On Tuesday (December 3), Amazon Web Services (AWS) officially announced the launch of its custom-built artificial intelligence chip, Trainium3. According to Amazon, this new chip boasts a computing speed four times faster than its previous-generation AI chips. Developed by AWS's in-house chip design division, Annapurna Labs, Trainium3 can slash the costs associated with training and operating AI models by up to 50% when compared to systems relying on equivalent GPUs. These chips are specifically engineered to empower software developers, like Dean Leitersdorf—co-founder and CEO of the AI video startup Decart—with enhanced computational capabilities. Leitersdorf remarked that after evaluating chips from various competitors, including Nvidia processors, his company witnessed a significant breakthrough with Trainium3. The introduction of Trainium3 marks Amazon's latest move to challenge Nvidia's dominance in the GPU market. Presently, a growing number of AI firms are looking to procure chips and other hardware from alternative suppliers besides Nvidia, aiming to diversify their sourcing.