Google and Meta launch TorchTPU, a native PyTorch backend for TPUs
emmanuelvivier · x · 2026-10-27
Google and Meta have open-sourced TorchTPU, a principled native PyTorch backend for Google TPUs that strengthens the compute ecosystem against Nvidia.
- Unlike the old PyTorch/XLA (Lazy Tensor-based), TorchTPU prioritizes eager execution and removes the need for intrusive source changes to mark compilable regions.
- It integrates torch.compile, cuts compile-time latency, delivers competitive training/inference performance, and enables out-of-the-box model migration to TPUs.
- The effort started at Google in mid-2025 in collaboration with Meta's PyTorch team.
More from Infra
- Local AI build dilemma: 2x used RTX 3090 vs RTX 5080 vs Radeon R9700 for image/video gen — shui95827 · 2026-10-27
- The real AI risk to Indian IT: every dollar earned comes with an OpenAI subscription — NirantK · 2026-10-27
- Local voice agent on a 2015 laptop hits 20-30s latency — advice needed — mesolimbic-bliss · 2026-10-27
- United CEO says passengers are switching airlines just for Starlink internet — XFreeze · 2026-10-27
- Naveen Rao: AI's next chapter is inference, a $10T market within a decade — ravikantagrawal · 2026-10-27
- Thermal imaging suggests IREN data halls 4 and 2 commissioning, 3-4 halls billing by year-end — McDonaghMatthew · 2026-10-27