Google Bets Up to 40 Billion Dollars on Anthropic as the AI Compute Race Reaches New Heights

The race to build the most advanced artificial intelligence has entered a new and staggeringly expensive phase. Google has agreed to invest up to 40 billion dollars in Anthropic, the company behind the Claude family of AI models, in a deal that highlights just how much capital and computing power the pursuit of frontier AI now demands from the industry's biggest players.

A Deal Worth Tens of Billions

The agreement, announced on April 24, 2026, is structured in two parts. Google is committing 10 billion dollars in cash upfront, while a further 30 billion dollars is tied to whether Anthropic meets certain performance benchmarks. The investment values the company at an eye watering 350 billion dollars, a figure that would have seemed unimaginable for an AI startup only a few years ago.

Google is far from the only backer betting on Anthropic. Around the same time, Amazon separately committed 5 billion dollars to the company, with an option to invest as much as 20 billion dollars more over time, valuing Anthropic at the same 350 billion dollars. This layering of investment from two of the world's largest technology firms speaks to the intensity of the competition.

A Long Standing Partnership

For Google, the move deepens a relationship that was already substantial. Before this latest deal, the search giant had invested roughly 3 billion dollars in Anthropic and pledged to provide up to one million of its custom tensor processing unit chips. The new commitment therefore represents a dramatic escalation rather than a first step into the partnership between the two companies.

The timing reflects surging demand for Anthropic's products, and in particular for Claude Code, the company's AI agent designed to help write and manage software. As businesses increasingly fold AI tools into their daily workflows, the appetite for reliable and capable models has grown quickly, and with it the need for enormous amounts of computing capacity.

A Race Measured in Gigawatts

Perhaps the most striking element of the arrangement is not the money but the electricity. As part of the deal, Google Cloud agreed to deliver five gigawatts of computing power to Anthropic over a five year window, with the possibility of adding several more gigawatts later. Such figures illustrate how the AI boom has become an infrastructure story as much as a software one.

This is not the only compute commitment Anthropic has secured. In a separate announcement, the company revealed a partnership with Google and the chipmaker Broadcom to build multiple gigawatts of next generation TPU capacity. That capacity is expected to come online in 2027, with the majority of it located within the United States, close to where much of the demand sits.

Building for Explosive Growth

Anthropic has framed these deals as a disciplined response to rapid expansion rather than reckless spending. Krishna Rao, the company's chief financial officer, described the partnership with Google and Broadcom as a continuation of a careful approach to scaling infrastructure, aimed at building the capacity needed to serve exponential growth in its customer base.

That growth is reflected in the company's financial trajectory. Anthropic's run rate revenue has surpassed 30 billion dollars, a sharp increase from roughly 9 billion dollars at the end of 2025. The number of business customers spending more than one million dollars a year with the company has climbed above 1,000, a total that doubled in less than two months.

Spreading the Bets Across Clouds

Despite the scale of the Google tie up, Anthropic has been careful not to depend on a single provider. Its Claude models are trained on a mix of hardware, including Amazon's Trainium chips, Google's TPUs and NVIDIA's graphics processors. This diversity helps the company avoid bottlenecks and reduces the risk of being locked into one supplier for its most critical resource.

The same philosophy extends to how customers reach Claude. The models are available across all three major cloud platforms, namely Amazon's Bedrock, Google Cloud's Vertex AI and Microsoft's Azure Foundry, while Amazon remains Anthropic's primary cloud provider and training partner. The result is a web of overlapping alliances that blur the line between partner and competitor.

What It Means for the Industry

Taken together, these announcements paint a picture of an industry where the cost of staying competitive has soared into the tens of billions. Earlier, in November 2025, Anthropic had already committed 50 billion dollars to American AI infrastructure, underlining a broader trend of huge, long term bets on data centres and chips right across the sector.

For observers, the message is clear. Building advanced artificial intelligence is no longer just about clever algorithms, but about securing the money, chips and raw power to run them at massive scale. As Google, Amazon and others pour capital into companies like Anthropic, the contours of the AI era are increasingly being drawn in dollars and gigawatts.