CALIFORNIA: Google is writing one of the biggest cheques in the history of artificial intelligence. The Alphabet subsidiary will invest $10bn in Anthropic, the AI startup behind the Claude family of models, with a further $30bn potentially to follow if the company hits performance targets, bringing the total commitment to $40bn. The deal, struck at a $350bn valuation, underscores just how ferociously the technology industry is scrambling to dominate the AI era.
The numbers are eye-watering, but the context is equally striking. Anthropic has been on a fundraising tear. Earlier this week Amazon injected $5bn into the startup, also at $350bn, with an option to add another $20bn over time. In February the firm raised $30bn; since then, investors have been clamouring to back it at valuations of $800bn or more. For a company not yet a decade old, that is a remarkable trajectory.
Google’s stake in the relationship goes well beyond cash. As part of the deal, Google Cloud will supply Anthropic with five gigawatts of computing capacity over the next five years, with several more potentially to follow. The arrangement expands an earlier agreement struck between Anthropic, Google and Broadcom. Google’s tensor processing units, among the best alternatives to Nvidia’s chips and fiercely sought by AI developers who consume computing power at a ferocious pace, are a crucial part of the package.
Anthropic has ramped up its fundraising on the back of Claude Code, an AI agent that has rapidly become the tool of choice for software engineers across Silicon Valley, including many at Google itself, a fact that has set alarm bells ringing at the search giant’s headquarters. The startup’s Cowork agent, which requires no coding skills and is designed for broader business use, is also growing fast.
The relationship between the two companies is tangled with history. Dario Amodei, Anthropic’s chief executive, is a former Google AI researcher who founded the startup in 2021 alongside a clutch of ex-OpenAI employees. Ties have remained close ever since. Last year Google committed to providing up to 1m of its TPU chips to Anthropic in a deal worth tens of billions of dollars, by which point it had already ploughed roughly $3bn into the startup.
Yet for all the financial intimacy, the two are racing each other hard. Google’s leaders have grown increasingly anxious about their position in the booming market for AI coding tools, an arena that Anthropic currently dominates. They are partners and rivals, and both know it.
Anthropic is not without vulnerabilities of its own. The Pentagon has designated the company a supply-chain risk, a label Anthropic is contesting in court, following a public row over how its technology might be deployed by the American military. Financial analysts have also flagged the circular logic embedded in deals of this kind, in which computing giants invest in AI startups and simultaneously sell those same startups chips and data-centre capacity. The money, in other words, has a habit of flowing straight back.
Anthropic is weighing an initial public offering as early as October. Whether the market will reward its stratospheric valuation will depend on whether Claude can keep its lead, and whether its biggest backer can live with coming second.
