Analysis
Three separate infrastructure rounds landed within days of each other this week, and together they say more about where AI capital is actually flowing than any single funding announcement. Valar Atomics closed a $1 billion Series B led by Sequoia at a $6 billion valuation -- triple its mark from months earlier -- to mass-produce small nuclear reactors for data centers, fresh off powering an Nvidia Blackwell cluster with its first reactor. The same week, Base Power raised a $1 billion Series D led by Ribbit, Addition, Valor Equity and JPMorgan's Strategic Investment Group at a $13 billion valuation to scale home batteries built at its Austin factory. And Tesla and SpaceX jointly committed $16.8 billion to Terafab, an initial-phase chip factory in Grimes County, Texas.
Add those three together and roughly $18.8 billion moved into power generation, grid storage and chip manufacturing in a single week -- more than most entire AI model-layer funding weeks this year, and none of it went to a chatbot or a foundation model.
The split is now familiar: model-layer valuations get the headlines, but capital backing physical capacity -- reactors, batteries, fabs -- is compounding just as fast, and arguably with less competitive risk, since a nuclear reactor or a battery factory doesn't face the same commoditization pressure a mid-tier LLM does.
Founders and GPs pricing infrastructure deals should treat this as the comp set: Valar's 3x markup in months and Base Power's jump to $13 billion both happened without a single model release attached to either company. What none of these three rounds answer yet is whether underlying AI data center demand holds at its current pace long enough to make the capacity these dollars are buying necessary rather than excess.