Google will invest at least $10 billion in Anthropic, with the figure rising to as much as $40 billion if Anthropic hits performance targets, Bloomberg reported. The deal values Anthropic at $350 billion and lands just days after Amazon committed another $5 billion in an initial tranche at the same valuation. Both investors have left the door open to further capital tied to milestones.
The pair of deals is less about cash and more about compute. Google is routing TPU capacity — including its TPU 8i chips — and cloud services to Anthropic, while Amazon is doing the same through AWS. The investments translate, in practice, into Anthropic buying compute from its backers at scale.
Anthropic's $350 billion mark is roughly 2x the valuation it carried earlier in its funding history and reflects the sharp run-up in Claude demand over the past year. The company has already pushed past the point where its existing compute footprint can absorb usage, with outages and rate limits becoming a regular feature of the service.
Key facts
- 01Google will invest a minimum of $10B in Anthropic, rising to as much as $40B tied to performance milestones.
- 02The deal values Anthropic at $350B, matching the mark set by Amazon's most recent round.
- 03Amazon put in $5B as an initial tranche a few days earlier, with room for more based on performance.
- 04Both investments are structured to route chips and cloud capacity back to Anthropic for Claude training and inference.
- 05Bloomberg first reported the Google commitment on April 24, 2026.
Claude Code, Anthropic's coding-focused agentic product, has been one of the main demand drivers. It lets developers and companies offload larger chunks of software work to the model, with results that Ars Technica notes vary from meaningful productivity gains to setbacks depending on how it is deployed. Claude Cowork, a newer product aimed at general knowledge work, is extending that same pattern outside of engineering teams.
“Google will put at least $10B into Anthropic, with the figure climbing to $40B if performance targets are hit — and both investors now value the company at $350B.”— Jaeden Schafer
Anthropic's recent growth has also been helped by turbulence at OpenAI, where controversies around ChatGPT have pushed some enterprise customers to diversify their model providers. More reliable agentic workflows in Claude have made that shift practical rather than theoretical.
The supply-demand gap has gotten severe enough that Anthropic has tested peak-hour limits and floated removing the most compute-intensive tools from cheaper plans. The Google and Amazon deals are, in effect, a funding mechanism to close that gap without requiring Anthropic to raise pure equity from financial investors alone.
The structure is now a familiar one in AI. Microsoft pioneered it with OpenAI: a hyperscaler invests, the AI lab spends that money on the hyperscaler's cloud and chips, and the hyperscaler books revenue while locking in a strategic foothold. Google and Amazon are both running that playbook with Anthropic simultaneously, which is unusual — most previous pairings have been exclusive.
This is also not Google's first check into Anthropic, despite the fact that Google's own Gemini models compete directly with Claude. That tension — backing a rival to keep it on your cloud — has become a defining feature of the current AI market rather than an exception.
The caveat sits in the structure itself. The $40 billion figure is a ceiling, not a commitment, and it is gated on performance targets that neither party has disclosed. If Claude growth slows, or if Anthropic's unit economics on agentic products deteriorate, the incremental tranches may never land. The $5 billion from Amazon and the $10 billion floor from Google are the numbers that are real today.
What this confirms is that the AI frontier is now a compute contest settled on hyperscaler balance sheets. Anthropic does not need to choose between Google and Amazon — it can take billions from both, spend it on their chips, and let them compete over who captures more of the inference workload. For a market that once looked like it would consolidate around one lab and one cloud, the Claude maker is quietly building the most diversified compute base of any frontier lab.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
