On the 11th, IBM announced a multiyear, $240 million agreement with AI-native cloud platform Together AI.
IBM will deploy an NVIDIA HGX B300 cluster on IBM Cloud to support Together AI's inference services for open-source models. The system is expected to enter service in Q1 2027.
IBM said this deployment is the first large-scale inference-only cluster built on IBM Cloud using NVIDIA HGX B300 systems and Spectrum-X Ethernet technology.
Together AI CEO Vipul Ved Prakash said:
Enterprises want the most advanced model performance without bearing the high costs of closed models, but this is only possible when the underlying infrastructure is fast, reliable, and capable of deployment at scale.
Our collaboration with IBM and NVIDIA lays the foundation for this. This cluster enables us to deliver production-grade inference capabilities to more companies, faster. It is an important step in making open-source AI the preferred choice for enterprises.
Together AI recently completed a $800 million Series C financing round at a valuation of $8.3 billion, with monthly processing capacity of 400T tokens.