IBM and Together AI Sign $240 Million Deal for Nvidia Blackwell AI Inference Cluster
IBM and Together AI signed a $240 million multiyear agreement to build a large AI inference cluster on IBM Cloud using Nvidia HGX B300 systems, Blackwell 300 processors and Spectrum-X networking.
Why it matters: The deal shows inference infrastructure becoming a major enterprise spending category and strengthens the role of open-weight models in production AI deployments.
IBM and Together AI have signed a $240 million multiyear agreement to build a large artificial-intelligence inference cluster on IBM Cloud, expanding enterprise access to open-weight AI models running on Nvidia infrastructure. The initial U.S. deployment will include about 2,000 Nvidia Blackwell 300 processors using HGX B300 systems and Spectrum-X networking. The project is focused on inference, the process of running trained models to generate responses, which has become one of the largest drivers of new computing demand as companies move AI applications into production. Together AI supports open models including DeepSeek, MiniMax and Kimi, giving enterprises additional options for controlling model costs and deployment architecture. The agreement also shows how cloud providers, chipmakers and AI platforms are increasingly competing around the economics of serving models continuously at enterprise scale.