Google Cloud and Inferact partner to expand TPU support in vLLM

Inferact announced a partnership with Google Cloud to make TPU a “first-class citizen” in vLLM, a model-serving engine. Inferact says the teams will share an engineering roadmap covering production serving features, optimized kernels and a native PyTorch path through TorchTPU. They aim to support frontier models on release day. The announcement also includes a community program offering shared TPU capacity to open-source contributors, plus review and design help from Inferact’s core vLLM maintainers. Inferact says everything the collaboration produces will be open source.