is only broad product ambition plus founder technical signal. SE004, SE006, SE027, SE028, SE029, SE023
CE033 Any Source Foundry product must ultimately compete against at least one of the publicly visible architecture families: optical projection, nanoimprint, FEL source innovation, or advanced-packaging system integration. SE008, SE012, SE014, SE017, SE019, SE026
CE034 The hardest technical risks are likely integration, reliability, alignment or overlay, uptime, and support, because comparator paths show those issues dominate real deployment timelines. SE010, SE012, SE015, SE021
CE035 The best-supported product-tech verdict is that Source Foundry has a broad but mostly undisclosed product envelope, strong evidence of ambition, and weak public evidence of shipped maturity or trust controls. SE001, SE002, SE006, SE027, SE028, SE029, SE023
CU001 Source Foundry's trademark scope points to enterprise semiconductor buyers rather than to mass-market or self-serve customers. SU004, SU005
CU002 The realistic near-term buyer universe is small because advanced lithography and wafer-processing infrastructure can only be evaluated by organizations with deep process capability and capital budgets. SU004, SU005, SU023
CU003 Process engineers and equipment teams would likely be the primary users, but payer authority would sit with fab, foundry, or program leadership. SU005, SU010, SU011, SU012
CU004 The service-mark scope implies that field support and maintenance are part of the customer value proposition, not a minor afterthought. SU005, SU020
CU005 Leading-edge foundries and IDMs are the highest-value buyer class because they already manage advanced process, packaging, and tooling decisions at meaningful scale. SU010, SU011, SU012, SU023
CU006 Pilot-line and research institutes are a plausible second buyer class because adjacent semiconductor tool innovators use them to prove prototype and process value before mass adoption. SU006, SU007, SU013, SU025
CU007 Foundry ecosystems and alliance programs are plausible commercial channels because buyers in this market often need design, packaging, mask, and support enablement before qualification. SU010, SU011, SU012, SU014, SU015, SU018
CU008 Government-backed semiconductor programs are plausible buyers or sponsors because domestic-capability buildouts can fund early infrastructure before broad commercial scale exists. SU007, SU008, SU024
CU009 Advanced packaging or heterogeneous-integration programs may matter as buyers if Source Foundry's product boundary extends beyond front-end lithography alone. SU011, SU017, SU018, SU022
CU010 No retained public source identifies a named Source Foundry production customer. SU001, SU002, SU003
CU011 No retained public source identifies a named Source Foundry pilot, evaluation, or prototype customer. SU001, SU002, SU003
CU012 The available funding coverage describes Source Foundry as stealth and does not supplement that profile with customer quotes, deployment sites, or case studies. SU001, SU002, SU003
CU013 Canon publicly named TIE as the destination for its NIL system and described the use case as advanced-semiconductor R&D and prototype production. SU006, SU025
CU014 That Canon-to-TIE disclosure is a useful benchmark because it contains the elements Source Foundry lacks publicly today: named buyer, deployment context, and stage clarity. SU006, SU025, SU001, SU002
CU015 TIE describes itself as a pure-play foundry and highlights advanced interconnect infrastructure, showing that institutional or consortium buyers can be meaningful early adopters in semiconductor infrastructure markets. SU007
CU016 TSMC's University FinFET Program shows that research and academic customers can be served through structured MPW and design-collateral programs before full commercial scale. SU013
CU017 TSMC's OIP, Samsung SAFE, and Intel Foundry Accelerator all present customer adoption as an ecosystem process involving partners, design support, and packaging or cloud integration. SU010, SU011, SU012
CU018 GF, Rapidus, and SkyWater similarly emphasize service models that help customers move from design or development to manufacturing and packaging outcomes. SU015, SU017, SU018, SU019
CU019 TSMC's mask-services page shows that some semiconductor customers buy into a full workflow where masks, quality, and prototyping services matter alongside core fabrication. SU014
CU020 UMC's public reference to customer products and SkyWater's trusted-foundry pitch reinforce that trust, yield, and service are customer requirements, not optional extras. SU016, SU018, SU019
CU021 No public Source Foundry retention metric such as NRR, GRR, churn, or renewal rate is disclosed in retained evidence. SU001, SU002, SU003
CU022 No public Source Foundry customer-count, account-location, or repeat-purchase metric is disclosed in retained evidence. SU001, SU002, SU003
CU023 Because no public installed base is visible, customer durability cannot be underwritten from public evidence alone. SU001, SU002, SU003, SU021
CU024 If the company truly provides installation, maintenance, and repair, then switching costs could become meaningful after qualification, but that remains a hypothesis until real accounts are disclosed. SU005, SU020, SU021
CU025 In semiconductor manufacturing infrastructure, field-service capability and uptime support often matter as much as first purchase, making repeat relationships structurally important if adoption occurs. SU005, SU020, SU021
CU026 Traditional satisfaction metrics are unavailable publicly, so diligence must rely on reference calls, pilot outcomes, and support obligations instead. SU001, SU002, SU020
CU027 A single anchor foundry, pilot-line, or national-program win could materially improve Source Foundry's commercial credibility and future customer funnel. SU006, SU007, SU010, SU011, SU012
CU028 The realistic near-term account universe is highly concentrated, so one delayed or failed qualification could shift revenue timing by years. SU010, SU011, SU012, SU017, SU023
CU029 Buyer concentration would likely remain high even in success because the number of globally relevant leading-edge process operators is limited. SU010, SU011, SU017, SU023
CU030 A packaging-adjacent or integration-focused wedge could expand the buyer map beyond classical front-end lithography owners. SU011, SU017, SU018, SU022
CU031 Public ecosystem programs suggest that customer conversion in this category depends heavily on enablement surfaces that reduce design, packaging, and qualification friction. SU010, SU011, SU012, SU015, SU017, SU018
CU032 If Source Foundry lacks similar ecosystem penetration, commercial conversion may be gated more by qualification throughput than by raw customer interest. SU010, SU011, SU012, SU017, SU018, SU019
CU033 Government-backed programs can accelerate early demand, but they can also make timing more policy-dependent and episodic than purely market-driven. SU007, SU008, SU024
CU034 Export-control and domestic-capability politics shape procurement decisions in semiconductor equipment, adding another layer of customer friction for any new entrant. SU008, SU024
CU035 The best-supported customer verdict is that Source Foundry may face a valuable but tiny buyer universe, with zero public proof yet of named adoption, retention, or diversification. SU001, SU002, SU003, SU004, SU005, SU010, SU011, SU012
CR001 The public trademark corpus shows broad semiconductor, software, mask, and service ambition, but not product-specific commercialization proof. SR001, SR002, SR003, SR004
CR002 The service-mark scope means maintenance and technical support are part of the promised customer experience if tools are ever deployed. SR002, SR004, SR029
CR003 Funding coverage and retained company-specific sources still do not provide a public product page, named product, or exposed quality surface for Source Foundry. SR005, SR006, SR007
CR004 No retained public source identifies a named Source Foundry customer or pilot deployment, leaving commercialization risk unusually high for the reported valuation. SR005, SR006, SR007
CR005 BIS says it updated export controls on semiconductor manufacturing equipment in order to tighten restrictions and close circumvention pathways. SR008
CR006 BIS later said it closed a loophole that had allowed certain foreign-owned fabs in China to export semiconductor manufacturing equipment and technology license-free. SR009
CR007 CSIS describes DUV and EUV equipment as active targets of industrial-policy and export-control competition, confirming that this is a policy-sensitive category. SR010
CR008 TSMC publicly states a zero-defect quality goal, which illustrates the quality bar that fab-facing entrants must ultimately meet. SR011
CR009 TSMC also publishes an ERM policy with risk appetite, governance, monitoring, and supplier-chain elements, highlighting how formalized mature operating control normally is. SR012
CR010 UMC publishes occupational safety, quality, business continuity, information-security, and environmental certifications, which benchmarks the control surface customers may expect. SR014
CR011 SkyWater publishes certifications and trusted accreditations across quality, defense, automotive, medical, environmental, and safety regimes, reinforcing the same benchmark. SR017
CR012 Source Foundry has not published any equivalent public quality, certification, or continuity surface in retained evidence. SR005, SR006, SR007
CR013 The absence of public quality-system evidence creates an operational trust risk independent of whether the core technology works. SR011, SR014, SR017, SR018, SR012
CR014 Canon’s NIL commercialization path demonstrates that alternative lithography architectures can take many years to reach market even when the technology is real. SR019, SR020, SR021
CR015 xLight’s NIST-backed prototype effort shows that even promising lithography-enabling technologies can still be at first-of-kind prototype stage rather than at production maturity. SR022, SR023
CR016 TIE’s role as a pilot and pure-play foundry model shows that institutional proof venues matter, but they are still a bridge to commercialization rather than commercialization itself. SR019, SR020, SR021
CR017 Intel, Samsung, and TSMC all frame customer adoption through ecosystem alliances and enablement, not through simple one-step equipment purchase. SR024, SR025, SR026
CR018 Source Foundry has no comparable public ecosystem or alliance disclosure in retained evidence, creating dependency risk around qualification and customer access. SR005, SR006, SR007
CR019 BIS and CSIS sources imply that supplier choice and customer geography can both become regulatory risk variables in semiconductor-equipment commercialization. SR008, SR009, SR010
CR020 TSMC business-contact materials show how mature foundry businesses expose structured routes for new customer engagement, partner access, and account management. SR013, SR026
CR021 Without a public product or partner disclosure, Source Foundry’s route to first silicon, pilot, or production remains operationally under-specified. SR013, SR024, SR025, SR026, SR027, SR028
CR022 The combination of mask, software, hardware, and service scope implies a complex integration challenge rather than a narrow standalone product risk. SR001, SR002, SR027, SR030
CR023 ASML and Nikon both emphasize hybrid hardware-software systems plus support coverage, underscoring that uptime and service are core operational requirements in this market. SR029, SR030
CR024 If Source Foundry lacks a visible field-support capability, customer adoption may stall even if technical performance is promising. SR002, SR029, SR030
CR025 The likely buyer universe is tiny and high-touch, so ecosystem and pilot dependencies are not marginal—they are likely central to the go-to-market path. SR019, SR024, SR025, SR026, SR028
CR026 Operational transition from prototype to repeatable production is a multi-system risk involving process control, quality, service, and supply chain. SR011, SR012, SR014, SR017, SR021
CR027 A reported $5B valuation raises the cost of execution misses because future rounds may demand much stronger customer and product proof. SR005, SR006, SR007
CR028 Given capital intensity and limited public proof, financing risk is not short-run solvency risk but proof-bar risk for future capital and mark support. SR005, SR006, SR007, SR027, SR028
CR029 A small number of anchor customers is likely to dominate near-term commercial outcomes, making concentration risk severe even in success. SR024, SR025, SR026, SR028
CR030 Diversification into pilot lines, packaging programs, or regional semiconductor initiatives could mitigate concentration, but no public Source Foundry evidence confirms that path yet. SR019, SR020, SR022, SR025, SR026
CR031 Policy-sensitive customers or suppliers can transmit risk directly into commercialization timing, not just into international expansion optionality. SR008, SR009, SR010
CR032 Founder technical signal is a positive input, but it does not by itself resolve manufacturing, support, compliance, or account-management risk. SR005, SR006, SR012, SR014, SR017
CR033 People risk is likely concentrated in manufacturing operations, quality, EHS, field support, and enterprise account ownership rather than only in core R&D. SR011, SR012, SR014, SR015, SR017
CR034 UMC’s enterprise-risk-management and certification surfaces show that industrial execution risk is usually controlled through formal systems rather than heroic effort. SR014, SR015
CR035 TSMC’s customer-claim management, supplier controls, and ERM framework imply that a semiconductor entrant needs operating systems that extend well beyond invention. SR011, SR012, SR013
CR036 If Source Foundry has not yet built those operating systems, execution risk should be treated as a top-tier thesis variable. SR011, SR012, SR014, SR017, SR018
CR037 A reasonable kill trigger is the continued absence of named customer, pilot, or ecosystem proof through the next major milestone window. SR004, SR005, SR006, SR007
CR038 A second kill trigger is failure to show any credible quality, uptime, or service-readiness framework during diligence. SR011, SR014, SR017, SR029, SR030
CR039 A third kill trigger is evidence that export-control or supply-chain rules materially block key inputs or customer routes. SR008, SR009, SR010
CR040 The best-supported overall verdict is that Source Foundry is a high-upside but high-residual-risk semiconductor infrastructure bet whose public proof lags the ambition implied by its funding and scope. SR001, SR002, SR005, SR006, SR007, SR008, SR011, SR017
CV001 Source Foundry received a fresh $400 million follow-on investment from Situational Awareness in early August 2026. SV001, SV002, SV003
CV002 Situational Awareness's total publicly disclosed commitment to Source Foundry is about $500 million. SV001, SV002, SV004
CV003 Multiple public sources peg Source Foundry's latest valuation at roughly $5 billion. SV001, SV002, SV004
CV004 Retained public sources do not disclose ownership sold, liquidation preferences, anti-dilution provisions, or board-right terms for the latest financing. SV001, SV005
CV005 A third-party funding profile reports an earlier $100 million initial investment in Source Foundry around January 2026 with valuation undisclosed. SV004
CV006 Source Foundry remains publicly stealthy, with no retained public source disclosing revenue, product SKU detail, or named customer as part of the financing narrative. SV001, SV002, SV004, SV005
CV007 Public backer framing positions Source Foundry as a challenge to ASML and to the lithography bottleneck upstream of AI chips. SV001, SV004
CV008 Sequoia's public framing emphasizes tooling for semiconductor manufacturing, starting in lithography, as the tightest bottleneck in the AI supply chain. SV001
CV009 Situational Awareness sold most of its public equity portfolio to Citadel after losses but kept its private positions intact. SV001, SV005
CV010 AI Weekly says Situational Awareness' assets under management fell from about $45 billion to about $10 billion after a 67% July loss. SV002, SV005
CV011 The August 2026 Source Foundry mark is best read as a concentrated sponsor conviction signal rather than as broad public price discovery. SV001, SV005
CV012 Source Foundry's public valuation history is shallow: one current ~$5 billion mark and one aggregator-reported earlier financing step rather than a long, independently documented round history. SV003, SV004
CV013 ASML trades at about 16.51x EV/Sales with roughly $40.29 billion of trailing-twelve-month revenue. SV006, SV007
CV014 Applied Materials trades at about 14.25x EV/Sales with roughly $29.02 billion of trailing-twelve-month revenue and $28.37 billion of FY2025 revenue. SV008, SV009
CV015 Lam Research trades at about 16.44x EV/Sales with about $23.23 billion of FY2026 revenue. SV010, SV011
CV016 KLA trades at about 18.64x EV/Sales with about $13.58 billion of FY2026 revenue. SV012, SV013
CV017 Intel trades at about 8.99x EV/Sales on roughly $57.03 billion of trailing revenue, providing a lower-multiple IDM anchor. SV014
CV018 TSMC trades at about 13.15x EV/Sales on roughly $139.57 billion of trailing revenue, providing a scaled foundry anchor. SV015
CV019 The public benchmark band across ASML, Applied Materials, Lam, KLA, Intel, and TSMC spans roughly 9x to 18.6x EV/Sales. SV006, SV008, SV010, SV012, SV014, SV015
CV020 A $5 billion valuation at Intel's 8.99x EV/Sales multiple implies about $556 million of annual revenue. SV014
CV021 A $5 billion valuation at TSMC's 13.15x EV/Sales multiple implies about $380 million of annual revenue. SV015
CV022 A $5 billion valuation at Applied Materials' 14.25x EV/Sales multiple implies about $351 million of annual revenue. SV008
CV023 A $5 billion valuation at ASML's 16.51x EV/Sales multiple implies about $303 million of annual revenue. SV006
CV024 A $5 billion valuation at Lam Research's 16.44x EV/Sales multiple implies about $304 million of annual revenue. SV010
CV025 A $5 billion valuation at KLA's 18.64x EV/Sales multiple implies about $268 million of annual revenue. SV012
CV026 Because Source Foundry discloses no public revenue, margin, backlog, or installed-base economics, public semicap multiples cannot validate the current mark directly. SV001, SV005
CV027 ASML's public investor surface exposes revenue history, valuation metrics, and annual-report access that illustrate how far Source Foundry is from mature priceability. SV006, SV007, SV032
CV028 SEMI, SIA, and IDC all point to a 2026 AI-driven semiconductor expansion that supports the category backdrop for Source Foundry. SV022, SV023, SV024
CV029 Crunchbase says 2026 semiconductor startup funding has already reached about $10.7 billion and is on track to exceed last year's levels. SV025
CV030 Crunchbase identifies Ayar Labs at a $500 million Series E and Etched at a reported $5 billion valuation, showing that strategic bottleneck narratives can command very large private marks in 2026. SV025
CV031 Hadrian's August 2026 Series D raised about $1.37 billion at roughly $7.9 billion, showing investors will pay premium sovereign-manufacturing valuations when operating proof is more public. SV027, SV028
CV032 Helion's January 2025 Series F valued the company at $5.425 billion post-money after named commercial counterparties and prototype progress had become public. SV031
CV033 Pacific Fusion's $900 million Series A was structured as a milestone-tranched commitment rather than fully accessible up-front capital. SV029, SV030
CV034 No retained public Source Foundry source discloses similarly explicit milestone-tranche structure, dilution protection, or financing contingencies for the reported $5 billion mark. SV001, SV005
CV035 A conventional DCF is not supportable from public evidence because Source Foundry discloses no revenue baseline, margin curve, capex schedule, cash burn, or customer conversion assumptions. SV001, SV004, SV005
CV036 The only defensible public-evidence method is scenario analysis tied to proof milestones and the revenue scale implied by benchmark multiples. SV006, SV008, SV012, SV015
CV037 A bear case of about $1.5 billion to $3.0 billion fits a world where Source Foundry proves only a wedge product or prototype without multi-customer qualification before the next financing window. SV025, SV028, SV005
CV038 A base case of about $3.0 billion to $5.0 billion requires credible prototype progress, at least one serious fab or ecosystem validation path, and a believable route toward roughly $200 million to $300 million of revenue over time. SV023, SV024, SV027
CV039 A bull case of about $5.0 billion to $8.0 billion requires Source Foundry to look like a true domestic semicap platform with public customer proof, meaningful manufacturing progress, and a credible path to hundreds of millions of revenue. SV006, SV012, SV027, SV031
CV040 At the current reported $5 billion mark, upside is back-end loaded and depends on proof landing before the next financing event rather than on macro enthusiasm alone. SV025, SV027, SV031
CV041 The price-sensitive recommendation is research-more because the public record does not yet support underwriting a $5 billion entry with conventional valuation tools. SV001, SV005, SV006, SV012
CV042 Confidence should be medium because the financing signal is real but company-specific valuation support remains mostly third-party and incomplete. SV001, SV002, SV004, SV005
CV043 Risk rating should be high because qualification risk, customer opacity, sponsor dependence, and multiple compression can all hit before public revenue appears. SV005, SV022, SV025, SV028
CV044 The valuation stance is expensive because Source Foundry is still stealth and pre-revenue publicly while already priced near late-stage hard-tech platforms with much more public proof. SV001, SV027, SV028, SV031
CV045 The stance would improve if Source Foundry disclosed a clear product boundary, named a pilot or design partner, and showed milestones that make $268 million to $380 million of eventual annual revenue plausible. SV012, SV015, SV001, SV005
CV046 The thesis breaks if the next financing clears below the current mark, if the sponsor base weakens further, or if no public customer or qualification proof appears before another major capital raise. SV001, SV005, SV028
CV047 The highest-priority diligence asks are cap-table terms, product architecture, prototype milestone plan, customer pipeline, and whether the latest $400 million was straightforward primary capital. SV002, SV004, SV005
CV048 The cleanest exit or validation read-throughs are a later round at or above the current mark, a strategic partnership with a U.S. foundry or manufacturing program, or evidence that Source Foundry can become a valuable subsystem supplier even without replacing ASML end-to-end. SV024, SV027, SV031
来源
编号出版方标题引文
SO001 Bizprofile Source Foundry Inc. San Francisco, CA - filing information Source Foundry Inc., a Stock Corporation, is a business entity located in San Francisco, CA. Officially filed on July 18, 2025.
SO002 Yahoo Finance / Investing.com Situational Awareness invested $500 million in chip startup Source Foundry Situational Awareness invested a total of $500 million in semiconductor manufacturing startup Source Foundry.
SO003 Yahoo Finance / Benzinga Situational Awareness Reportedly Bet $400 Million on a $5 Billion Stealth Chip Startup— Weeks After the Most Catastrophic Hedge Fund Blowup of the Year Situational Awareness has invested $500 million in Source Foundry, including a new $400 million infusion this week.
SO004 Quartz Situational Awareness bets $400M on chip startup Source Foundry The AI-focused hedge fund has now committed $500 million to Source Foundry, a Stanford-founded startup aiming to challenge ASML in chip manufacturing equipment.
SO005 MLQ Situational Awareness puts $400 million into stealth chip-manufacturing startup Source Foundry Public reporting does not establish whether the financing is equity, debt or a committed facility, and does not disclose ownership, liquidation preferences or other terms.
SO006 TechCrunch Embattled hedge fund Situational Awareness invests $400M in chip startup Source Foundry This week, the fund invested $400 million into Source Foundry... That brings its total investment in Source Foundry to $500 million.
SO007 The Daily Hodl Hedge Fund Situational Awareness Commits $500,000,000 to Source Foundry Chip Startup If you trace the AI supply chain upstream... Source Foundry tackles the tightest bottleneck: tooling for semiconductor manufacturing, starting in lithography.
SO008 Creati.ai Source Foundry Reportedly Raises $400 Million at $5 Billion Valuation in ASML Challenge Source Foundry reportedly raises $400 million at $5 billion valuation in ASML challenge.
SO009 Stanford Bio-X Abdulmalik Obaid - Bruce and Elizabeth Dunlevie Fellow Home Department: Materials Science & Engineering... bringing together materials science and microfabrication techniques with electrophysiology.
SO010 Stanford Wu Tsai Neurosciences Institute Stanford device brings silicon computing power to brain research and prosthetics We had to throw out what we already know about conventional chip fabrication and design new processes to bring silicon electronics into the third dimension.
SO011 Stanford Dauskardt Group Joe Burg My research interests are to develop computational methods to address the fundamental relationships between molecular structure and resulting mechanical properties of organosilicate glasses.
SO012 Justia Trademarks Source Foundry Inc. Trademarks Page 1 Source Foundry Inc. Trademarks... Machines for manufacturing semiconductors... Installation, maintenance, and repair services... Custom manufacturing... Providing training... Research and design...
SO013 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682288 Semiconductors; semiconductor chips; semiconductor wafers; downloadable computer software... for the development of advanced semiconductor lithography; lithographic masks and reticles.
SO014 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682289 Installation, maintenance, and repair services for semiconductor manufacturing machines and lithography systems.
SO015 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Trademark Registration Research and development of advanced lithography technology for semiconductor manufacturing; providing online non-downloadable computer software... custom design and engineering of semiconductor facilities.
SO016 Benzinga Situational Awareness Invests $400 Million in Source Foundry Source Foundry was founded last year by Stanford researchers Abdulmalik Obaid and Joe Burg. The startup has also previously received backing from Sequoia Capital.
SO017 ASML ASML 2025 Annual Report Total net sales €32.7bn... Our holistic approach to lithography.
SO018 ASML High NA EUV reaches new readiness milestone with first high-volume Logic product Intel Foundry has entered high-volume manufacturing... using ASML’s EXE High NA EUV technology.
SO019 Intel Intel 18A | See Our Biggest Process Innovation Intel 18A is ready for customer projects... In high-volume production in the United States.
SO020 Intel Newsroom Intel Foundry Details Process Milestones and Future Innovation at VLSI Symposium Intel 18A, Intel Foundry’s leading-edge process node, entered into production in 2025.
SO021 Canon Nanoimprint lithography semiconductor manufacturing system that covers diverse applications with simple patterning mechanism Canon's NIL technology enables patterning with a minimum linewidth of 14 nm.
SO022 Canon U.S.A. Canon Inc. Delivers FPA -1200NZ2C Nanoimprint Lithography System for Semiconductor Manufacturing to the Texas Institute for Electronics Canon became the first in the world to commercialize a semiconductor manufacturing system that uses NIL technology.
SO023 Nikon Semiconductor Lithography Systems | Nikon Business Introduction to the lithography systems, alignment stations, and metrology/inspection systems.
SO024 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography The U.S. Department of Commerce... announced today the signing of a final award of $150 million... to xLight, Inc.
SO025 Business Wire xLight Finalizes $150M CHIPS Incentives with U.S. Department of Commerce xLight’s FEL systems are built outside the fab and deliver significantly more power to up to 16 scanners at once.
SM001 SEMI Global Semiconductor Equipment Sales Forecast to Reach a Record $229 Billion in 2028, SEMI Reports Global sales of total semiconductor manufacturing equipment... are forecast to reach a record high of $165.9 billion in 2026.
SM002 Semiconductor Industry Association Market Data Global semiconductor sales were $403.3 billion during the second quarter of 2026.
SM003 IDC Semiconductor Market to Surge Past the Trillion-Dollar Threshold: AI Infrastructure Drives Market Growth Semiconductor Market to Surge Past the Trillion-Dollar Threshold: AI Infrastructure Drives Market Growth.
SM004 ASML ASML Annual reports As we are committed to transparency, we provide access to annual reports from 1995 onwards.
SM005 ASML Products Our lithography systems, metrology and inspection, and software solutions help chipmakers produce faster, cheaper, more energy-efficient chips.
SM006 Cadence / Samsung Foundry Cadence and Samsung Foundry Deepen 2nm and 3D‑IC Collaboration to Meet Surging AI Infrastructure and Physical AI Demand Customers are increasingly drawn to Samsung Foundry’s second-generation 2nm for leading-edge AI designs.
SM007 Samsung Semiconductor Foundry Overview Samsung Foundry offers competitive processes... and high-volume manufacturing capability for customers.
SM008 TSMC TSMC Files Annual Report on Form 20-F for 2025 TSMC deployed 305 distinct process technologies, and manufactured 12,682 products for 534 customers in 2025.
SM009 Mordor Intelligence Semiconductor Lithography Equipment Market Analysis | Industry Growth, Size & Forecast Report The semiconductor lithography equipment market size is projected to expand... to USD 30.44 billion in 2026.
SM010 IEEE Spectrum Canon Delivers Nanoimprint Lithography to Compete With EUV The NIL system offers advantages that may challenge the US $150 million machines that dominate today’s advanced chipmaking.
SM011 Intel Intel 18A | See Our Biggest Process Innovation Intel 18A is ready for customer projects.
SM012 Intel Newsroom Intel Foundry Details Process Milestones and Future Innovation at VLSI Symposium Intel 18A‑P, the first performance enhancement in the Intel 18A family, has entered risk production.
SM013 ASML High NA EUV reaches new readiness milestone with first high-volume Logic product Intel Foundry is using ASML’s High NA EUV technology on the Intel 18A process node.
SM014 Business Wire xLight Finalizes $150M CHIPS Incentives with U.S. Department of Commerce xLight’s FEL systems are built outside the fab and deliver significantly more power to up to 16 scanners at once.
SM015 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography These incentives are for the construction and demonstration of a first-of-its-kind free-electron laser prototype.
SM016 ASML ASML 2025 Annual Report Total net sales €32.7bn.
SM017 Canon Nanoimprint lithography semiconductor manufacturing system that covers diverse applications with simple patterning mechanism Canon’s NIL technology enables patterning with a minimum linewidth of 14 nm.
SM018 Canon U.S.A. Canon Inc. Delivers FPA -1200NZ2C Nanoimprint Lithography System for Semiconductor Manufacturing to the Texas Institute for Electronics The FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SM019 Nikon Semiconductor Lithography Systems | Nikon Business Introduction to the lithography systems, alignment stations, and metrology/inspection systems.
SM020 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682288 downloadable computer software for use in connection with processing semiconductor wafers
SM021 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Trademark Registration research and development of advanced lithography technology for semiconductor manufacturing
SM022 Yahoo Finance / Investing.com Situational Awareness invested $500 million in chip startup Source Foundry AI demand is growing on an exponential software curve, while semiconductor manufacturing capacity is growing on a linear industrial-equipment curve.
SM023 MLQ Situational Awareness puts $400 million into stealth chip-manufacturing startup Source Foundry Source Foundry’s trademark filings describe lithography machines, semiconductor-processing equipment, wafer software and facility-engineering services.
SM024 Yahoo Finance / Benzinga Situational Awareness Reportedly Bet $400 Million on a $5 Billion Stealth Chip Startup ASML’s machines... can cost more than $400 million each.
SM025 Benzinga Situational Awareness Invests $400 Million in Source Foundry ASML’s machines are the size of a school bus... and can cost more than $400 million each.
SP001 ASML ASML 2025 Annual Report Total net sales €32.7bn.
SP002 ASML ASML EUV lithography systems Both types of technology will be required in parallel for many years to come.
SP003 ASML See ASML's DUV lithography systems Our latest NXT machines have shown the ability to run in excess of 6,000 wafers per day.
SP004 ASML Our technology - Supplying the semiconductor industry Our lithography systems are now a hybrid of high-tech hardware and advanced software.
SP005 ASML High NA EUV reaches new readiness milestone with first high-volume Logic product Intel Foundry is using ASML’s High NA EUV technology on the Intel 18A process node.
SP006 Nikon Precision Nikon Precision lithography equipment sales, service and support Nikon Precision semiconductor manufacturing products include DUV & immersion scanners, i-line steppers, & steppers for MEMS, LED & packaging.
SP007 Nikon Semiconductor Lithography Systems | Nikon Business Introduction to the lithography systems, alignment stations, and metrology/inspection systems.
SP008 Nikon Precision Nikon Announces First Back-End Process Lithography System The DSP-100 uses a spatial light modulator (SLM) to directly project circuit patterns onto substrates without the need for a photomask.
SP009 Canon Nanoimprint lithography semiconductor manufacturing system that covers diverse applications with simple patterning mechanism Canon's NIL technology enables patterning with a minimum linewidth of 14 nm.
SP010 Canon U.S.A. Canon Inc. Delivers FPA-1200NZ2C Nanoimprint Lithography System for Semiconductor Manufacturing to the Texas Institute for Electronics The FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SP011 IEEE Spectrum Canon Delivers Nanoimprint Lithography to Compete With EUV The NIL system offers advantages that may challenge the US $150 million machines that dominate today’s advanced chipmaking.
SP012 Business Wire xLight Finalizes $150M CHIPS Incentives with U.S. Department of Commerce xLight’s FEL systems are built outside the fab and deliver significantly more power to up to 16 scanners at once.
SP013 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography These incentives are for the construction and demonstration of a first-of-its-kind free-electron laser prototype.
SP014 xLight xLight: World's Most Powerful Lasers to Revive Moore's Law 4X more EUV power, 50% lower costs, better chip quality.
SP015 EV Group Lithography EVG’s key competencies in lithographic technology lie in the high-throughput contact and proximity exposure capabilities of its mask aligners and its newly developed... maskless exposure lithography systems.
SP016 Intel Semiconductor Manufacturing Process | Intel 14A, 18A, and 3 See how Intel is enabling tomorrow's cutting-edge semiconductor manufacturing process technologies to deliver a systems foundry for the AI era.
SP017 Intel Intel 18A | See Our Biggest Process Innovation Intel 18A is ready for customer projects.
SP018 Samsung Semiconductor Process Technology | Foundry | Samsung Semiconductor Global As a dedicated foundry that has pioneered many industry milestones—from HKMG and FinFET to EUV and GAA—we empower customers to push innovation further.
SP019 Samsung Semiconductor Foundry Overview | Samsung Semiconductor Global Samsung Foundry offers competitive processes, design technologies, IP, and high-volume manufacturing capability for customers.
SP020 TSMC Fab Capacity Annual capacity... exceeded 17 million 12-inch equivalent wafers in 2025.
SP021 TSMC Advanced Packaging Services TSMC advanced packaging services create the best solutions... offering integrated turnkey services.
SP022 CISES The Expiring Chokepoint: EUV Lithography and the Coming Test of Western Technological Leverage The chokepoint is not a company but instead, a coalition.
SP023 Bloomberg US Tells ASML China May Have Top Chip Tool US Commerce Secretary Howard Lutnick warned its leaders that one of the company’s advanced lithography machines may have reached China.
SP024 Benzinga Situational Awareness Invests $400 Million in Source Foundry Source Foundry, a chipmaking startup reportedly valued at $5 billion, has secured fresh investment...
SP025 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682289 Installation, maintenance, and repair services for semiconductor manufacturing machines and lithography systems.
SP026 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682288 downloadable computer software for use in connection with processing semiconductor wafers
SI001 Benzinga Situational Awareness Invests $400 Million in Source Foundry Situational Awareness has invested $500 million in Source Foundry, including a new $400 million infusion this week.
SI002 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682288 downloadable computer software for use in connection with processing semiconductor wafers
SI003 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682289 Installation, maintenance, and repair services for semiconductor manufacturing machines and lithography systems.
SI004 TechCrunch AI hedge fund Situational Awareness may have sold its public portfolio, but it still has its Anthropic shares Situational Awareness has sold the majority of its public stock portfolio to Ken Griffin’s Citadel following steep losses over the past month.
SI005 Yahoo Finance / Reuters Citadel buys most of Situational's stock holdings after AI share rout, sources say The fund was down around 67% so far in July after incurring heavy losses on AI stocks.
SI006 ASML ASML 2025 Annual Report Total net sales €32.7bn.
SI007 ASML Financial Results We release complete financial results each quarter as well as a full annual report at the end of the year.
SI008 ASML Q1 2026 financial results €8.8bn Total net sales; 53.0% Gross margin; €1.2bn R&D costs.
SI009 ASML Q2 2026 financial results €9.3bn Total net sales; 54.0% Gross margin; €1.3bn R&D costs.
SI010 Intel 10-K - 01/23/2026 - Intel Corporation The significant, long-term and inherently risky investments we are making in R&D and manufacturing facilities may not realize a favorable return.
SI011 Intel Newsroom Intel Reports Second-Quarter 2026 Financial Results Intel’s second-quarter 2026 earnings news release and presentation are available on the company’s Investor Relations website.
SI012 Intel Semiconductor Manufacturing Process | Intel 14A, 18A, and 3 See how Intel is enabling tomorrow's cutting-edge semiconductor manufacturing process technologies to deliver a systems foundry for the AI era.
SI013 Intel Intel 18A | See Our Biggest Process Innovation Intel 18A is ready for customer projects.
SI014 TSMC Annual Reports - Taiwan Semiconductor Manufacturing Company Limited TSMC Annual Report.
SI015 TSMC Financial Results - 2026Q2 Financial Results -2026Q2.
SI016 TSMC TSMC Files Annual Report on Form 20-F for 2025 TSMC deployed 305 distinct process technologies, and manufactured 12,682 products for 534 customers in 2025.
SI017 TSMC Logic Technology TSMC’s 2nm (N2) technology has started volume production in 4Q25 as planned.
SI018 TSMC Advanced Packaging Services TSMC advanced packaging services create the best solutions... offering integrated turnkey services.
SI019 TSMC Fab Capacity Annual capacity... exceeded 17 million 12-inch equivalent wafers in 2025.
SI020 Samsung Global IR Earnings Releases | Financial Information | Investor Relations | Samsung Global We provide you with the main data release material of Samsung Electronics.
SI021 Samsung Semiconductor Foundry Overview | Samsung Semiconductor Global Samsung Foundry offers competitive processes, design technologies, IP, and high-volume manufacturing capability for customers.
SI022 Samsung Semiconductor Process Technology | Foundry | Samsung Semiconductor Global As a dedicated foundry that has pioneered many industry milestones—from HKMG and FinFET to EUV and GAA...
SI023 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography These incentives are for the construction and demonstration of a first-of-its-kind free-electron laser prototype.
SI024 Business Wire xLight Finalizes $150M CHIPS Incentives with U.S. Department of Commerce xLight helps unlock manufacturing capacity and drive down cost by transforming how light is delivered to the fab: from an individual tool feature to a utility-scale service.
SI025 Nikon Precision Nikon Announces First Back-End Process Lithography System The system offers high-precision correction for substrate warpage and deformation, reducing production costs with maskless technology and minimizing maintenance costs with solid state light sources.
SE001 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Trademark Registration Trademark registration is intended to cover ... software for the development of advanced semiconductor lithography; lithographic masks and reticles.
SE002 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Service Mark Trademark registration is intended to cover ... installation, maintenance, and repair services for semiconductor manufacturing machines and lithography systems.
SE003 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682288 downloadable computer software for use in connection with processing semiconductor wafers
SE004 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682289 technical support services for semiconductor manufacturing equipment
SE005 ASML ASML Annual reports As we are committed to transparency, we provide access to annual reports from 1995 onwards.
SE006 Wu Tsai Neurosciences Institute / Stanford Stanford device brings silicon computing power to brain research and prosthetics We had to throw out what we already know about conventional chip fabrication and design new processes to bring silicon electronics into the third dimension.
SE007 Canon Canon completes acquisition of Molecular Imprints Canon aims to realize an early market entry through the accelerated development of next-generation semiconductor lithography systems.
SE008 Canon Canon delivers FPA-1200NZ2C nanoimprint lithography system to the Texas Institute for Electronics This FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SE009 Canon Nanoimprint lithography semiconductor manufacturing system that covers diverse applications with simple patterning mechanism Canon's NIL technology enables patterning with a minimum linewidth of 14 nm.
SE010 IEEE Spectrum Canon Delivers Nanoimprint Lithography to Compete With EUV Yet even with the addition of MII ... it has taken 20 years to bring the technology to market.
SE011 ASML Our technology - Supplying the semiconductor industry Our lithography systems are now a hybrid of high-tech hardware and advanced software.
SE012 ASML ASML EUV lithography systems Both types of technology will be required in parallel for many years to come.
SE013 ASML See ASML's DUV lithography systems In a given chip, there may be one or two more complicated layers that are made using an EUV lithography machine, but the rest can often be printed using older technology such as dry lithography systems.
SE014 Nikon Semiconductor Lithography Systems | Nikon Business Introduction to the lithography systems, alignment stations, and metrology/inspection systems.
SE015 Nikon Precision Nikon Announces First Back-End Process Lithography System The DSP-100 uses a spatial light modulator (SLM) to directly project circuit patterns onto substrates without the need for a photomask.
SE016 Nikon Precision Nikon Precision lithography equipment sales, service and support Nikon Precision semiconductor manufacturing products include DUV & immersion scanners ... and packaging.
SE017 xLight xLight: World's Most Powerful Lasers to Revive Moore's Law xLight is building the world’s most powerful Free Electron Lasers (FELs) to replace LPP.
SE018 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography These incentives are for the construction and demonstration of a first-of-its-kind free-electron laser prototype.
SE019 TSMC 3DFabric | TSMC Packaging technologies ... are critical to a product's performance, function, and cost.
SE020 TSMC Advanced Packaging Services TSMC advanced packaging services create the best solutions to unleash our customer's innovations by advancing the core technology, offering integrated turnkey services.
SE021 CSIS The True Impact of Allied Export Controls on the U.S. and Chinese Semiconductor Manufacturing Equipment Industries Domestic production of immersion deep ultraviolet (DUV) lithography machines before 2025; domestic production of extreme ultraviolet (EUV) lithography equipment before 2030.
SE022 CISES The Expiring Chokepoint: EUV Lithography and the Coming Test of Western Technological Leverage The chokepoint is not a company but instead, a coalition.
SE023 Benzinga Situational Awareness Invests $400 Million in Source Foundry Source Foundry ... seeks to reinvent semiconductor manufacturing.
SE024 ASML ASML 2025 Annual Report Our holistic approach to lithography.
SE025 Samsung Semiconductor Process Technology | Foundry | Samsung Semiconductor Global As a dedicated foundry that has pioneered many industry milestones—from HKMG and FinFET to EUV and GAA...
SE026 Intel Intel 18A | See Our Biggest Process Innovation Intel 18A-PT ... purpose built for advanced 3DIC designs.
SE027 Stanford Bio-X Abdulmalik Obaid - Bruce and Elizabeth Dunlevie Fellow Abdulmalik proposes bringing together materials science and microfabrication techniques with electrophysiology ... Abdulmalik is working at a stealth startup developing microelectronic devices in the Bay Area.
SE028 Science Advances Massively parallel microwire arrays integrated with CMOS chips for neural recording We present a new strategy to interface silicon-based chips with three-dimensional microwire arrays, providing the link between rapidly-developing electronics and high density neural interfaces.
SE029 PubMed Central Massively parallel microwire arrays integrated with CMOS chips for neural recording We present a new strategy to interface silicon-based chips with three-dimensional microwire arrays ... The modular design enables a variety of microwire types and sizes to be integrated with different types of pixel arrays.
SU001 Benzinga Situational Awareness Invests $400 Million in Source Foundry Source Foundry ... seeks to reinvent semiconductor manufacturing.
SU002 MLQ.ai Situational Awareness puts $400 million into stealth chip-manufacturing startup Source Foundry Source Foundry ... has yet to launch a public website or product.
SU003 Yahoo Finance Situational Awareness reportedly bet $400 million on chip startup Source Foundry The company is taking aim at a supply-chain chokepoint dominated by ASML.
SU004 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Trademark Registration Trademark registration is intended to cover ... semiconductor wafers ... software ... lithographic masks and reticles.
SU005 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Service Mark Trademark registration is intended to cover ... installation, maintenance, and repair services for semiconductor manufacturing machines and lithography systems.
SU006 Canon U.S.A. Canon Inc. Delivers FPA -1200NZ2C Nanoimprint Lithography System for Semiconductor Manufacturing to the Texas Institute for Electronics The FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SU007 Texas Institute for Electronics Texas Institute for Electronics TIE is a pure-play foundry ... Our advanced mixed-material interconnect technology and standardized EDA flows unlock up to 100x improvements in microsystem performance.
SU008 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography These incentives are for the construction and demonstration of a first-of-its-kind free-electron laser prototype.
SU009 xLight xLight: World's Most Powerful Lasers to Revive Moore's Law xLight is building the world’s most powerful Free Electron Lasers (FELs) to replace LPP.
SU010 Intel Semiconductor Manufacturing Company for the AI Era Intel Foundry Accelerator Ecosystem Alliances ... comprehensive support in key areas from ecosystem partners.
SU011 Samsung Semiconductor SAFE™ Program | Foundry | Samsung Semiconductor Global Samsung’s SAFE™ program fosters seamless collaboration between Samsung Foundry, ecosystem partners, and customers.
SU012 TSMC Open Innovation Platform Overview OIP ... lower design barriers, improve design cycle times, and accelerate first-time silicon success.
SU013 TSMC TSMC University FinFET Program The program offers ... MPW services and design collateral ... for research projects leading to silicon test chips.
SU014 TSMC Mask Services TSMC's mask service is the interface between designers and our world class fabs.
SU015 GlobalFoundries Manufacturing services | GlobalFoundries Our manufacturing services are designed to seamlessly get you from chip design to wafer fabrication to final assembly and test.
SU016 UMC Technologies - UMC UMC became the first foundry in the world to deliver 65nm customer products.
SU017 Rapidus Rapidus Corporation | World's Most Advanced 2nm Semiconductor Rapidus will create new industries together with our customers through the development and provision of services to shorten cycle times in design, wafer processes, 3D packaging and more.
SU018 SkyWater Technology SkyWater Technology: U.S. Semiconductor Foundry Company SkyWater’s TaaS model streamlines the path to production for customers with development services, volume production and heterogeneous integration solutions.
SU019 SkyWater Technology Services Page - Skywater Technology TaaS is built up of 2 main components. Advanced Technology Services (ATS). Wafer Services (WS).
SU020 Nikon Precision Nikon Precision lithography equipment sales, service and support Nikon Precision semiconductor manufacturing products include DUV & immersion scanners ... and packaging.
SU021 ASML Our technology - Supplying the semiconductor industry Our lithography systems are now a hybrid of high-tech hardware and advanced software.
SU022 TSMC 3DFabric | TSMC Packaging technologies ... are critical to a product's performance, function, and cost.
SU023 Samsung Semiconductor Process Technology | Foundry | Samsung Semiconductor Global As a dedicated foundry that has pioneered many industry milestones—from HKMG and FinFET to EUV and GAA...
SU024 CSIS The True Impact of Allied Export Controls on the U.S. and Chinese Semiconductor Manufacturing Equipment Industries Domestic production of immersion deep ultraviolet (DUV) lithography machines before 2025; domestic production of extreme ultraviolet (EUV) lithography equipment before 2030.
SU025 Canon Canon delivers FPA-1200NZ2C nanoimprint lithography system for semiconductor manufacturing to the Texas Institute for Electronics The FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SR001 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Trademark Registration Trademark registration is intended to cover ... software ... lithographic masks and reticles.
SR002 USPTO.report SOURCE FOUNDRY - Source Foundry Inc. Service Mark Trademark registration is intended to cover ... installation, maintenance, and repair services ...
SR003 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682288 downloadable computer software for use in connection with processing semiconductor wafers
SR004 Justia Trademarks SOURCE FOUNDRY Trademark Application - Serial Number 99682289 technical support services for semiconductor manufacturing equipment
SR005 Benzinga Situational Awareness Invests $400 Million in Source Foundry Source Foundry ... seeks to reinvent semiconductor manufacturing.
SR006 MLQ.ai Situational Awareness puts $400 million into stealth chip-manufacturing startup Source Foundry Source Foundry ... has yet to launch a public website or product.
SR007 Yahoo Finance Situational Awareness reportedly bet $400 million on chip startup Source Foundry The company is taking aim at a supply-chain chokepoint dominated by ASML.
SR008 BIS Commerce Strengthens Restrictions on Advanced Computing Semiconductors and Semiconductor Manufacturing Equipment BIS released a package of rules ... on advanced computing semiconductors and semiconductor manufacturing equipment.
SR009 BIS Department of Commerce closes export controls loophole for foreign-owned semiconductor fabs in China BIS closed a loophole that allowed a handful of foreign companies to export semiconductor manufacturing equipment and technology to China license-free.
SR010 CSIS The True Impact of Allied Export Controls on the U.S. and Chinese Semiconductor Manufacturing Equipment Industries Domestic production of immersion deep ultraviolet (DUV) lithography machines before 2025; domestic production of extreme ultraviolet (EUV) lithography equipment before 2030.
SR011 TSMC Quality Policy There is only one ultimate goal: Zero defect - in everything we do.
SR012 TSMC Risk Management Policy TSMC is committed to maintaining a proactive and robust Enterprise Risk Management (ERM) system.
SR013 TSMC Business Contacts TSMC is a dedicated contract semiconductor manufacturer that provides wafer fabrication and advance packaging solutions.
SR014 UMC Awards and Certificates - UMC ISO 45001 Occupational Health and Safety Management System.
SR015 UMC Risk Management System - UMC robust risk management and appropriate crisis management are the major cornerstones to ensure sustainable operation.
SR016 UMC Technologies - UMC UMC became the first foundry in the world to deliver 65nm customer products.
SR017 SkyWater Technology Certification & Policies | SkyWater Technology SkyWater’s commitment to quality is backed by internationally recognized certifications.
SR018 SkyWater Technology SkyWater Technology: U.S. Semiconductor Foundry Company We protect our customers’ IP and we build trust by being dependable.
SR019 Canon U.S.A. Canon Inc. Delivers FPA -1200NZ2C Nanoimprint Lithography System for Semiconductor Manufacturing to the Texas Institute for Electronics The FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SR020 Canon Canon delivers FPA-1200NZ2C nanoimprint lithography system for semiconductor manufacturing to the Texas Institute for Electronics The FPA-1200NZ2C will be used at TIE for the research and development of advanced semiconductors and production of prototypes.
SR021 IEEE Spectrum Canon Delivers Nanoimprint Lithography to Compete With EUV it has taken 20 years to bring the technology to market
SR022 NIST Department of Commerce Announces Finalization of CHIPS Incentives with xLight to Support Next-Generation Light Source for Lithography construction and demonstration of a first-of-its-kind free-electron laser prototype
SR023 xLight xLight: World's Most Powerful Lasers to Revive Moore's Law xLight is building the world’s most powerful Free Electron Lasers (FELs) to replace LPP.
SR024 Intel Semiconductor Manufacturing Company for the AI Era Intel Foundry Accelerator Ecosystem Alliances
SR025 Samsung Semiconductor SAFE™ Program | Foundry | Samsung Semiconductor Global SAFE™ program fosters seamless collaboration between Samsung Foundry, ecosystem partners, and customers.
SR026 TSMC Open Innovation Platform Overview lower design barriers, improve design cycle times, and accelerate first-time silicon success
SR027 GlobalFoundries Manufacturing services | GlobalFoundries from chip design to wafer fabrication to final assembly and test
SR028 Rapidus Rapidus Corporation | World's Most Advanced 2nm Semiconductor services to shorten cycle times in design, wafer processes, 3D packaging and more
SR029 Nikon Precision Nikon Precision lithography equipment sales, service and support products include DUV & immersion scanners ... and packaging
SR030 ASML Our technology - Supplying the semiconductor industry Our lithography systems are now a hybrid of high-tech hardware and advanced software.
SV001 Yahoo Finance Situational Awareness bets $400M on chip startup Source Foundry The investment brings Situational Awareness's total stake in Source Foundry to $500 million.
SV002 AI Weekly Aschenbrenner Pours $400M Into ASML Challenger Source Foundry Source Foundry, incorporated in California in July 2025 by Stanford alumni Abdulmalik Obaid and Joe Burg, is now valued at $5 billion.
SV003 Best AI Situational Awareness Fund Invests $400M in Stealth Chip Startup Source Foundry Amidst Halved AUM
SV004 toflow.ai What Funding Has Source Foundry Raised? Full Round History & Key Investors Source Foundry received an initial $100 million investment from hedge fund Situational Awareness followed by a $400 million follow-on, bringing total known investment to $500 million at an approximately $5 billion valuation.
SV005 TechCrunch AI hedge fund Situational Awareness may have sold its public portfolio, but it still has its Anthropic shares After Citadel bought the bulk of those holdings, Situational Awareness’ overall assets fell to roughly $10 billion.
SV006 StockAnalysis ASML Holding (ASML) Statistics & Valuation
SV007 StockAnalysis ASML Holding (ASML) Revenue 2005-2026
SV008 StockAnalysis Applied Materials (AMAT) Statistics & Valuation
SV009 StockAnalysis Applied Materials (AMAT) Revenue 2005-2026
SV010 StockAnalysis Lam Research (LRCX) Statistics & Valuation
SV011 StockAnalysis Lam Research (LRCX) Revenue 2005-2026
SV012 StockAnalysis KLA Corporation (KLAC) Statistics & Valuation
SV013 StockAnalysis KLA Corporation (KLAC) Revenue 2005-2026
SV014 StockAnalysis Intel (INTC) Statistics & Valuation
SV015 StockAnalysis Taiwan Semiconductor Manufacturing Company (TSM) Statistics & Valuation
SV016 U.S. Securities and Exchange Commission ASML Holding NV SEC XBRL companyfacts
SV017 U.S. Securities and Exchange Commission Applied Materials SEC XBRL companyfacts
SV018 U.S. Securities and Exchange Commission Lam Research SEC XBRL companyfacts
SV019 U.S. Securities and Exchange Commission KLA SEC XBRL companyfacts
SV020 U.S. Securities and Exchange Commission Intel SEC XBRL companyfacts
SV021 U.S. Securities and Exchange Commission Taiwan Semiconductor SEC XBRL companyfacts
SV022 SEMI Global Semiconductor Equipment Sales Forecast to Reach a Record $229 Billion in 2028, SEMI Reports Global sales of total semiconductor manufacturing equipment are forecast to reach a record high of $165.9 billion in 2026.
SV023 Semiconductor Industry Association Market Data Global semiconductor sales were $403.3 billion during the second quarter of 2026.
SV024 IDC Semiconductor Market to Surge Past the Trillion-Dollar Threshold: AI Infrastructure Drives Market Growth Semiconductor Market to Surge Past the Trillion-Dollar Threshold: AI Infrastructure Drives Market Growth.
SV025 Crunchbase News Sector Snapshot: Semiconductor Startup Funding Still Running Hot So far in 2026, investors have poured around $10.7 billion into seed through pre-IPO rounds for companies in Crunchbase’s semiconductor category.
SV026 U.S. Securities and Exchange Commission Hadrian Automation Series C Form D
SV027 PR Newswire / Hadrian Hadrian Raises $1.37B Series D to Build Highly Automated Factories to Accelerate America's Industrial Renewal The raise values Hadrian at $7.87 billion.
SV028 CNBC Hadrian valued at nearly $8 billion after fresh funding as money pours into defense tech Hadrian said on Thursday that it has closed its latest round of funding at a $7.9 billion valuation.
SV029 POWER Magazine Bill Gates' Energy Group, Former Google CEO Investing in Pacific Fusion Startup The Series A funding is a phased approach. The entire $900 million is committed, but only unlocked as the company achieves pre-defined milestones.
SV030 TechCrunch Exclusive: Here’s how Pacific Fusion plans to build a fusion power plant The funding round, while massive by Series A standards, isn’t accessible all at once. It’s built to pay out serially as the company hits certain milestones.
SV031 Helion Helion Announces $425M Series F Investment to Scale Commercialized Fusion Power This latest round of funding will bring the total invested in Helion to over $1 billion and values the company at $5.425 billion post-money.
SV032 ASML ASML Annual reports As we are committed to transparency, we provide access to annual reports from 1995 onwards.