Amazon Unveils Open-Source TorchNeuron Tool for AWS Trainium AI Chips, Shares Rise 5%

Amazon $AMZN unveiled TorchNeuron on October 19, 2026, an open-source backend that allows developers to run existing PyTorch code on AWS Trainium chips with a single-line device change. Amazon shares climbed as much as 5% intraday following the announcement. Built by Amazon's custom chip division Annapurna Labs and released under the Apache 2.0 license, the tool registers Trainium as a native device within PyTorch. Developers can point existing workflows to Trainium2 or Trainium3 instances without a lengthy codebase rewrite. The public beta opened on October 20, 2026, at the PyTorch Conference in San Jose, supporting eager execution, distributed training APIs, and the Neuron Kernel Interface for custom kernels. More than 30 early customers across robotics, voice AI, and drug discovery are already using TorchNeuron and have reported financial benefits from running workloads on Trainium hardware, with Annapurna Labs VP of software Colin Brace citing cost savings as the primary driver for switching.