Amazon’s Nvidia Bet Swells as AI Demand Outruns Forecasts

Amazon’s Nvidia Bet Swells as AI Demand Outruns Forecasts
Amazon and Nvidia cast their enlarged alliance as a response to real customer demand and a route to production-ready AI. But the scale of the order also underlines how far AWS still depends on Nvidia even as Amazon builds chips meant to reduce that reliance.

The buildup began in March, when AWS said it would add more than one million Nvidia GPUs from 2026. That forecast did not last: customer appetite for accelerated computing moved faster than expected, prompting plans for a further two million Blackwell Ultra, Rubin and Rubin Ultra GPUs across AWS infrastructure in 2027 and 2028.

Nvidia CEO Jensen Huang framed the escalation as evidence that the AI boom has outrun the industry’s own projections. “NVIDIA and AWS have built one of the great growth engines of the AI era, and demand is running ahead of every forecast,” he said. The expanded deal reaches beyond graphics processors, adding Nvidia networking, Vera CPUs, open models, data-processing software and robotics tools to AWS’s stack.

For AWS, the pitch is choice rather than lock-in. Chief executive Matt Garman said customers want the “freedom to choose the best tools for their AI workloads” and confidence that those tools work together. AWS plans to offer Nvidia’s technology for frontier labs, companies and governments, including secure federal AI infrastructure, while Amazon Robotics adopts Nvidia platforms including Jetson, Omniverse and Isaac.

Yet the partnership arrives as Amazon presses ahead with its own Trainium AI chips and Graviton server CPUs. The company says its custom-chip business has reached a $25 billion annualized revenue run rate, but the fresh GPU commitment suggests Nvidia remains the immediate answer for customers chasing large-scale model training and inference.

The commercial wager is immense. Nvidia reported $96.2 billion in second-quarter sales, with $89 billion from data centers, and argues that more computing capacity can generate more profitable AI services. Whether that logic holds as infrastructure spending mounts will be the harder test.

Continue reading https://foxvector.com/stories/01a04e51-fc50-2ebe-72f0-3433b84952cf