Amazon CEO Andy Jassy disclosed during the company's Q1 2026 earnings call that commitments for its custom Trainium chips have reached $225 billion. The figure, which covers multi-year contracts and reservations, signals strong appetite for Amazon's in-house AI silicon.
Who's buying Trainium chips
Jassy named three major customers driving the demand: AI startups Anthropic and OpenAI, and ride-hailing giant Uber. All three are using Trainium to train and run large-scale machine learning models. The CEO did not break down how much each company has committed, but the total dwarfs earlier projections for Amazon's chip business.
Why demand is outstripping supply
Jassy said Trainium demand is currently outpacing supply. He did not specify the size of the gap or when Amazon expects to catch up. The company has been ramping production at its chip design centers and with manufacturing partners, but the pace of orders has accelerated faster than anticipated.
The customers behind the numbers
Anthropic, which has a close partnership with Amazon and uses Trainium for its Claude models, is one of the largest known customers. OpenAI, traditionally reliant on Nvidia GPUs, has diversified its hardware stack with Trainium. Uber is using the chips for real-time inference in its logistics and autonomous vehicle systems. The three companies represent a cross-section of AI workloads — from frontier model training to large-scale deployment.
Supply constraints and next steps
Amazon has not detailed how it plans to close the supply-demand gap. The company is investing in new fabrication capacity and has hinted at a next-generation Trainium design, but no timeline has been given. The $225 billion in commitments suggests customers are betting on Amazon's ability to deliver, even as Nvidia continues to dominate the AI chip market. Investors will be watching for updates on production milestones in the coming quarters.
