AITom's Hardwareabout 1 hour ago
Google TPUs Now Run in Meta Data Centers
3 min read
Google announced that its custom tensor processing units now serve AI workloads in Meta's data centers via a new PyTorch-native SGLang engine. This collaboration, involving Meta, Google, and RadixArk, allows deployment without rewriting code. The move aims to loosen Nvidia's software lock-in and follows Amazon's recent open-sourcing of TorchNeuron.
Level
Hype check
Tap to vote and see what everyone thinks.
