Amazon adds $25 billion investment in Anthropic, AI infrastructure "arms race" fully escalates

Author: Claude, Deep Tide TechFlow

Deep Tide Guide: Amazon announced on Monday that it will invest up to $25 billion in Anthropic (with $5 billion immediately available), and has committed to over $100 billion in AWS spending over the next ten years.

This is the second time in two months that Amazon has written a billion-dollar check to a leading AI lab—just after it invested $50 billion in OpenAI.

Anthropic’s annual revenue has surpassed $30 billion, but computing power bottlenecks are dragging down user experience. The core goal of this deal is to address the capacity crisis.

Amazon is betting heavily on two major AI labs simultaneously, with increasing stakes.

According to reports from CNBC, Bloomberg, and others on April 20, Amazon announced an additional investment of up to $25 billion in Anthropic, with $5 billion immediately transferred and the remaining $20 billion tied to specific business milestones. This investment is based on Anthropic’s February valuation of $380 billion during its Series G funding round, and combined with the previous $8 billion invested, Amazon’s total investment commitment to Anthropic has reached a cap of $33 billion.

Two months ago, Amazon also invested $50 billion in Anthropic’s main competitor, OpenAI, and reached a similar-scale cloud service agreement. Amazon CEO Andy Jassy stated in a release that Anthropic commits to running large language models on AWS Trainium for ten years, “reflecting our progress in custom chip development.”

Following the announcement, Amazon’s stock price rose about 2.5% after hours.

A hundred-billion-dollar cloud commitment for 5 gigawatts of compute power in response to OpenAI’s “insufficient compute” accusations

The core of this deal is not just equity investment but a deeply integrated infrastructure agreement.

Anthropic has committed to investing over $100 billion on AWS technology over the next decade, including Amazon’s custom AI chips Trainium (from Trainium2 to Trainium4 and future generations) and tens of millions of Graviton CPU cores. In return, Anthropic will gain up to 5 gigawatts of compute capacity for training and deploying the Claude models. According to Anthropic’s blog, the company is currently using over 1 million Trainium2 chips to train and serve Claude, with plans to bring nearly 1 gigawatt of Trainium2 and Trainium3 capacity online by the end of 2026.

This expansion of compute capacity directly responds to recent public criticisms from OpenAI. OpenAI’s Chief Revenue Officer Denise Dresser stated in an internal memo last week that Anthropic made a “strategic mistake by not acquiring enough compute,” and predicted that by 2030, OpenAI will have 30 gigawatts of compute, while Anthropic will only have 7 to 8 gigawatts by the end of 2027. In the announcement, Anthropic admitted that demand for Claude from enterprises and developers is accelerating, with “sharp increases” in consumer usage, causing “inevitable pressure” on infrastructure, affecting reliability and performance during peak times.

Anthropic CEO Dario Amodei said in a statement: “Users tell us that Claude is becoming increasingly important to their work, and we need to build infrastructure to keep up with this rapid growth.”

Amazon writes billion-dollar checks to two AI labs in two months

Amazon’s investment strategy is now very clear: simultaneously betting on two leading players in the AI race.

In February, Amazon announced a $50 billion investment in OpenAI, along with a $100 billion commitment to AWS cloud services. The structure of the deal with Anthropic is almost identical—$25 billion investment plus over $100 billion in cloud spending locked in. According to GeekWire, Amazon is executing “the same script” for both labs.

Both AI companies are also competing to demonstrate their strength to investors. CNBC reports that both Anthropic and OpenAI are preparing for potential IPOs, possibly as early as this year. OpenAI’s latest funding round valued it at over $850 billion, while Anthropic’s valuation is $380 billion. Anthropic claims its annual revenue has surpassed $30 billion (about $9 billion by the end of 2025), but OpenAI’s memo alleges that this figure has been inflated by about $8 billion because Anthropic counts cloud revenue from Amazon and Google as gross rather than net.

Microsoft is also betting on both sides—having previously invested over $13B in OpenAI, and in November 2025, investing up to $100k in Anthropic, which has committed to purchasing $30 billion worth of Azure compute.

Claude platform integrated into AWS, a battle for over 100k customers

Beyond investments, product integration is deepening.

According to the announcement, Anthropic’s native Claude platform will be directly embedded into AWS, allowing users to access the full Claude console through their existing AWS accounts, with permissions and billing systems intact, without additional registration or contracts. This is a step further than the previous availability of Claude via Amazon Bedrock marketplace. Amazon disclosed that over 100k organizations are currently running Claude models on Amazon Bedrock.

Anthropic also emphasized in its blog that Claude is the only cutting-edge AI model available on all three major cloud platforms (AWS Bedrock, Google Cloud Vertex AI, Microsoft Azure Foundry). This multi-platform strategy offers enterprise customers flexible deployment options and is one of Anthropic’s differentiators in competing with OpenAI.

On the client side, Lyft has reduced its average customer service resolution time by 87% after deploying Claude via Amazon Bedrock. Pfizer uses Claude to assist scientists with voice searches in drug development documents, saving approximately 16,000 hours annually.

AI infrastructure race: Amazon’s capital expenditure expected to reach $200 billion this year

The broader context of this deal is the AI infrastructure arms race among cloud giants.

Amazon announced in February that its capital expenditure is projected to reach about $200 billion by 2026, with the majority allocated to AI infrastructure. Previously, the Project Rainier collaboration (a massive compute cluster with nearly 500k Trainium2 chips) was one of the largest AI compute clusters globally, used by Anthropic to train and deploy current and future versions of Claude.

Earlier this month, Anthropic expanded its collaborations with Google and Broadcom, securing “multi-gigawatt” level compute capacity expected to go online starting in 2027. Along with the recent $5 billion agreement with Amazon, Anthropic is expanding its compute reserves across multiple lines.

Amazon’s custom chip business is also accelerating. Jassy recently revealed that this business’s annual revenue has exceeded $20 billion, doubling from the $10 billion reported earlier this year, describing it as “exceptionally hot.”