An exclusive tour of Amazon’s Trainium lab, the chip that’s won over Anthropic, OpenAI, even Apple

2026-03-22 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Emerging Technologies & Innovation · Depth: Intermediate, extended

Summary

Amazon Web Services (AWS) is significantly advancing its custom AI chip development, underscored by a \$50 billion investment deal with OpenAI that includes supplying 2 gigawatts of Trainium computing capacity. The latest liquid-cooled 3-nanometer Trainium3 chips are optimized for both AI model training and inference, aiming to cut operational costs by up to 50% compared to traditional cloud servers. Developed by the Annapurna Labs team, these chips are extensively used by Anthropic's Claude and Amazon's Bedrock service, with Trainium now supporting PyTorch to simplify developer transitions from Nvidia's ecosystem. AWS further enhances performance and cost efficiency through custom server hardware, including Nitro virtualization and Neuron switches. This strategic investment positions AWS as a strong competitor to Nvidia, with Trainium already a multi-billion dollar business and Bedrock poised to rival EC2 in scale.

Key takeaway

AWS's Trainium3 chips, enhanced by liquid cooling and Neuron switches, deliver up to 50% lower operational costs for AI training and inference at comparable performance to classic cloud servers. Supporting PyTorch with a "one-line change" for migration, Trainium powers Anthropic's Claude and Amazon Bedrock, directly challenging Nvidia's market dominance. This offers AI/ML professionals a scalable, cost-efficient compute alternative crucial for large-scale model deployment and development.

Topics

• AWS Custom AI Chips
• Trainium
• AI Inference Optimization
• Cloud AI Infrastructure
• NVIDIA Market Competition

Best for: CTO, Investor, VP of Engineering/Data, AI Engineer, MLOps Engineer, Tech Journalist

Related on AIssential

• Anthropic and Amazon expand collaboration for up to 5 gigawatts of new compute — Anthropic secures massive AWS compute and investment to scale Claude's training and inference capabilities.
• In another wild turn for AI chips, Meta signs deal for millions of Amazon AI CPUs — AWS secured a significant Meta deal for Graviton CPUs, intensifying cloud and custom AI chip competition.
• Anthropic takes $5B from Amazon and pledges $100B in cloud spending in return — Major AI developers are securing massive compute capacity through cloud infrastructure deals with tech giants.
• AWS and NVIDIA deepen strategic collaboration to accelerate AI from pilot to production — AWS and NVIDIA are deepening their partnership to provide integrated, scalable, and secure AI infrastructure and services.
• Why Everyone Is Moving Away from NVIDIA — Hyperscalers are vertically integrating AI infrastructure, developing custom silicon and energy solutions to overcome GPU bottlenecks.
• Google Cloud launches two new AI chips to compete with Nvidia — Google Cloud's new specialized TPUs enhance AI training and inference while complementing, not replacing, Nvidia's hardware.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.