Anthropic, the San Francisco-based AI lab, is making a historic investment in Google’s chips and cloud services as it seeks to secure the computing power required to meet skyrocketing demand for its AI tools, according to reporting by the Financial Times. The company announced on Monday that it has committed to deploying multiple gigawatts of capacity from Google’s Tensor Processing Units (TPUs), which rival Nvidia’s dominant GPUs, as well as from the search giant’s cloud infrastructure.
The investment will provide Anthropic access to nearly five gigawatts of new computing capacity over the coming years, with roughly 3.5 gigawatts coming through a partnership with Broadcom starting next year. Industry sources say that the hardware and infrastructure required to develop a single gigawatt of capacity—roughly equivalent to the output of a nuclear reactor—costs between $35 billion and $50 billion, suggesting that Anthropic’s total outlay could run into hundreds of billions of dollars.
The unprecedented spending reflects the company’s urgency to scale operations in response to explosive growth. Annualized revenue for Anthropic has surged from $9 billion at the end of 2025 to $30 billion at the end of March 2026, based on revenues over the past 28 days projected over a full year. The growth is largely driven by its popular coding agent, Claude Code, and other AI tools that are attracting a rapidly expanding customer base.
“We are building the capacity necessary to serve the exponential growth we have seen in our customer base while also enabling Claude to define the frontier of AI development,” said Krishna Rao, Anthropic’s chief financial officer. The announcement underscores the company’s strategy to dominate the next generation of AI platforms by locking in the computational resources needed for model training and deployment at scale.
Broadcom, which will supply custom TPUs for Google as part of a long-term agreement through 2031, saw its shares rise nearly three percent following the market close on Monday. Google’s move to expand sales of its in-house chips comes as it seeks to power its own Gemini AI models and compete more directly with Nvidia, the world’s largest semiconductor company.
The arrangement highlights a trend among AI labs to secure vast amounts of computing power through multiple partnerships. Last year, Anthropic rival OpenAI struck deals with Broadcom, Nvidia, AMD, and other suppliers to lock in capacity for its AI tools. Critics have noted the circularity in such arrangements, with Big Tech companies simultaneously acting as investors, suppliers, and customers. Google, for example, holds a 14 percent stake in Anthropic following multi-billion-dollar investments, according to legal filings.
The financial scale of these deals has drawn scrutiny. Both Anthropic and its competitors have repeatedly turned to venture capital and sovereign wealth funds to raise tens of billions of dollars, relying on promises that market dominance and sufficient scale will eventually lead to profitability. In February 2026, Anthropic raised $30 billion in a funding round that valued the company at $380 billion, reflecting investor confidence in the start-up’s aggressive expansion strategy.
The new agreement with Google and Broadcom expands on a partnership announced last October, which was initially described as being worth tens of billions and expected to deliver over one gigawatt of capacity online in 2026. The deal also complements other major commitments, including $50 billion in new data centers with cloud computing group Fluidstack in Texas and New York, and a further $30 billion in capacity purchases from Microsoft and Nvidia.
“Anthropic’s deal with Broadcom is dependent on its continued commercial success,” Broadcom said in a filing, adding that discussions with operational and financial partners are ongoing. The statement underscores the risk inherent in the company’s rapid scaling efforts, as its capacity investments are contingent on maintaining revenue growth and market demand for AI services.
Analysts say that Anthropic’s strategy reflects a broader arms race for AI compute, as labs seek to secure the resources needed to train ever-larger models. By controlling significant cloud and TPU capacity, the company aims to gain a competitive edge over rivals and establish Claude as a leading platform in AI development. The scale of the investment also signals the growing role of hardware partnerships in shaping the AI industry, as companies compete for both talent and technological infrastructure.
As demand for AI tools continues to grow, the stakes for Anthropic and its investors are enormous. The company’s ambitious spending on chips, cloud services, and data centers demonstrates a willingness to leverage capital markets and strategic partnerships to achieve technological dominance. At the same time, the arrangement raises questions about concentration of power in the AI ecosystem, the interdependence of Big Tech and start-ups, and the sustainability of such unprecedented capital outlays.
Anthropic’s unprecedented scale-up highlights a pivotal moment in the AI industry. By securing multi-gigawatt capacity from Google and Broadcom, the company positions itself at the forefront of AI development, with a revenue trajectory that has already tripled in just a few months. The investments reflect both the promise and the risk of a sector in which computational power has become the critical bottleneck for innovation and market leadership.
With AI adoption expanding across enterprise, coding, and research applications, Anthropic’s strategy may set a blueprint for future competitors seeking to combine software innovation with access to high-end hardware infrastructure. The success or failure of this gamble could shape the industry’s competitive landscape for years to come, underscoring the intersection of technology, finance, and geopolitics in the emerging AI economy.
