档案库 · 气候与能源 · 技术决策 · 2018–2026
CFS安装首块SPARC磁体，押注HTS磁体实现聚变Q>1
CFS，MIT衍生公司，押注高温超导磁体缩小托卡马克；2026年1月安装首块SPARC磁体，目标2027年首次等离子体，已融资30亿美元，2030年代初并网。
Commonwealth Fusion Systems
做的是什么生意
CFS makes fusion power plants: first build the SPARC demonstration machine to verify that magnetically confined fusion can generate electricity, then build ARC (400MW, near Richmond, Virginia), which it claims will be the world's first stable grid-connected fusion power plant.
启动资金：Cumulative funding of about $3B, including the $863M Series B2 in 2025-08, participated in by about 30 investors including Nvidia and Google
起因
In 2018, Commonwealth Fusion Systems (CFS) was spun off from the MIT Plasma Science and Fusion Center (MIT 等离子体科学与聚变中心) and founded by Bob Mumgaard and others, betting on high-temperature superconducting (REBCO) magnets: with stronger magnetic fields, a tokamak can be much smaller than ITER and much lower in construction cost. The company first built the SPARC demonstration machine outside Boston, with the goal of first proving scientific feasibility, then scaling it up into a commercial power plant.
经过
In 2025-03, the 24-foot-wide, 75-ton stainless steel cryostat for SPARC was put in place; on 2026-01-06, at CES, CFS announced that it had installed the first of 18 24-ton D-shaped high-temperature superconducting magnets (capable of producing 20 tesla, about 13 times the magnetic field of a conventional MRI, cooled to -253°C and carrying 30,000 amps of current). CEO Bob Mumgaard said all 18 magnets would be installed by late summer that year, the reactor would be largely complete by year-end, and first plasma would occur in 2027. In the same period, CFS announced collaborations with Nvidia (Omniverse digital twin) and Siemens (design and manufacturing software), and disclosed cumulative funding of about $3B.
结果
Still on the way: as of 2026-09-01, SPARC is still under assembly, with first plasma expected in 2027; the first commercial power plant ARC (400MW, enough for about 300,000 households) is planned to connect to the grid in the early 2030s outside Richmond, Virginia, and if successful would be the world's first fusion power plant stably supplying electricity to the grid.
背景
Commonwealth Fusion Systems (CFS)制造聚变电站：2018年从MIT等离子体科学与聚变中心剥离，押注高温超导磁体能建造比ITER更小但更强的托卡马克。公司先在波士顿外建SPARC示范机，证明聚变产出能量大于消耗（Q>1），再扩展为400MW商业电站ARC。
2025年3月，SPARC的75吨不锈钢低温恒温器就位；2026年1月6日，CFS在CES上宣布已安装18块D形高温超导磁体中的第一块。每块磁体重24吨，能产生20特斯拉磁场（约为常规MRI的13倍），冷却至-253°C后承载30000安培电流，将等离子体加热至超过1亿度。CEO Bob Mumgaard表示当年夏末将安装全部18块磁体，年底前反应堆基本完工，2027年实现首次等离子体。
为在真机点火前消除风险，CFS与英伟达（Omniverse数字孪生）和西门子（设计与制造软件）合作，用虚拟模型和物理机器并行学习。资金方面，公司累计融资约30亿美元，包括2025年8月的8.63亿美元B2轮，英伟达和谷歌等约30家投资者参投。
截至2026年9月1日，SPARC仍在组装，尚未点火；若示范成功，首个商业电站ARC计划在2030年代初于弗吉尼亚州里士满附近并网，届时将成为全球首个稳定向电网供电的聚变电站。这一押注的终局尚未到来。
这件事要成立，得有什么
- 高温超导磁体是关键；若性能不足，公司失败。因此自产磁体，再组装整机。
- SPARC仅验证发电；商业风险推迟到ARC。里程碑与融资节奏匹配。
- 与英伟达/西门子合作数字孪生，与物理机器并行学习，压缩验证周期。
- 30亿美元押注2030年代初并网，寄望于AI数据中心电力需求。
可借鉴之处
将聚变分解为可验证的里程碑：先证明磁体，再证明发电。使用数字孪生降低风险和成本。这套方法将深度技术从实验室推向商业电站。
后续进展
截至2026年9月1日：SPARC组装按计划推进，预计2026年夏末安装全部18块磁体，年底前反应堆基本完工，2027年首次等离子体；ARC（400MW）选址弗吉尼亚州里士满附近，目标在2030年代初成为全球首个稳定并网的聚变电站；公司累计融资约30亿美元，正与英伟达和西门子共建数字孪生。聚变尚未发电，押注未结束。
资料来源
- Commonwealth Fusion Systems installs reactor magnet, lands deal with Nvidia
- Fusion power nearly ready for prime time as Commonwealth builds first pilot for limitless, clean energy with AI help from Siemens, Nvidia
发现哪里写错了？告诉我们。
轮到你了
你刚读完一家。说说你在做什么，看看谁在赌同一件事。
免费账号 · 3 次免费提问 · 不用绑卡