Amazon Releases TorchNeuron Beta to Run PyTorch on Trainium Chips

Amazon is releasing TorchNeuron as a free, open-source public beta that lets developers run existing PyTorch code on its Trainium AI chips after a configuration change, without rewriting the code. The tool is intended to lower the software barrier that has helped Nvidia retain an advantage through its widely used CUDA ecosystem, where adapting applications to other hardware can require substantial engineering work. Amazon says more than 30 early customers in areas including robotics, voice AI and world models have tested TorchNeuron, and its chip unit says Trainium may offer cost savings. TorchNeuron adds to Amazon’s broader Neuron software effort, which includes tools for programming and optimizing workloads on its custom chips. Google and Microsoft are also pursuing ways to shift AI workloads toward their own processors, intensifying efforts across the industry to reduce reliance on Nvidia.