Google Commits $40B to Anthropic at $350B Valuation — With Another $30B Tied to Milestones
Google is investing up to $40 billion in Anthropic. The deal closes in two tranches: $10 billion wired immediately at a $350 billion valuation — the same figure from Anthropic’s February round — with the remaining $30 billion conditional on performance targets that Anthropic has not publicly disclosed. The announcement came on the same day Amazon confirmed an additional $5 billion tranche of its own investment, bringing total pledged capital from Google and Amazon alone to $65 billion.
The Structure
The milestone-linked architecture is significant. Performance targets of this scale typically gate on revenue growth, product deployment, or safety benchmarks — not on model capability in isolation. Anthropic reached a $30 billion annualised run rate in early 2026; the next logical thresholds would be crossing $50B or demonstrating commercial Mythos deployments at enterprise scale.
The $10B tranche is at the February round valuation of $350 billion — flat to that round’s pricing. At $40B, Google would hold a position substantially larger than any prior Anthropic investor. OpenAI’s equivalent structure for Microsoft capped at $13 billion before the relationship restructured in 2025.
Compute Is the Real Story
The investment comes bundled with a commitment to provide 5 gigawatts of computing capacity over the next five years via Google Cloud. That is not a rounding error. Five gigawatts is roughly five times the total compute capacity of the largest single hyperscale data centre campus in operation today.
Anthropic already relies heavily on Google Cloud for inference, and struck a multi-gigawatt TPU deal with Google and Broadcom in April 2026 that brings custom silicon capacity online in 2027. The new investment effectively cements Google as Anthropic’s primary infrastructure supplier through the rest of the decade.
Amazon’s $100 billion AWS compute agreement sits on the other side of the ledger. Anthropic is now positioned to run frontier workloads against both the largest GPU network (AWS) and the largest TPU network (Google Cloud) simultaneously — a hedge that no other lab currently has.
Why Google Is Writing a $40B Cheque to a Competitor
Google DeepMind’s Gemini 3.1 Pro Preview currently ties Claude Opus 4.7 at the top of the Artificial Analysis Intelligence Index at 57. The two companies are in direct competition at the frontier. Yet Google Cloud’s revenue from Anthropic inference workloads — already in the billions annually — makes Anthropic one of its largest customers. The investment deepens that dependency.
There is also a defensive component. Microsoft’s OpenAI investment made Azure the default inference layer for the most widely deployed AI products in the enterprise. Google cannot afford to see Anthropic shift to AWS infrastructure exclusively and take that customer base with it.
The timing is not coincidental. Mythos, Anthropic’s restricted cybersecurity model, launched in early April with government and enterprise access. Japan’s financial regulators convened emergency sessions over it. The NSA is deploying it for offensive cyber work. A model with that profile generates institutional demand that requires sovereign infrastructure guarantees — the kind that only a $40B equity partner with five gigawatts of capacity can credibly make.
Anthropic’s Capital Position
Anthropic has now raised more capital than any private AI company in history on a single investment cycle. The full tally since January 2026:
| Investor | Amount |
|---|---|
| Google (new) | $10B (+ $30B milestone) |
| Amazon (additional) | $5B |
| Amazon (existing commitment) | $25B |
| OpenAI Round (various) | — |
The $65B combined figure from Google and Amazon alone dwarfs OpenAI’s $122 billion total round at $852 billion valuation in absolute dollar terms. Whether Anthropic’s $350 billion valuation holds through the milestone tranches depends on whether the commercial deployment of Mythos converts into the kind of revenue that justifies it.