🕶️ Google’s $40B bet on Anthropic’s next move

AI has entered its infrastructure era: Helsinki’s cloud upstart is hiring like demand is already booked, sovereign alliances are trying to route around Big Tech, and even creator tooling is suddenly worth $500M when control becomes a feature, not a footnote.

Meanwhile, X-energy’s $1B Amazon-backed IPO suggests the real growth market may be less ‘models’ than power, permits, and anyone willing to underwrite 5 gigawatts of ambition. In startup terms: the roadmap got very physical, very expensive, and oddly confident.

The hidden team behind unicorn growth for Suno, Jeeves, and Picsart.

We’ve hired agencies. Most sucked. So we built one for founders like us.

Free growth plan for The Startup Club readers. We’ll show you the levers to scale fast in your niche over 6 months.

X-energy raised $1 billion after pricing 44.3 million shares at $23, above the $16–$19 range, roughly 20% more capital than planned. The listing rode investor demand tied to data center power needs, coming to market through an upsized $1B IPO, and saw a first-day 27% jump. The company’s small modular reactors target both electricity and industrial heat, with 80-megawatt units and early customers in chemicals and cloud.

Suppliers of TRISO fuel and helium-cooled reactor components now have a clearer path to scale, which could shorten lead times for high‑temperature process heat beyond power generation. Hyperscalers and heavy industry will lean on long-term PPAs (Power Purchase Agreements) to secure firm, zero‑carbon capacity, pressuring grid interconnection queues and local permitting.

For founders in energy development software, interconnection tooling, site selection, or heat‑recovery hardware, model around multi‑unit campuses and offtake tied to pledged 5 GW capacity by 2039. Watch procurement criteria from cloud buyers, emerging credit structures that back nuclear offtake, and opportunities to colocate reactors with industrial parks where waste heat can be monetized.

In AI, friendship now starts at a $40B check. Google is committing up to $40B in cash and compute to Anthropic, $10B now at a $350B valuation, with $30B contingent, paired with another 5 GW of TPU capacity over five years, extending its earlier TPU and Broadcom chip pact. Commits up to $40B is less a funding round than a long-dated infrastructure treaty.

This locks Anthropic deeper into a tri-cloud, multi-chip strategy: Amazon’s $5B for up to $100B in future spend, CoreWeave data centers, and now Google as both rival model lab and primary landlord. The catalyst is Mythos, a highly capable, cybersecurity-heavy model that is expensive to run and already leaking into unauthorized use. Scramble for safe deployment and massive chip preorders are turning frontier AI into a capital expenditure arms race where cloud access, not algorithms, sets the frontier.