Summary: Google is investing up to $40 billion in Anthropic ($10B cash now at $350B valuation, $30B conditional on performance targets) plus five gigawatts of compute over five years, bringing its total commitment to ~$43B. The deal arrives as Anthropic’s revenue hit $30B ARR (up from $1B in January 2025) and its secondary market valuation reached $1 trillion, while Google DeepMind has assembled a “strike team” to catch up to Claude in coding. Amazon committed up to $33B in Anthropic days earlier, making both cloud giants simultaneously competing to lock in the same AI company.

Google is preparing to invest up to $40 billion in Anthropic, the company whose flagship model Claude is outselling Gemini in the enterprise market that Google considers its own. Bloomberg reported on Thursday that the deal comprises $10 billion in cash at Anthropic’s existing $350 billion valuation, with an additional $30 billion contingent on undisclosed performance targets. Alongside the equity, Google will provide five gigawatts of computing power over five years, including access to up to one million of its seventh-generation Ironwood TPU chips. The investment would bring Google’s total commitment to Anthropic to approximately $43 billion, making it the largest single-company bet on an AI rival in history. It is also, by any honest reading, an admission that Gemini alone is not enough.

Anthropic’s annualised revenue run rate hit $30 billion in April, up from $1 billion in January 2025. No company in American technology history has grown at that rate. Claude holds 32% of the enterprise large language model API market, ahead of OpenAI’s GPT-4o at 25%. Eight of the Fortune 10 are Claude customers. More than 1,000 businesses spend over $1 million annually on the platform, a figure that doubled since February. Anthropic’s implied valuation reached $1 trillion on secondary markets just three months after its $380 billion Series G, overtaking OpenAI’s secondary pricing at $880 billion. Google invested its initial $300 million in 2023 when Anthropic was worth approximately $5 billion. The $350 billion price tag on the current deal represents a 70-fold increase in under three years.

The structure tells the story

Google’s existing stake in Anthropic stands at approximately 14%, acquired through more than $3 billion in prior investments. It is contractually capped at 15% and carries no voting rights, no board seats, and no board observer privileges. The new $10 billion pushes Google close to that ceiling without breaching it, a structure that appears designed to avoid triggering merger review thresholds while maintaining meaningful economic exposure. The conditional $30 billion, tied to performance milestones that neither company has disclosed, functions less like equity investment and more like a structured compute-and-capital arrangement: if Anthropic hits its targets, Google pays more and gets more, but the relationship remains that of investor and customer, not parent and subsidiary.

The deal arrived four days after Amazon announced its own escalation. Amazon is now committed to up to $33 billion in Anthropic, comprising $8 billion previously invested plus $5 billion in new capital and $20 billion in conditional future investment. In return, Anthropic pledged to spend $100 billion on Amazon Web Services over the next decade. Amazon’s $25 billion investment in Anthropic and Google’s $40 billion sit on the same cap table, meaning the two largest cloud providers are simultaneously competing to lock in the same company as their most important AI customer. Amazon has separately invested up to $50 billion in OpenAI. The cloud wars have become AI custody battles, and the parents are paying record-breaking sums.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

What Google is actually buying

The five-gigawatt compute commitment is more strategically significant than the equity cheque. Anthropic’s major compute deal with Google and Broadcom, announced earlier this month, already gave Anthropic access to approximately 3.5 gigawatts of Google TPU capacity manufactured via Broadcom starting in 2027, on top of one gigawatt supplied this year. The new deal layers another five gigawatts on top. Anthropic trains and deploys Claude on three chip platforms simultaneously, Google TPUs, Amazon Trainium, and Nvidia GPUs, maintaining strategic independence from any single provider. Google’s play is to make its TPUs so deeply embedded in Anthropic’s infrastructure that switching becomes prohibitively expensive, regardless of what the equity stake allows.

Google Cloud already distributes Claude through its Vertex AI platform to thousands of enterprise customers, including Coinbase, Cursor, Palo Alto Networks, Replit, and Shopify. Every API call routed through Vertex generates cloud revenue for Google whether or not Gemini is involved. The $40 billion investment ensures that as Anthropic’s usage grows, a substantial share of that compute spend flows through Google’s infrastructure. It is a cloud computing play disguised as an AI investment, and it is structured to generate returns even if Gemini loses every head-to-head benchmark against Claude for the next decade.

The Gemini problem

Google does not frame the investment as a concession. It argues that owning all four layers of the AI stack, custom silicon, frontier models, cloud platform, and enterprise distribution through Workspace’s three billion users, makes it uniquely positioned to profit from AI regardless of which model leads. The framing is reasonable. The internal reality is less comfortable. Google DeepMind has assembled what employees describe as a “strike team” specifically to close the gap between Gemini and Claude in coding, after internal assessments showed Claude’s superiority in the category that drives the most enterprise revenue. Google co-founder Sergey Brin personally raised alarms. Claude Code, Anthropic’s coding product, reached $1 billion in annual run rate within six months of launch and now generates over $2.5 billion.

Microsoft launched its own AI models to compete with OpenAI after renegotiating its contract in September 2025, a move that signalled the beginning of the end of unconditional loyalty between cloud providers and their AI partners. Google is taking the opposite approach: rather than building distance from its AI investment, it is doubling down, pouring capital into Anthropic at the same time it is trying to make Gemini competitive. The two strategies are not contradictory, but they reveal different assessments of risk. Microsoft is hedging against OpenAI’s independence. Google is hedging against Gemini’s inadequacy.

The regulatory question

The investment lands in a regulatory environment that is paying closer attention to AI partnerships than it was a year ago. The FTC’s 2025 study on AI partnerships warned that cloud-provider investments in AI developers risk “locking in the market dominance of large incumbent technology firms.” Senators Elizabeth Warren and Ron Wyden have sent letters probing both the Google-Anthropic and Microsoft-OpenAI relationships over antitrust concerns. The Department of Justice and the FTC launched a joint public inquiry in February on updated guidance for competitor collaborations, citing emerging technologies explicitly. Google’s 15% ownership cap and absence of board representation were almost certainly structured with these proceedings in mind. But $43 billion in total investment and a multi-gigawatt compute dependency create a degree of economic entanglement that no ownership percentage can fully capture.

Anthropic has rejected venture capital offers valuing it at $800 billion or more, preferring to raise from strategic partners who bring compute rather than just capital. That preference gives Google and Amazon structural advantages over financial investors but also concentrates Anthropic’s dependency on two companies that compete with each other and with Anthropic’s own enterprise customers. Google’s $750 million fund for agentic AI deployments, announced at Cloud Next this week, makes the integration even deeper: Accenture, Deloitte, KPMG, and PwC will build AI agents on Google’s platform, and those agents will increasingly use Claude alongside Gemini.

The price of honesty

Google launched Gemini 3.1 Pro two days before the Anthropic deal was reported. The timing was not accidental. Google needs the market to see it as an AI leader in its own right, not merely as the company that writes cheques to the AI leader. But the $40 billion tells a different story. You do not invest that sum in a competitor’s product because you are confident in your own. You invest it because the competitor’s product is generating $30 billion in annual revenue and growing faster than anything you have ever built, and because the alternative, letting Amazon or someone else become that company’s primary infrastructure provider, is worse than the admission the investment implies.

Anthropic projects positive free cash flow by 2027, years ahead of OpenAI’s target of 2030. If that projection holds, Google’s equity stake will generate returns independent of the cloud revenue. If it does not hold, Google still captures billions in compute spending through the infrastructure commitments. The deal is structured so that Google wins whether Anthropic succeeds spectacularly or merely succeeds. The only scenario in which Google loses is the one in which it did not invest and Anthropic’s $30 billion in revenue flowed exclusively through Amazon’s cloud. That is not an investment thesis. It is a defensive necessity dressed up as a strategic partnership, and $40 billion is the price of making sure the best AI company in the world keeps running on your chips.