Why Anthropic has Signed Huge Deal with Google Cloud

Anthropic is expanding its AI infrastructure through a multi-billion dollar agreement with Google Cloud.

The company plans to deploy up to one million of Google's tensor processing units (TPUs) to meet the escalating demand for its Claude AI models and to further its research and development efforts.

This move highlights the intense competition for computing resources within the artificial intelligence industry.

The expansion is a response to rapid growth.

According to Anthropic, it now serves over 300,000 business customers and its largest accounts, which are those paying more than US$100,000 annually, have grown nearly sevenfold in the last year alone.

The additional infrastructure set to bring more than 1GW of new capacity online in 2026 is designed to keep pace with this trajectory.

The role of TPUs in AI infrastructure

Google's TPUs are specialised processors designed for AI workloads. Unlike general-purpose CPUs or even GPUs these chips are optimised for the matrix multiplications that are fundamental to neural networks.

This specialisation can make them more efficient and faster for training and operating large language models.

Google is currently developing its seventh-generation TPU, codenamed Ironwood, which Anthropic will use as part of this extended partnership.

Thomas Kurian, CEO of Google Cloud, says: “Anthropic’s choice to significantly expand its usage of TPUs reflects the strong price-performance and efficiency its teams have seen with TPUs for several years.

"We are continuing to innovate and drive further efficiencies and increased capacity of our TPUs building on our already mature AI accelerator portfolio, including our seventh generation TPU Ironwood.”

The decision follows a multi-year period where Anthropic evaluated the performance of TPUs for its specific AI tasks.

A diversified compute strategy

Despite this major commitment to Google Cloud, Anthropic maintains a diversified approach to its computing infrastructure. Anthropic strategically runs its workloads across three distinct chip platforms:

• Google’s TPUs
• Amazon’s Trainium processors
• Nvidia’s graphics processing units (GPUs)

This multi-provider strategy is a pragmatic measure to mitigate risks associated with supply chain constraints and to maintain negotiating leverage in a market where computing power is a critical and often scarce resource.

Beyond serving customer demand the expanded capacity will support Anthropic's research into AI alignment and safety that has been a core element of its mission since its founding by former OpenAI executives.

Krishna Rao, Anthropic’s Chief Financial Officer (CFO), explains: “Anthropic and Google have a longstanding partnership and this latest expansion will help us continue to grow the compute we need to define the frontier of AI. Our customers – from Fortune 500 companies to AI-native startups – depend on Claude for their most important work and this expanded capacity ensures we can meet our exponentially growing demand while keeping our models at the cutting edge of the industry.”

Balancing AWS and Google Cloud partnerships

While deepening its ties with Google, which has held an equity stake in Anthropic since 2023, Anthropic has clarified that Amazon Web Services (AWS) remains its primary partner for training its AI models.

Work is ongoing with AWS on Project Rainier, a major compute cluster involving hundreds of thousands of AI chips located in multiple US data centres.

This dual-front strategy allows Anthropic to leverage the strengths of two of the largest cloud providers simultaneously.

The scale of the planned 1GW capacity increase is substantial, equivalent to the power required for a small city.

This energy consumption is dedicated entirely to training and running AI models, drawing attention to the substantial power and cooling requirements of data centres operating at this scale.

As the industry grows the environmental and logistical challenges of powering AI are becoming increasingly prominent issues for leading companies.

“Anthropic will continue to invest in additional compute capacity to ensure our models and capabilities remain at the frontier,” says Krishna.