Amid skyrocketing demand, Anthropic has extended its compute contracts with Google and Broadcom

Introduction

“On Monday, Anthropic disclosed a fresh partnership with Google and Broadcom aimed at expanding the computational power available for its Claude AI suite. This strategic infrastructure boost is a response to the steadily increasing popularity of their models.” Amid skyrocketing demand, Anthropic has extended its compute contracts with Google and Broadcom

This agreement broadens Anthropic’s utilization of Google Cloud’s Tensor Processing Units (TPUs)—Google’s proprietary AI chips—building upon the partnership established in October 2025, which originally provided over a gigawatt of compute capacity. Amid skyrocketing demand, Anthropic has extended its compute contracts with Google and Broadcom

“As outlined in Anthropic’s blog post, this expanded computing power is expected to be fully functional by 2027.  While the company did not disclose specific details regarding the scope of the expansion, a recent SEC filing from Broadcom indicates that the agreement includes 3.5 gigawatts of compute power. Amid skyrocketing demand, Anthropic has extended its compute contracts with Google and Broadcom

“Anthropic mentioned that the bulk of this computing power will be hosted in the U.S., functioning as part of their broader $50 billion investment commitment. in domestic compute infrastructure. Amid skyrocketing demand, Anthropic has extended its compute contracts with Google and Broadcom

“Our collaboration with Google and Broadcom serves as a testament to our calculated strategy for scaling infrastructure,” stated Anthropic CFO Krishna Rao in the official release. “We are building the capacity needed to serve our exponentially growing customer base while ensuring Claude remains at the frontier of AI development.

This is our most significant compute commitment yet, essential for keeping pace with our unprecedented growth.” Amid skyrocketing demand, Anthropic has extended its compute contracts with Google and Broadcom

Anthropic did not provide a comment in response to TechCrunch’s inquiry. The company has experienced a surge in demand for its Claude models in recent months, driven largely by enterprise adoption, even as the U.S. Defense Department has flagged it as a supply-chain risk.

Additionally, Anthropic recently finalized a $30 billion Series G funding round, bringing its valuation to $380 billion.

“Anthropic reported a major jump in run rate revenue, reaching $30 billion—a significant rise from the $9 billion recorded at the close of 2025. Additionally, the firm has now surpassed 1,000 enterprise clients, with each contributing over $1 million in annual spending.” .

What are Google’s TPUS?

 Tensor Processing Units, commonly known as TPUs, are specialized AI processors developed by Google. Unlike general-purpose processors, TPUs are designed specifically for machine-learning workloads.

They can be used for both training AI models and running AI applications at scale. For a company such as Anthropic, access to large amounts of TPU capacity can provide another option for handling demanding AI workloads.

Anthropic does not rely on a single type of hardware. The company says it uses AWS Trainium, Google TPUs, and NVIDIA GPUs. This multi-platform strategy can help the company match different workloads with suitable hardware while improving infrastructure resilience.

Claude is driving infrastructure demand

Anthropic’s Claude family has become an important part of the company’s infrastructure strategy. Businesses are increasingly using Claude for software development, writing, research, analysis, and AI-powered agents.

As usage increases, Anthropic needs additional computing resources not only to train future models but also to serve millions of requests from customers.

This creates a difficult challenge for AI companies: building advanced models is only one part of the equation. They also need enough data-center capacity, networking equipment, specialized processors, electricity, and cloud infrastructure to make those models available at scale.

The Google and Broadcom agreement is therefore more than a simple chip-supply deal. It is part of Anthropic’s broader effort to build the infrastructure required for long-term AI growth.

Why Broadcom is important

 Broadcom plays a key role in the custom AI-chip ecosystem. The company works with Google on the development and supply of Google’s TPU technology.

Broadcom’s expanded relationship with Anthropic gives the AI company access to additional Google TPU-based computing capacity while also strengthening Broadcom’s position in the rapidly expanding AI infrastructure market.

The agreement also comes as Broadcom continues developing future generations of Google’s TPUs and related networking components. A separate long-term agreement between Broadcom and Google extends this work through 2031.

 What this means for the AI industry

Anthropic’s expanded compute agreement highlights a major trend in the AI industry: access to computing power is becoming just as important as model development.

AI companies are competing not only on model quality but also on their ability to secure chips, data centers, electricity, and networking infrastructure.

The scale of Anthropic’s latest commitment shows how quickly infrastructure requirements are growing. Recent industry developments also show that financing massive AI infrastructure projects has become a major challenge, with companies and investors exploring new ways to fund data centers and specialized AI hardware.

Anthropic is taking a multi chip approach

One interesting aspect of Anthropic’s strategy is that the company is not depending entirely on Google TPUs.

Anthropic continues to work with AWS and uses NVIDIA GPUs alongside Google’s TPUs and AWS Trainium processors. This gives the company flexibility when deciding which hardware should handle specific AI workloads.

That approach could become increasingly important as demand for AI computing continues to rise. Instead of depending on one supplier, Anthropic can use multiple hardware platforms to improve availability, performance, and resilience.

The bigger picture

Anthropic’s expanded partnership with Google and Broadcom is a clear example of how quickly the AI infrastructure market is evolving.

The company needs enormous amounts of computing power to keep Claude competitive while serving a rapidly growing customer base. Google’s TPUs provide Anthropic with another important source of AI processing capacity, while Broadcom helps make the underlying custom-chip infrastructure possible.

The additional capacity expected from 2027 could give Anthropic more room to train larger models, operate AI agents, and support enterprise customers at scale.

At the same time, the deal demonstrates a broader reality: the next phase of artificial intelligence will depend not only on smarter algorithms but also on massive investments in hardware, data centers, energy, and networking.

Final thoughts

Anthropic’s expanded compute agreement with Google and Broadcom reflects the extraordinary growth of the AI industry in 2026. With Claude demand increasing rapidly, securing future computing capacity has become a strategic priority for the company.

The planned multi-gigawatt TPU expansion, expected to begin coming online in 2027, could provide Anthropic with the infrastructure needed to develop future Claude models and serve a growing global customer base.

As AI competition becomes more intense, companies that can secure reliable and scalable computing infrastructure may have a significant advantage. Anthropic’s latest deal shows that the race for the future of AI is increasingly becoming a race for compute.