Money that can flow back
Of the $40 billion, Google is giving only $10 billion in cash upfront — at a valuation of Anthropic of around $350 billion. The remaining $30 billion is tied to performance milestones and will be released in tranches if Anthropic achieves the set goals. Bloomberg ↗The real catch of the deal isn't in the cash. In practice, part of the capital can flow back to Google through computing capacity contracts. In early April, Anthropic announced a deal with Google and Broadcom for multi-gigawatt capacity of next-generation TPU chips, which will begin rolling out from 2027 and will go almost entirely to American data centers. In November 2025, Anthropic pledged to invest $50 billion in American infrastructure. Anthropic ↗
The mechanism is simple: Google sends the money, Anthropic spends it at Google Cloud on TPU chips and server time, and part of the money returns to Google Cloud's revenues. The investment functions as an equity-financed cloud contract — Alphabet (Google's parent company) gains an equity stake, Anthropic gets capital for expansion, and Google Cloud gets higher revenues.
Why Gemini isn't enough
The question of why Google is pouring billions into a company with a competing chatbot has its answer in the enterprise market. According to the portal TheNextWeb, Claude holds 32 percent of the enterprise LLM API market, while OpenAI's GPT-4o holds 25 percent according to the same source. TheNextWeb also states that eight out of ten companies on the Fortune 10 list use Claude, and more than a thousand companies spend over a million dollars a year on Anthropic's platform. TheNextWeb ↗Google distributes Claude through its Vertex AI platform to enterprise customers including Coinbase, Shopify, Replit, Cursor, and Palo Alto Networks. Without Anthropic, these customers would end up with a competitor — on Amazon Bedrock or Microsoft Azure. For Google Cloud, which lags behind both AWS and Azure, Anthropic is a key magnet for enterprise contracts.
The strategy has precedent. Microsoft operates its own Copilot product while simultaneously maintaining an exclusive tie with OpenAI via Azure. Google is choosing similar logic: its own Gemini plus equity in Anthropic. If Gemini loses the race for the strongest model, or if the enterprise market definitively favors Claude, Alphabet has a backup plan ready.
Three coalitions that control AI
After the deal, the AI sector is clearly consolidating into three coalitions: Microsoft with OpenAI, Google with Anthropic, and Amazon, which was originally supposed to be Anthropic's primary patron and is now losing the exclusivity of its strategic narrative. Anthropic is a key reference customer for AWS Trainium, especially through the Project Rainier program — Amazon had previously pledged $5 billion in cash and another $20 billion conditionally. After Google's offensive, AWS's position no longer seems as dominant. TechCrunch ↗For Brussels, the situation is regulatorily new. The European Commission has already examined the Microsoft-OpenAI tie under competition rules. The Google + Anthropic deal has the same structure — equity without direct ownership, but with a tied cloud contract. Slovak and Czech entrepreneurs building applications on top of cloud APIs are, in practice, now moving only between three hyperscalers.
What hotinfo is watching
- Performance milestones — when Google releases the remaining $30 billion (an Anthropic announcement, Alphabet's 10-Q)
- The EU's response — the European Commission formally begins assessing the Google–Anthropic tie-up (by analogy with Microsoft–OpenAI)
- Amazon's position — whether AWS answers with another tranche or shifts to its own Bedrock/Trainium
- The price of cloud for EU firms — Vertex AI, AWS Bedrock and Azure OpenAI API pricing
- Anthropic revenue — whether the Q1 2026 annualised run rate of $30bn holds or rises
- The Microsoft–OpenAI EU decision — a precedent for Google–Anthropic
- A fourth player — Meta, xAI or another frontier startup joins a hyperscaler alliance
