- AWS encourages customers to switch from nvidia to its cheaper trainer chip
- It says its hardware offers the same performance with a cost saving of 25 percent
- Amazon’s pitch happened when NVIDIA showcased her new hardware on GTC 2025
While NVIDIA hosted its annual GTC 2025 conference, which showed new products such as DGX Spark and DGX Station AI Supercomputer, Amazon tried to convince his clouds that they could save money by moving away from Pexy Nvidia -Darware and embracing Amazon’s own AI chips.
The information The AWS claims beat at least one of its cloud customers to consider renting servers driven by Amazon’s Trainium Chip and claimed they could enjoy the same service as NVIDIA’s H100, but to 25 percent of costs.
Tractium is one of several internal chips developed by Amazon (together with graviton and inferentia), built for training machine learning models in the AWS cloud and offers a cheaper alternative to GPU-based systems. Amazon’s silicon is not intended as a similar-to-like replacement for Nvidia’s more advanced products, but it doesn’t have to be.
Part of the AI conversation
Amazon’s offerings appear to be part of a wider shift across the ski market, where providers such as AWS and Google develop their own chips and offer them to customers as a way to avoid costs – and scarcity – of Nvidia’s much sought after GPUs.
“What AWS is doing is smart,” Matt Kimball, VP and Principal Analyst for Data Compute and Storage at Moor Insights & Strategy, told NetworkWorld. “It tells the world that there is a cost-effective alternative that is also performing for AI training needs. It inserts itself into the AI conversation.”
The course here is of course access. AWS allows customers to experiment with training and inference workloads without having to wait months for an NVIDIA GPU or pay top dollar for it.
While a 25 percent saving is definitely not to be sniffed on, and something that will undoubtedly appeal to a number of AWS customers, there are obvious disadvantages for buyers to consider.
Seam NetworkWorld Remarks, “Companies used to work with Nvidia’s Compute Unified Device Architecture (CUDA) have to think about the cost of switching to a brand new platform like Trainium. In addition, Trainium is only available on AWS so users can be locked inside.”
