Amazon unveils Trainium3 custom AI training chip, expanding proprietary accelerator portfolio for AWS infrastructure.

Amazon.com Inc (NASDAQ:AMZN) on Tuesday unveiled its latest artificial intelligence semiconductor, Trainium3, at AWS's annual re:Invent conference, signaling a major effort by the e-commerce and cloud giant to challenge Nvidia's dominance in the booming AI hardware market. AWS executives detailed the chip's capabilities and the company's broader AI strategy during the event. The announcement reflects Amazon's continued investment in custom silicon to reduce dependence on Nvidia and improve performance control.

Servers equipped with Trainium3 deliver up to four times the speed and energy efficiency of the previous generation, allowing enterprises to train large language models and multimodal systems more quickly while cutting operational costs. AWS vice president Dave Brown emphasized the strategic importance of diversification, stating that "Trainium-based instances typically provide 30% to 40% cost savings compared with Nvidia GPU clusters—a meaningful advantage as AI model training expenses can reach hundreds of millions of dollars." This positions Trainium3 as a compelling alternative for cost-conscious organizations scaling their AI infrastructure.

AWS CEO Matt Garman highlighted the scale of Amazon's commitment, noting that "Trainium already represents a multibillion-dollar business today and continues to grow really rapidly." Beyond the chip itself, the company outlined Project Rainier, a large-scale data center initiative designed specifically for AI workloads. OpenAI competitor Anthropic is expected to use roughly one million Trainium chips across Rainier and other AWS data centers by the end of 2025, and reportedly contributed to the chip's design. With nearly 500,000 Trainium2 chips already active and plans to double capacity by 2027, AWS is positioning AI infrastructure as a major growth driver.

Wedbush analysts viewed the announcements favorably, stating, "We come away from re:Invent with increased conviction in the trajectory of growth and the health of Amazon's competitive positioning relative to peers," citing momentum from backlog growth and a higher capital expenditure guide for 2025. This assessment reflects broader confidence in Amazon's ability to compete effectively in the AI hardware and infrastructure markets.