Google will invest $10 billion in Anthropic, with an option to put in a further $30 billion later, deepening ties between two companies that remain both partners and rivals in the race to build frontier artificial-intelligence systems.
Anthropic said it will receive $10 billion in cash from Google at a valuation of $350 billion—the same level as in its latest funding round in February, excluding the new capital being raised. A further $30 billion may be invested if certain milestones are met. At the same time, the agreement provides for a substantial expansion of the startup’s computing capacity.
The aggressive capital raising is tied to the rapid growth in the popularity of Claude Code, an AI agent that accelerates software development. Earlier this week, Anthropic announced another $5 billion from Amazon at the same $350 billion valuation, with room for follow-on investment of up to $20 billion. In February, the company had already raised $30 billion, while some investors put its potential value at more than $800 billion.
Anthropic remains a major Google customer in cloud services and chips—businesses the company is leaning on as growth in advertising slows. Under the agreement, Google Cloud will provide the startup with 5 gigawatts of computing capacity over five years, with the possibility of further expansion. For comparison, one gigawatt can supply electricity to about 750,000 homes in the United States.
After the deal was announced, Google shares rose more than 1% and were trading at $341 by midday.
A central element of the partnership remains Google’s specialized chips—tensor processing units—which are seen as one of the few alternatives to Nvidia’s products and are a scarce resource for AI developers.
Anthropic, which is considering an IPO as early as October, is rapidly scaling its infrastructure as demand rises. Claude Code has quickly become a popular tool among engineers in Silicon Valley, including employees at Google, intensifying competition across the industry. Further growth is also being driven by the Cowork agent, which requires no programming skills and can perform a broader range of tasks.
Anthropic chief executive Dario Amodei previously worked at Google as an AI researcher. The companies have maintained close ties since the startup was founded in 2021 by former OpenAI employees. As recently as last year, Google said it was prepared to provide up to 1 million of its TPU chips, and total investment in the company had by then reached about $3 billion.
At the same time, the partnership is accompanied by direct competition: both companies are seeking to be the first to build AI systems comparable to humans in capability and to secure their position in the corporate market. According to current and former employees, Google’s leadership is increasingly concerned about the company’s standing in AI for programming, where Anthropic is dominant.
Risks also remain for Anthropic itself. The Pentagon has classified the company as a supply-chain risk—a decision it is challenging in court amid public disputes over the possible use of its technology by the U.S. military.
Some analysts also point to the risks of so-called closed-loop deals, in which technology giants simultaneously invest in AI startups and sell them computing resources, including chips and data-center capacity.