Google, Amazon challenge Nvidia’s AI dominance with new PyTorch tools
Google launched TorchTPU on Tuesday, following Amazon’s TorchNeuron release Monday, to simplify moving PyTorch applications to their AI chips. The tools could lower switching costs from Nvidia, though adoption will depend on performance, reliability and operating costs.
1. Tools target chip migration
Google launched TorchTPU on Tuesday, following Amazon’s TorchNeuron release Monday, to simplify migration of PyTorch applications to their proprietary AI chips. Amazon is also submitting changes to PyTorch to reduce Nvidia-specific dependencies. Adoption will depend on performance, reliability and operating costs, while Nvidia’s established software ecosystem remains a competitive advantage.
