All articles tagged with #trainium
Exclusive planning documents show Amazon is redesigning a rural Indiana AI data‑center campus into a large, interconnected AGI SuperCluster, consolidating facilities, upgrading networking and storage, and deploying thousands of Trainium-powered servers to accelerate its next frontier AI model and improve efficiency, all while expanding capacity to meet AWS compute demands amid industry constraints.
Amazon’s AWS silicon unit, Trainium, is quietly scaling toward a roughly $50 billion annual revenue run rate with over $225 billion in commitments, while AWS revenue continues to grow and earnings margins stay robust. The market largely values AMZN as a retailer, but this in‑house chip business could unlock meaningful upside and help offset capital spend if AI demand stays strong, according to the analysis.
Amazon and Anthropic deepen their AI partnership, committing up to 5 GW of Trainium capacity to train and run Claude on AWS, expanding the Claude Platform on AWS and enabling Claude access within customers’ AWS accounts. Amazon will invest $5 billion now and up to $20 billion later, with Anthropic slated to spend over $100 billion on AWS technologies over the next decade. The deal sustains Project Rainier, expands international inference to Asia and Europe, and leverages Trainium and Graviton across Bedrock, powering real‑world deployments for Lyft, Pfizer and others while keeping AWS as Anthropic’s primary training and cloud provider.
Amazon is moving its AI workloads onto its own Trainium and Inferentia chips to cut AI operating costs and gain more control over hardware, potentially changing the economics of AI infrastructure without an immediate replacement of Nvidia; the effort aligns with a broader 2026 capex push for data centers and custom chips, and Wall Street consensus remains strongly bullish on AMZN with upside potential.
AWS and OpenAI announce a multi-year, roughly $100 billion expansion to accelerate enterprise AI, delivering a Stateful Runtime Environment on Amazon Bedrock, making AWS the exclusive cloud distributor of OpenAI Frontier, committing ~2 gigawatts of Trainium compute, and jointly developing customized models for Amazon apps.
Microsoft unveiled Maia 200, an AI inference accelerator for Azure, claiming it delivers over 10 petaflops at FP4 and 5 PFLOPS at FP8, with 3x FP4 performance versus Amazon’s Trainium Gen3 and FP8 performance above Google’s TPU Gen7. Built on TSMC’s 3-nanometer process with about 100 billion transistors, Maia 200 is designed for data-center inference to speed Copilot and Azure OpenAI workloads, featuring a memory system to keep model weights local and is currently deployed in a US data center with broader Azure availability planned in the future.
Amazon is developing an AI supercomputer using its in-house Trainium chips and a new server for AI startup Anthropic, following a $4 billion investment in the company. This move by Amazon Web Services aims to position Trainium as a competitive alternative to Nvidia's GPUs. The "Ultracluster," designed by Amazon's chip lab, will feature hundreds of thousands of Trainium2 chips, significantly enhancing AI model training capabilities.
Amazon has announced a $4 billion investment in AI startup Anthropic, aiming to increase the use of its Trainium AI chips and challenge Nvidia's dominance in the AI chip market. This brings Amazon's total investment in Anthropic to $8 billion, with Anthropic agreeing to use AWS as its primary cloud and training partner. Despite the investment, Anthropic CEO Dario Amodei expressed a balanced approach, using chips from Nvidia, Google, and Amazon. The collaboration includes developing Amazon's AI-model platform, AWS Neuron, to compete with Nvidia's CUDA software stack.