Amazon says its AI chips will work with popular coding language, challenging Nvidia

Amazon is rolling out a new tool that enables its custom AI chips to execute code written in the most widely-used AI coding language, PyTorch, without the need for any modifications. This move challenges the market dominance of Nvidia, which enjoys an edge due to its proprietary software called CUDA that optimizes PyTorch code for its own chips.

Amazon's TorchNeuron tool allows developers to run PyTorch code on Amazon's Trainium AI chips with a single configuration change, eliminating the need for any code rewriting. The software will be made available as a free and open-source offering at a developer conference in Silicon Valley on Monday. Amazon's Annapurna Labs chip unit, led by Colin Brace, vice president of software, said the tool is designed to streamline the process for AI developers to adopt Amazon's hardware.

Early adopters, including those in robotics, world models, and voice AI, are reportedly experiencing significant cost savings by switching to Amazon's chips.