.png){kind=logo}
Amazon is making it easier for AI developers to use its own AI chips instead of relying on NVIDIA. The company is set to introduce a software tool that allows developers to run existing AI code on Amazon's Trainium processors with minimal changes, potentially lowering one of the biggest barriers to switching chip providers.
The new tool, called TorchNeuron, supports PyTorch, an open-source framework widely used to build and train AI models. According to Amazon's Annapurna Labs chip unit, developers can run PyTorch code on Trainium chips with a single configuration change, without rewriting their applications.
Amazon plans to release the software as a free, open-source offering. The announcement marks another attempt by a major cloud provider to challenge NVIDIA's position in the AI computing market.
NVIDIA's advantage comes from more than its powerful graphics processing units (GPUs). Its CUDA software platform has spent years building support across AI development tools, making it easier for developers to run their workloads on NVIDIA hardware.
Switching to another chip platform can require code changes, testing, and additional engineering work. This can discourage companies from switching, even when competing hardware offers potential cost savings.
TorchNeuron aims to reduce that workload by allowing developers to use familiar PyTorch code on Amazon's Trainium chips. The company hopes the simpler process will encourage more AI teams to try its processors.
Amazon says more than 30 early customers are already using TorchNeuron. These users work across areas including robotics, voice-based AI, and world models that help AI systems understand and simulate aspects of the real world.
Colin Brace, vice president of software at Amazon's Annapurna Labs, told Reuters that the initiative is designed to reduce the steps developers must take to experiment with Amazon's chips. He also said moving workloads to Trainium could deliver substantial cost savings, although the actual benefits will depend on the application.
Also Read: Amazon’s USD 7 Billion Decart Bid Could Change How AI Models Use Chips
Amazon is not alone in trying to loosen NVIDIA's grip on AI computing. Google is also developing tools to help developers run PyTorch workloads on its custom AI processors. For Amazon, making Trainium easier to use could attract customers seeking alternatives to NVIDIA hardware. However, wider adoption will depend on performance, software support, and the overall cost of running AI workloads.
The latest move highlights a growing battle in the AI chip market: companies are competing not only on processor speed, but also on how easily developers can use their technology.
