Google commits $10B immediately plus $30B conditional on performance targets, valuing Anthropic at $350B and bundling a fresh 5GW of Google Cloud compute over five years.
Key Takeaways
Mythos, Anthropic’s most powerful model to date, has significant cybersecurity applications but was restricted to select partners due to misuse risk and has already leaked to unsanctioned parties.
Google Cloud’s new 5GW commitment stacks on the existing Anthropic-Google-Broadcom TPU deal (3.5GW starting 2027), making Google both a direct AI competitor and a primary infrastructure supplier to Anthropic.
Amazon invested an additional $5B this week under a separate agreement where Anthropic is expected to spend up to $100B for roughly 5GW of capacity; CoreWeave also has a data center deal in place.
Anthropic’s valuation was $350B in February; investors are now reportedly willing to back the company at $800B or more, with an IPO potentially as soon as October 2026.
Widespread Claude use limit complaints in recent weeks preceded this week’s infrastructure announcements, suggesting compute scarcity is an active operational constraint, not just a capital story.
Hacker News Comment Review
This thread is a duplicate; the bulk of technical discussion was consolidated in an earlier thread, so signal here is thin.
One commenter noted Google was already an existing investor, framing this as a straightforward re-up rather than a new strategic shift, and characterizing Anthropic’s cap table as a coalition of major tech incumbents recycling capital into the AI field.
Notable Comments
@state_less: frames Anthropic as “a MicroAmaGooVidia amalgamation” of incumbent capital, arguing the $10B hedge is an easy call for Google given its existing position.
