Meta Engineer Confirms: Meta Is Running AI Inference on Google TPUs in Production. On Oct 20, Google published a TPU version of the SGLang inference engine on its developer blog, quoting Meta's PyTorch-T...
In an Oct 20 developer-blog post announcing new vLLM and SGLang inference engines for TPU, Google quoted Meta's PyTorch-TPU lead Zongwei Zhou saying 'SGLang-TorchTPU is already running in Meta's production environment' — the first first-party confirmation that Meta runs AI inference on Google TPUs in production. The Information reported in February on a multi-year, multi-billion-dollar TPU rental contract between the two companies; scale, products and terms were not disclosed. Why it matters: even the biggest GPU buyers are now publicly diversifying into TPUs, a visible dent in the CUDA moat.
SOURCES
- 宝玉 (Baoyu)2026-10-21 06:09