IBM, Together AI ink $240 million deal for Nvidia-powered AI inference cluster

IBM and startup Together AI signed a $240 million, multi-year deal to build a large-scale AI inference cluster on IBM Cloud using Nvidia hardware, the companies said Tuesday. The system will run inference for open-source AI models, targeting demand from businesses trying to rein in AI costs while also weighing cybersecurity concerns around models from Anthropic, OpenAI and Meta. Inference has become a major driver of cloud capacity. The IBM Cloud cluster will use Nvidia HGX B300 systems paired with Spectrum-X Ethernet networking. The first site will include about 2,000 Nvidia Blackwell 300 chips in the U.S., with Together AI’s Kai Mak saying it expects the capacity to be sold out 2–3 months early.