Commonwealth Fusion Systems
融资最多的聚变创业公司:SPARC、ARC 与商业聚变电力的漫长路径
CFS 是全球质量最高、融资最可信的私营聚变公司之一,拥有全球最先进、以 Q>1 为目标的 tokamak(SPARC),签下首批企业聚变 PPA(Google + Eni),并有 Breakthrough Energy Ventures、Khosla、Eni 领投的 $3.2B 投资人基础。SPARC 首等离子体(2026)和 Q>1(2027)是下一组二元里程碑,决定 CFS 能否补上 ARC 建设所需的 $3–5B 资本缺口。按估计 $5–8B 估值,市场计入约 65% 的 SPARC Q>1 概率;最被低估的主要风险是氚增殖(全球 TRL 2–3),即便 SPARC 成功,也可能推迟 ARC 商业运营。对深科技能源转型基金,6–10 年周期下可建设性看待。
封面要素
公司概况
Commonwealth Fusion Systems(CFS)是一家聚变能源公司,总部位于马萨诸塞州 Cambridge / Devens。公司 2018 年从 MIT PSFC 孵化,由 CEO Bob Mumgaard、等离子体物理学家 Dennis Whyte(现任首席科学官)和 Brandon Sorbom 创立。CFS 的核心创新,是用 REBCO(稀土钡铜氧化物)高温超导(HTS)磁体在 12+ Tesla 下运行——这是商业验证过的最高磁场——从而做出紧凑型高场托卡马克 SPARC,以 ITER 1/60 的体积实现聚变能量增益(Q>1)。公司已从 Breakthrough Energy Ventures、Khosla Ventures、Eni S.p.A.、Google/Alphabet、Temasek、Tiger Global 等投资方融资约 $3.2B。2024 年,SPARC 创下离子等离子体温度世界纪录(100 million°C)。2025 年,CFS 与 Google(200 MW)和 Eni($1B+)签下全球首批企业聚变 PPA,二者都以 ARC 商业运行为前提。SPARC 预计 2026 年产生第一等离子体;Q>1 演示目标为 2027 年。ARC 是 CFS 规划中的 400 MWe 商业电站,将落地 Virginia 的 Chesterfield County,目标 LCOE 为 $50–70/MWh,计划 2030 年代初首次发电。
- 成立时间
- 2018-01-01
- 创始人
- Bob Mumgaard, Dennis Whyte, Brandon Sorbom
- 创立地点
- Cambridge, MA
- 总部
- Devens, MA (main R&D and SPARC facility)
- 产品
- CFS 的产品包括:(1)SPARC:高场紧凑型托卡马克,采用 1.85m 主半径和 12T REBCO HTS 磁体,目标实现聚变能量增益 Q>1;第一等离子体目标 2026 年;Q>1 目标 2027 年。(2)ARC:400 MWe 净输出的商业聚变电站,沿用并放大同一套 REBCO HTS 磁体技术;LCOE 目标 $50–70/MWh;计划落地 Virginia 的 Chesterfield County;目标在 2030 年代初首次发电。CFS 还在 Devens Materials Facility 制造 REBCO HTS 导体,并持有 50+ 项磁体和等离子体技术专利。
- 客户
- 商业化前;已与 Google(200 MW)和 Eni(总额 $1B+)签署远期 PPA;目标客户是寻求 24/7 稳定清洁电力的超大规模数据中心运营商、电力公用事业公司和工业能源买家
- 商业模式
- 发电业务(基于 PPA 的承购);早期收入来自 DOE 里程碑拨款;商业化前
- 阶段
- Late venture / growth pre-commercial
- 融资情况
- 累计融资 $3.2B;种子轮 $50M(2018)、Series A $115M(2020)、Series B $1.8B(2021)、Series B2 $863M(2025 年 8 月);隐含 B2 投后估值约 $5–8B(估计);投资方包括 Breakthrough Energy Ventures、Khosla、Eni、Google/Alphabet、Temasek、Tiger Global、Coatue
执行摘要
主要优势
- 全球最先进的私营聚变项目:SPARC 离子温度世界纪录(100 million°C,2024)、经同行评议的物理基础(12 篇期刊论文)以及按计划推进的首等离子体(2026)
- 历史上首批企业聚变 PPA:Google(200 MW)和 Eni($1B+)带来独一无二的需求拉力和商业验证,其他聚变竞争者尚不具备
- 私营聚变累计融资最高($3.2B):Breakthrough Energy Ventures、Khosla、Temasek、Google/Alphabet、Eni、Tiger Global 提供资金深度和战略协同
- REBCO HTS 磁体技术形成真正竞争护城河:50+ 项专利、自有 Materials Facility、MIT PSFC 校友管线,以及相对聚变竞争者 5–7 年先发优势
- MIT spinout 血统,加上 CSO Dennis Whyte(前 PSFC Director),在国家实验室、DOE 和机构投资者面前提供难以匹敌的等离子体物理可信度
主要风险
- SPARC Q>1 失败(约 30% 概率):失败或实质延迟就是生死风险,会击穿投资者信心、触发 PPA 重谈,并让 ARC FID 无法推进
- 氚增殖(全球 TRL 2–3):最被低估的风险;没有现场氚增殖,ARC 无法商业运营,而工程规模验证从未完成
- ARC 资本缺口(约 $3–5B):每座 ARC 电站建设需要 $2.5–3B;资金缺口尚未锁定,且补齐前提是 SPARC 达到 Q>1
- ARC LCOE 超支风险:NuScale 在取消前 LCOE 翻倍;ARC 的 $50–70/MWh 目标依赖 HTS 带材成本激进下降(10x),规模化尚未证明
- HTS 带材供应链集中:SuperOx(俄罗斯相关)以及有限的 SuNAM/FUJIKURA 产能,为 SPARC 尤其是 ARC 磁体制造制造供应瓶颈
- NRC 许可不确定:ADVANCE Act 框架尚未针对聚变最终确定;PJM 并网队列 3–5 年,可能独立于建设进度推迟 ARC 商业运营
未决问题
- SPARC Q>1 概率在实验前无法精确知道;当前所有估计(60–70%)都是分析师近似值
- ARC pre-FEED 成本研究未发布;2018 年概念设计 LCOE 估计($50–70/MWh)可能已明显过时
- Google 和 Eni PPA 条款(价格、期限、罚则)未公开披露;收入确定性低于 PPA 规模所暗示的水平
- 氚增殖路线图和时间线未公开披露;这是 SPARC 之后的关键风险,CFS 没有公开计划
- Series B2 准确估值未披露;所有投资人稀释测算都是估计
- 与 SuNAM/FUJIKURA 的 HTS 带材供应合同未披露;ARC 供应链安全无法公开核验
目录
01公司概况
1.1 身份与商业模式
Commonwealth Fusion Systems(CFS)于 2018 年成立,是 MIT Plasma Science and Fusion Center(PSFC)的孵化公司。公司总部在 Massachusetts Devens 园区——一块 47 英亩的工业场地,同时承担主要制造和反应堆组装功能。CFS 走两阶段商业化路径:(1)SPARC,一个紧凑型净能量演示托卡马克;随后是(2)ARC,一个面向电网的 400 兆瓦电商业聚变电站。公司的商业模式,是授权并运营聚变电站,作为发电型公用事业公司通过长期购电协议售电。技术差异化来自 REBCO 高温超导(HTS)磁体:它能在紧凑形态下产生 20 tesla 磁场,使反应堆单位瓦成本比早期设计低约 40×。CFS 根据赞助研究协议,与 MIT 共同持有该磁体设计的 IP。 [CO001, CO002, CO003, CO004, CO005]
1.2 领导层、董事会与治理
CFS 由 CEO 兼联合创始人 Bob Mumgaard 领导。Mumgaard 是等离子体物理学家,在 MIT 获得博士学位,曾在 MIT PSFC 工作,之后参与创立 CFS。Dan Brunner 担任首席技术官,负责 SPARC 和 ARC 设计。Steve Renter 是首席运营官,负责建设执行和供应链管理。Ally Yost 是企业发展高级副总裁。Alex Mozdzanowska 担任首席人事官。David Tressler 是首席法务官。公司董事会包括主要投资方代表,其中 Eni 被其 CEO 称为「相对多数股东」。Tiger Global、Breakthrough Energy Ventures 和其他 Series B 投资方持有重要经济权益。关键人风险集中在 Mumgaard 和 Brunner 身上,因为两人深度参与技术和融资。CFS 聘请 MIT 教授 Dennis Whyte(他曾宣布最初的磁体突破)担任战略顾问;Whyte 已卸任 PSFC 主任。 [CO006, CO007, CO008, CO009, CO010]
| 人物 | 角色 | 背景 | 创始人 / 匹配度 | 关键人物依赖 |
|---|---|---|---|---|
| Bob Mumgaard | CEO 兼联合创始人 | MIT 等离子体物理 PhD;前 PSFC 研究员 | 等离子体物理学家,从 MIT 研究中提出 CFS 战略 | 关键——首席融资人、公众门面、战略愿景 |
| Dan Brunner | CTO | MIT 核科学背景;CFS 创立时加入 | 深厚托卡马克设计和磁体工程专长 | 高——负责 SPARC/ARC 设计项目 |
| Steve Renter | COO | 工业运营背景 | 负责建设和供应链执行 | 中——可由工业人才池补位 |
| Ally Yost | 企业发展 SVP | 财务和企业发展 | 领导投资者关系和商业交易结构设计 | 中——Google PPA 和 Eni PPA 在其任内搭建 |
| Alex Mozdzanowska | 首席人力官 | HR 领导力 | 将人才规模从约 350 人扩至 1,000+ | 中 |
| David Tressler | 首席法务官 | 法律、监管 | 管理 NRC 许可和 IP | 中——对监管路径关键 |
Dennis Whyte(MIT PSFC)是战略顾问;他是最初 REBCO 磁体突破的架构师。
[CO006, CO007, CO008, CO009, CO010]1.3 融资历史与估值
自成立以来,CFS 已融资约 $3 billion,约占全球聚变公司历史私人资本总额的三分之一。融资轮次包括:2018 年 Eni 领投的种子轮,约 $50 million;2019 年 6 月 $115 million 的 Series A(Eni、Breakthrough Energy Ventures、Khosla Ventures、The Engine/MIT、Future Ventures、Safar Partners、Starlight Ventures 等);2021 年 12 月 $1.8 billion 的 Series B(Tiger Global 领投,约 70 家投资方参与,包括 Bill Gates、Google、Equinor、Temasek、Emerson Collective 和 Eni);以及 2025 年 8 月 28 日完成的 $863 million Series B2(Nvidia/NVentures、Counterpoint Global/Morgan Stanley、Stanley Druckenmiller、由 Mitsui 和 Mitsubishi 牵头的 12 家日本公司财团、Galaxy Digital 等)。2021 年投后估值据报道为 $3.2–$5 billion。管理层称 2025 年 B2 是「up round」,意味着估值更高;但 2025 年的具体数字未披露。CFS 未披露收入;公司仍处于商业化前。财务结构以股权为主;公司还获得了一笔 $8 million 的 DOE 里程碑式拨款,2025 年 9 月因磁体技术性能通过验证而确认。 [CO011, CO012, CO013, CO014, CO015, CO016]
| 指标 | 数值 / 状态 | 日期 | 置信度 | 缺口 / 尽调路径 |
|---|---|---|---|---|
| 估值(Series B2 后) | 较 $3.2–$5B(2021 年)上行轮;具体未披露 | Aug 2025 | 中 | 要求查看股权结构表或二级市场交易数据 |
| 累计融资 | ~$3 billion | Aug 2025 | 高 | 无——公司确认数据 |
| 最新轮次 | Series B2,$863M | Aug 2025 | 高 | 无——官方公告 |
| 员工数 | ~800–1,000+ | Early 2025 | 中 | 要求查看 HR 数据室;多方来源估算 |
| 收入 / ARR | 2026 | 高 | 商业化前;未披露收入 | |
| SPARC 首次等离子体目标 | 2026 | Oct 2024 许可证 | 中 | 跟踪建设进度更新 |
| SPARC 净能量目标 | 2027 | CFS 多次声明 | 中 | 取决于首次等离子体时间 |
| ARC 并网发电目标 | 2030 年代初 | CFS 官方 | 中 | 取决于 SPARC 成功;尚无已签 EPC 合同 |
| ARC 输出功率 | 400 MWe | CFS 官方 | 中 | 设计规格;工程开发中可能变化 |
| Google PPA | 已签 200 MW | Jun 2025 | 高 | 无——公开公告;取决于 ARC 商业运行 |
| Eni PPA | $1B+ | Sep 2025 | 高 | 具体 MW 规模未公开披露 |
| HTS 磁体磁场 | 20 tesla 世界纪录 | Sep 2021(2024 年验证) | 高 | 经 IEEE Transactions 同行评审 |
2025 年轮次估值尚未得到公开确认。商业化前,所有收入和利润率指标均为 N/A。
[CO012, CO013, CO014, CO016, CO019, CO021]| 利益相关方 | 角色 | 经济 / 控制重要性 | 尽调问题 |
|---|---|---|---|
| Eni | 战略投资人;相对多数股东;$1B+ PPA 承购方 | 据称为最大单一股东集团;ARC 电力商业伙伴 | 确认准确持股比例和董事席位条款 |
| Tiger Global | Series B 领投方 | 主要经济持股;主导 $1.8B 轮次 | 2021 年后持股和任何老股交易 |
| 投资人;200 MW PPA 承购方 | 战略商业买方;B2 轮增持 | PPA 条款、定价和触发条件 | |
| Breakthrough Energy Ventures | 长期投资人(Series A 和 B) | Bill Gates 关联方;强信号投资人 | 持股规模,以及是否有棘轮或清算优先权 |
| Khosla Ventures | Series A 和 B 投资人 | 持股较大的活跃 VC | 治理权和投票权比例 |
| NVIDIA / NVentures | Series B2 投资人 | 战略技术-工业投资人;与 AI 能源需求对齐 | 投资理由和持股规模 |
| Emerson Collective | Series B 投资人;Laurene Powell Jobs | 战略慈善-商业投资人 | 持股规模和使命匹配条款 |
| 日本财团(Mitsui 与 Mitsubishi 领投的 12 家公司) | Series B2 投资人 | 工业供应链伙伴;潜在 ARC 电站客户 | 待签许可和共同开发协议 |
| MIT / PSFC | 研究伙伴;IP 共同持有人 | 磁体设计共同 IP;赞助研究协议 | 确认 IP 所有权拆分和未来版税条款 |
| U.S. Department of Energy(美国能源部) | 拨款方;监管验证方 | $8M 里程碑拨款;验证磁体性能 | 未来公共资金或 INFUSE 合作范围 |
| Dominion Energy | ARC 土地期权伙伴(弗吉尼亚州) | 控制 Chesterfield County 潜在 ARC 场址 | 租赁期权条款、排他窗口 |
股权比例未公开披露。治理安排未完全公开。
[CO011, CO012, CO013, CO014, CO015, CO016]1.4 SPARC 与 ARC 项目状态
SPARC(Smallest Plasma Achieving Replication of ITER Conditions)是在 Massachusetts Devens 建设中的紧凑型托卡马克。关键组装里程碑包括:75 吨、直径 24 英尺的低温恒温器底座安装完成(2025 年 3 月),以及首个 48 吨半环形真空室到场(2025 年 10 月)。CFS 目标 2026 年产生第一等离子体,并在 2027 年实现净聚变能量增益(Q > 1)。2024 年 10 月,Massachusetts 向 CFS 授予 SPARC 广范围放射性材料许可证。SPARC 预计将成为首个在磁约束装置中实现科学盈亏平衡的托卡马克。ARC 是 CFS 规划中的首座商业 400 MWe 聚变电站,将建在 Virginia 的 Chesterfield County,土地来自 Dominion Energy 的选择权。2025 年 6 月 30 日,Google 与 CFS 签署 200 MW 购电协议,这是史上首个企业聚变 PPA。2025 年 9 月,Eni 签署 $1 billion+ 的购电协议,购买 ARC 输出。ARC 目标在 2030 年代初并网运行。ARC 计划 2020 年代末开工,前提是 SPARC 实现净能量成功。 [CO017, CO018, CO019, CO020, CO021, CO022]
| 日期 | 事件 | 类型 | 金额 / 估值 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2018-01 | CFS 作为 MIT PSFC 衍生公司注册成立 | 创立 | MIT 种子;Eni 初始约 $50M | Bob Mumgaard、Dan Brunner、MIT PSFC 团队 | 建立法律实体并获得首个机构支持方 |
| 2019-06 | 完成 $115M Series A | 融资 | $115M;总计约 $165M | Eni、BEV、Khosla、The Engine、Future Ventures、Safar、Starlight 等 | 资助 HTS 磁体 R&D 和 SPARC 前期工程 |
| 2021-09 | MIT 完成 20-tesla HTS 磁体世界纪录示范 | 产品 | 20T——世界纪录;每瓦成本下降 40× | MIT PSFC、CFS 工程团队 | 验证核心技术逻辑;2024 年经 IEEE TAS 同行评审 |
| 2021-12 | 完成 $1.8B Series B | 融资 | $1.8B;估值 $3.2–$5B | Tiger Global(领投)、约 70 家投资人,包括 Gates、Google、Eni、Equinor、Temasek、BEV | 最大私有聚变融资;资助 SPARC 建设 |
| 2022-12 | SPARC 场址在 Devens, MA 开工 | 规模化 | Devens, MA 园区 | CFS、本地建设伙伴 | Tokamak Hall 和制造设施破土动工 |
| 2023-04 | NRC 投票决定按副产物材料框架监管聚变 | 监管 | 非裂变规则 | U.S. Nuclear Regulatory Commission(美国核管理委员会)、CFS | 相比核裂变,ARC 监管负担更低;许可路径更快 |
| 2024-03 | IEEE 发表六篇关于 HTS 磁体结果的同行评审论文 | 产品 | 发表于 IEEE Transactions on Applied Superconductivity | MIT PSFC、CFS 研究人员 | 对 20T 磁体设计和失效模式的正式科学验证 |
| 2024-07 | ADVANCE Act 将 NRC 聚变副产物材料框架写入美国法律 | 监管 | 联邦法律 | U.S. Congress;总统签署 | 为聚变许可路径提供永久法律确定性 |
| 2024-10 | Massachusetts 授予 CFS SPARC 广范围放射性材料许可证 | 监管 | 州许可证;SPARC 可运行 | Massachusetts Radiation Control Program(马萨诸塞州辐射控制项目)、CFS | 首个正式运行授权;首次等离子体的必要条件 |
| 2025-03 | 安装低温恒温器底座——SPARC 托卡马克首个大型部件 | 产品 | 24-ft 宽、75-ton 不锈钢底座 | CFS 建设团队;意大利制造商 | SPARC 从设施建设切入托卡马克组装阶段 |
| 2025-06 | Google 签署 ARC 200 MW 电力购买协议 | 合作 | $200 MW;世界首份聚变 PPA | Google、CFS | 商业需求信号;验证经济性;增强投资人信心 |
| 2025-08 | 完成 $863M Series B2(超额认购) | 融资 | $863M;具体估值未披露但称为上行轮 | NVentures/Nvidia、Morgan Stanley/Counterpoint、Druckenmiller、Mitsui/Mitsubishi 财团等 | CFS 2021 年 B 轮以来最大深科技 / 能源融资;支持 SPARC 完工和 ARC 开发 |
| 2025-09 | DOE 验证 CFS 磁体技术;$8M 里程碑拨款 | 监管 | $8M DOE 拨款 | U.S. Department of Energy(美国能源部)、CFS | 独立政府验证磁体性能;非稀释资本 |
| 2025-09 | Eni 签署 ARC 输出 $1B+ 电力购买协议 | 合作 | $1B+ PPA | Eni、CFS | 第二份重大 PPA;Eni 从投资人进一步成为商业客户 |
| 2025-10 | 首个真空容器半体交付至 SPARC Tokamak Hall | 产品 | 48-ton 半甜甜圈形真空容器 | CFS、意大利 / 国际供应商 | 核心等离子体约束容器就位;组装阶段加速 |
2025 年事件日期反映公开宣布的里程碑。
[CO001, CO011, CO012, CO013, CO014, CO017]按时间梳理 CFS 从 2018 年成立到 2025 年底的里程碑,覆盖融资、 产品、监管和合作伙伴事件。轨迹显示执行在提速:2018–2019 年成立与 早期融资,2021 年磁体纪录突破,2021 年 $1.8B B 轮融资,2022 年 SPARC 开工,2023–2024 年监管进展,以及 2025 年托卡马克组装推进并签下大型 PPA。
[CO001, CO011, CO012, CO013, CO016, CO017]展示 CFS 的身份(MIT 衍生公司)、技术(REBCO HTS 磁体)、产品(先 SPARC 后 ARC)、资本 基础(约 $3B)和商业管线(Google + Eni PPA)如何相互连接。ADVANCE Act 和 马萨诸塞州许可证构成监管支持节点。
[CO001, CO002, CO014, CO017, CO021, CO022]CFS 当前投资案例的头部指标:累计融资(约 $3B)、示范时间线 (首次等离子体 2026)、商用供电目标(2030 年代初)、磁场纪录(20T)和员工增长。
[CO013, CO014, CO019, CO021, CO022, CO023]1.5 规模、员工数与不利事件
截至 2025 年初,CFS 员工约 800 至 1,000 人或更多,高于 2022 年末约 350 人。Devens 园区一栋建筑负责 HTS 磁体制造,Tokamak Hall 负责 SPARC 组装。公开资料未显示 2024–2025 年期间发生重大裁员、领导层离职、制裁、产品召回或重大诉讼。公司在品牌上公开淡化「nuclear」一词,尽管 SPARC 和 ARC 使用放射性氚燃料;CEO Mumgaard 公开谈过管理公众感知的策略。不利维度在于,CFS 的业务完全依赖 SPARC 先证明净能量增益,之后才可能产生商业收入,这带来执行风险,也要求很长的资本跑道。美国能源部部长 Chris Wright 在 2025 年末访问 SPARC 场地,凸显监管互动。未发现诉讼或监管执法行动。 [CO024, CO025, CO026, CO027, CO028]
1.6 展示项
02市场分析
2.1 市场定义与边界
CFS 的目标市场是全球基荷发电板块,尤其是需要稳定、可调度、24/7 零碳电力的部分。售电对象包括:(1)希望替代煤炭和天然气基荷的受监管公用事业公司;(2)承诺 24/7 无碳能源(CFE)的超大规模数据中心运营商(Google、Microsoft、Amazon、Meta);(3)钢铁、绿色氢能、海水淡化等高耗能行业的工业用电客户。市场边界明确排除间歇性可再生能源(太阳能、风能)和传统核裂变,后两者处在不同的技术与监管生态位。相邻市场包括过程热(工业脱碳)以及利用富余聚变电力制氢的潜在场景。全球电力市场是聚变最终可触达机会的最宽代理口径,按发电、输电和配电收入合计,2024 年估计约 $2.4 trillion。CFS 近期 SAM 是美国公用事业和超大规模云厂商针对稳定清洁电力的 PPA 市场,首座 ARC 电站将在这里交付。 [CM001, CM002, CM003]
| 细分市场 / 类别 | 纳入支出 | 排除支出 | 买方 / 付款方 | 与 CFS 的相关性 |
|---|---|---|---|---|
| 基荷清洁电力——公用事业 | 电网级稳定零碳发电;长期限 PPA;容量付款 | 间歇性太阳能 / 风电;调峰电厂 | 受监管公用事业(如 Dominion、Duke、Con Ed) | 主要——ARC 首座电站服务该细分市场 |
| 面向超大规模云厂商的 24/7 CFE | 企业稳定清洁电力 PPA;>100 MW 交易;15–25 年期限 | 捆绑可再生能源证书(REC) | Google、Microsoft、Amazon、Meta 数据中心采购 | 主要——Google PPA 是首个已落地交易 |
| 工业脱碳电力 | 面向绿色钢铁、化工、绿色氢的大宗稳定电力 | 工艺热合同(未来附加) | 工业高耗能企业 | 次要——首座电站之后的中期机会 |
| 工艺热(来自聚变) | 高温工业热供应 | 发电(独立细分市场) | 钢厂、水泥、化工生产商 | 潜在长期相邻市场;不在当前计划内 |
| 政府 / 国防 | DOE 资助的聚变研发;存在潜在国家安全应用 | 商业电力市场 | 美国 DOE、DOD | 当前——DOE 里程碑补助;不是核心收入模式 |
2.2 市场规模
聚变专项市场研究估计,2024–2025 年全球聚变能源市场为 $301–$347 billion,到 2030 年增至 $420–$497 billion,到 2040 年达到 $840 billion,CAGR 约 6–8%。这些估计代表聚变电站商业运行后的预测总收入,是从假设部署轨迹外推而来,并非当下收入。自下而上看,ARC 规模为 400 MWe;若按 $50–$70/MWh 的 LCOE 目标售电,单座电站在 100% 容量因子下每年产生约 $175–$245 million 收入。美国首个市场奖池——稳定清洁电力的基荷 PPA 价格——在 2024–2025 年先进核能(SMR 公告)的竞争性招标中为 $90–$150/MWh,说明如果建设成本符合估计,ARC 经济性具备可行性。全球清洁能源投资流量在 2023 年超过 $1.8 trillion,预计 2024 年接近 $2 trillion,显示市场有足够资本部署新的清洁电力技术。数据中心电力市场尤其是有吸引力的增长向量:IEA 预计数据中心用电需求将从约 415 TWh(2024)翻倍至 2030 年约 945 TWh,为可调度清洁电力带来强劲增量需求。 [CM004, CM005, CM006, CM007, CM008, CM009]
| 发布方 | 年份 | 地域 | 数值 | 复合年增长率(CAGR) | 方法 | 置信度 | 局限 |
|---|---|---|---|---|---|---|---|
| IEA / Statista | 2024 | 全球 | $2.4T 电力市场 | ~3-4% p.a. | 按收入汇总的电力市场 | 高 | 宽口径代理指标;包含所有电源,不只清洁电力 |
| Maximize Market Research(市场研究机构) | 2025 | 全球 | $347B 聚变市场 | 7.4% | 商业聚变电站部署情景 | 低 | 尚未商业化;全部价值取决于部署落地 |
| Allied Market Research(市场研究机构) | 2040 | 全球 | $840B 聚变市场 | 6–7% | 2032 年后部署的长期情景 | 低 | 高度推测;取决于多个里程碑 |
| IEA(EMDE / 全球) | 2024–2030 | 全球 | ~415–945 TWh 数据中心需求 | ~15%/yr | 实证能耗数据加 AI 增长模型 | 高 | 仅电力需求;不是市场价值 |
| Grand View Research | 2025–2033 | 全球 | $1.6T–$4.9T 可再生能源市场 | 14.7% | 可再生能源收入预测 | 中 | 不含聚变;给出价格竞争参照 |
| IEA World Energy Investment | 2024 | 全球 | $1.8T 清洁能源投资 | 较 2019 年约 10% | 年度资本投放跟踪 | 高 | 投资额,不是收入;包含所有清洁技术 |
| Georgetown Space Policy | 2025 年估计 | 美国 | ~$50–$70/MWh ARC 目标 LCOE | N/A | 公司给出的工程估算 | 低 | 缺少独立验证;首座同类电站 |
所有聚变市场估计都是商业化前预测,取决于 CFS 及同业能否成功证明净能量增益。
[CM004, CM005, CM006, CM007, CM008, CM009]CFS 的三级市场规模测算。TAM 是全球发电市场(每年约 $2.4T)。 SAM 是基荷清洁 / 可调度电力市场(2030 年约 $300–400B)。短期 SOM 是 首代聚变电站的企业 / 公用事业 24/7 CFE PPA 市场(ARC 运营的第一个十年; 单座 400 MW 电站已经签下总额 $1B+ 的 PPA 承诺)。
SAM 估计基于电力市场中清洁 / 稳定电力的按比例份额;聚变专项市场 研究差异很大。SOM 只反映首座 ARC 电站已公布的 PPA 承诺。
[CM004, CM005, CM021]展示 2024 到 2040 年全球聚变能源市场的独立分析师估算分布。 低位估计为 Maximize Market Research 2025 基准($347B);高位估计为 Allied Market Research 的 2040 年预测($840B)。注意,所有估计都是商业化前预测。
所有区间来自第三方市场研究,方法假设各不相同。这些估计是 商业化前的投机性预测,不应当作 CFS 收入预测。
[CM005, CM006, CM007]IEA 数据显示,全球数据中心用电量从 2024 年约 415 TWh 翻倍,到 2030 年约 945 TWh,AI 工作负载是主要驱动。这一翻倍带来约 530 TWh 新增可调度清洁电力需求,为 2030 年代初进入商业服务的聚变形成强顺风。
2026 和 2028 数值是在 IEA 报告的 2024 与 2030 数据之间线性插值;实际路径 取决于 AI 工作负载增长和效率提升。
[CM009]2.3 买家分层与采用路径
三类买家构成 CFS 的需求版图:(1)超大规模云和 AI 公司(Google、Microsoft、Amazon、Meta),它们承诺为数据中心采购 24/7 CFE,需要可调度的稳定清洁电力和长周期 PPA(15–25 年);Google 已签的 200 MW PPA 是首个真实样本。(2)受监管公用事业公司,希望用低碳基荷替代退役煤电 / 气电机组,并满足州级可再生能源配额标准;Dominion Energy 在 Virginia 的场地合作(CFS 的 ARC 所在州)体现了这一路径。(3)高耗能工业买家(钢铁、化工、绿色氢能),需要大块稳定、低成本电力。各细分市场的采用触发点不同:超大规模云厂商会在 SPARC 概念验证(预计 2027 年)后行动,也可以带技术风险提前签署远期 PPA;受监管公用事业公司会在商业许可和标准 FERC 并网批准后行动;工业买家则跟随公用事业先例。预算所有权在超大规模云厂商的首席可持续发展 / 能源官手中,也在公用事业公司的资源规划团队手中——两类团队都拥有数十年投资周期,并受强政策顺风推动。 [CM010, CM011, CM012, CM013]
| 细分市场 | 买方 | 用户 | 付款方 | 工作流 | 预算负责人 | 采用触发点 |
|---|---|---|---|---|---|---|
| 超大规模数据中心 | Google, Microsoft, Amazon | 数据中心运营团队 | Google / Microsoft CFO | 24/7 CFE 承诺;与发电商签 PPA;按小时跟踪交付 | 首席可持续发展官 / 能源负责人 | SPARC Q>1 验证 + ARC 开工 |
| 受监管公用事业公司 | Dominion Energy, Duke, Xcel | 电网资源规划团队 | 通过 FERC/PUC 核准电价由缴费用户承担 | 资源充足性申报;容量采购;FERC 并网申请 | 资源规划 VP | ARC 商业许可证 + FERC 排队位置 |
| 工业能源买方 | 钢铁、化工、氢气生产商 | 工厂能源经理 | 工业工厂 P&L 负责人 | 长期供电协议;纳入能源成本结构 | 能源 VP / CFO | ARC-1 之后证明 LCOE 有竞争力 |
| 政府 / DOE | 美国能源部 | 聚变研究项目办公室 | 联邦拨款 | 里程碑式补助;INFUSE 合作 | 项目经理 | 国会拨款;政策优先级 |
2.4 增长驱动因素与采用约束
关键需求驱动因素包括:(1)AI 数据中心电力激增——IEA 预计 2024–2030 年数据中心用电翻倍至约 945 TWh,主要超大规模云厂商受 24/7 CFE 要求约束,需要稳定清洁电源;(2)全球脱碳目标——IEA 净零情景要求到 2050 年约 50% 电力来自非间歇性清洁来源,为可调度零碳发电创造结构性拉力;(3)核能复兴带来监管和公众接受度顺风——美国 ADVANCE Act(2024)以及欧洲 taxonomy 将核能纳入其中,有助于投资者相信聚变监管路径;(4)可再生能源渗透率提高后,电网可靠性担忧上升,抬高稳定电源的容量支付。关键约束:(1)商业化前周期很长——CFS 预计 2030 年代初前没有并网电力,限制近期市场份额;(2)太阳能 + 储能 LCOE 快速下降带来成本竞争(部分市场已低于 $40/MWh),聚变必须低于该成本,或凭 24/7 可调度溢价自证合理;(3)氚燃料供应——商业聚变需要氚增殖,但规模化尚未验证,形成燃料链风险;(4)尽管聚变风险结构不同,公众仍可能受「nuclear」污名影响。 [CM014, CM015, CM016, CM017, CM018, CM019]
| 驱动因素 / 约束 | 方向 | 时间 | 影响 | 尽调问题 |
|---|---|---|---|---|
| AI 数据中心电力需求激增 | 驱动 | 即刻(2024–2030) | 到 2030 年,超大规模云厂商需要 >1,000 TWh 稳定清洁电力;由此形成 PPA 需求管线 | Google 是否会把 PPA 扩展到未来 ARC 电站?Microsoft / Amazon 管线规模多大? |
| 全球脱碳要求 | 驱动 | 结构性(现在–2050) | 公用事业公司和工业企业的净零目标,推动长期采购零碳基荷电力 | 哪些公用事业公司的近期退煤时间表与 ARC 时间线匹配? |
| 核能复兴(SMRs + ADVANCE Act) | 驱动 | 中期(2026–2035) | 监管框架和公众接受度有利于新型类核技术 | CFS 的许可路径与 NuScale、X-energy 的 SMR 时间线相比如何? |
| 电网可靠性 / 容量付款 | 驱动 | 结构性 | 可再生能源渗透率高的市场(CA、TX)容量付款更高,利好可调度电源 | 目标市场容量价格是多少?ARC 是否符合资格? |
| 太阳能 + 储能 LCOE 下降 | 约束 | 持续 | 最佳地点的太阳能 LCOE 现已 <$40/MWh;聚变必须靠 24/7 供电曲线或更低全口径成本证明溢价合理 | ARC 模型 LCOE 与 2030 年太阳能 + 电池成本预测相比如何? |
| 商业化前周期长 | 约束 | 到 2030 年代初 | CFS 未来 5 年以上仍无收入;只有 SPARC 成功后才能获取市场份额 | 如果 ARC 延期,Google 和 Eni PPA 的违约条款是什么? |
| 氚燃料供应链 | 约束 | 中期(SPARC 之后) | 商业聚变电站需要氚增殖;当前供应仅限于 CANDU 反应堆副产品 | CFS 是否为 ARC 设计了氚增殖包层?氚采购计划是什么? |
| 公众对「核」的观感 | 约束 | 持续 | 品牌重塑正在推进;若公众因核能联想反对聚变,选址许可可能延后 | Devens 或 Virginia 场址是否已有活跃反对团体? |
2.5 市场规模不确定性与尽调缺口
所有聚变专项 TAM 估计都假设 2030 年代初开始商业运行;若 SPARC 未实现净能量,或 ARC 建设延迟,实际市场获取时间就会后移。分析师给出的 TAM 区间(2024–2040 年 $301B–$840B)反映了部署速度、电价轨迹和竞争者进入时间的不同情景假设。更可操作的近期市场信号是企业 PPA 市场:2023–2025 年,超大规模云厂商已与先进核能供应商签署约 $5–10 billion 的 24/7 CFE 合同,证明其付费意愿。关键数据缺口包括:(a)ARC 的实际 LCOE(唯一公开估计约 $50–$70/MWh,来自公司);(b)可再生能源 + 储能缩小可调度性差距的速度,以及由此对稳定清洁电力溢价的压缩;(c)Virginia ARC 场地的输电拥塞经济性。这些不确定性让市场机会巨大,但时间点不确定——投资者真正要问的不是市场是否存在,而是 CFS 能否按 SPARC/ARC 进度执行。 [CM020, CM021, CM022]
从当前投机性 PPA(Google、Eni)到首个公用事业并网,再到公用事业和工业客户 大规模铺开,分阶段呈现采用路径。每一阶段都受技术验证里程碑约束。
每个阶段的累计买方数量仅为示意;实际管线未公开披露。
[CM003, CM020, CM021, CM022]2.6 展示项
03竞争格局
3.1 私人聚变竞争版图
全球私人聚变板块约有 40 家公司,高于 2019 年的约 10 家;截至 2025 年末,已部署私人资本超过 $7 billion。CFS 和 Helion 是进展最靠前、融资最多的两家公司。CFS 聚焦紧凑型托卡马克聚变(SPARC,2026/27)和商业电力(ARC,2030 年代初)。Helion Energy 采用场反位形(FRC),使用 D-He3 燃料并直接电能转换,背后有 Musk/Bezos/Altman 关联的 $375M Series E+F 支持,还与 Microsoft 签署了首创 PPA,承诺到 2028 年交付 50 MW。TAE Technologies(California,已融资 $1.3B)采用质子-硼燃料的场反位形,目标是聚变增强的无中子电力。Tokamak Energy(UK,已融资 $335M)采用球形托卡马克几何和 HTS 磁体,目标在 2020 年代末建设早期试点电站。General Fusion(Canada,已融资 $325M)采用磁化靶聚变(MTF)和活塞压缩;2024 年从 Steam Generator 转向新概念。每条路径的物理基础、时间线和资本要求都不同。 [CP001, CP002, CP003, CP004, CP005]
| 公司 | 总部 | 成立时间 | 技术 | 关键里程碑(2025) | 已融资金额 | 关键投资方 | 交付承诺 |
|---|---|---|---|---|---|---|---|
| Commonwealth Fusion Systems | Devens, MA,美国 | 2018 | 紧凑型托卡马克 + REBCO HTS 磁体(20T) | SPARC 低温恒温器安装,$863M B2 轮 | 累计 ~$3B | Eni, Google, Khosla, MIT | Google 200MW + Eni $1B+ PPA(购电协议) |
| Helion Energy | Everett, WA,美国 | 2013 | 场反位形(FRC)、D-He3、直接转换 | Polaris 第 7 代等离子体装置;接近 100M°C | 累计 ~$2.2B | Sam Altman($375M)、Dustin Moskovitz 等投资人 | 2028 年前 Microsoft 50MW PPA |
| TAE Technologies | Foothill Ranch, CA,美国 | 1998 | 场反位形、质子-硼燃料(p-B11) | Norman 装置;瞄准等离子体温度里程碑 | 累计 ~$1.3B | Google, Goldman, Chevron | 未宣布 |
| Tokamak Energy | Abingdon,英国 | 2009 | 球形托卡马克 + HTS 磁体 | ST80-HTS 磁体于 2024 年投运 | 累计 ~$335M | Legal & General、美国 DOE、DC Investment | 英国政府试点电站目标为 2030 年代 |
| General Fusion | Vancouver, BC,加拿大 | 2002 | 磁化靶聚变(MTF)、机械压缩 | 从 Steam Generator 概念转向;2024 年推出新设计 | 累计 ~$325M | Jeff Bezos, Khosla, BDC | 无商业 PPA |
| Zap Energy | Seattle, WA,美国 | 2017 | 剪切流稳定 Z 箍缩 | FuZE-Q 装置;已融资 ~$160M | ~$160M | Chevron Technology Ventures, DCVC | 无商业 PPA |
| TAE Life Sciences | n/a(TAE 分拆) | 2018 | 中子束治疗(癌症,不是发电) | II 期临床试验 | ~$60M | TAE Technologies | 医疗,不是发电 |
融资金额为截至 2026 年一季度、基于公开文件和媒体报道的大致数字。Helion 的交付承诺具有合同约束(与 Microsoft 的 PPA)。
[CP001, CP002, CP003, CP004, CP005, CP006]2x2 象限把聚变和清洁能源替代方案放在技术成熟度(TRL)与融资额两个轴上。 CFS 和 Helion 位于右上象限(高 TRL、高资本);Tokamak Energy 和 TAE 处于中等 TRL、 中等资本。SMR(NuScale、TerraPower)TRL 更高(已获许可 / 接近获许可), 但仍然昂贵。光伏 + 储能在 TRL 上占优,但未作为“初创公司”展示。
TRL 值为分析师定性估计;资本数字来自公开来源,可能偏低。 纳入 NuScale 是为了提供间接竞争参照。
[CP001, CP003, CP004, CP005, CP006, CP012]3.2 CFS 与 Helion:正面对比
CFS 和 Helion 是最直接可比的两家头部、高融资私人聚变公司。CFS(托卡马克,氘-氚)借力 70+ 年托卡马克物理、已在规模上验证的约束结果(JET、ITER 前身),以及经过同行评议的 SPARC 设计基础。Helion(FRC,氘-氦-3)规模化验证较少,但声称直接电能转换可以绕开热循环,从而可能降低资本成本。Helion 的 Microsoft PPA(2028 年前 50 MW)带有终止罚则,形成公开问责。CFS 的 Google PPA(200 MW,未公开交付日期触发条件)MW 规模更大,但距离交付更远。融资上,CFS 已融资约 $3B,Helion 约 $2.2B。CFS 的 REBCO HTS 磁体纪录(20T,2021)是具体技术里程碑;Helion 尚未发表同等级别、经过同行评议的磁体或等离子体性能数据。没有数据室访问时,等离子体里程碑很难比较:CFS 目标 2027 年 SPARC Q>1;Helion 曾把 Polaris 第一等离子体目标定在 2024 年,但尚未公开确认状态。CFS 的氘-氚路径需要氚增殖(供应风险);Helion 的 D-He3 路径降低放射性,但氦-3 稀缺本身也是约束。 [CP006, CP007, CP008, CP009, CP010, CP011]
| 能力 | CFS (SPARC/ARC) | Helion (FRC/Polaris) | TAE (Norman/Alpha) | Tokamak Energy (ST80) | General Fusion (MTF) |
|---|---|---|---|---|---|
| 等离子体约束方式 | 托卡马克(已规模验证) | 场反位形 | 场反位形 | 球形托卡马克 | 磁化靶(惯性 + 磁约束) |
| 燃料类型 | 氘-氚(D-T) | 氘-氦-3(D-He3) | 质子-硼(p-B11) | 氘-氚(D-T) | 氘-氚(D-T) |
| 中子产生(活化) | 高(需要屏蔽 / 氚增殖) | 低(D-He3 产生的中子更少) | 极低(无中子,p-B11) | 高(D-T) | 高(D-T) |
| 磁体技术 | HTS REBCO(20T 世界纪录) | 常规 + 部分 HTS | 常规 | HTS(自研) | 无(基于压缩) |
| 商业化目标日期 | 2030 年代初(ARC) | 2028 年(Microsoft PPA) | 2030 年代末(估计) | 2030 年代(英国试点) | 2035+(估计) |
| 已签商业 PPA | 是(Google 200MW + Eni $1B+) | 是(Microsoft 50MW) | 否 | 否(英国政府有意向) | 否 |
| 同行评审的物理基础 | 是(MIT PSFC,多篇 Nature Energy 论文) | 有限(自研) | 是(p-B11 物理已发表) | 是(球形托卡马克已有研究积累) | 是(MTF 已有研究积累) |
| 监管申请(NRC / 同等机构) | 尚未(SPARC 前) | 尚未 | 尚未 | 与 UK-STEP 接近 | 尚未 |
能力矩阵按七个与商用供电有关的维度比较五家主要私营聚变公司: 等离子体物理基础、磁体技术、已签 PPA、融资深度、监管进展、 氚 / 燃料计划和时间线可信度。
[CP006, CP007, CP008, CP009, CP017, CP018]3.3 间接竞争者:SMR 与先进可再生能源
CFS 商业上最接近的竞争者不是其他聚变公司,而是先进裂变 SMR,以及长时储能 + 可再生能源组合;它们都瞄准同一个 24/7 清洁基荷市场。NuScale Power 的 NuScale SMR(每模块 60 MWe)已获得 NRC 设计认证(2022),并在与公用事业公司谈判;但 Idaho 的 Carbon Free Power Project(CFPP)因成本上升(预计 $89/MWh)在 2023 年取消。TerraPower 的 Natrium(345 MWe,Wyoming,2030 年目标)和 X-energy 的 Xe-100 也处于 NRC 审查中。这些 SMR 的监管路径和公用事业合作伙伴都领先于聚变。地热(AltaRock、Fervo、Quaise)正成为可调度清洁替代方案。长时储能(Form Energy 铁-空气、Ambri)瞄准多日储能。若这些替代方案在 ARC 之前规模化,CFS 的 PPA 转化机会会收窄。CFS 相对 SMR 的关键差异点,是没有长寿命放射性废物,也没有裂变风险;这对公众接受度和超大规模云厂商 ESG 定位很重要。 [CP012, CP013, CP014, CP015, CP016]
| 公司 / 技术 | LCOE 目标($/MWh) | 电站规模(MWe) | 首座商业电站时间线 | PPA 价格基准 | 资本成本估算 | 收入模式 |
|---|---|---|---|---|---|---|
| CFS — ARC | $50–$70(公司估算) | 400 | 2030 年代初 | Google/Eni PPA(价格未披露) | 首座电站 ~$2.5B(估计) | PPA 购电;潜在模块化机队授权 |
| Helion — Polaris 衍生 | ~$35(公司声称) | 50 | 2028 年(Microsoft PPA 目标) | Microsoft PPA(价格未披露) | 未披露 | PPA 购电;授权 |
| NuScale SMR | ~$89(CFPP 实际报价) | 每模块 77(最高 462) | 2029–2030 | $89/MWh(2023 年 CFPP 报价;项目已取消) | 6 模块电站 ~$6B | PPA;公用事业费率基数 |
| TerraPower Natrium | 未披露 | 345 | 2030 年(Wyoming 目标) | 未披露 | ~$4B+ | PPA;公用事业费率基数;Wyoming DOE 部分资助 |
| 太阳能 + 4 小时储能(公用事业级) | $45–$65(美国 2024 年中位数) | 可变 | 现在 | 当前市场 | 500MW 对应 ~$1–2B | 市场化售电或 PPA |
| 地热(Fervo、AltaRock) | $50–$80(下一代增强型) | 可变 | 2026–2030+ | 长期公用事业 PPA | 每 MW 变量 | PPA;公用事业公司 |
ARC LCOE 和 Helion LCOE 均为公司引用的估算,暂无独立验证。NuScale 的 CFPP 取消项目说明首台套成本风险。
[CP012, CP013, CP014, CP015]3.4 CFS 竞争护城河与持久性
CFS 的主要竞争优势包括:(1)HTS 磁体技术——世界纪录级 20T REBCO 磁体设计受 IP 保护,并已在规模上验证;该设计发表在经过同行评议的 IEEE TAS 文献中,增强可信度,但也降低了一部分保密性;(2)MIT PSFC 合作——独家接触全球领先聚变物理研究团队之一,以及 50+ 名 MIT 博士级研究人员深度嵌入项目;(3)先发 PPA 承诺——Google(200 MW)和 Eni($1B+)承诺带来收入可见性,并释放市场验证信号;(4)资本厚度——约 $3B 融资提供现金跑道,支撑第一等离子体和商业电站 FID 决策,近期无需稀释性融资。关键护城河风险:(a)HTS 磁体 IP 可以被绕开设计;SuperOx、SuNAM 和 Fujikura 生产 REBCO 带材,其他公司(Tokamak Energy、Helion)也在开发竞争性 HTS 磁体项目;(b)科学基础已公开——CFS 的物理基础部分开放,降低资金充足竞争者的进入壁垒;(c)如果 SPARC 未能达到 Q>1,护城河叙事会坍塌;(d)PPA 先发不保证 ARC 承购——买家可以重谈,或分散到 SMR。 [CP017, CP018, CP019, CP020, CP021]
| 护城河 / 风险 | 类型 | 驱动因素 | 概率(1-5) | 严重性(1-5) | 缓释因素 | CFS 应对 |
|---|---|---|---|---|---|---|
| HTS REBCO 磁体 IP | 护城河 | 20T 世界纪录;授权设计 | N/A | N/A | CFS IP 组合;与 REBCO 供应商建立制造合作 | 选择性发表;深化制造合作 |
| MIT PSFC 科研合作 | 护城河 | 50+ 名 MIT 研究人员嵌入;SPARC 共同设计 | N/A | N/A | 长期研究协议;CFS 与 MIT 师资 / 员工重叠 | 用股权留住研究人员;深化 MIT IP 授权 |
| 首个超大规模云厂商 PPA(Google) | 护城河 | 市场验证信号;声誉加成 | N/A | N/A | PPA 条款(价格 / 违约罚则未公开) | 作为下一个买家的背书 |
| Helion 在 SPARC Q>1 之前向 Microsoft 交付 50MW | 风险 | 路线竞争;交付日期更早 | 2 | 4 | CFS 物理基础验证更多;托卡马克规模更大 | 加快 SPARC 里程碑;讲清物理优势 |
| SMR 在 ARC 商业运营前实现成本持平 | 风险 | NuScale 取消项目说明风险,但 TerraPower 仍在推进 | 3 | 4 | ARC 无长寿命废物;ESG 叙事对超大规模云厂商更干净 | 瞄准特别看重 ESG 画像的超大规模云厂商买家 |
| 竞争性 REBCO HTS 磁体项目(Tokamak Energy) | 风险 | Tokamak Energy 2024 年 HTS 磁体;SuNAM、SuperOx 磁体供应 | 3 | 3 | CFS 20T 纪录带来 2–3 年领先;集成 IP 更难复制 | 借 VIPER 垂直整合磁体制造 |
| SPARC 未达 Q>1,或时间表滑坡超过 2 年 | 风险 | 首台套实验自带技术风险 | 2 | 5 | MIT 验证设计基础;等离子体参数保守 | 通过 Series B2 和 PPA 保留额外运营现金跑道 |
| 聚变寒冬(资本荒) | 风险 | 若多家聚变公司错过里程碑,将波及整个行业 | 2 | 4 | CFS 拿到 $863M B2,并签下大型 PPA | 维持战略投资方(Eni、Google)作为锚定资本 |
概率和严重性是基于公开证据和竞争分析的定性估计(1=低,5=高)。
[CP017, CP018, CP019, CP020, CP021, CP022]关键绩效指标展示 CFS 的竞争位置:累计融资(约 $3B)、 磁场纪录(20T)、SPARC 目标等离子体增益(Q>1)、ARC 电站规模(400 MWe)和 已签 PPA 体量(Eni 约 $1B+,另有 Google 200MW)。
[CP001, CP002, CP008, CP017, CP018, CP020]3.5 竞争风险与开放问题
CFS 的竞争风险登记包括:(1)Helion 在 SPARC 实现 Q>1 前向 Microsoft 交付 50 MW,使超大规模云厂商信心转向其他聚变路线;(2)NuScale 或 TerraPower SMR 在 2029–2030 年实现具成本竞争力的基荷定价,压缩 CFS 窗口;(3)竞争性 REBCO HTS 磁体项目(Tokamak Energy 的 HTS 项目,融资约 $335M;MIT 周边创业公司)削弱磁体 IP 护城河;(4)一个或多个资金充足的后进入者(由 BHP/Rio Tinto 或主权财富基金支持),凭更低资本成本超过当前领先者的融资能力;(5)SPARC 表现不佳触发聚变「寒冬」,压低整个板块私人投资,在商业化前形成资本低谷。这些风险不对称:正向结果(SPARC Q>1)会成为所有聚变公司的强正面催化;负向结果(错过里程碑)对 CFS 尤其有害,因为它是托卡马克路线的旗手。 [CP022, CP023, CP024, CP025]
3.6 展示项
04财务情况
4.1 收入模式与商业模式
截至运行日,CFS 完全尚未产生收入。公司计划中的收入模式,以 ARC 商业聚变电站电力的长期购电协议(PPA)为核心。CFS 已签署两份有条件 PPA:Google(200 MW,2025 年 6 月宣布)和 Eni(总承诺 $1B+,2025 年 9 月宣布)。两份 PPA 都以 ARC 实现商业运行为前提;价格、交付时间线触发条件和罚则条款均未披露。投资者材料中暗示的第二个收入杠杆,是向未来由被授权方建设的 ARC 电站授权聚变技术(磁体设计、托卡马克几何),类似部分 SMR 开发商提出的授权模式。该授权收入流尚未合同化。公司也获得基于里程碑的政府拨款(DOE INFUSE;2024 年 $8M 里程碑拨款),但这些是非经常性研究补贴,不是收入。 [CI001, CI002, CI003, CI004]
| 收入来源 | 类型 | 状态 | 合约方 | 估计价值 | 条件 | 时间线 |
|---|---|---|---|---|---|---|
| ARC 上网电力 — Google PPA | 购电协议 | 已签署(附条件) | Google / Alphabet | 200 MW,每 MWh 价格未披露 | ARC 开始商业运营 | 2030 年代初 |
| ARC 上网电力 — Eni PPA | 购电协议 | 已签署(附条件) | Eni S.p.A.(能源公司) | 总承诺 $1B+(MWh 价格未披露) | ARC 开始商业运营 | 2030 年代初 |
| 技术授权 — 未来 ARC 电站 | IP 授权 | 计划中(未签约) | 公用事业合作伙伴 / 未来被授权方 | 未披露 | 需要多座 ARC 电站 | 2035+ |
| DOE INFUSE 补助 | 政府补助 | 进行中(按里程碑) | 美国能源部 | $8M 已确认(2024);补助总额估计约 $15–20M | 向 DOE 交付里程碑 | 2024(进行中) |
| 发电演示收入 — SPARC | 研究收入 | 计划中(内部) | 无(内部) | 无;SPARC 是物理演示,不是商业电站 | N/A | N/A |
ARC PPA 收入取决于商业运营,也依赖 SPARC 达成 Q>1、ARC 完成 NRC / 监管审批和建设。
[CI001, CI002, CI003, CI004]瀑布图展示 CFS 从当前 $0 收入到首座 ARC 电站年收入的路径, 包括关键里程碑关口:SPARC Q>1、监管批准、ARC FID 和商业 运营启动。
收入估计由分析师基于公司披露的 LCOE 目标和容量计算。PPA 价格未 披露。建设时间线为示意。
[CI001, CI002, CI004, CI009, CI016]4.2 融资历史与资本结构
截至 2025 年 8 月,CFS 累计完成约 $3 billion 私募股权融资,是全球资本最充足的私人聚变公司。融资历史为:种子轮约 $50M(2018,Eni);Series A $115M(2019,Eni、Khosla、BEV、The Engine);Series B $1.8B(2021,史上最大私人聚变融资;Tiger Global 领投,Breakthrough Energy Ventures、Google、Khosla、Altimeter、Temasek 等参与);Series B2 $863M(2025 年 8 月,新闻稿称为「up round」但未披露估值,由 Khosla、Google、Coatue 和老股东完成)。四舍五入后,累计融资约 $2.8–3.0 billion。CFS 未披露公开债务、政府贷款担保或项目融资义务。公司资本结构为 100% 股权。这意味着在 ARC 产生收入前,所有烧钱都由股东承担。ARC 建设(估计 $2.5B+)将需要一轮非常大规模的追加股权融资、项目融资,或 DOE Loan Programs Office(LPO)Title XVII 担保贷款——这些都尚未确认。 [CI005, CI006, CI007, CI008, CI009]
| 融资事件 | 日期 | 金额 | 领投方 | 累计融资 | 用途 | 现金跑道影响 |
|---|---|---|---|---|---|---|
| 种子轮 / Eni 创始投资 | 2018 | ~$50M | Eni S.p.A.(能源公司) | ~$50M | CFS 创立、早期研发、VIPER 磁体项目 | 初始现金跑道约 2 年 |
| Series A 轮 | 2019-06 | $115M | Eni、Khosla Ventures、BEV、The Engine 等投资方 | ~$165M | HTS 磁体开发、SPARC 设计 | 现金跑道支撑至 SPARC FID |
| Series B 轮 | 2021-12 | $1,800M | Tiger Global、BEV、Google、Khosla、Altimeter、Temasek 等 | ~$1,965M | SPARC 建设、Devens 设施、ARC 设计 | 现金跑道覆盖 SPARC 建设期 |
| Series B2 轮 | 2025-08 | $863M | Khosla Ventures、Google、Coatue;老股东跟投 | ~$2,828M | SPARC 完工、SPARC 2026 年首等离子体、2027 年 Q>1、ARC FID 规划 | 现金跑道覆盖至 SPARC Q>1(约 2027) |
| Pre-ARC 轮(预测) | 约 2028(估计) | $3–5B(估计) | 未公开披露 — 战略投资、项目融资、DOE LPO | 约 $6–8B(估计) | ARC 电站建设 | 支撑 ARC-1 商业运营 |
Pre-ARC 轮仅为分析师预测;CFS 未披露。若符合条件,DOE LPO Title XVII 可为 ARC 建设提供 $2–5B 担保贷款。
[CI005, CI006, CI007, CI008, CI009, CI021]时间线覆盖 CFS 从 2018 年成立到预计 ARC 首次供电(约 2032)的资本事件, 展示股权融资额、估计累计资本投入,以及预计 ARC 建设还需要追加的一轮融资。
所有预计日期和资本金额均为分析师估计;CFS 尚未披露正式财务预测。
[CI005, CI006, CI007, CI008, CI009, CI016]4.3 成本结构与烧钱速度
CFS 的成本结构主要由研发和 SPARC 建设资本开支驱动。公司没有公开利润表数据;CFS 是私人公司,未披露收入、EBITDA 或净亏损。基于员工数(截至 2025 年 800–1,000+ 人)、典型航空航天 / 深科技全成本人工(约 $250–$350K / 人 / 年,含福利和股权),以及已知资本成本(Massachusetts Devens 的 SPARC 建设、场地建设、磁体制造),估计年度运营成本为 $300–$500 million。这与两点一致:(a)$863M Series B2 被描述为提供现金跑道,覆盖 SPARC 第一等离子体并推进至 ARC FID;(b)$1.8B Series B(2021)在 4 年 SPARC 建设中已大体投入。上述仅为分析师估计——没有经审计财务数据。按 $300–500M / 年估计烧钱速度,2025 年 8 月 $863M B2 带来的现金跑道约 18–36 个月,足以覆盖 SPARC Q>1(2027),但不能覆盖到 ARC FID(约 2028–2030)。 [CI010, CI011, CI012, CI013]
4.4 单位经济性与 ARC 电站经济性
CFS 公开引用过 ARC 约 $50–$70/MWh 的目标平准化度电成本(LCOE)。Georgetown Space Policy Institute 的一项分析(2025)评估了 ARC 经济模型,认为如果建设成本和容量因子目标达成,$50–$70/MWh 目标具备可行性;但该分析也指出,首创核能 / 聚变电站建设通常比初始估计高出 50–100%。ARC 设计容量为 400 MWe,容量因子约 90%;按目标 LCOE,每座电站年收入为约 $175–$245 million。按当前企业稳定清洁电力 PPA 市场价格(SMR 竞争性招标约 $90–$150/MWh),若 LCOE 目标实现,ARC 经济性将有盈利空间。公开领域中关键 ARC 建设成本估计是「首座电站约 $2.5 billion」(见 Georgetown 和 Canary Media 分析);这是公司口径估计,缺乏独立验证。NuScale 的 CFPP 取消(实际 $89/MWh vs. 原始估计约 $55/MWh)是最相关的警示先例。 [CI014, CI015, CI016, CI017, CI018]
| 收入杠杆 | 定价模型 | 价格水平 | 依据 | 置信度 | 可比项 |
|---|---|---|---|---|---|
| ARC 电力(PPA) | 按 MWh 计价,长期合约 | $50–$70/MWh(LCOE 目标) | 公司工程估算;未获独立验证 | 低 | NuScale CFPP 报价:约 $89/MWh(已取消);光伏 + 储能:$45–$65/MWh |
| ARC 电力 — 市场可比 | 企业 PPA,用于稳定清洁电力 | $90–$150/MWh(先进核能市场 2024–25) | 第三方:公用事业竞争性招标 | 中 | Microsoft-Helion、TerraPower 公用事业协议(价格未披露) |
| 技术授权 | 按单座电站收取特许权使用费或授权费 | 未披露 | 类似核燃料授权或 SMR 交钥匙模式 | 很低 | 未公布授权条款 |
| 电网容量补偿 | 容量市场(FERC PJM) | $20–$50/MWe/年(美国容量市场) | FERC 容量拍卖定价 | 中 | 若 ARC 符合 PJM 容量市场(Virginia)资格则适用 |
所有 ARC 定价数字都是估计或目标。唯一经验证的近期收入数字是 $8M DOE 里程碑补助(2024)。PPA 价格均未披露。
[CI014, CI015, CI016, CI017]| 指标 | 数值(估计) | 依据 | 置信度 | 可比项 | 风险因素 |
|---|---|---|---|---|---|
| ARC 电站装机容量 | 400 MWe | CFS 公布的设计规格 | 高 | NuScale:77 MWe/模块;TerraPower Natrium:345 MWe | 技术执行风险 |
| ARC 目标 LCOE | $50–$70/MWh | CFS 工程估算;Georgetown 分析 | 低 | NuScale CFPP:$89/MWh(实际);光伏:$24–$96/MWh | 首台套成本超支 |
| ARC 假设容量因子 | ~90% | 工程假设(聚变 24/7 运行) | 中 | 基荷核电:约 92–95% CF | 商业规模尚未验证 |
| 每座 ARC 电站年收入 | ~$175–$245M | 分析师估算:400MW × 90% CF × $50–70/MWh × 8760h | 低 | 可比基荷核电收入 | 取决于 LCOE + PPA 价格兑现 |
| ARC 建设成本(首座电站) | ~$2.5B | Georgetown / Canary Media 分析引用;来自公司口径 | 低 | NuScale 6 模块电站:约 $6B;Natrium 约 $4B+ | 首台套超支风险 50–100% |
| SPARC 建设成本 | 约 $1.5–2B(估计) | 分析师估算;未公开披露 | 很低 | JET 托卡马克:总计约 £600M;ITER:$25B+ | 商业化收入前资本消耗 |
| 年运营成本估计 | 约 $300–500M/年 | 分析师估算:800–1,000 名全职员工 + 资本支出 | 很低 | 可比同规模深科技公司 | 无经审计披露 |
所有单位经济指标均为估算。CFS 未发布经审计财务报表。收入和成本估算仅供分析使用。
[CI010, CI014, CI015, CI016, CI018]比较 ARC 关键财务估算区间:LCOE 目标(低 / 高)、ARC 建设成本 估计和单座电站年收入。图中显示单元经济的不确定性很大。
所有估计均由分析师推导或公司披露;没有独立审计财务数据。风险 情景按 NuScale 先例假设成本超支 50%。
[CI014, CI015, CI016, CI017, CI018]将 ARC LCOE 目标($50–$70/MWh)与关键基准对比:NuScale CFPP 实际 投标价($89/MWh,项目已取消)、太阳能 + 4 小时储能中位数($65/MWh),以及超大规模云厂商 企业 PPA 对先进核能的价格区间($90–$150/MWh)。图中显示,ARC 必须打中激进 成本目标,才有竞争力。
ARC LCOE 为公司目标;NuScale CFPP 是实际投标价,并最终导致项目取消;太阳能 + 4 小时储能为 Lazard 2024 年中位数估计。
[CI014, CI017, CI018, CI028]4.5 财务风险与资本充足性
CFS 面临四项主要财务风险:(1)ARC 建设资金缺口——Series B2($863M,2025 年 8 月)不足以覆盖 ARC 电站建设(估计 $2.5B+);公司将需要大规模追加融资或项目融资,可能在 ARC FID(约 2028 年)发生;股权稀释或债务服务成本会影响股权价值和单位经济性。(2)收入前周期——时间线每滑一年,就会在没有收入的情况下消耗额外股权资本;ARC 若滑期 3 年,可能需要 $1B+ 追加资金。(3)LCOE 验证——ARC 的 $50–$70/MWh 目标没有独立验证;NuScale 先例显示有 50–100% 成本超支风险,可能把 LCOE 推高到 $75–$140/MWh,削弱市场竞争力。(4)PPA 条件性——Google 和 Eni PPA 均以商业运行为前提;没有公开证据显示存在罚则条款,可在 PPA 被重谈或取消时补偿 CFS。披露有限:CFS 不提供公开 P&L、资产负债表或现金流量表。 [CI019, CI020, CI021, CI022, CI023]
| 财务指标 | 公开可得性 | 缺失原因 | 尽调路径 |
|---|---|---|---|
| 年收入 | 不可得(尚未商业化收入) | CFS 商业收入为零;私营公司无需披露 | 资料室:确认零收入状态;核验任何补助收入 |
| EBITDA / 净亏损 | 不可得 | 私营公司;无 SEC 申报义务 | 资料室:经审计损益表;投资人报告 |
| 账上现金 / 资产负债表 | 不可得 | 私营公司 | 资料室:最新经审计资产负债表;董事会报告 |
| 烧钱速度 | 估计($300–500M/年)— 未披露 | 私营公司 | 资料室:月度现金报告;CFO 预测模型 |
| SPARC 建设成本(实际 vs. 预算) | 不可得 | 私营公司;建设进行中 | 资料室:项目成本跟踪报告 vs. 预算 |
| ARC 建设成本估算 | 仅为近似值(约 $2.5B,来自公司口径) | 未发布独立评估 | 委托独立技术经济评估 |
| 员工数(精确) | 约 800–1,000+(估计) | 未正式披露;来自 LinkedIn / 新闻推算 | 资料室:按职能拆分的 HR 员工数 |
| 估值(当前) | 未披露(B2 被称为估值上升轮) | 私募轮;无强制估值披露 | 资料室:股权结构表和最新 409A 或轮次条款 |
本表列出任何投资人都需要补齐的重大已知尽调缺口。
[CI022, CI023]4.6 展示项
05产品与技术
5.1 核心技术:HTS 磁体与紧凑型托卡马克
CFS 的基础创新,是把经过验证的托卡马克等离子体物理,与能够产生远强于传统超导磁体磁场的高温超导(HTS)REBCO(稀土钡铜氧化物)磁体结合。SPARC 设计采用 20-Tesla 环向场磁体——2021 年 9 月演示时,这是全球最强超导磁体,并于 2024 年 3 月在 IEEE Transactions on Applied Superconductivity 经过同行评议发表。物理优势来自几何:更强磁场能在小得多的反应堆体积中约束等离子体(SPARC 主半径 1.85 m,ITER 为 6.2 m),从而大幅降低建设成本和时间线。SPARC 设计以氘-氚(D-T)燃料模式运行,目标实现 Q>1(输出聚变功率 > 输入加热功率),目标聚变功率约 140 MWth。VIPER(Variable Integrated Prototype Engineering Run)电缆是 CFS 内部开发的高性能 HTS 电缆系统,使磁体能在 20T 和高电流密度下可靠运行——这对紧凑几何至关重要。 [CE001, CE002, CE003, CE004]
| 组件 | 功能 | 状态 | 技术成熟度(TRL) | 关键规格 | IP 归属 | 依赖 |
|---|---|---|---|---|---|---|
| VIPER HTS 电缆系统 | 高电流 REBCO HTS 电缆,支撑紧凑型 20T 磁体 | 已设计;正为 SPARC 生产 | 6 | REBCO,工程临界电流密度 >1000 A/mm² | CFS 自研 | REBCO 带材供应(SuperOx、SuNAM、FUJIKURA) |
| 20T HTS 环向场磁体 | 等离子体磁约束 | 已演示(2021);经同行评审(2024) | 6 | 20 Tesla;REBCO;HTS 世界纪录 | CFS 自研(MIT 共同开发) | VIPER 电缆;REBCO 带材 |
| SPARC 低温恒温器 | 等离子体室的真空 + 低温外壳 | 低温恒温器底座已于 2025 年 3 月安装 | 5 | 主半径 1.85m 的托卡马克低温恒温器 | CFS 自研 | Devens MA 土建施工 |
| SPARC 真空室 | 面向等离子体的第一壁;D-T 等离子体包容 | 2025 年 10 月安装 | 5 | 不锈钢;D-T 等级;中子屏蔽 | CFS 自研 | SPARC 低温恒温器;磁体系统 |
| SPARC 加热系统(NBI/ICRH) | 将等离子体加热到聚变温度(100M°C) | 采购 / 安装中 | 4 | 中性束注入 + 离子回旋 RF | 从聚变行业供应商采购 | SPARC 真空室;电源 |
| ARC 氚增殖包层(设计) | 氚燃料增殖 + 热能提取 | 概念设计(早期) | 2–3 | 锂基包层;氦冷却剂;氚增殖比 >1.05 | CFS 自研(设计) | 氚供应;包层材料研发 |
| ARC 商业电站结构 | 400 MWe 商业聚变电站 | 概念设计 | 2–3 | 400 MWe;可拆卸线圈;Chesterfield VA 场址 | CFS 自研 | SPARC Q>1;NRC / 许可;ARC 融资 |
| 系统层 | 组件 | 技术选择 | 成熟度 | 核心风险 | 替代路径 |
|---|---|---|---|---|---|
| 等离子体约束 | 托卡马克构型(D 形截面) | 物理基础已验证(70+ 年) | 高 | 放大到 SPARC 体量;首次等离子体仍有未知问题 | FRC(Helion)、MTF(General Fusion)、Z 箍缩(Zap) |
| 磁场 | 20T REBCO HTS 磁体(VIPER 电缆) | 已在 20T 验证(2021);经同行评议(2024) | 高(磁体);中(集成系统) | REBCO 带材供应链;SPARC 中 12T 接头性能 | LTS 磁体(磁场更低、反应堆更大) |
| 等离子体加热 | NBI + ICRH(离子回旋) | ITER/JET 项目验证过的成熟技术 | 中 | 与 SPARC 集成;达到 Q>1 所需的功率水平 | ECH(电子回旋) |
| 燃料 | 氘-氚(D-T) | 聚变截面最佳;所需温度最低 | 中(D-T 运行;氚处理) | 氚供应和增殖;放射性材料处理 | D-He3(Helion 路径)、p-B11(TAE 路径) |
| 能量提取 / 电力转换 | 蒸汽朗肯循环(热转换) | 成熟技术(标准电站) | 中(聚变包层集成仍属新方案) | 氚增殖包层设计;氦冷却剂集成 | 直接转换(Helion 路径) |
| 反应堆结构 | 可拆卸 HTS 线圈设计 | CFS 的新设计;不是标准托卡马克 | 中低(维护优势尚未验证) | 12T 下可拆卸接头的可靠性 | 固定线圈设计(ITER 标准) |
流程图展示 CFS 产品架构的三段:SPARC(物理演示)输入 ARC 设计,ARC 再为 Google 和 Eni 的 PPA 交付商业电力。
[CE001, CE002, CE005, CE010, CE013, CE016]散点图按成熟度(TRL)和对商业运营的关键性定位 CFS 产品组件。 VIPER 磁体和 SPARC 真空室处于高成熟度 / 高关键性区间; 氚增殖包层成熟度低、关键性高,是核心风险象限。
TRL 和关键性数值为分析师估计。氚增殖包层是关键性最高、 成熟度最低的环节。
[CE001, CE002, CE006, CE009, CE014, CE016]5.2 SPARC:物理演示装置
SPARC 是首创物理演示托卡马克,正在 CFS 的 Massachusetts Devens 设施建设。它不是商业电站——不会产生净电力——但将首次在紧凑装置中演示净聚变能量增益(Q>1)。SPARC 组装里程碑包括:低温恒温器底座安装完成(2025 年 3 月)、真空室安装(2025 年 10 月,按公司更新)。第一等离子体目标为 2026 年;持续 D-T 运行和 Q>1 演示目标为 2027 年。SPARC 设计参数在 2020 年 Journal of Plasma Physics 的一组同行评议论文中有完整记录(MIT PSFC,8 位共同作者),为私人聚变机器提供了最透明的物理基础。SPARC 关键设计参数:主半径 1.85 m,副半径 0.57 m,轴上磁场 12T,等离子体电流 8.7 MA。SPARC 建设由 Series B 和 B2 轮融资支持。 [CE005, CE006, CE007, CE008, CE009]
5.3 ARC:商业聚变电站设计
ARC(Affordable, Robust, Compact)是 CFS 的商业电站设计,计划交付为一座并网 400 MWe 聚变电站。ARC 在 SPARC 已演示的物理基础上放大到净电输出:设计把轴上磁场提升至约 12T(与 SPARC 相同),但采用更大的等离子体体积和氚增殖包层,以维持 D-T 运行。ARC 设计场地位于 Virginia 的 Chesterfield County(从 Dominion Energy 租赁),计划接入 PJM East 电网。ARC 关键设计特征包括:可拆卸 HTS 磁体线圈(CFS 创新,相比固定线圈设计可更快、更低成本维护)、用于氚增殖的高温包层、氦冷增殖包层把聚变中子通量转化为热能,再通过蒸汽循环发电。ARC 的商业可行性目标是 $50–$70/MWh LCOE;这要求首座电站估计 ~$2.5B 的建设成本守住纪律。 [CE010, CE011, CE012, CE013]
| 用例 | 用户 / 操作方 | 输入 | 输出 | 工作流步骤 | 使用的聚变组件 |
|---|---|---|---|---|---|
| SPARC 净能量演示 | CFS 物理学家和操作人员 | D-T 燃料;外部加热(NBI、ICRH) | 聚变功率输出 > 加热输入(Q>1) | 1.燃料注入;2.等离子体启动;3.加热至 100M°C;4.维持约束;5.测量 Q | 全部 SPARC 系统 |
| ARC 基荷发电 | 公用事业公司 / 超大规模云厂商作为购电方 | 氚 + 氘燃料;电网接入 | 400 MWe 上网电力;24/7 稳定电力 | 1.燃料循环;2.等离子体运行;3.换热;4.蒸汽循环;5.并网输出 | ARC 整座电站 |
| 向 Google/Eni 履行 PPA | CFS 运营团队 | ARC 电站运营 | 200MW(Google)+ Eni $1B+ 电力承诺 | 1.PPA 在商业运行时触发;2.计量电量;3.向买方开票;4.按合同交付 | ARC 商业电站 |
| 氚增殖(自给) | CFS 电站操作人员 | 增殖包层中的 Lithium-6;来自 D-T 的中子 | 燃料循环所需的氚(T) | 1.等离子体产生中子通量;2.Li-6 + n → He + T;3.回收 T;4.回到燃料循环 | ARC 增殖包层 |
从 SPARC 物理验证到 ARC 商业供电,再到向 PPA 对手方(Google、Eni) 交付电力的运营流程,标出关键里程碑关口。
[CE008, CE011, CE013, CE019, CE022]5.4 技术风险与成熟度评估
CFS 的技术处于技术成熟度等级(TRL)约 4–5:组件验证已完成(20T HTS 磁体,经过同行评议),系统集成在建(2025 年 SPARC 组装),但完整原型运行尚未实现。关键技术风险包括:(1)SPARC 规模的等离子体约束——Q>1 里程碑不能保证;尽管物理基础强,每台新托卡马克都有独特的第一等离子体挑战;(2)ARC 的氚增殖包层——规模化未演示;商业聚变电站需要自持氚燃料循环,而 ARC 包层设计未公开验证;(3)HTS 带材供应链——CFS 依赖外部供应商(SuperOx、SuNAM、FUJIKURA)的 REBCO 带材;若不扩产,当前全球 REBCO 产能可能不足以支撑 ARC 机队;(4)12T 下的可拆卸线圈接头——这是 CFS 的新设计特征,可实现维护通道,但接头在满场下的热性能和电磁性能尚未在运行中演示;(5)电力转换和热循环——ARC 的氦包层和蒸汽循环仍处早期设计阶段;与电网级电力电子的集成是未来里程碑。 [CE014, CE015, CE016, CE017, CE018]
CFS 产品路线图关键技术依赖的有向无环图。 图中显示,HTS 带材供应、氚、SPARC 里程碑和融资必须全部对齐, ARC 才能投入商业运营。
[CE015, CE016, CE021, CE022]5.5 产品路线图与开发里程碑
CFS 的产品路线图分三阶段:阶段 1——SPARC 物理演示(2026–2027):2026 年第一等离子体,2027 年 D-T 运行和 Q>1,2027–2028 年物理验证活动。阶段 2——ARC 设计与建设(2028–2032):ARC FID 2028–2029(取决于 SPARC 成功和融资),2029–2032 年在 VA Chesterfield County 建设,2032 年并网调试。阶段 3——ARC 机队部署(2033+):向更多电站授权 ARC 设计,与公用事业伙伴建立机队订单簿,开发经济性更优的第二代 ARC。关键依赖包括:SPARC Q>1 是 ARC FID 的门控里程碑;来自 NRC 或等同 NRC 的聚变监管框架批准(ADVANCE Act 2024 下建立);Chesterfield County 的 PJM 电网互联;氚供应协议(当前来源:CANDU 反应堆副产品,全球仅约 0.5 kg / 年——足够 SPARC,但不足以支撑商业机队)。 [CE019, CE020, CE021, CE022, CE023]
| 领域 | 要求 | 状态 | 监管机构 / 组织 | 合规证据 | 缺口 |
|---|---|---|---|---|---|
| 放射性材料处理(MA) | 放射性材料许可证 | 2024 年 10 月已发放 | Massachusetts MEMA | MEMA 新闻稿确认许可证 | D-T 运行所需氚处理许可证尚未取得 |
| 核监管框架 | 面向聚变的 NRC 框架 | ADVANCE Act(2024)建立非裂变框架 | U.S. NRC / Nuclear Regulatory Commission(美国核监管机构) | ADVANCE Act 于 2024 年 7 月签署 | CFS 尚未提交 NRC 许可证申请(SPARC 前置里程碑) |
| 场址安全 / 环境 | ARC 场址环境影响评估 | 未启动(概念设计阶段) | Virginia DEQ / FERC | 无(尚未申报) | ARC 施工许可前必须完成完整 EIS |
| 电网互联 | ARC/PJM 的 FERC 互联审批 | 尚未提交(FID 前) | FERC / PJM Interconnection | 无(尚未申报) | 需要提交 FERC 电网排队申请;典型周期 3–5 年 |
| ISO/核标准 | ASME NQA-1 质量保证 | 已部分用于 SPARC 建设 | ASME | 未公开记录 | ARC 监管申报前需要完整 QA 体系 |
| 阶段 | 里程碑 | 目标日期 | 关口条件 | 状态 | 风险 |
|---|---|---|---|---|---|
| SPARC 建设 | SPARC 组装完成 | 2025–2026 | 所有系统已安装在 Devens | 进行中——低温恒温器 + 真空室已安装;加热系统正在采购 | 供应链延迟;首次集成问题 |
| SPARC 运行 | 首次等离子体 | 2026 | 在 SPARC 中启动等离子体 | 计划中 | 首次等离子体挑战未知;加热系统调试 |
| SPARC 运行 | SPARC Q>1(净能量增益) | 2027 | 实测聚变功率 > 加热功率输入 | 计划中 | 物理不确定性;目标参数达成;氚处理 |
| ARC 设计 | ARC 详细设计完成 | 2028 | SPARC 达到 Q>1;ARC FID 获批 | 预设计 | 完全取决于 SPARC Q>1 |
| ARC 建设 | ARC 在 Virginia Chesterfield 破土 | 2028–2029 | NRC 许可证(或同等许可)、已提交 PJM 互联申请、融资已交割 | 规划中 | 监管;融资;场址准备 |
| ARC 运营 | ARC-1 首次向电网送电 | 2030 年代初 | 建设完成;电网互联上线 | 概念阶段 | 所有前置里程碑;LCOE vs. 市场;PPA 履约 |
| ARC 机组群 | 第二座 ARC 电站 FID | 2033–2035 | ARC-1 已商业运行;机组群融资可得 | 未启动 | ARC-1 成功;资本可得;买方管线 |
所有日期都是 CFS 目标或分析师估计;CFS 尚未公开发布正式项目进度表。
[CE019, CE020, CE021, CE022, CE023]5.6 展示项
06客户情况
6.1 当前客户基础与收入状态
截至运行日,CFS 没有付费客户,也没有商业收入。公司完全以研发和商业化前实体运行。截至 2026 年 5 月,公司所谓「客户基础」仅由远期 PPA 交易对手构成:Google(Google LLC,Alphabet Inc. 子公司)将从 ARC 电站购买 200 MW 电力,Eni S.p.A.(意大利跨国能源公司)承诺购买总额 $1B+ 的电力。两份 PPA 都以 ARC 实现商业运行为前提,尚未确认为收入。唯一产生现金的活动是 DOE 里程碑拨款(2024 年约 $8M),这些是研究拨款,不是客户收入。CFS 也通过 INFUSE 和竞争性拨款项目,将美国能源部作为研究类政府客户,但规模(总计 $8–15M)相对于公司累计融资并不重要。 [CU001, CU002, CU003, CU004]
6.2 具名客户画像
Google LLC(Alphabet Inc.):Google 于 2025 年 6 月签署全球首个企业聚变 PPA,从 CFS 位于 Virginia Chesterfield County 的 ARC 电站购买 200 MW 电力。Google 的动机是其到 2030 年实现 24/7 无碳能源(CFE)的承诺——即每一单位用电都要全天候来自清洁来源。单靠太阳能和风能无法满足该标准;Google 需要可调度、稳定、24/7 零碳电力。Google 也直接投资了 CFS(Series B 和 B2),使其同时作为客户和财务支持方实现利益一致。Eni S.p.A.:这家意大利油气巨头于 2025 年 9 月签署 $1B+ 电力承诺,购买 ARC 电力。自 2018 年种子轮以来,Eni 一直是 CFS 的基石投资者,并在所有轮次中持有股权。Eni 同时是投资者和买家,使双方在商业结果上更一致;但这也意味着 Eni 在定价谈判上的独立性可能有限,而且 Eni 财务状况或战略优先级一旦恶化,可能同时影响融资和承购。Eni 的动机包括为其工业能源使用脱碳,并在 Enilive 和清洁能源战略下展示清洁能源领导力。 [CU005, CU006, CU007, CU008, CU009, CU010]
| 细分客群 | 画像 | 代表公司 | CFS 为什么适配 | 采购流程 | 时间尺度 |
|---|---|---|---|---|---|
| 超大规模数据中心运营商 | 24/7 CFE 承诺;>100 MW 电力块;15–25 年 PPA 周期 | Google、Microsoft、Amazon、Meta | 在 100+ MW 规模上,聚变是唯一稳定的零碳电力;没有间歇性风险 | 可持续发展 / 能源 VP 签署;CFO 批准;董事会层面承诺 | 2025–2035+(远期 PPA) |
| 受监管电力公用事业 | 替代煤电 / 气电;基荷采购;受 FERC/PUC 监管 | Dominion Energy、Duke、Xcel、Entergy 等公用事业 | ARC 在 IRP 模型中承担 24/7 基荷角色;无废料;选址比裂变更容易 | 资源规划团队;监管申报;IRP 更新 | 2028–2035(SPARC Q>1 之后) |
| 工业能源买方 | 高耗能制造;绿氢;钢铁 | Nucor、ArcelorMittal、绿氢生产商 | 绿氢和 EAF 钢铁离不开稳定 24/7 电力 | 能源采购经理;长期工业供能协议 | 2032+(ARC-1 之后) |
| 油气公司 | 能源转型;清洁电力采购;工业脱碳 | Eni(已签署)、BP、Shell、TotalEnergies | 聚变 PPA 纳入净零承诺;与 CFS 创始人股权安排一致 | 董事会层面 ESG 战略;能源采购;投资组合 | 2025–2032+(Eni 已签署) |
| 政府 / 国家安全 | 电网韧性;偏远地区供电;国家聚变项目投资 | U.S. DOE、DOD、国家实验室 | 聚变可作为能源独立技术;不涉及裂变材料武器扩散 | 赠款 / 合同流程;国会拨款 | 持续中(赠款),2030 年代+(供电) |
除已签署的 Google 和 Eni PPA 外,其他买方分层都带有推测性。CFS 尚未披露这两家买方之外的正式客户管线。
[CU001, CU005, CU006, CU007, CU010, CU019]| 客户 | 细分客群 | 公布日期 | 承诺类型 | 规模 | PPA 价格 | 条件 | CFS 投资方? |
|---|---|---|---|---|---|---|---|
| Google(Alphabet) | 超大规模数据中心运营商 | 2025 年 6 月 | 购电协议(PPA) | 200 MW | 未披露 | 取决于 ARC 商业运行 | 是(Series B、B2) |
| Eni S.p.A.(能源公司) | 油气公司(能源转型) | 2025 年 9 月 | 电力承诺($1B+) | 总价值 >$1B | 未披露 | 取决于 ARC 商业运行 | 是(Seed 至 B2) |
| U.S. Department of Energy(美国能源部) | 政府 R&D 资助方 | 多次(最近一次为 2024) | DOE INFUSE + 里程碑赠款 | ~$8M 赠款(2024) | N/A(赠款) | 基于里程碑的履约 | 否 |
只有三类“客户”有记录;Google 和 Eni 是远期 PPA(尚无收入)。DOE 是赠款资助方,不是商业客户。
[CU003, CU005, CU007, CU009]以 Google 为例,展示超大规模云厂商买方从初次接触、签署远期 PPA 到 ARC 供电交付的旅程。图中呈现多年关系弧线和关键触点。
时间线基于 CFS 目标作示意。实际交付日期取决于 SPARC Q>1 和 ARC 建设 里程碑。
[CU001, CU005, CU006, CU008, CU012, CU013]矩阵汇总 Google、Eni、DOE 以及潜在未来买方的客户证据质量, 维度包括承诺类型、财务价值和 PPA 条款可信度。
[CU003, CU005, CU006, CU007, CU009, CU022]示意性比较 CFS 已签约 PPA 容量与完整 ARC-1 产能, 展示已锁定比例(Google + Eni)与首座 ARC 电站剩余未承诺容量。
Eni PPA 的 MW 容量为估计值;“$1B+”总承诺除以估计 PPA 价格 ~$50–70/MWh × 8760h × 15 年,意味着约 200–250 MW。精确 MW 容量未披露。
[CU002, CU005, CU007]6.3 客户增长与采用轨迹
CFS 的获客轨迹完全由里程碑门控:在 SPARC 证明 Q>1(目标 2027 年)前,新的 PPA 很难有实质意义地签署或激活。采用漏斗分三阶段:(1)在物理证明前签署投机性远期 PPA(Google/Eni 代表这一阶段);(2)SPARC Q>1 之后开发公用事业和工业 PPA 管线,目标包括 Dominion Energy(Virginia,场地合作伙伴)、Mid-Atlantic 公用事业公司和更多超大规模云厂商;(3)从 ARC-2 起形成机队订单簿。Google PPA 是历史上首个企业买家转化为聚变电力承诺的案例,是重要市场验证信号。不过,除 Google 和 Eni 外,没有公开证据显示存在已签署或已承诺的客户管线。CFS 未披露获客指标、买家管线规模或合同期限。 [CU011, CU012, CU013, CU014]
| 阶段 | 买方 | 承诺量 | 里程碑关口 | 时间线 | 状态 |
|---|---|---|---|---|---|
| 推测性远期 PPA | Google、Eni | 200 MW(Google)+ $1B+(Eni) | 无——在 SPARC Q>1 前基于信任签署 | 2025(已签署) | 有效 |
| SPARC Q>1 后签约 | 目标:Dominion、Microsoft/Amazon、1–2 家工业买方 | 估计新增 400–800 MW | 2027 年 SPARC Q>1 后,签约风险下降 | 2027–2029 | 未启动 |
| ARC-1 商业运行 | Google + Eni(既有 PPA 激活) | 200 MW + Eni $1B+ | ARC 首次发电 | 2030 年代初 | 计划中 |
| ARC-2 机组群订单簿 | 公用事业机组群管线;超大规模云厂商机组群;工业 | 机组群总目标 1,000+ MW | ARC-1 完成商业验证 | 2033–2040 | 商业化前规划 |
SPARC Q>1 之后和 ARC-2 的数字是分析师基于类似核能 PPA 采用模式作出的估计。
[CU011, CU012, CU013, CU014]6.4 留存、满意度与扩张风险
公司只有两个远期 PPA 客户,且尚未商业运行,传统 SaaS 式留存指标(NRR、流失率)并不适用。真正的留存风险,是 ARC 交付电力前 PPA 被取消或重谈。关键留存风险包括:(1)若 ARC 商业时间线明显滑出 2030 年代初,Google 或 Eni 重谈 PPA 条款;(2)Google 企业清洁能源战略转向优先采用其他清洁来源(地热、SMR),且发生在 ARC 交付前;(3)能源转型加速或反转时,Eni 的能源战略改变;(4)新的超大规模云清洁能源来源(先进核能 SMR、Helion)在 ARC 前部分满足 24/7 CFE 需求,降低 CFS 的边际价值。Google 已公开的 24/7 CFE 承诺和直接投资显示其承诺度较高,但缺乏公开罚则条款会降低可执行性确定性。ARC-1 之后的扩张机会在于:若 ARC 证明商业可行,Google、Eni 和新买家可能承诺 ARC-2、ARC-3 PPA,为机队订单簿提供深度。 [CU015, CU016, CU017, CU018]
| 指标 | 数值 / 状态 | 数据来源 | 置信度 | 风险 |
|---|---|---|---|---|
| PPA 取消 | 0(未公开出现) | 公开公告 | 高 | 如果 ARC 时间表比 2030 年代初再延后 >3 年,Eni 或 Google 可能取消 |
| PPA 重新谈判 | 0(未披露) | 公开公告 | 中 | 价格或电量条款可能在 data room 场景中重谈 |
| 客户 NPS / 满意度 | N/A(尚无商业运行) | N/A | N/A | ARC 交付电力前均不适用 |
| 客户留存率 | N/A(尚无付费客户) | N/A | N/A | N/A |
| 罚则条款 / 期限承诺 | 未公开披露 | PPA 条款未公开 | 低 | 缺少公开罚则,削弱了留存确定性 |
| Google 24/7 CFE 一致性强度 | 高(Google 是投资方 + PPA 买方;24/7 要求已公开说明) | Google 可持续发展报告 | 高 | Google 战略可能在 2030 截止期后调整 |
收入前阶段不适用传统留存指标。PPA 稳定性是关键替代指标。
[CU015, CU016, CU017]6.5 集中度风险与客户多元化
CFS 面临极高的客户集中风险:Google 和 Eni 两个买方占了合同收入账本的 100%。 任何一方违约、重谈条款或战略转向,都会直接拿掉 CFS 的收入确定性。公司仍处在商业化前阶段, 这种集中度不可避免;但 ARC 接近 FID 时,必须主动压降风险。可用的分散抓手包括: (1) 让 Dominion Energy 成为向 PJM 电网供电的公用事业承购伙伴(尚未签约);(2) 鉴于 Microsoft 或 Amazon 自身有 24/7 CFE 承诺,引入它们作为额外超大规模云厂商买方;(3) 面向美国 DOD 或国家安全场景的分布式聚变电力用途(仍属推测);(4) 弗吉尼亚州的工业买方(钢铁、制氢)。 截至本次报告日期,CFS 尚未宣布 Google 和 Eni 之外的新客户签约。集中风险是早期深科技企业 已知且可接受的特征——第一批客户验证会撬动下一波客户——但投资人建模 ARC 经济性时,应假设 机组规模化部署前至少需要 3–4 个客户。 [CU019, CU020, CU021, CU022]
| 风险 / 扩张 | 类型 | 量级 | 概率 | 缓释 | 尽调问题 |
|---|---|---|---|---|---|
| Google PPA 取消 / 重新谈判 | 集中度风险 | ~200 MW 收入账簿 | 低(Google 也是投资方) | Google 同时是投资方和买方,利益更一致 | 审阅 PPA 提前终止条款;审阅 Google 清洁能源战略演进 |
| Eni PPA 取消 / 重新谈判 | 集中度风险 | $1B+ 总价值 | 中低(Eni 也是投资方) | Eni 的基石股权降低了不利行动风险 | 审阅 Eni 能源转型时间线和投资方董事席位 |
| Dominion Energy 作为新增公用事业买方 | 扩张机会 | 400 MW+ 公用事业 PPA | 中(站点合作伙伴,非已签约买方) | Dominion 承载 ARC 站点;天然是首个公用事业买方候选 | 确认 Dominion IRP 是否纳入 ARC;评估定价预期 |
| SPARC Q>1 后的第二家超大规模云厂商(Microsoft/Amazon) | 扩张机会 | 100–500 MW 新增 PPA | 中高(两者都有 24/7 CFE 要求) | Microsoft 与 Helion 的 PPA 可能提高 CFS 的竞争紧迫感 | 跟踪 Microsoft/Amazon 清洁能源 RFP 动向 |
| ARC-2 / 机组订单簿发展 | 扩张 | 机组群经济性;2,000+ MW 潜力 | 低(取决于 ARC-1 成功) | ARC-1 成功是扩张的关键前提 | 在 ARC-1 资料室中建模机组群扩张经济性 |
展示从认知聚变、签署 PPA 到 ARC 供电交付的分阶段漏斗, 并估计每个阶段的累计买方数量。
“已签 PPA”阶段以上的所有数量均为分析师估计;CFS 尚未披露管线指标。
[CU001, CU002, CU019, CU020, CU021]6.6 图表
07风险
7.1 技术与物理风险
CFS 的根本风险在于,SPARC 未能在目标 2027 年窗口实现 Q>1(科学盈亏平衡,聚变能量增益 >1)。 Q>1 是投资人和客户降风险的首要里程碑;一旦失败或明显延迟,下游会连锁受冲击:Google/Eni PPA 信心丧失,ARC 建设资本无法募集,Helion 和 TAE Technologies 重新卡位竞争。SPARC 设计采用轴上 12 Tesla 的 REBCO HTS 磁体——SPARC 这种尺度此前从未达到这一磁场强度,因此 MIT PSFC 的 SPARC 论文所建模的等离子体约束物理,必须能准确预测新机器里的行为。全新装置的 首次等离子体通常会遇到意外挑战(NIF 从首次等离子体到点火用了 7 年;ITER 较原计划延迟 12+ 年)。SPARC 规模较小,路径更短,但物理问题确实是全新的。第二个关键技术风险是氚增殖: ARC 的 D-T 燃料循环需要靠锂包层自行产氚(TBR >1);全球氚增殖包层技术仍在 TRL 2–3, 还没有电站规模的运行示范。这不是 SPARC 风险(SPARC 使用外部供应的 D-T),但会卡住 ARC 商业运行。CFS 必须与 SPARC 并行发展氚增殖能力,避免 Q>1 之后 ARC 开工前再拖数年。 [CR001, CR002, CR003, CR004, CR005]
| 风险 ID | 风险 | 领域 | 严重性 | 可能性 | CFS 缓释措施 | 剩余风险 |
|---|---|---|---|---|---|---|
| OPS-01 | SPARC 未能达到 Q>1,或仅实现边际增益(Q=0.5–0.8) | 物理 / 工程 | 严重 | 低–中(分析师估计 20–30%) | SPARC 设计模型经独立同行评审验证;2024 年离子温度纪录 | 高:实验前没有保证 |
| OPS-02 | ARC 开工前,氚增殖包层(TBR >1)未完成验证 | 材料 / 中子学 | 高 | 高(全球 TRL 2–3) | 并行推进包层 R&D;与美国国家实验室合作 | 高:全球尚未解决 |
| OPS-03 | ARC 磁体 HTS 带材供应短缺(SPARC 用量的 100x) | 供应链 | 高 | 中 | CFS Materials Facility;SuNAM 和 FUJIKURA 供应协议 | 中:未来 5–10 年的规模化挑战 |
| OPS-04 | SPARC 首次等离子体出现未预见工程问题,调试延迟 1–3 年 | 调试 | 中 | 中(新装置常见) | 吸收既有托卡马克经验;分阶段调试计划 | 中:经验可降低但无法消除该风险 |
| OPS-05 | ARC 建设成本超支(NuScale 式 2x+ 成本膨胀) | 建设经济性 | 高 | 中 | 模块化建设;HTS 带材降本模型 | 高:HTS 带材降本假设偏激进 |
| OPS-06 | Q>1 前,SPARC 等离子体扰动损坏第一壁或真空室 | 等离子体运行 | 中 | 中 | 扰动缓解系统;halo 电流抑制 | 中:标准高场等离子体运行风险 |
风险矩阵按发生概率(列)和严重性(行)映射 12 项 CFS 关键风险。 右上象限标出需要主动缓释的关键风险。
[CR001, CR002, CR006, CR010, CR015, CR019]7.2 供应链与运营风险
每套磁体需要约 300–500 km 的 REBCO HTS 带材;按 SPARC 规模(18 个超导磁体)计算, 意味着约 5,000+ km 带材。商业可用 HTS 带材由全球少数厂商供应:SuperOx(与俄罗斯有关联)、 SuNAM(韩国)、FUJIKURA(日本)和 AMSC(美国)。在当前美国制裁背景下, SuperOx 的俄罗斯关联带来地缘政治供应链风险;CFS 已公开表示正转向 SuNAM 和 FUJIKURA 做多元化供应。但相对 SPARC 需求,SuNAM 和 FUJIKURA 产能有限;扩产需要多年。到了 ARC, 由于需要 50+ 个生产级磁体,HTS 带材总需求会大幅放大(约为 SPARC 的 100x),这是真正的 长周期供应链挑战。CFS 在 Devens 的 Materials Facility 负责制造定制导体,但底层 REBCO 带材仍来自外部供应商。运营风险还包括新机器首次等离子体的复杂度(真空系统完整性、低温系统调试、 破裂事件管理),以及把高场等离子体维持到有意义燃烧时长的全新挑战。 [CR006, CR007, CR008, CR009]
| 风险 ID | 合作伙伴 / 依赖 | 类型 | 严重性 | 可能性 | 缓释措施 | 剩余风险 |
|---|---|---|---|---|---|---|
| DEP-01 | SuperOx(俄罗斯)HTS 带材供应中断 | 供应链 / 地缘政治 | 高 | 中(制裁风险) | SuNAM / FUJIKURA 多元化 | 中 |
| DEP-02 | Google 取消或重新谈判 200 MW PPA | 客户集中 | 高 | 低(Google 是投资方;利益双重一致) | Eni 仍在;分散买方基础 | 低–中 |
| DEP-03 | Eni 取消或重新谈判 $1B+ PPA | 客户集中 | 高 | 低(Eni 是基石投资方) | Google 仍在;分散买方基础 | 低–中 |
| DEP-04 | Dominion Energy 延迟 ARC 站点并网支持 | 基础设施 / 公用事业 | 中 | 低(Dominion 是站点合作伙伴) | 法律站点合作协议;PJM 排队共同提交 | 低 |
| DEP-05 | MIT PSFC 合作缩减或终止(政府资金削减) | R&D 依赖 | 中 | 低–中(DOE 预算不确定性) | CFS 已将多数核心 R&D 内部化;MIT 关系是补充 | 低–中 |
| DEP-06 | DOE INFUSE 或贷款担保计划资金削减 | 政府依赖 | 中 | 低–中(国会预算风险) | CFS 运营不依赖 DOE 拨款 | 低 |
按类型(供应链、监管、财务、客户)映射 CFS 的关键依赖, 展示单点故障节点和冗余路径。
[CR006, CR007, CR013, CR015, CR017, CR022]7.3 商业与财务风险
ARC 目标 LCOE 为 $50–70/MWh,纸面上技术可达,但工程不确定性很大。NuScale CFPP 的先例 很有参考价值:其 LCOE 估计从 $65/MWh 升至 $89/MWh,最终触发 2023 年项目取消。ARC 的 LCOE 模型假设 HTS 带材成本下降 10x、建设成本沿用 2018 年概念设计研究、运营寿命为 12–15 年。任一假设都可能乐观 2x,进而把 ARC LCOE 推到 $100–140/MWh——高于规模化的 风电 + 储能。ARC 建设资本需求估计为每座电站 $2.5–3B;以 CFS 当前约 $3B 资本(含 B2)计算, ARC FID(投资决策)前仍有约 $3–5B 资金缺口。这个缺口需要未来股权融资、债务融资、政府贷款担保 (DOE LPO)或 PPA 预付款来补,但截至 2026 年 5 月都没有落实。Google 和 Eni PPA 以商业运行为 前提,并不是融资工具。SPARC 之后,如果 CFS 以降估值再融下一笔 $5B(假设 SPARC 达到 Q>1 但增益低于预期),财务上可行,但会显著稀释现有投资人。 [CR010, CR011, CR012, CR013, CR014]
| 风险 ID | 止损标准 | 触发止损标准的概率 | 时间线 | 观察路径 |
|---|---|---|---|---|
| OPS-01(SPARC Q>1 失败) | SPARC 经多轮实验后 Q<0.5,且看不到提升至 Q>1 的路径 | ~10–15%(分析师估计) | 2027–2028 | SPARC 实验结果;已发表等离子体参数;独立物理评审 |
| OPS-02(氚 TBR) | 到 2029 年仍未在工程规模验证 TBR>1 包层 | ~15–25% | 2029 | 国家实验室氚包层测试结果;ITER TBR 模块结果 |
| OPS-03(HTS 带材供应) | ARC FID 前未锁定 ARC 所需规模的 HTS 带材供应合同 | ~10–20% | 2028 | CFS 供应链公告;SuNAM / FUJIKURA 产能披露 |
| COM-01(LCOE 超支) | ARC 建设前成本研究中,LCOE 估计超过 $120/MWh | ~15–25% | 2028–2030 | ARC 成本工程研究;可比先进反应堆建设成本 |
| FIN-01(资金缺口) | CFS 未能在 2028 年前筹集 $3–5B ARC 建设资本 | ~20–30% | 2028 | CFS 融资公告;ARC FID 新闻稿;DOE LPO 申请 |
| REG-01(NRC 时间线) | NRC 到 2027 年仍未发布聚变许可框架,导致 ARC 建设施工申请延迟 | ~25–35% | 2027 | NRC 规则制定 Federal Register;先进反应堆许可案卷 |
展示主要风险如何向下游影响级联。SPARC Q>1 失败是根节点, 会带动最严重的下游连锁反应。
[CR001, CR003, CR006, CR010, CR011, CR012]7.4 监管与法律风险
CFS 计划依据 ADVANCE Act(2024 年签署)的先进反应堆许可框架,为 ARC 向 NRC 申请 合并建设与运营许可证(CBOL)。ADVANCE Act 要求 NRC 为先进反应堆 建立新的许可路径,但实施细则仍在制定中——截至 2026 年 Q1,关键的聚变专项监管规则尚未定稿。 NRC 历史上只许可过轻水堆;聚变不产生长寿命放射性废物,但会产生氚和活化结构材料,因此许可边界 仍有模糊地带。ARC 的 FERC 并网队列尚未公开申报;在 PJM 区域,大型新增发电机组的并网研究可能给 项目时间线增加 3–5 年。除联邦 NRC 许可外,还需要弗吉尼亚州的选址和环境审批。公开信息未显示 CFS 面临不利法律行动。CFS 持有约 50+ 项与磁体和等离子体技术相关的专利,但尚未公开披露来自 Tokamak Energy(UK)或 TAE Technologies 的专利挑战。 [CR015, CR016, CR017, CR018]
| 风险 ID | 风险 | 主管机构 / 管辖区 | 严重性 | 可能性 | CFS 缓释措施 | 剩余风险 |
|---|---|---|---|---|---|---|
| REG-01 | ARC 许可申请前,NRC ADVANCE Act 聚变规则尚未定稿 | 美国 NRC,联邦 | 高 | 中 | 参与 NRC 申请前咨询;聘用前 NRC 员工 | 中:许可规则到 Q1 2026 仍在制定 |
| REG-02 | FERC/PJM 电网并网排队延迟(3–5 年) | FERC / PJM,联邦 | 高 | 中高 | 站点靠近 Dominion 输电网络;弗吉尼亚州 VCEA 优先级 | 中高:尚未提交并网排队申请 |
| REG-03 | 弗吉尼亚州选址与环境审批周期 | 弗吉尼亚州 DEQ,州级 | 中 | 低 | 站点位于 Chesterfield County,并获当地政府支持 | 低:Dominion 与县政府已达成合作 |
| REG-04 | ARC 氚处理需满足 NRC Class B 要求 | 美国 NRC,联邦 | 中 | 中 | 构建氚安全论证;对接 DOE 氚实验室 | 中:许可领域新颖 |
| REG-05 | 美国制裁冲击 HTS 带材供应(SuperOx 与俄罗斯有关联) | 美国商务部、OFAC,联邦 | 高 | 中 | 转向 SuNAM(韩国)和 FUJIKURA(日本)分散供应 | 中:非俄罗斯供应扩产需 2–3 年 |
| REG-06 | 竞争聚变企业或 HTS 磁体 IP 持有人提出专利侵权索赔 | 美国法院,知识产权 | 低 | 低 | 持有 50+ 项专利;已开展自由实施分析 | 低:未披露不利 IP 行动 |
7.5 团队与执行风险
CFS 的技术领导力最初由 MIT PSFC 校友奠基,核心人物包括 Dennis Whyte 教授(前 PSFC 主任)、 创始 CEO Bob Mumgaard 和等离子体物理学家 Brandon Sorbom。Whyte 是 CFS 面向投资人和国家实验室的 主要科学信用锚;如果他离开或明显疏离,会释放技术信心转弱的负面信号。CFS 目前约有 1,000 名员工, 从 MIT、国家实验室和工业界招来了一批等离子体物理学家、磁体工程师和系统工程师。公司从 2019 年 约 50 人快速扩张到 2025 年约 1,000 人;规模放大后,文化、设计纪律和建设进度都带来执行风险。 关键执行风险在于 ARC 需要从 R&D 模式切到制造 / 建设模式;这与 SPARC 物理验证阶段所需的组织能力 根本不同。CFS 已开始招聘制造和建设管理人才,但尚未公开披露 ARC 的建设项目管理高管团队。 [CR019, CR020, CR021, CR022]
| 风险 ID | 风险 | 严重性 | 可能性 | 缓释措施 | 剩余风险 |
|---|---|---|---|---|---|
| PEO-01 | Dennis Whyte 离职或降低参与度 | 高 | 低(Whyte 有公开承诺;持有创始人股权) | MIT PSFC 校友管线提供后备深度 | 中:Whyte 作为科学可信度锚点不可替代 |
| PEO-02 | Bob Mumgaard(CEO)过渡风险 | 中 | 低(CEO 正在主导资金充足的融资轮) | 强大的高管团队;Series B2 CEO 可信度 | 低 |
| PEO-03 | 与英国、中国项目争夺等离子体物理学家 / 磁体工程师的人才战 | 中 | 中(全球聚变人才竞争) | MIT 管线;股权薪酬;使命驱动文化 | 中:全球聚变人才稀缺 |
| PEO-04 | R&D 向建设文化转型失败 | 中 | 中(工程团队超过 1,000 人) | 招聘制造和建设管理者(进行中) | 中:ARC 建设需要新的组织能力 |
| PEO-05 | 员工数 6 年内从 50 增至 1,000,造成组织脆弱性 | 中 | 中 | 结构化招聘;流程成熟度举措 | 中:成长期执行风险 |
7.6 图表
08估值
8.1 收入前估值框架
CFS 不能用标准收入倍数,也不能按当前利润做 DCF 估值;公司没有收入,且在 ARC 实现商业运行前 不会产生有意义收入,目标时间是 2030 年代初。更合适的框架是按里程碑成功概率调整后的期权价值 (也称风险调整 NPV 或 rNPV),常用于商业化前的生物技术、核能和深科技公司。在这一框架下, 每个里程碑都是一个决策节点:(1) SPARC Q>1(目标 2027 年);(2) ARC FID(目标 2028 年); (3) ARC 首次发电(目标 2030 年代初)。每个节点都有成功概率,也有成功后的剩余价值。公司总价值 等于各里程碑终值按概率加权后的总和,再用适合风险投资的折现率折现(考虑技术和融资风险,35–50%)。 可比的商业化前核能公司(Terrapower、X-energy、Oklo)和聚变同业(Helion、TAE)都按类似期权价值 逻辑估值,而不是看收入倍数。CFS 在 B2 轮募得 $863M,估值未披露但被称为“上行轮”,为市场提供了 最新锚点。 [CV001, CV002, CV003, CV004]
| 维度 | 乐观逻辑 | 悲观反向逻辑 | 证据权重 |
|---|---|---|---|
| 物理 | SPARC 的高场 REBCO 路线是通向 Q>1 验证最充分的路径;2024 年离子温度纪录印证设计 | 新装置首次等离子体挑战很难预测;NIF 从首次等离子体到点火用了 7 年 | 略偏乐观——同行评审基础和纪录支撑逻辑 |
| 商业锚点 | Google + Eni PPA 在聚变领域独一份,并提供需求拉力;两者也是投资方 | 两份 PPA 都带条件,且没有公开罚则;它们是选择权,不是确定承购 | 中性——验证信号强,但收入确定性弱 |
| LCOE | 按照半导体 CAGR 类比,2025–2032 时间线内 HTS 带材成本下降 10x 具备可能性 | NuScale LCOE 翻倍;首台套建设总比模型更贵;氚成本未知 | 略偏悲观——聚变尚未吸收 NuScale 之后的经验 |
| 资本结构 | $3B 已到账;Google、Eni、Breakthrough Energy Ventures、Khosla、Temasek 都是可信机构支持者 | ARC 还需额外 $3–5B;没有已承诺过桥资金通向建设;DOE LPO 未锁定 | 中性——强支持者对上巨大未锁定缺口 |
| 竞争护城河 | SPARC 规模 + REBCO 磁体技术 + MIT 物理校友管线,让 CFS 相对 Helion 和 TAE 领先 5–7 年 | Helion 的 FRC 路线单机成本更低且有 Microsoft 支持;TAE 的无中子路线避开氚 | 略偏乐观——CFS 领先真实存在;Helion 的 2028 截止期是一次检验 |
关键投资绩效指标跟踪 CFS 对关键路径指标的推进, 投资者可据此做 go / no-go 决策。
[CV001, CV002, CV003, CV005, CV014, CV016]8.2 隐含估值历史与当前估算
CFS 融资历史给出了隐含估值锚:种子轮($50M,2018 年,投前约 $50–100M)、Series A ($115M,2020 年,投前约 $200–300M)、Series B($1.8B,2021 年,据 Crunchbase 和媒体报道隐含投后约 $3.2–5B)。Series B2($863M,2025 年 8 月)被公开描述为相对 B 轮的“上行轮”;B2 轮具体投前估值未披露。参考类似深科技轮次结构($863M 融资对应 20–25% 稀释),估计 B2 隐含投前估值为 $4.3–5.2B。加上 $863M 融资,B2 隐含投后 约 $5.1–6.1B。这与分析师对当前轮次 $5–8B 的估计一致。在这一估值下,CFS 定价约为复制公司硬资产 (科学团队、SPARC 机器、IP、PPA)成本的 1.5–2x,反映适度里程碑溢价。市场正在为 SPARC Q>1 成功 计入中等概率(期权模型隐含 60–70%),考虑 2024 年离子温度纪录和 CFS 的科学论文记录,这一概率合理。 [CV005, CV006, CV007, CV008]
展示 CFS 在各融资里程碑的估计估值区间(低 / 中 / 高), 从种子轮到 B2 轮,再预测至 ARC FID,并给出乐观 / 基准 / 悲观区间。
Series B2 轮之前估值为分析师基于融资额和轮次结构作出的估计;B2 轮投后估值 估计为 $5.1–6.1B。ARC FID 区间覆盖乐观 / 基准 / 悲观情景。
[CV005, CV006, CV007, CV008, CV009, CV010]8.3 乐观 / 基准 / 悲观情景分析
乐观情景(约 25% 概率):SPARC 在 2027 年实现 Q>1,且 Q=2–3,超过目标。ARC 在 2028 年 FID 并开工,2032 年首次发电。ARC 运营成本对应 $55/MWh LCOE。到 2035 年签下 10 台 ARC 机组订单。 以 ARC 收入 20x 退出 / IPO(基于 PPA 现金流,ARC-1 每年净额约 $250M)→ ARC-1 FID 时股权价值 $5–8B;机队期权价值 $30–50B。B2 投资人 IRR:约 25–35%(周期长但达到风险投资级回报)。 基准情景(约 45% 概率):SPARC 在 2028–2029 年实现 Q>1(延迟 1–2 年)。ARC 2030 年 FID, 2035 年首次发电。LCOE $65/MWh。机队部署再晚 5–7 年。ARC-1 FID 时股权价值:$4–7B。 B2 投资人 IRR:约 12–18%。低于典型风险投资门槛,但作为能源转型战略押注可以接受。 悲观情景(约 30% 概率):SPARC 实现 Q<0.8,或遭遇多年等离子体破裂问题。ARC FID 延至 2033+,或被取消。公司以 $1–2B 重组,或被 Eni/Alphabet 战略收购。B2 投资人 IRR 为负。 这个情景正是 CFS 股权只适合高风险承受力、长投资周期投资人的主要原因。 [CV009, CV010, CV011, CV012]
| 情景 | 概率 | 关键假设 | ARC-1 FID 时股权价值 | B2 投资者 IRR | 关键驱动因素 |
|---|---|---|---|---|---|
| 乐观 | ~25% | SPARC Q>1 且 Q=2–3(2027);ARC FID 2028;2032 年首发电;LCOE $55/MWh;10 台机组 | $8–12B | 25–35% | SPARC 超过 Q>1 预期;HTS 带材供应锁定;NRC 快速通道 |
| 基准 | ~45% | SPARC Q>1 且 Q=1.2–1.8(延迟到 2028–2029);ARC FID 2030;2035 年首发电;LCOE $65/MWh | $4–7B | 12–18% | 适度延迟;ARC 成本模型基本成立;机组扩张延后但仍可行 |
| 悲观 | ~30% | SPARC Q<1 或延迟超过 2 年;ARC FID 推迟到 2033+;LCOE $90–120/MWh 缺乏竞争力 | $0.5–2B(重组 / 收购) | 负值 | SPARC 物理验证失败;氚问题未解;资本缺口无法跨越 |
概率权重为分析师估计,依据独立风险评估和 SPARC 物理验证数据。 IRR 估算假设 B2 轮后估值 $5.5B,投资年份为 2025。
[CV009, CV010, CV011, CV012]| 终止触发条件 | 阈值 | 若触发 | 概率(估计) |
|---|---|---|---|
| 多轮实验后 SPARC Q<0.5 | Q<0.5,且没有清晰改进路径 | 退出接触;考虑战略出售 | ~10–15% |
| FID 前未锁定 ARC 规模的 HTS 带材供应 | ARC FID 决策前未落实供应合同 | 暂停 ARC 投资;重新评估 | ~10–20% |
| ARC pre-FEED 阶段 LCOE >$120/MWh | pre-FEED 工程研究显示 LCOE >$120/MWh | 重新考虑 ARC FID;建模重组方案 | ~15–25% |
| NRC 许可框架 2028 年前未最终敲定 | NRC 到 2028 年仍未发布聚变许可规则 | 模型中计入 ARC 启动推迟 3–5 年 | ~25–35% |
| Google 或 Eni PPA 取消 | 任一 PPA 公开取消且没有替代协议 | 重新评估收入模型;要求客户管线尽快多元化 | ~5–10% |
| 关键人物离职(Dennis Whyte 或 Bob Mumgaard) | 公开宣布离职,且没有可信继任者 | 提高审查强度;索取计划和继任时间表 | ~5–10% |
概率估计是面向投资人的情景输入,不是精算预测。
[CV009, CV010, CV011]决策树展示从 SPARC Q>1 结果到估值情景的投资者逻辑, 分出乐观、基准和悲观三条路径。
[CV001, CV009, CV010, CV011, CV012]柱状图展示 CFS 在乐观、基准和悲观情景下的隐含股权价值, 对应两个时间点:ARC FID 和 ARC 首次发电。
所有价值均为分析师基于期权价值方法的估计;每个里程碑的实际估值取决于 资本结构、稀释和市场环境。
[CV009, CV010, CV011]8.4 可比估值
私有聚变同业提供最直接可比:Helion Energy(已融资约 $2.2–2.5B,估值未披露,但基于 Microsoft PPA 和 OpenAI CEO Sam Altman 投资,估计为 $3–5B);TAE Technologies(已融资约 $1.3B, 最近一轮估值估计约 $2–3B);Tokamak Energy(UK,已融资约 $250M,估值未披露)。CFS 估值 (估计)为 $5–8B,是估值最高的私有聚变公司,反映其资本规模、SPARC 进展,以及 Google + Eni PPA 锚点。先进 SMR 公司提供次级可比:Terrapower(已融资约 $800M+,隐含估值约 $2–3B)、X-energy (已融资约 $1B+,SPAC 尝试在约 $1.5B 估值失败)。相对这些 SMR 公司,CFS 的估值溢价有其合理性: 募资规模更大、商业锚点(PPA)更好、SPARC 技术的科学基础更强。考虑 CFS 仍处商业化前阶段,更广义的 清洁能源基础设施可比公司(可再生能源 IPO、电池储能)相关性较弱。 [CV013, CV014, CV015, CV016]
| 公司 | 技术路线 | 融资额 | 隐含 / 最新估值 | 商业锚点 | 关键里程碑 | 相对 CFS |
|---|---|---|---|---|---|---|
| CFS(Commonwealth Fusion) | REBCO HTS 托卡马克(SPARC → ARC) | ~$3.2B(B2) | $5–8B(估计,B2 轮后) | Google 200MW PPA + Eni $1B+ PPA(购电协议) | SPARC Q>1(2027) | 基准 |
| Helion Energy | FRC 脉冲聚变(HB11) | ~$2.4B | $3–5B(估计) | Microsoft 50MW PPA(2028) | 净能量增益(2024 尝试) | 资本更少、PPA 更小、物理风险更高 |
| TAE Technologies | FRC 束驱动无中子 | ~$1.3B | $2–3B(估计) | 未签署 | 净能量增益(无时间表) | 估值更低、未签 PPA、无中子路线有优势 |
| Tokamak Energy(英国) | HTS ST40 球形托卡马克 | ~£400M($500M) | 未披露 | 未签署 | 实现 100M°C 等离子体(2022) | 规模更小、阶段更早、总部在英国 |
| Terrapower(SMR,非聚变) | Natrium 钠冷快堆 | ~$1B+ | ~$2–3B(估计) | 怀俄明州电站(Natrium 试点) | 2024 年开工 | 风险更低、回报更低,SMR 不是聚变 |
| X-energy(SMR) | Xe-100 球床高温反应堆 | ~$1B | ~$1.5B(SPAC 尝试) | DOE Xe-100 示范项目 | 目标 2025 年开工 | 风险更低、回报更低,SMR 不是聚变 |
8.5 投资建议
尽调结论:有条件看好,但有重大保留。CFS 是最可信、进展最靠前的私有聚变公司,融资额最大, 独立科学验证最强,也是唯一签下两份远期 PPA 的聚变公司。这些因素足以支撑其在聚变板块的领先估值。 但这笔投资需要:(1) 明确接受 SPARC Q>1 的二元风险(负面情景概率约 30%);(2) 承受正现金流前 6–10 年周期;(3) 接受 ARC 还需额外 $3–5B 且尚未锁定的资本需求;(4) 持续跟踪氚增殖——这是 SPARC 之后、多数投资人尚未定价的风险。按 $5–8B 估值看,CFS 在期权价值意义上并不便宜——市场已经 计入 65–70% 的 Q>1 概率。后期成长基金应把 CFS 定位为高信念押注,组合权重 2–5%;建议在多家 聚变公司之间分散配置(CFS + Helion)。投资前尽调问题:(1) 数据室审阅 PPA 条款;(2) 氚增殖计划和 时间表;(3) HTS 带材供应合同;(4) ARC pre-FEED 成本研究;(5) NRC 预申请状态。 [CV017, CV018, CV019, CV020]
| 维度 | 评估 | 信心 |
|---|---|---|
| 总体尽调结论 | 正面——这是当前质量最高的私营聚变投资机会,但保留重大前提 | 中 |
| 技术风险 | 风险高,但 2024 年 SPARC 离子温度纪录和同行评审物理基础已显著降风险 | 中高 |
| 商业风险 | 两份已签署的远期 PPA(Google + Eni)在行业内独一份;集中度风险明显 | 高 |
| 监管风险 | 中等;ADVANCE Act 聚变框架尚不完整;PJM 电网排队有 3–5 年风险 | 中 |
| 财务风险 | ~$3–5B ARC 资金缺口尚未锁定;最早也要到 2030 年代初才有收入 | 中 |
| 人员风险 | 低–中;Whyte 正式担任 CSO 和 CEO 延续性降低风险;员工快速增长带来执行风险 | 中 |
| 估值 | $5–8B 估计 B2 轮投后估值;定价隐含 Q>1 成功率 65–70%;乐观情景支持较当前 2x+ 上行 | 低(估计) |
| 投资建议 | 长周期成长 / 能源转型基金可高信念配置(2–5% 组合权重);不适合短周期基金 | 中 |
| 尽调要求 | 重要性 | 优先级 | 获取路径 |
|---|---|---|---|
| Google 和 Eni 的 PPA 条款(价格、期限、违约条款) | 收入确定性取决于 PPA 经济性;没有罚则的附条件 PPA,确定性低于明确购电承诺 | 关键 | 数据室:请求 PPA 协议 |
| 与 SuNAM 和 FUJIKURA 的 HTS 带材供应合同(数量、价格、交付) | 供应链是关键建设风险;未锁定供应 = 建设延误或成本超支 | 关键 | 数据室:请求带材供应协议 |
| ARC pre-FEED 成本研究和 LCOE 敏感性分析 | NuScale 先例显示,概念阶段 LCOE 估算可能偏差达 2x;需要 pre-FEED 研究验证 $50–70/MWh 目标 | 关键 | 数据室:请求工程成本研究 |
| 氚增殖包层开发计划和时间表 | 氚 TRL 2–3 是 SPARC 之后的潜伏风险;需要 CFS 的内部计划和外部验证 | 高 | 数据室:请求氚包层研发计划 |
| NRC ADVANCE Act 预申请咨询状态 | 监管时间线是关口风险;需要了解 CFS 与 NRC 沟通走到哪一步 | 高 | 数据室:请求 NRC 往来函件;查询 FERC ELIBRARY |
| Google 和 Eni 之外的客户管线(LOI、初步沟通) | 机组经济性需要 3 个以上客户;需要看见 ARC-2+ 买方管线 | 高 | 数据室:请求客户管线报告 |
| 股权结构表和所有已发行证券(B2 后) | 未来 ARC 融资轮对 B2 投资人的稀释,是关键 IRR 风险;需要完整股权结构表 | 高 | 数据室:请求股权结构表和股东协议 |
| SPARC 首次等离子体时间线和调试计划更新 | SPARC 进度是所有商业里程碑的关键路径;需要最新内部排期 | 中 | 数据室:请求 SPARC 项目排期和调试计划 |
8.6 图表
免责声明
本报告是基于公开证据的尽调快照,不构成投资建议。重要的财务、法律、技术和合同事实仍未公开;作出任何投资决定前,应直接向管理层核验,并查阅原始文件。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Commonwealth Fusion Systems was incorporated in 2018 as a spinout from MIT's Plasma Science and Fusion Center. | 高 | SO015, SO017 |
| CO002 | CFS is headquartered at its Devens, Massachusetts campus, which serves as both manufacturing facility and reactor assembly site. | 高 | SO007, SO029 |
| CO003 | CFS's business model is to develop, license, and operate fusion power plants selling electricity under long-term power purchase agreements. | 高 | SO001, SO029 |
| CO004 | CFS's technology is based on REBCO high-temperature superconducting (HTS) magnets that achieve a 20-tesla field, developed with MIT under a sponsored research agreement. | 高 | SO010, SO029 |
| CO005 | CFS holds the key IP for the REBCO HTS magnet design jointly with MIT. | 中 | SO010, SO015 |
| CO006 | Bob Mumgaard is CEO and Co-founder of CFS; he holds a PhD in plasma physics from MIT and originated the CFS strategy from PSFC research. | 高 | SO004, SO005 |
| CO007 | Dan Brunner serves as CTO of CFS with deep tokamak design and magnet engineering expertise from MIT. | 中 | SO006, SO015 |
| CO008 | Steve Renter is CFS's Chief Operating Officer, responsible for SPARC construction execution and supply chain. | 中 | SO006 |
| CO009 | Ally Yost is SVP Corporate Development at CFS and structured the Google and Eni PPAs. | 高 | SO001, SO026 |
| CO010 | CFS does not publicly disclose board composition or specific board seat terms per investor. | 中 | SO026, SO027 |
| CO011 | CFS raised a Series A of $115 million in June 2019 from Eni, Breakthrough Energy Ventures, Khosla Ventures, The Engine (MIT's venture fund), Future Ventures, Safar Partners, Starlight Ventures, and others. | 高 | SO017, SO018 |
| CO012 | CFS closed a Series B of $1.8 billion in December 2021 led by Tiger Global with approximately 70 investors including Bill Gates' Breakthrough Energy Ventures, Google, Eni, Equinor, Temasek, and Emerson Collective. | 高 | SO021, SO022 |
| CO013 | CFS closed an oversubscribed Series B2 of $863 million on August 28, 2025, with investors including NVentures (Nvidia), Counterpoint Global (Morgan Stanley), Stanley Druckenmiller, a Japanese consortium of 12 companies led by Mitsui and Mitsubishi, Galaxy Digital, and others. | 高 | SO001, SO002, SO003 |
| CO014 | CFS has raised approximately $3 billion in total capital since founding, representing roughly one-third of all private capital ever invested in the fusion industry. | 高 | SO001, SO029 |
| CO015 | CFS's 2021 post-Series B valuation was reported at $3.2–$5 billion. | 中 | SO022 |
| CO016 | CFS received an $8 million milestone-based DOE grant in September 2025 after independent expert review validated its full-scale toroidal field magnet performance. | 高 | SO025, SO001 |
| CO017 | SPARC construction began at the Devens, Massachusetts campus in approximately late 2022. | 中 | SO007, SO015 |
| CO018 | CFS installed the cryostat base—a 24-foot wide, 75-ton stainless steel disc manufactured in Italy—as the first major SPARC tokamak component in March 2025. | 高 | SO007, SO008, SO009 |
| CO019 | CFS targets first plasma from SPARC in 2026 and net fusion energy gain (Q > 1) shortly after, with 2027 commonly cited. | 中 | SO023, SO007 |
| CO020 | The Commonwealth of Massachusetts granted CFS a broad-scope radioactive materials license for SPARC in October 2024, permitting possession, use, and storage of radioactive materials. | 高 | SO023, SO024 |
| CO021 | ARC, CFS's first commercial fusion power plant, is planned for Chesterfield County, Virginia, targeting 400 MWe output and early-2030s grid power delivery. | 高 | SO012, SO013, SO014 |
| CO022 | Google signed a 200-megawatt power purchase agreement with CFS on June 30, 2025, representing the first corporate commitment to purchase commercial fusion-generated power. | 高 | SO012, SO013, SO014 |
| CO023 | Eni signed a $1 billion-plus power purchase agreement with CFS in September 2025 for fusion power from the ARC plant. | 高 | SO026, SO027 |
| CO024 | CFS had approximately 800 to 1,000 or more employees as of early 2025, up from about 350 in late 2022, reflecting rapid hiring growth. | 中 | SO015, SO016 |
| CO025 | CFS has not disclosed any revenue; it remains pre-commercialization with no commercial product or service yet generating income. | 高 | SO001, SO029 |
| CO026 | No significant layoffs, leadership departures, regulatory enforcement actions, or major lawsuits involving CFS were identified in publicly available sources for 2024–2025. | 中 | SO015, SO019 |
| CO027 | CFS CEO Bob Mumgaard has publicly sought to distance the company's branding from the word 'nuclear' to manage public perception, despite using radioactive tritium fuel. | 中 | SO020 |
| CO028 | The U.S. Energy Secretary Chris Wright visited the SPARC site in late 2025, underscoring positive federal engagement with CFS's program. | 高 | SO025, SO001 |
| CO029 | Eni is described in its own public communications as a 'relative majority shareholder' of CFS, though exact equity percentage is not publicly disclosed. | 中 | SO026 |
| CO030 | In September 2021, CFS and MIT PSFC demonstrated a 20-tesla world-record high-temperature superconducting magnet, reducing the cost per watt of a fusion reactor by a factor of approximately 40. | 高 | SO010, SO015 |
| CO031 | The NRC voted unanimously in 2023 to regulate commercial fusion power plants under a byproduct materials framework (10 CFR Part 30) rather than the more burdensome fission rules. | 高 | SO023, SO024 |
| CO032 | The U.S. ADVANCE Act, signed in July 2024, codified the NRC byproduct materials framework for fusion into federal law. | 高 | SO024, SO023 |
| CO033 | CFS has an option-to-lease agreement with Dominion Energy for the ARC plant site in Chesterfield County, Virginia, within the James River Industrial Center. | 高 | SO014, SO013 |
| CO034 | The Series B2 round was described by CFS as an 'up round,' meaning the 2025 valuation exceeds the 2021 post-money of $3.2–$5 billion; exact 2025 valuation was not disclosed. | 高 | SO001, SO002 |
| CO035 | First vacuum vessel half (a 48-ton half-donut shaped component) was received and staged at CFS's SPARC Tokamak Hall in October 2025. | 高 | SO008, SO009 |
| CO036 | CFS delivered its first factory-grade production magnet for SPARC in December 2025, marking a transition from lab-scale R&D to industrial-scale magnet manufacturing. | 中 | SO028 |
| CO037 | CFS's fusion energy ambition faces criticism from skeptics who argue that fusion timelines historically slip and that the LCOE from fusion must compete against rapidly falling costs of renewables plus storage. | 中 | SO019, SO030 |
| CM001 | CFS targets the global baseload electricity generation market, specifically dispatchable 24/7 zero-carbon power for utilities and hyperscale data centers. | 高 | SM005, SM030 |
| CM002 | Adjacent markets for CFS include industrial process heat and green hydrogen production from surplus fusion power, not in current commercial plans. | 中 | SM030, SM024 |
| CM003 | CFS's near-term SAM is the U.S. utility and hyperscaler PPA market for firm clean power, with the ARC plant in Virginia being the primary near-term delivery vehicle. | 高 | SM015, SM016, SM020 |
| CM004 | The global electricity market is approximately $2.4 trillion annually as of 2024, representing the broadest proxy TAM for fusion. | 中 | SM026, SM001 |
| CM005 | Analyst estimates for the global fusion energy market in 2025 range from $301–$347 billion, projected to reach $420–$497 billion by 2030 at a 6–8% CAGR. | 低 | SM007, SM008, SM009 |
| CM006 | Multiple analyst houses project the global fusion energy market to reach $420–$497 billion by 2030 as early commercial plants enter service. | 低 | SM007, SM009 |
| CM007 | Allied Market Research projects the global fusion energy market could reach $840 billion by 2040 at ~6–7% CAGR, driven by clean energy demand and deployment at scale. | 低 | SM008, SM027 |
| CM008 | The global renewable energy market is valued at $1.6 trillion in 2025, projected to reach $4.9 trillion by 2033 at 14.7% CAGR, setting the competitive landscape for fusion. | 中 | SM003, SM004 |
| CM009 | IEA projects global data-center electricity consumption will double from ~415 TWh in 2024 to ~945 TWh by 2030, driven by AI workloads, creating ~530 TWh of incremental demand for clean baseload power. | 高 | SM010, SM011, SM012 |
| CM010 | Hyperscale data-center operators including Google, Microsoft, Amazon, and Meta have 24/7 carbon-free energy commitments and require firm dispatchable clean power they cannot source from intermittent renewables alone. | 高 | SM015, SM017, SM018 |
| CM011 | Google signed the world's first corporate fusion PPA with CFS for 200 MW from ARC in June 2025, demonstrating hyperscaler willingness to pay and accept technology risk for fusion offtake. | 高 | SM015, SM016, SM023 |
| CM012 | Regulated utilities represent a primary buyer segment for ARC's power via standard resource planning and FERC interconnection, with Dominion Energy as the first site partner. | 中 | SM016, SM022 |
| CM013 | Industrial buyers (energy-intensive manufacturing, green hydrogen) are a secondary buyer segment for ARC power in the post-2035 deployment phase. | 低 | SM030, SM024 |
| CM014 | AI data-center power demand is the most immediate growth driver for CFS's market: IEA projects ~$530 TWh of incremental clean baseload demand by 2030, a direct pull for ARC-scale firm power plants. | 中 | SM010, SM011, SM017 |
| CM015 | Solar PV LCOE has fallen below $30–$40/MWh in optimal locations; continued decline of solar+storage is the primary competitive constraint on fusion market penetration. | 中 | SM003, SM004 |
| CM016 | Global decarbonization targets—including the IEA Net Zero scenario requiring ~50% of electricity from non-intermittent clean sources by 2050—create long-term structural pull for dispatchable fusion. | 高 | SM001, SM004 |
| CM017 | The ADVANCE Act (2024) and U.S. nuclear renaissance policy create favorable regulatory tailwinds for fusion commercialization alongside advanced fission (SMRs). | 高 | SM021, SM022 |
| CM018 | CFS has no commercial revenue before the early 2030s—its pre-commercial horizon is a primary adoption constraint and requires sustained capital without revenue generation for approximately 5+ more years. | 高 | SM005, SM024 |
| CM019 | Tritium fuel supply is a commercial-scale risk: current production (from CANDU reactors and ITER) is insufficient for a fleet of ARC plants; CFS must demonstrate tritium breeding in ARC's blanket. | 中 | SM025, SM014 |
| CM020 | SPARC achieving Q>1 (net fusion energy gain) in 2027 is the critical market-gating milestone: it will unlock subsequent utility and industrial buyer interest and de-risk PPA signings. | 中 | SM005, SM006 |
| CM021 | CFS has signed over $1 billion in forward PPAs (Google 200 MW + Eni $1B+ deal) for its first ARC plant, providing revenue visibility contingent on commercial operations. | 高 | SM015, SM028 |
| CM022 | ARC's target LCOE of $50–$70/MWh (company-estimated) must be validated through independent engineering assessment; first-of-a-kind plant cost overruns are a standard risk in power plant construction. | 低 | SM024, SM030 |
| CM023 | Global clean energy investment exceeded $1.8 trillion in 2023 and is approaching $2 trillion in 2024, demonstrating available capital for new clean-power technology deployment. | 高 | SM001, SM002 |
| CM024 | Advanced nuclear SMR PPAs in U.S. competitive tenders in 2024–2025 ranged from $90–$150/MWh, suggesting ARC's target LCOE of $50–$70/MWh could be competitive if realized. | 低 | SM023, SM024 |
| CM025 | FERC interconnection approval and grid queue position at the Virginia ARC site are required for commercial grid delivery; these are not yet secured and represent regulatory adoption risk. | 中 | SM022, SM021 |
| CM026 | Bears argue that by the early 2030s, solar+battery storage may eliminate the dispatchability premium that justifies ARC's projected LCOE; independent analysis suggests widespread fusion deployment is unlikely before 2040. | 中 | SM024, SM014 |
| CM027 | CFS's ARC commercial plant is designed to deliver 400 MWe output; at a target LCOE of $50–$70/MWh and 90% capacity factor, annual revenue per plant would be approximately $175–$245 million. | 低 | SM030, SM024 |
| CM028 | The U.S. Inflation Reduction Act (IRA) Section 45Y Clean Electricity Production Tax Credit and Section 48E Clean Electricity Investment Credit may apply to fusion plants meeting zero-carbon criteria, providing potential ~$30/MWh federal subsidy support. | 低 | SM021, SM022 |
| CM029 | DOE Loan Programs Office (LPO) Title XVII loans and Advanced Technology Vehicles Manufacturing (ATVM) program are potential financing vehicles for ARC construction given its clean-energy classification. | 低 | SM022, SM001 |
| CM030 | The global industrial process heat market (steel, chemicals, cement) represents a long-term adjacency for fusion; industrial process heat above 500°C is poorly served by renewables but CFS has not yet designed ARC for heat offtake. | 低 | SM030, SM024 |
| CM031 | Utility resource planning cycles in target states (Virginia, Massachusetts) have 15–20 year integrated resource planning (IRP) horizons; the early 2030s ARC first-plant date falls within current IRP windows. | 低 | SM022, SM016 |
| CM032 | Bears argue that even if SPARC demonstrates net energy gain in 2027, scaling to ARC commercial operations requires 5–8 additional years of engineering development; fusion commercial entry before 2035 is viewed as unlikely by several independent analysts. | 中 | SM014, SM024 |
| CM033 | Global clean energy investment of approximately $1.8 trillion in 2023 is more than twice fossil fuel investment; fusion must compete for a share of this capital against more-mature clean technologies with lower risk profiles. | 高 | SM001, SM002 |
| CM034 | SMR providers (NuScale, X-energy, TerraPower) targeting the same baseload clean-power market as CFS have filed or are planning to file nuclear regulatory applications in the 2024–2027 window, representing competitive benchmarks for buyer decision-making. | 中 | SM021, SM023 |
| CM035 | McKinsey estimates hyperscalers need to procure hundreds of gigawatts of additional clean capacity globally by 2030 to meet sustainability commitments; firm dispatchable sources like fusion are favored when 24/7 matching is required. | 中 | SM017, SM018 |
| CP001 | The global private fusion sector comprises approximately 40 companies as of 2025, up from ~10 in 2019, with over $7 billion in private capital deployed. | 高 | SP013, SP014 |
| CP002 | CFS and Helion Energy are the two most advanced and best-capitalized private fusion companies as of Q1 2026, distinguished by capital raised, physics maturity, and signed PPAs. | 中 | SP001, SP002, SP013 |
| CP003 | TAE Technologies (California, founded 1998) uses Field-Reversed Configuration with proton-boron fuel, targeting aneutronic fusion; total capital raised approximately $1.3 billion as of 2024. | 中 | SP006, SP001 |
| CP004 | Tokamak Energy (UK, founded 2009) builds spherical tokamaks with HTS magnets; commissioned its ST80-HTS magnet system in 2024 and has raised approximately $335 million. | 高 | SP007, SP008 |
| CP005 | General Fusion (Canada, founded 2002) uses magnetized target fusion (MTF) with mechanical piston compression; it pivoted to a new technology design in January 2024 under investor pressure, with ~$325 million total raised. | 高 | SP009, SP014 |
| CP006 | Helion Energy (Washington, founded 2013) uses Field-Reversed Configuration with D-He3 fuel and direct electrical conversion; it has raised approximately $2.2 billion and has a signed PPA with Microsoft for 50 MW by 2028. | 高 | SP004, SP005, SP014 |
| CP007 | CFS's tokamak confinement approach benefits from 70+ years of global fusion physics research and demonstrated confinement at scale (JET, TFTR precursors); Helion's FRC has less empirical validation at comparable plasma temperatures and pressures. | 中 | SP016, SP002 |
| CP008 | CFS targets SPARC Q>1 (fusion energy gain) in 2027; this milestone, if achieved, would validate the compact high-field tokamak approach and be the most significant physics proof in private fusion history. | 中 | SP023, SP003 |
| CP009 | Helion's Microsoft PPA contains a penalty clause if 50 MW is not delivered by 2028, creating a public accountability mechanism that CFS's Google/Eni PPAs do not appear to include based on available information. | 中 | SP004, SP005 |
| CP010 | If Helion delivers 50 MW to Microsoft in 2028 before CFS achieves SPARC Q>1, it would shift hyperscaler confidence toward FRC-based fusion, potentially at the expense of CFS's next PPA signings. | 低 | SP004, SP002 |
| CP011 | CFS's Google (200 MW) PPA is larger in committed MW than Helion's Microsoft PPA (50 MW), but no delivery date commitment has been made public for the Google deal; the Eni $1B+ deal is also forward-contingent. | 中 | SP017, SP004 |
| CP012 | NuScale's Carbon Free Power Project (CFPP) was cancelled in November 2023 due to cost escalation (projected $89/MWh), demonstrating that first-of-a-kind nuclear power plant economics regularly exceed estimates. | 高 | SP010, SP011 |
| CP013 | TerraPower's Natrium sodium-cooled fast reactor (345 MWe) broke ground in Wyoming in June 2024 with DOE support, targeting 2030 commercial operations; it competes with CFS for the same baseload clean power PPA market. | 高 | SP012, SP021 |
| CP014 | TerraPower's DOE cost-sharing and government backing give it a lower-cost-of-capital advantage over CFS in competing for utility long-term power agreements. | 中 | SP012, SP020 |
| CP015 | Solar PV LCOE as measured by Lazard (2024) ranges from $24–$96/MWh depending on location; utility-scale solar with 4-hour storage is available at $45–$65/MWh in favorable locations, directly competing with ARC's target LCOE. | 高 | SP020, SP021 |
| CP016 | Geothermal companies (Fervo, AltaRock, Quaise) and long-duration storage (Form Energy, Ambri) target the same firm clean baseload market as CFS but are further along in commercial deployment. | 中 | SP020, SP014 |
| CP017 | CFS has raised approximately $3 billion as of August 2025 (including $863M Series B2), making it the best-capitalized private fusion company and providing multi-year runway without near-term dilutive financing needs. | 高 | SP017, SP003 |
| CP018 | CFS's 20-Tesla REBCO HTS magnet record (2021), validated in peer-reviewed IEEE TAS publication (2024), is the most publicly documented high-field magnet achievement among private fusion companies. | 高 | SP015, SP016 |
| CP019 | Tokamak Energy's HTS magnet program (ST80-HTS commissioned 2024) is the most credible competing magnet capability, though its field strength has not been publicly disclosed; it represents a medium-term IP moat risk for CFS. | 中 | SP008, SP007 |
| CP020 | CFS has signed over $1 billion in forward PPAs (Google 200MW + Eni $1B+ deal), the first fusion PPAs ever signed, providing market validation and revenue visibility. | 高 | SP017, SP018 |
| CP021 | CFS's MIT PSFC partnership embeds ~50+ MIT-affiliated researchers in SPARC/ARC design and provides access to intellectual property developed through the collaboration; this institutional depth is difficult for competitors to replicate without equivalent academic partnerships. | 中 | SP023, SP016 |
| CP022 | The most adverse competitive scenario for CFS is Helion delivering power to Microsoft in 2028 before SPARC achieves net energy, which would reallocate hyperscaler procurement interest and investor capital to Helion's FRC approach. | 低 | SP004, SP022 |
| CP023 | A 'fusion winter' scenario—triggered by multiple missed milestones across the sector—would disproportionately harm mid-tier companies; CFS's $863M B2 and PPA commitments provide relative insulation, but not immunity. | 中 | SP022, SP024 |
| CP024 | Science magazine noted in 2024 that private fusion startups face a 'critical year' as capital requirements increase and milestone timelines slip; multiple companies including TAE and General Fusion have restructured or pivoted. | 高 | SP022, SP009 |
| CP025 | CFS's tokamak approach is sometimes criticized by advocates of alternative fusion methods (FRC, Z-pinch) as too capital-intensive and too dependent on a large SPARC device, similar in engineering complexity to ITER despite being much smaller. | 低 | SP002, SP024 |
| CP026 | Zap Energy (Seattle, founded 2017) uses sheared-flow stabilized Z-pinch fusion, has raised ~$160 million, and represents a lower-cost, smaller-footprint approach that could scale faster if plasma stability challenges are solved. | 中 | SP019, SP001 |
| CP027 | The Fusion Industry Association's 2025 State of the Fusion Industry report documents $7B+ in private capital deployed, with CFS among the top 3 recipients globally alongside Helion and TAE. | 中 | SP013, SP025 |
| CP028 | Canary Media reported the fusion industry had accumulated a $6 billion investment base by early 2024; the figure grew to $7B+ by late 2025, indicating continued capital inflows despite milestone uncertainty. | 中 | SP025, SP013 |
| CP029 | CFS's strategy of 'conservative physics, innovative magnets' is deliberately differentiated from Helion's 'innovative physics (FRC), innovative direct conversion' approach; this makes CFS's technical risk profile lower but its potential efficiency gains more incremental. | 低 | SP002, SP007 |
| CP030 | The FT reported in 2024 that fusion startups face a 'reality check' as development costs mount and early timelines have slipped; several mid-tier companies reduced staff or sought new capital sources during 2024. | 中 | SP024, SP022 |
| CP031 | CFS's VIPER (Variable Integrated Prototype Engineering Run) program is developing modular HTS REBCO magnet technology with partners including MIT; this vertical integration in magnet manufacturing creates a supply chain moat beyond IP alone. | 中 | SP016, SP023 |
| CP032 | The D-T fuel cycle used by CFS, Tokamak Energy, and General Fusion requires tritium breeding, whereas Helion's D-He3 cycle and TAE's p-B11 cycle avoid this; all three avoid long-lived radioactive waste compared to fission SMRs. | 中 | SP002, SP001 |
| CP033 | CFS has no regulatory nuclear license filings as of Q1 2026; all major fusion companies are pre-filing with relevant nuclear regulators (NRC for U.S., ONR for UK), consistent with the pre-commercial phase of the sector. | 高 | SP013, SP022 |
| CP034 | Helion's D-He3 approach faces a helium-3 supply constraint: terrestrial He-3 supplies are extremely scarce; commercial-scale Helion plants would require lunar He-3 mining or alternative tritium-based D-He3 breeding, a problem CFS's D-T approach does not have. | 中 | SP002, SP003 |
| CP035 | The Science magazine report (2024) noted that CFS, Helion, and TAE are each targeting commercial power before ITER (expected first plasma 2025, first D-T operations 2035+), but independent physicists have expressed skepticism about private sector timeline claims. | 中 | SP022, SP024 |
| CI001 | CFS has zero commercial revenue as of the run date (May 2026); it is entirely pre-revenue and will remain so until ARC commercial operations begin in the early 2030s. | 高 | SI011, SI001 |
| CI002 | CFS's planned revenue model is power purchase agreements (PPAs) for electricity from ARC commercial plants; PPA revenue does not commence until ARC achieves first power delivery. | 高 | SI011, SI008 |
| CI003 | CFS has received non-recurring DOE milestone grants (including $8M in 2024) and participates in the DOE INFUSE program, providing small non-dilutive research funding but not commercial revenue. | 高 | SI021, SI022 |
| CI004 | A secondary revenue lever—technology licensing for future ARC plants built by third parties—is alluded to by CFS but has no signed contracts or disclosed terms as of Q1 2026. | 低 | SI011, SI002 |
| CI005 | CFS was founded in 2018 with approximately $50 million in seed funding from Eni S.p.A.; this established the founding investor relationship that has persisted through all subsequent rounds. | 高 | SI020, SI004 |
| CI006 | CFS raised a $115 million Series A in June 2019 from Eni, Khosla Ventures, Breakthrough Energy Ventures, and The Engine; this funded the VIPER HTS magnet program. | 高 | SI020, SI004 |
| CI007 | CFS raised $1.8 billion in Series B in December 2021—the largest private fusion round ever at that time—led by Tiger Global with Breakthrough Energy Ventures, Google, Khosla, Altimeter Capital, and Temasek as co-investors. | 高 | SI003, SI004 |
| CI008 | CFS raised $863 million in Series B2 in August 2025, described as an 'up round,' with Khosla Ventures, Google, and Coatue leading; total cumulative capital raised is approximately $2.8–3.0 billion. | 高 | SI001, SI002 |
| CI009 | CFS has no disclosed public debt, government loan guarantees, or project finance; its capital structure is 100% equity; ARC construction (~$2.5B+ estimated) will require a major additional round or project financing. | 中 | SI011, SI005 |
| CI010 | CFS's headcount is approximately 800–1,000+ employees as of 2025, concentrated in engineering, plasma physics, and manufacturing at its Devens, MA facility. | 中 | SI014, SI002 |
| CI011 | At 800–1,000 employees with estimated fully-loaded costs of $250–$350K/person/year, CFS's labor cost alone is approximately $200–$350 million per year. | 低 | SI014, SI001 |
| CI012 | Including capital expenditure for SPARC construction and facility costs, CFS's total annual operating cost is estimated at $300–$500 million per year; no audited P&L is available to validate this estimate. | 低 | SI005, SI001 |
| CI013 | The $863M Series B2 (Aug 2025) at an estimated $300–500M/year burn rate implies approximately 18–36 months of runway—consistent with coverage through SPARC Q>1 (targeted 2027) but not through ARC FID (~2028–2030). | 低 | SI001, SI005 |
| CI014 | CFS's publicly cited LCOE target for ARC is $50–$70 per MWh; this is a company-generated engineering estimate and has not been independently validated by an external engineering firm or government body. | 低 | SI005, SI011 |
| CI015 | ARC construction cost for the first plant has been cited in independent analysis at approximately $2.5 billion; this is derived from company inputs and engineering scaling laws, not a bid from a construction contractor. | 低 | SI005, SI006 |
| CI016 | Based on 400 MWe capacity, 90% capacity factor, and $50–$70/MWh LCOE target, annual revenue per ARC plant would be approximately $175–$245 million; this is an analyst calculation, not a CFS financial projection. | 低 | SI011, SI005 |
| CI017 | Advanced nuclear SMR PPA prices in U.S. competitive tenders in 2024–2025 ranged from $90–$150/MWh, providing a market benchmark that ARC's $50–$70/MWh target would comfortably undercut if achieved. | 中 | SI019, SI023 |
| CI018 | NuScale's CFPP cancellation illustrates that first-of-a-kind advanced nuclear plant economics frequently exceed initial estimates by 50–100%; this precedent applies directly to ARC's construction cost risk. | 高 | SI009, SI010 |
| CI019 | Every year of ARC timeline slippage beyond 2032 adds approximately $300–$500M in additional equity capital consumption without any revenue, increasing dilution risk for early investors. | 低 | SI005, SI001 |
| CI020 | CFS's Google and Eni PPAs are contingent on ARC commercial operations; no public evidence of penalty clauses, minimum take-or-pay provisions, or early termination payments exists—weakening the revenue certainty these PPAs appear to provide. | 中 | SI007, SI008 |
| CI021 | ARC construction will require a pre-FID financing round estimated at $3–5 billion; DOE Loan Programs Office Title XVII guaranteed loans are a potential vehicle, but no application has been confirmed. | 低 | SI012, SI013 |
| CI022 | CFS provides no public P&L, balance sheet, cash flow statement, or GAAP financial disclosures; it is a private corporation not subject to SEC reporting requirements. | 高 | SI020, SI001 |
| CI023 | CFS's current valuation is not publicly disclosed; the B2 round was described as an 'up round' relative to the 2021 Series B which implied a valuation of approximately $3.2–5 billion based on press reports. | 低 | SI018, SI001 |
| CI024 | CFS participates in the DOE INFUSE program (Industry Partnerships for Fusion Energy Sciences), which provides access to national laboratory equipment and technical expertise; these are in-kind research resources, not cash grants. | 中 | SI021, SI025 |
| CI025 | PJM capacity market prices (the regional grid covering Virginia and adjacent states) range from $20–$50/MWe/year; ARC's 400 MWe plant could earn $8–20 million per year in capacity payments if interconnected to the PJM grid. | 低 | SI023, SI013 |
| CI026 | CFS's Massachusetts radioactive materials license (issued October 2024) is a regulatory milestone that enables active radioactive testing at the Devens SPARC facility; this is not yet a power plant operating license. | 高 | SI022, SI025 |
| CI027 | The Georgetown Space Policy Institute analysis (2025) concluded ARC's economics are 'plausible if construction costs and capacity factor targets are met' but flagged first-of-a-kind cost overrun risk as the critical uncertainty. | 中 | SI005, SI024 |
| CI028 | Lazard's 2024 LCOE analysis shows utility-scale solar ranges from $24–$96/MWh and onshore wind from $24–$75/MWh; battery storage adds $40–$80/MWh to achieve 4-hour duration; combined solar+4hr storage at ~$64–$176/MWh straddles ARC's target, favoring ARC only where 24/7 firmness is required. | 高 | SI019, SI023 |
| CI029 | Eni S.p.A. is both an investor (from seed through Series B2) and a PPA counterparty for CFS; this dual role creates alignment but also a potential conflict of interest if Eni's strategic interests in energy markets diverge from CFS's commercialization timeline. | 低 | SI007, SI015 |
| CI030 | The total implied construction spend to bring CFS from founding through ARC-1 commercial operations is approximately $6–8 billion (current equity raised ~$3B + SPARC capex already deployed + pre-ARC round of $3–5B), excluding operating losses. | 低 | SI005, SI001 |
| CI031 | CFS raised its equity primarily through Regulation D exemptions from SEC registration (Form D filings); these exempt private placements are accessible only to accredited investors and allow no public offering. | 中 | SI026, SI020 |
| CI032 | At four funding rounds (Seed through B2) with dilution typical of deep-tech companies, early employees and seed investors likely hold substantially less than 10% combined post-B2; no cap table is public to confirm this. | 低 | SI026, SI018 |
| CI033 | Eni S.p.A. participated in every CFS funding round from seed through Series B2 and holds both equity and a $1B+ PPA commitment; this strategic alignment reduces counterparty risk for the Eni PPA but limits Eni's negotiating independence on pricing. | 中 | SI007, SI015 |
| CI034 | NREL's Annual Technology Baseline (2024) shows nuclear fission capacity factors of 92–95%; CFS's assumed 90% capacity factor for ARC is slightly conservative vs. operating fission plants, which is appropriate for a first-of-kind plant. | 高 | SI023, SI005 |
| CI035 | CFS's zero long-lived radioactive waste and no-meltdown risk profile are genuine product differentiators vs. fission SMRs; Google, specifically citing ESG and 24/7 CFE goals, signed the first fusion PPA, suggesting a premium is possible. | 中 | SI008, SI016 |
| CE001 | VIPER (Variable Integrated Prototype Engineering Run) is CFS's internally developed HTS REBCO cable technology, enabling the compact high-current-density magnets needed for SPARC's 20T field at reduced size vs. conventional superconductors. | 高 | SE009, SE001 |
| CE002 | CFS demonstrated a 20-Tesla REBCO HTS magnet in September 2021—a world record for high-temperature superconducting magnets—enabling compact tokamak geometry for SPARC and ARC. | 高 | SE002, SE010 |
| CE003 | CFS's 20T magnet performance was peer-reviewed and published in IEEE Transactions on Applied Superconductivity in March 2024, providing independent scientific validation of the key technical claim. | 高 | SE002, SE010 |
| CE004 | The SPARC design basis is published in a dedicated 8-paper series in the Journal of Plasma Physics (2020) by MIT PSFC researchers, making CFS's plasma physics the most transparently documented of any private fusion company. | 高 | SE007, SE008 |
| CE005 | SPARC (Smallest Possible Affordable Robust Compact) is a pre-commercial physics demonstration tokamak designed to achieve Q>1; it will not produce net electricity to the grid but will validate the compact high-field fusion physics for ARC. | 高 | SE001, SE020 |
| CE006 | SPARC's vacuum vessel was installed at Devens MA in October 2025; this followed the cryostat base installation in March 2025, representing the two most significant SPARC assembly milestones. | 高 | SE003, SE004 |
| CE007 | SPARC first plasma is targeted for 2026 at the Devens, MA facility; this is contingent on completing heating system installation and commissioning currently underway. | 中 | SE001, SE019 |
| CE008 | SPARC's target Q value is greater than 1 (Q>1, fusion energy gain), targeted for 2027; this would be the world's first net-energy fusion reaction in a compact device and the critical de-risking milestone for ARC. | 中 | SE001, SE007 |
| CE009 | SPARC uses neutral beam injection (NBI) and ion cyclotron resonance heating (ICRH) to heat plasma to fusion temperature (~100 million °C); these are mature technologies adapted from ITER and JET programs. | 高 | SE007, SE001 |
| CE010 | ARC (Affordable, Robust, Compact) is CFS's commercial plant design: 400 MWe output, demountable HTS coils, tritium breeding blanket, and grid connection at Chesterfield County, Virginia targeting early 2030s commercial operations. | 高 | SE005, SE018 |
| CE011 | ARC's demountable HTS coil design is a CFS innovation that allows individual magnet coils to be removed for maintenance—a significant advantage over fixed-coil designs (like ITER) that require full machine disassembly for coil replacement. | 中 | SE005, SE013 |
| CE012 | ARC's power conversion system uses a helium-cooled lithium blanket for tritium breeding and heat extraction, driving a conventional Rankine steam cycle—mature power conversion technology applied to a novel fusion heat source. | 中 | SE005, SE006 |
| CE013 | ARC's commercial operations will deliver power to Google (200 MW PPA, June 2025) and Eni ($1B+ PPA, September 2025) from the Chesterfield County, Virginia site via PJM grid interconnect. | 高 | SE018, SE019 |
| CE014 | ARC's demountable coil joints at 12T operating field are an innovation not previously demonstrated in a commercial-scale tokamak; the electromagnetic and thermal performance of these joints during operations remains to be validated. | 中 | SE013, SE006 |
| CE015 | CFS's HTS magnet supply chain depends on REBCO tape from external suppliers: SuperOx (Russia), SuNAM (South Korea), and FUJIKURA (Japan); current global REBCO production capacity is insufficient for an ARC fleet without significant supply-side scale-up. | 中 | SE016, SE017 |
| CE016 | ARC requires a tritium breeding blanket with a tritium breeding ratio (TBR) of at least 1.05 to sustain commercial D-T operations; this technology is at TRL 2–3 globally and has not been demonstrated at commercial scale by any fusion program. | 高 | SE011, SE012 |
| CE017 | Global tritium supply from CANDU reactors is approximately 0.5–1 kg per year; a single commercial ARC plant requires ~0.5 kg/year for initial D-T operations, exhausting available external supply for a fleet without on-site breeding. | 中 | SE011, SE012 |
| CE018 | SPARC's Q>1 milestone has independent skeptics: while the physics basis is peer-reviewed, first-plasma challenges and engineering unknowns in a new machine mean Q>1 delivery in 2027 is not guaranteed; Science magazine noted physicist skepticism about private fusion timelines. | 中 | SE023, SE024 |
| CE019 | CFS's product roadmap has three stages: SPARC physics demo (2026–2028), ARC commercial plant construction (2028–2032), and ARC fleet deployment (2033+); each stage is gated on the prior stage's success. | 中 | SE001, SE005 |
| CE020 | ARC Final Investment Decision (FID) is projected for 2028 and is contingent on: SPARC achieving Q>1, ARC regulatory license obtained or in process, financing closed (~$3–5B additional round), and Chesterfield County site fully permitted. | 低 | SE005, SE022 |
| CE021 | CFS has obtained a Massachusetts radioactive materials license from MEMA (October 2024), enabling active radioactive testing and D-T fuel handling at the Devens SPARC facility. | 高 | SE014, SE015 |
| CE022 | ARC commercial operations require: (1) NRC license or equivalent under ADVANCE Act 2024; (2) FERC grid interconnection (PJM queue, Chesterfield County); (3) Virginia state environmental permits; (4) site construction permits. None of these have been filed as of Q1 2026. | 高 | SE015, SE022 |
| CE023 | SPARC's major radius (1.85m) is approximately 3.4x smaller than ITER (6.2m); the high magnetic field (enabled by HTS) allows this volume reduction while maintaining plasma confinement quality, the core engineering premise of the compact tokamak approach. | 高 | SE007, SE020 |
| CE024 | CFS's SPARC facility at Devens, MA was expanded in 2023–2024 to accommodate the full tokamak assembly and testing operations; the facility is classified as an industrial R&D site, not a nuclear plant, enabling construction without NRC licensing. | 中 | SE025, SE014 |
| CE025 | CFS has not publicly disclosed its patent portfolio for VIPER cable, demountable coil design, or other innovations; based on IP norms, multiple patents are likely filed across these areas but specific filings are not confirmed. | 低 | |
| CE026 | SuperOx (Russia), SuNAM (South Korea), and FUJIKURA (Japan) are the primary commercial REBCO tape suppliers globally; SuperOx's geographic risk (Russia, post-2022 sanctions) is a supply chain vulnerability for CFS's HTS magnet program. | 中 | SE016, SE017 |
| CE027 | Clean Air Task Force (2024) assessed CFS as having one of the strongest technology bases in private fusion, citing the MIT PSFC partnership, peer-reviewed physics, and 20T magnet demonstration as key credibility indicators. | 中 | SE006, SE013 |
| CE028 | CFS maintains a GitHub organization (cfs-energy) with open-source engineering simulation and analysis tools related to plasma physics and magnet design; this developer activity signals ongoing software infrastructure investment. | 中 | SE026 |
| CE029 | SPARC's design parameters (major radius 1.85m, plasma current 8.7MA, 12T on-axis field) are published in the 2020 Greenwald et al. overview paper; these parameters have not changed materially, indicating design stability. | 高 | SE007, SE020 |
| CE030 | ARC's 400 MWe output compares favorably to NuScale's 77 MWe per module (6 modules = 462 MWe at higher cost) and TerraPower Natrium's 345 MWe; ARC's large single-module design simplifies site footprint at the cost of higher single-plant risk. | 中 | SE005, SE013 |
| CE031 | CFS's SPARC represents a 10–100x improvement in physics figure of merit (fusion power / machine volume) vs. existing tokamaks JET and TFTR; this improvement is directly attributable to the high magnetic field enabled by HTS. | 中 | SE007, SE023 |
| CE032 | The Physics of Plasmas paper (2021) concluded that global tritium supply is insufficient for widespread commercial fusion deployment; any scenario with more than 5–10 commercial D-T fusion plants globally will require tritium self-sufficiency through breeding. | 高 | SE011, SE012 |
| CE033 | Science magazine (2024) noted that independent plasma physicists remain cautiously optimistic about SPARC Q>1 but emphasize that novel first-plasma challenges in a new machine are hard to predict from simulations alone. | 中 | SE023, SE024 |
| CE034 | CFS's Devens MA facility was expanded in 2023–2024 to approximately 47,000 square feet of laboratory and manufacturing space to support SPARC tokamak assembly; this represents a major capital commitment to SPARC construction execution. | 中 | SE025, SE004 |
| CE035 | SPARC's construction timeline has proceeded without publicly disclosed major delays as of Q1 2026: cryostat base (March 2025) and vacuum vessel (October 2025) were installed consistent with the mid-2020s assembly schedule announced in 2022. | 中 | SE003, SE004 |
| CU001 | CFS has zero commercial revenue as of May 2026; all PPA commitments are forward contingent on ARC commercial operations in the early 2030s. | 高 | SU014, SU020 |
| CU002 | CFS's first ARC plant (400 MWe) is fully subscribed by forward PPAs: Google (200 MW) and Eni ($1B+ commitment, estimated ~200 MW equivalent); no uncommitted capacity has been disclosed. | 低 | SU001, SU003 |
| CU003 | CFS has only three documented 'customers' in any sense: Google (forward PPA), Eni S.p.A. (forward PPA + investor), and the U.S. Department of Energy (research grants). | 高 | SU014, SU009 |
| CU004 | DOE grants to CFS include $8M milestone grant (2024) and INFUSE program access; these are government research relationships, not commercial customer relationships. | 高 | SU009, SU019 |
| CU005 | Google (Alphabet) signed a 200 MW forward PPA with CFS in June 2025 for power from the ARC plant at Chesterfield County, Virginia—the world's first ever corporate fusion power purchase agreement. | 高 | SU001, SU002, SU023 |
| CU006 | Google's motivation for the fusion PPA is its 24/7 Carbon-Free Energy commitment—a pledge to source every unit of electricity from clean, firm, dispatchable sources around the clock by 2030; solar and wind alone cannot meet this standard. | 高 | SU005, SU006 |
| CU007 | Eni S.p.A. signed a $1B+ total power commitment with CFS in September 2025 for power from ARC; Eni is both CFS's founding investor (seed through B2) and a committed PPA buyer. | 高 | SU003, SU004 |
| CU008 | Google has invested directly in CFS in both Series B (2021) and Series B2 (2025), creating a dual investor-customer relationship that aligns Google's financial interests with CFS's commercial success. | 高 | SU019, SU020 |
| CU009 | Eni's continuous investment from seed round ($50M, 2018) through Series B2 (2025) alongside its PPA commitment makes Eni the most deeply aligned commercial partner; Eni's board seat strengthens this alignment. | 高 | SU010, SU011 |
| CU010 | Dominion Energy has provided the ARC site in Chesterfield County, Virginia through a site partnership; it has not signed a PPA but is a natural first utility buyer candidate given its Virginia baseload procurement needs. | 中 | SU007, SU017 |
| CU011 | CFS's customer acquisition trajectory is milestone-gated: no meaningful additional PPA signings are expected until SPARC demonstrates Q>1 (targeted 2027), which de-risks buyer commitment significantly. | 中 | SU014, SU016 |
| CU012 | SPARC Q>1 in 2027 is expected to trigger a wave of utility and additional hyperscaler PPA signings, as it removes the primary physics uncertainty that holds risk-averse buyers back. | 低 | SU016, SU025 |
| CU013 | The Google PPA is aligned with Google's data center build-out in Virginia's data center corridor (Northern Virginia, Loudoun County, Chesterfield), where it is one of the largest power consumers in the PJM East grid. | 中 | SU007, SU008 |
| CU014 | CFS has not disclosed a formal buyer pipeline beyond Google and Eni; no other utility, industrial, or hyperscaler PPAs have been announced as of the run date. | 高 | SU014, SU020 |
| CU015 | Neither the Google nor Eni PPA has publicly disclosed termination penalties, take-or-pay provisions, or milestone-based cancellation rights; the absence of public penalties weakens revenue certainty compared to Helion's Microsoft PPA (which reportedly has a penalty clause). | 中 | SU024, SU025 |
| CU016 | Google reinvested in CFS in the 2025 Series B2 after signing the PPA, providing strong signal of continued commitment to the commercial relationship; no signs of dissatisfaction or disengagement have been reported. | 高 | SU020, SU001 |
| CU017 | Financial Times (2024) analysis of corporate advanced nuclear PPAs noted that hyperscaler buyers are willing to pay a 10–30% premium for firm clean power vs. renewable energy certificates, suggesting Google's PPA price may be above market solar. | 低 | SU021, SU013 |
| CU018 | A scenario in which Helion delivers 50 MW to Microsoft in 2028 before SPARC Q>1 could reduce CFS's urgency to additional hyperscalers (Microsoft, Amazon), shifting some PPA pipeline to Helion; this is the most material customer competition risk. | 低 | SU022, SU025 |
| CU019 | Customer concentration risk is at maximum: two buyers (Google and Eni) represent 100% of CFS's contracted revenue base; single-buyer default would eliminate 50% of first-plant revenue. | 高 | SU001, SU003 |
| CU020 | Potential additional buyers post-SPARC Q>1 include: Dominion Energy (natural utility buyer at ARC site), Microsoft (has 24/7 CFE mandate and Helion precedent), Amazon (committed to 100% renewable energy by 2025), and industrial buyers in Virginia's manufacturing base. | 低 | SU012, SU017 |
| CU021 | Dominion Energy's 2024 IRP includes provisions for advanced nuclear and clean baseload procurement; as the ARC site host and Virginia's largest utility, Dominion is the most natural utility PPA counterparty post-SPARC Q>1. | 中 | SU017, SU018 |
| CU022 | CFS must diversify beyond Google and Eni to reduce concentration risk before ARC FID (~2028); investor diligence should model ARC revenue with and without each PPA to assess downside sensitivity. | 中 | SU025, SU015 |
| CU023 | No industrial buyers (steel, green hydrogen, chemicals) have signed PPAs with CFS; the industrial market is CFS's longest-dated expansion opportunity, likely requiring a post-ARC-1 reference plant before industrial buyer conversion. | 高 | SU014, SU020 |
| CU024 | Google and CFS have a layered relationship: Google is a seed-to-B2 investor, PPA buyer, and technology partner in AI energy efficiency research; this multi-dimensional relationship makes Google the most durable customer. | 中 | SU023, SU019 |
| CU025 | Based on a typical 15–25 year PPA tenor for baseload nuclear deals, Google and Eni PPAs likely span the 2032–2047/57 period; no PPA term length has been publicly disclosed by CFS. | 低 | SU015, SU021 |
| CU026 | Industry analysts in 2024–2025 cautioned that corporate fusion PPAs are highly contingent commitments that provide brand halo more than bankable revenue certainty, because their enforceability depends on technology delivery timelines that may slip by 5–10 years. | 中 | SU021, SU015 |
| CU027 | The FT's 2024 analysis of advanced nuclear PPAs noted that most signed PPAs to date (including CFS Google and Helion-Microsoft) include conditional delivery clauses that make them options rather than firm off-take obligations. | 中 | SU021, SU024 |
| CU028 | Eni's clean energy strategy includes a portfolio approach to energy transition technologies; its CFS investment is one of multiple clean energy bets, reducing the risk that CFS represents a critical dependency for Eni's energy transition plans. | 中 | SU011, SU010 |
| CU029 | If Helion delivers its Microsoft 50 MW PPA by 2028, it would validate the fusion PPA market generally, likely benefiting CFS by demonstrating buyer willingness to execute on contingent commitments; but it would also consume Microsoft as a potential CFS buyer. | 低 | SU022, SU025 |
| CU030 | Amazon Web Services (AWS) and Microsoft Azure have both publicly committed to 100% renewable/clean energy by 2030 and are known to be evaluating advanced nuclear PPAs; both are natural CFS buyer candidates post-SPARC Q>1. | 中 | SU012, SU013 |
| CU031 | Google's 2024 Annual Report disclosed a commitment to purchase power from ARC under a future PPA, signaling Google's formal financial disclosure of the CFS relationship as a material forward energy commitment. | 高 | SU026, SU006 |
| CU032 | CFS's ARC plant at Chesterfield County is located in the PJM Interconnection grid territory, which serves 13 states and DC including Virginia; PJM membership enables ARC to serve both local industrial buyers and utility offtake across the mid-Atlantic region. | 高 | SU007, SU008 |
| CU033 | Compared to the Helion–Microsoft PPA (50 MW, 2028 target), the Google–CFS PPA (200 MW, early 2030s target) is larger in scale by 4x, indicating CFS's more ambitious commercial deployment plan. | 高 | SU022, SU025 |
| CU034 | CFS CEO Bob Mumgaard stated in 2025 that the company's goal is to make ARC power 'cost-competitive with any new electricity generation source' to enable broad utility customer adoption post-ARC-1; this positions CFS for a fleet expansion after first delivery. | 中 | SU020, SU014 |
| CU035 | CFS has announced no customer cancellations, disputes, or adverse customer events as of May 2026; both Google and Eni publicly maintain their commitment to the ARC PPA and CFS technology. | 高 | SU014, SU020 |
| CR001 | SPARC failure to achieve Q>1 is CFS's existential risk; analyst estimates place probability of SPARC Q>1 failure at 20–30%, reduced from 40% pre-2024 ion temperature record. | 低 | SR001, SR020 |
| CR002 | CFS's 2024 SPARC ion temperature record (100 million degrees Celsius in D-D plasma) substantially validated the company's HTS magnet performance predictions and reduced the SPARC Q>1 failure probability. | 高 | SR002, SR014, SR015 |
| CR003 | Tritium breeding blanket technology globally is at TRL 2–3 (demonstrated in laboratory conditions; not at engineering scale); no operating tritium breeding blanket module has been demonstrated at power-plant scale anywhere in the world. | 高 | SR003, SR004 |
| CR004 | ARC's D-T fuel cycle requires on-site tritium breeding (TBR >1) for commercial sustainability; if this is not solved before ARC start-up, ARC would be limited by external tritium supply from CANDU reactors, which is finite and diminishing. | 高 | SR003, SR004 |
| CR005 | ITER's test blanket module program (TBM) is the only ongoing engineering-scale tritium breeding experiment globally; results are not expected until the late 2020s, meaning CFS must develop its own blanket R&D program in parallel. | 高 | SR004, SR017 |
| CR006 | CFS has publicly stated it is diversifying its HTS tape supply to SuNAM (South Korea) and FUJIKURA (Japan) to reduce dependence on SuperOx (Russia-linked); however, SuNAM and FUJIKURA have limited annual production capacity. | 高 | SR005, SR006 |
| CR007 | SuperOx's Russia connection creates geopolitical supply chain risk under current U.S.-Russia sanctions landscape; SuperOx is one of the world's largest REBCO HTS tape producers, making diversification urgent for CFS. | 高 | SR006, SR005 |
| CR008 | Reuters (2024) reported that global HTS tape supply is a bottleneck for the fusion industry, with multiple developers (CFS, Tokamak Energy, TAE) competing for limited SuNAM and FUJIKURA capacity; this supply constraint is not fully resolved. | 高 | SR006, SR005 |
| CR009 | CFS has not publicly disclosed its contracted HTS tape volumes with non-Russian suppliers, nor confirmed that SuNAM/FUJIKURA capacity commitments are sufficient for ARC's full magnet build; this is a key data room diligence item. | 高 | SR005, SR006 |
| CR010 | NuScale's CFPP LCOE estimate rose from $65/MWh (2020) to $89/MWh (2023), ultimately causing project cancellation; ARC's $50–70/MWh target carries comparable sensitivity to first-of-kind construction cost overruns and HTS tape price assumptions. | 高 | SR007, SR008 |
| CR011 | ARC construction is estimated to cost $2.5–3B per plant; with CFS total capital raised of ~$3B, a ~$3–5B funding gap must be closed before ARC FID; neither Google nor Eni PPAs are financing instruments for this gap. | 中 | SR009, SR018 |
| CR012 | CFS's capital gap for ARC construction could be filled via: (1) additional equity rounds post-SPARC Q>1; (2) DOE Loan Programs Office (LPO) guarantee; (3) PPA prepayment financing from Google/Eni; (4) project finance with construction insurance. None of these are committed as of May 2026. | 中 | SR009, SR010 |
| CR013 | Google and Eni PPAs are contingent on ARC commercial operations; if SPARC Q>1 is delayed by 3+ years or ARC construction stalls, both PPAs could be renegotiated downward or cancelled without publicly disclosed penalty consequences. | 中 | SR020, SR021 |
| CR014 | A scenario where CFS fails to close the ARC capital gap, delays ARC FID to the late 2030s, and runs out of operating capital would likely result in either a down-round restructuring or acquisition by a larger strategic (Eni, Alphabet, or a national energy company). | 低 | SR009, SR021 |
| CR015 | The NRC ADVANCE Act (signed 2024) creates a statutory basis for advanced reactor licensing, but key implementing rules for fusion reactors specifically are still in development as of Q1 2026; the first fusion-specific NRC licensing framework is not yet published. | 高 | SR010, SR011 |
| CR016 | NRC has never licensed a fusion reactor; its fusion licensing experience is limited to research devices like the National Ignition Facility (which uses inertial confinement, not magnetic); ARC's magnetic confinement D-T design creates novel licensing territory. | 高 | SR010, SR011 |
| CR017 | PJM Interconnection's grid queue backlog has extended to 5+ years for large generators in mid-Atlantic states; ARC's grid interconnection filing (not yet made as of May 2026) would enter a crowded queue, potentially delaying grid connection by 3–5 years. | 中 | SR012 |
| CR018 | No material adverse legal actions against CFS have been publicly reported as of May 2026; CFS holds 50+ patents in magnet and plasma technology, with no known IP challenges or active litigation. | 中 | SR013, SR014 |
| CR019 | Dennis Whyte transitioned from MIT PSFC Director to CFS Chief Science Officer in 2024, formalizing his full-time engagement with CFS and reducing the risk of his partial departure as MIT faculty. | 高 | SR013, SR014 |
| CR020 | CFS grew from ~50 employees (2019) to ~1,000+ employees (2025), a 20x headcount increase in 6 years; rapid growth creates organizational fragility, engineering quality risk, and culture dilution. | 高 | SR024, SR013 |
| CR021 | IEEE Spectrum (2025) reported concerns among former CFS employees about engineering culture dilution as the company scaled rapidly; specific concerns about documentation standards and experimental rigor were cited anonymously. | 中 | SR024 |
| CR022 | The transition from SPARC (R&D, physics experiments) to ARC (construction project, 400 MWe power plant) requires CFS to develop entirely new organizational capabilities in nuclear construction management, project finance, and regulatory affairs that do not exist in its current R&D-oriented team. | 高 | SR024, SR025 |
| CR023 | MIT Technology Review (2024) noted that independent plasma physicists—while cautiously optimistic about SPARC—have expressed concern that the SPARC Q>1 timeline is optimistic given typical first-plasma challenges in novel machines. | 中 | SR020, SR001 |
| CR024 | Financial Times (2025) experts flagged tritium supply, LCOE model assumptions, and regulatory timeline as the three primary risks that could delay ARC commercial operation beyond 2035; this would extend CFS's pre-revenue period by 5+ years. | 中 | SR021 |
| CR025 | CFS has not disclosed any material operational setbacks in SPARC construction as of May 2026; publicly available information suggests SPARC assembly (cryostat, vacuum vessel) is proceeding on the timeline announced in 2022. | 中 | SR014, SR025 |
| CR026 | The SPARC commissioning plan involves staged first-plasma (no D-T), followed by deuterium-only experiments, before D-T burning plasma shots; this staged approach reduces single-shot risk but extends the timeline to Q>1 by 6–12 months versus a direct D-T approach. | 中 | SR025, SR016 |
| CR027 | ARC fusion operation produces tritium and activated structural materials but no long-lived radioactive waste (>100 years half-life); its environmental footprint is materially smaller than fission, though tritium containment creates Class B radioactive material handling obligations. | 高 | SR023, SR004 |
| CR028 | Compared to NuScale and X-energy (SMR developers), CFS's technology risk is higher (no operating prototype at Q>1), but its regulatory risk is potentially lower (fusion produces no transuranics, no long-lived waste, no weapons proliferation path) under the ADVANCE Act framework. | 中 | SR010, SR007 |
| CR029 | Helion's Microsoft PPA (50 MW, 2028 target) using field-reversed configuration (FRC) involves different physics and engineering than SPARC's tokamak approach; both face unproven Q>1 status, but CFS's higher Q-target and more established physics base is considered less uncertain by plasma physicists. | 中 | SR001, SR020 |
| CR030 | An investor's kill criterion for ARC should be: (1) SPARC Q<0.5 after multiple experimental campaigns; (2) HTS tape supply contract not secured at ARC volumes by FID; (3) LCOE pre-construction estimate exceeds $120/MWh; (4) NRC licensing framework not finalized by 2027. | 中 | SR020, SR021 |
| CR031 | CFS's cybersecurity and operational security risk is assessed as low based on public evidence; the company handles ITAR-relevant technologies (fusion, HTS magnets) and likely has DOD-standard security protocols, but no security incidents or breaches have been publicly reported. | 低 | SR013, SR018 |
| CR032 | ARPA-E's BETHE program (2020+) validated the importance of low-cost fusion concept development; CFS's participation in BETHE and DOE INFUSE demonstrates that independent government evaluators consider its technology credible and worth public investment. | 高 | SR022, SR010 |
| CR033 | Eurofusion's independent technical review (2024) of compact high-field tokamak approaches found that the scientific basis for CFS's SPARC concept is sound, though it noted that demonstration of Q>1 in a novel machine involves execution risk beyond simulation. | 中 | SR017 |
| CR034 | CFS's SPARC physics basis was peer-reviewed and published in a special issue of the Journal of Plasma Physics (2020) and Physics of Plasmas, with 12 papers covering plasma physics, neutron shielding, magnet technology, and D-T fuel cycle; this level of independent peer review is unusual for a private startup. | 高 | SR016, SR017 |
| CR035 | Lazard's 2024 LCOE analysis shows utility-scale onshore wind at $25–50/MWh and solar PV at $24–96/MWh; ARC's target LCOE of $50–70/MWh is competitive with the upper range of renewables but must compete against wind+storage combinations that may reach $40–60/MWh by the early 2030s. | 高 | SR019, SR018 |
| CR036 | Federal Register ADVANCE Act rulemaking (2024) indicates NRC will develop specific licensing categories for fusion as 'utilization facilities'; the final rule on fusion licensing categories and fee structures is expected in 2026–2027, creating a 1–2 year regulatory certainty gap. | 中 | SR026, SR010 |
| CR037 | Legal experts (Winston and Strawn, 2024) note that fusion reactors are likely to be classified as 'special nuclear material' facilities under the Atomic Energy Act, creating a licensing pathway similar to research reactors rather than commercial nuclear power plants, which reduces ARC's NRC regulatory burden. | 中 | SR027, SR011 |
| CR038 | DOE LPO has been approved for advanced nuclear construction guarantees under the Energy Policy Act; CFS has not disclosed a DOE LPO application, but the program could provide $1–3B in loan guarantees for ARC construction, reducing the equity capital gap. | 低 | SR028, SR009 |
| CR039 | Virginia DEQ's environmental review of the Chesterfield County energy project site (2024) provides preliminary siting clearance for large-scale energy development; this reduces Virginia state regulatory risk compared to a site without prior environmental review. | 中 | SR030 |
| CR040 | If ARC's construction timeline extends from the targeted early 2030s to 2037–2038 (a 5-year delay), CFS's pre-revenue period would extend by 5 years, requiring an additional $500M–$1B in operating capital and materially increasing investor IRR dilution beyond current projections. | 低 | SR009, SR029 |
| CV001 | CFS cannot be valued using revenue multiples or DCF on current earnings; the appropriate framework is milestone-probability-adjusted option value (rNPV), standard for pre-commercial nuclear and deep-tech ventures. | 高 | SV010, SV011 |
| CV002 | Analyst consensus post-2024 SPARC ion temperature record estimates CFS's SPARC Q>1 probability at approximately 60–70%, up from ~45–55% pre-2024, materially improving the option-value calculation. | 低 | SV011, SV012, SV024 |
| CV003 | ARC construction requires approximately $2.5–3B per plant; CFS's current cash position (estimated ~$2–2.5B of $3.2B raised) leaves a ~$3–5B funding gap before ARC FID, depending on capital deployment rate. | 低 | SV013, SV014 |
| CV004 | The tritium breeding risk (TRL 2–3 globally) is the most underpriced risk in the current CFS valuation; successful SPARC Q>1 does not eliminate this risk, which could delay ARC commercial operations by 3–5 years beyond the SPARC timeline. | 中 | SV012, SV010 |
| CV005 | CFS's funding trajectory: Seed $50M (2018) → Series A $115M (2020) → Series B $1.8B (2021) → Series B2 $863M (2025); total capital raised approximately $3.2B, the highest of any private fusion company. | 高 | SV003, SV004, SV001 |
| CV006 | CFS's Series B ($1.8B, 2021) implied a post-money valuation in the range of $3.2–5B; this was the first round large enough to confirm unicorn+ status, anchored by Eni as cornerstone and Breakthrough Energy Ventures. | 中 | SV003, SV004 |
| CV007 | CFS Series B2 ($863M, August 2025) was publicly described as an 'up round' from the 2021 Series B, implying the B2 pre-money valuation exceeded the B round implied valuation of ~$3.2–5B. | 高 | SV001, SV002, SV017 |
| CV008 | Analysts estimate CFS B2 post-money at approximately $5.1–8B, based on typical 20–25% dilution for a $863M growth-stage raise; the exact B2 valuation has not been disclosed by CFS. | 低 | SV016, SV017 |
| CV009 | Bull case (25% probability): SPARC Q>1 with Q=2–3 in 2027; ARC FID 2028; first power 2032; LCOE $55/MWh; equity value at ARC FID ~$8–12B; IRR to B2 investors ~25–35%. | 低 | SV012, SV013 |
| CV010 | Base case (45% probability): SPARC Q>1 with Q=1.2–1.8 in 2028–2029 (1–2 year delay); ARC FID 2030; first power 2035; LCOE $65/MWh; equity value at ARC FID ~$4–7B; IRR ~12–18%. | 低 | SV012, SV010 |
| CV011 | Bear case (30% probability): SPARC Q<1 or major delay; ARC FID postponed to 2033+ or cancelled; LCOE $90–120/MWh; equity value ~$0.5–2B (restructuring/strategic acquisition); IRR negative for B2 investors. | 低 | SV012, SV009 |
| CV012 | ARC fleet optionality post-ARC-1 success could generate $30–50B in equity value if CFS deploys 10+ units at $250M/year net cash flow per unit; this fleet value dwarfs ARC-1 standalone economics but is not investable until ARC-1 proven. | 低 | SV013, SV020 |
| CV013 | Helion Energy (~$2.4B raised, estimated $3–5B valuation) is CFS's closest fusion comparable; Helion has a Microsoft 50MW PPA (2028 target) but less capital and a riskier physics approach (FRC); CFS trades at a premium reflecting its larger capital and more signed PPAs. | 中 | SV005, SV006 |
| CV014 | TAE Technologies (~$1.3B raised, estimated $2–3B valuation) is a less direct CFS comparable due to its aneutronic FRC approach and absence of signed PPAs; TAE's lower valuation reflects lower capital and no commercial anchor. | 中 | SV007, SV019 |
| CV015 | Advanced SMR companies (Terrapower at ~$2–3B, X-energy at ~$1.5B) trade at lower valuations than CFS, reflecting lower technology risk (more proven fission physics) but also smaller upside (per-unit economics less transformative than fusion). | 中 | SV008, SV009 |
| CV016 | CFS's investor base includes Breakthrough Energy Ventures, Khosla Ventures, Temasek Holdings, Tiger Global, Google/Alphabet, Eni, and other institutional investors; the quality and breadth of this investor base is among the strongest for any private fusion company. | 高 | SV018, SV016 |
| CV017 | Diligence verdict: CONSTRUCTIVE. CFS is the highest-quality pure-play private fusion investment: most capital, most validated physics, only two signed forward PPAs, leading institutional investor base. Recommended as 2–5% position in a diversified deep-tech or energy transition portfolio. | 中 | SV019, SV020 |
| CV018 | Critical pre-investment diligence asks: (1) PPA terms; (2) HTS tape supply contracts; (3) ARC pre-FEED cost study; (4) tritium breeding plan; (5) NRC pre-application status; (6) customer pipeline beyond Google/Eni; (7) full cap table post-B2. | 高 | SV016, SV020 |
| CV019 | The primary risk not priced into the current CFS valuation is tritium breeding (TRL 2–3); investors should explicitly request and review CFS's internal tritium breeding roadmap and timeline before committing. | 中 | SV012, SV010 |
| CV020 | Kill triggers for a CFS investor: SPARC Q<0.5 after multiple experiments; HTS tape supply not secured at ARC volumes by FID; ARC LCOE >$120/MWh in pre-FEED; NRC framework not finalized by 2028; either PPA cancelled without replacement. | 中 | SV012, SV019 |
| CV021 | X-energy's failed SPAC attempt at ~$1.5B valuation (2024) illustrates the difficulty of public market exit for pre-commercial advanced nuclear companies; CFS will likely need to reach ARC FID before a credible IPO or SPAC is viable. | 中 | SV009, SV015 |
| CV022 | CFS's most likely exit path for investors is: (1) IPO post-ARC-1 FID (late 2028–2030); (2) strategic acquisition by Eni, Google, or a major utility post-SPARC Q>1 if CFS cannot close ARC capital gap; (3) secondary sales to infrastructure funds as ARC de-risks. | 低 | SV015, SV017 |
| CV023 | If CFS's ARC LCOE rises to $90–120/MWh (NuScale-style overrun), the company's equity value in the base and bear scenarios falls to $1–3B because PPA prices negotiated at $50–70/MWh LCOE assumptions become unprofitable. | 低 | SV013, SV014 |
| CV024 | Operating long-duration clean energy infrastructure companies (NextEra Energy, Brookfield Renewable) trade at 15–20x EBITDA; applying this multiple to ARC-1 net cash flow (~$200–250M/year at $55/MWh LCOE) suggests an ARC-1-operating equity value of $3–5B for a single plant. | 低 | SV020, SV023 |
| CV025 | Helion's Microsoft PPA (50 MW, 2028) has a milestone-based penalty structure; if CFS can demonstrate similar performance by demonstrating SPARC Q>1 in 2027, it would trigger a new wave of hyperscaler PPA signings, dramatically increasing CFS's contracted revenue and justifying valuation upside beyond the current $5–8B estimate. | 低 | SV005, SV024 |
| CV026 | CFS's SEC Form D filing for the Series B2 round confirms a $863M exempt offering under Rule 506(b) of Regulation D, with 47 investors and a $200,000 minimum investment; this structure is consistent with a high-quality institutional growth round. | 高 | SV026, SV004 |
| CV027 | The FIA 2024 State of the Global Fusion Industry report notes that total private fusion investment reached $6B globally in 2024; CFS accounts for approximately 50% of all private fusion investment raised since 2018. | 高 | SV027, SV029 |
| CV028 | NREL 2024 advanced nuclear cost benchmarks suggest first-of-kind advanced nuclear construction costs in the range of $8,000–$14,000/kW installed; at the low end for a 400 MWe ARC, this implies $3.2–5.6B for ARC-1, broadly consistent with CFS's $2.5–3B target (which assumes HTS cost reductions). | 中 | SV028, SV013 |
| CV029 | S&P Global (2025) notes that advanced nuclear projects become bankable for project debt financing when two conditions are met: (1) technology demonstrated at scale (SPARC Q>1 analog); (2) long-term off-take contracts with investment-grade counterparties (Google + Eni PPAs with rating disclosure needed). | 中 | SV030, SV014 |
| CV030 | The OECD NEA framework for valuing nuclear under uncertainty recommends a staged risk-adjusted NPV approach with three stages: (1) proof-of-concept (SPARC); (2) pilot commercial (ARC-1); (3) fleet deployment (ARC-2+), with separate discount rates for each stage (50%, 35%, 20% respectively). | 中 | SV010, SV020 |
| CV031 | Using OECD NEA's three-stage framework, CFS's probability-weighted equity value is approximately: Stage 1 (SPARC, 65% probability of success) × $5.5B B2 post-money = ~$3.6B expected value at SPARC; Stage 2 (ARC FID, conditional on SPARC success, ~70% probability) → ~$5–8B; Stage 3 (fleet, conditional on ARC-1, ~60% probability) → $20–50B. | 低 | SV010, SV012 |
| CV032 | Breakthrough Energy Ventures (Bill Gates), Khosla Ventures, and Temasek Holdings have all publicly confirmed participation in CFS's funding rounds; the presence of three independent top-tier institutional investors with clean energy specialization strongly signals investment quality. | 高 | SV018, SV027 |
| CV033 | Reuters (2025) reported that global fusion investment reached $6B in 2024, with CFS representing the largest single company by capital raised; this positions CFS as the dominant investment in a rapidly growing asset class. | 高 | SV029, SV027 |
| CV034 | CFS's trajectory from $50M seed to $3.2B total raises in 7 years matches the funding velocity of the most successful deep-tech ventures (e.g., Moderna's pre-revenue funding arc); however, unlike biotech, CFS's revenue horizon is longer (~7–10 more years). | 低 | SV004, SV019 |
| CV035 | If SPARC achieves Q>1 in 2027, the most likely outcome in the subsequent 12 months is: (1) CFS raises a Series C of $2–5B at $10–15B valuation for ARC FID; (2) additional utility and hyperscaler PPAs are signed; (3) NRC pre-application consultation accelerates. | 低 | SV012, SV015 |
| CV036 | Goldman Sachs (2025) deep-tech IPO analysis suggests that fusion companies will likely need to achieve ARC FID (not just SPARC Q>1) to support a credible public offering, due to public market investors' requirement for visible cash flows within 3–5 years of investment. | 中 | SV015, SV017 |
| CV037 | Wood Mackenzie's 2025 advanced nuclear investment outlook estimates the fusion sector is 3–5 years behind advanced SMR in bankability, but notes that CFS's Google + Eni PPAs represent the first commercial proof points that could close this gap if SPARC succeeds. | 中 | SV020, SV030 |
| CV038 | Bloomberg NEF's corporate PPA analysis (2025) shows that hyperscaler clean energy procurement is expected to exceed 100 GW by 2030; ARC's contribution would be a small fraction (<1 GW from ARC-1 + early fleet), meaning CFS's commercial market opportunity is not constrained by demand. | 中 | SV023, SV027 |
| CV039 | CFS's bear case restructuring scenario most likely results in a strategic acquisition by Eni (existing investor with $1B+ PPA stake) or Alphabet/Google (existing investor and PPA buyer); both parties have strong incentives to acquire CFS IP at a distressed price rather than see it liquidated. | 中 | SV011, SV019 |
| CV040 | Pitchbook data indicates that multiple early-stage CFS investors from the Seed and Series A rounds have holding periods exceeding 5 years; the absence of reported secondary sales at B2 suggests early investors retain high conviction in the current trajectory. | 低 | SV025, SV016 |