Google Commits Up to $40 Billion to Anthropic in Its Biggest AI Bet Yet

Google has agreed to invest up to $40 billion in Anthropic, the AI safety company behind Claude, in what would be the largest single investment in an AI lab to date. The deal commits $10 billion immediately at a valuation of $350 billion, with a further $30 billion tied to performance milestones.

The money comes with serious compute. Google Cloud will supply Anthropic with around 5 gigawatts of computing capacity over five years, giving the company the processing power to train and run frontier models without depending on the open market for GPU time.

Amazon moved in the same direction this week, adding $5 billion to its existing Anthropic stake as part of a broader infrastructure agreement. The lab is expected to spend up to $100 billion on Amazon infrastructure over time, representing another 5 gigawatts. Between the two cloud giants, Anthropic has locked in the compute capacity that most AI labs can only hope for.

That matters because training frontier models requires processing power at a scale that standard cloud services cannot readily provision. Access to dedicated compute has become as important to competitive position in AI as the research itself, which is why these infrastructure deals carry as much weight as the equity figures.

The timing matters for other reasons too. Anthropic recently released Mythos, its most powerful model, to a limited set of partners. The model carries significant cybersecurity capabilities and has already attracted attention from unauthorised users. The company is reportedly considering an IPO as early as October 2026, and some investor valuations are now running at $800 billion or higher.

For Google, the investment is both a competitive hedge and a bet on infrastructure. Anthropic's Claude competes directly with Gemini in the enterprise market. By embedding itself as Anthropic's primary cloud provider, Google gains influence over a rival while locking in years of compute spending.

Anthropic has no shortage of momentum on other fronts. The company outspent OpenAI on lobbying in the first quarter of 2026, and Mythos has been at the centre of debate over how much access the public should have to the most capable AI systems. The $40 billion from Google confirms that whatever questions remain around safety and deployment, the financial commitment to frontier AI is not going anywhere.