Google launches TorchTPU to challenge Nvidia dominance

Google has released TorchTPU, a specialized version of the open-source PyTorch framework designed to make it easier for developers to run AI applications on Google's Tensor Processing Units rather than Nvidia GPUs. Since over 85% of AI apps are built with PyTorch, which was originally designed around Nvidia hardware, Google aims to reduce that lock-in. TorchTPU allows existing apps to migrate with minimal code changes. Google also launched a TPU agent suite to automate cloud migration and updated its LiteRT tool for on-device compatibility. Amazon Web Services made a similar move this week with its own PyTorch variant, TorchNeuron.