Amazon’s US$25bn Anthropic Bet Fuels AI Infrastructure Race

As large language model developers scramble to secure compute capacity, Amazon’s massive investment highlights the critical infrastructure battle defining the AI development frontier.

Amazon has announced plans to invest up to US$25bn in Anthropic on top of the US$8bn the company has spent on the AI start up in recent years, as part of an expanded agreement that could reshape the competitive landscape for AI infrastructure and custom chip development.

In a company announcement, Anthropic states it plans to spend US$100bn on AWS over the next decade, securing up to 5GW of new capacity to train and run its large language model-powered AI assistant Claude and develop current and future generations of Trainium, Amazon’s custom AI chips.

The deal represents more than just financial backing. It positions AWS’s custom silicon as a credible alternative to dominant GPU providers in the race to build the infrastructure needed for frontier AI development.

“Anthropic’s commitment to run its large language models on AWS Trainium for the next decade reflects the progress we’ve made together on custom silicon, as we continue delivering the technology and infrastructure our customers need to build with generative AI,” Amazon CEO Andy Jassy says in the statement.

Large language model economics drive infrastructure demands

Amazon’s investment comes just two months after the company agreed to invest up to US$50bn in OpenAI, Anthropic’s main competitor. OpenAI has stated previously that it will in turn spend more than US$100bn on Amazon’s services.

Both Anthropic and OpenAI have been racing to convince investors of their strong market position ahead of potential IPOs that could land as soon as 2026.

AI spending across all industries has hit a record high between 2023 and 2025, with worldwide AI spending estimated to reach US$2.5tn in 2026 according to Gartner, despite industry concerns over the long-term sustainability of the technology by large companies like Amazon.

In the company announcement, Anthropic says that enterprise and developer demand for Claude, as well as a “sharp rise in consumer usage,” has led to “inevitable strain” on its infrastructure, affecting the software’s reliability and performance.

“Our users tell us Claude is increasingly essential to how they work, and we need to build the infrastructure to keep pace with rapidly growing demand,” Anthropic CEO Dario Amodei adds.

“Our collaboration with Amazon will allow us to continue advancing AI research while delivering Claude to our customers, including the more than 100,000 building on AWS,” Dario says.

Custom AI chips challenge established players

The agreement’s focus on Trainium development signals Amazon’s ambition to reduce dependence on established AI chip manufacturers. By committing to develop future generations of custom AI chips alongside Anthropic’s large language models, AWS could potentially offer more cost-effective infrastructure for AI training and inference workloads.

This strategic partnership represents a significant challenge to established GPU manufacturers who have dominated the AI chip market. The collaboration allows both companies to optimise hardware and software in tandem, potentially delivering performance advantages that generic solutions cannot match.

Anthropic states that this agreement will help expand its available capacity and will deliver “meaningful compute in the next three months and nearly 1GW in total before the end of the year”. The company’s capacity expansions and diversified hardware are part of an effort to build the infrastructure needed to improve the Claude platform and reliably serve its self-reported base of 300,000 customers.

The custom chip development cycle typically spans multiple years, making long-term commitments like this essential for ensuring that hardware capabilities align with the evolving demands of frontier AI models.

Multi-cloud strategy reflects competitive dynamics

Anthropic is known for its family of Claude AI models and its early success in selling its software to larger enterprises like Amazon. However, the AI start up has maintained relationships across the competitive landscape of cloud infrastructure providers.

Anthropic named AWS as its primary cloud provider in 2023 and its primary training partner in 2024, though the company has also made agreements with competing providers, such as Microsoft and Google.

In November 2025, Microsoft agreed to invest up to US$5bn into Anthropic with the AI start up saying it was committed to purchasing US$30bn of Azure compute capacity. In March 2026, Anthropic also expanded its partnerships with Google and Broadcom for “multiple gigawatts“ of capacity, according to a company press release.

Krishna Rao, CFO of Anthropic, discussed the deal in the same announcement, saying: “This ground breaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure. 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. We are making our most significant compute commitment to date to keep pace with our unprecedented growth.”

Krishna’s comments underscore the massive financial commitments required to compete at the frontier of AI development. The aggressive infrastructure buildout across multiple providers reflects the technical and financial realities of operating at the AI development frontier, where access to massive compute capacity has become as critical as algorithmic innovation for large language model advancement.