Details
- IBM and Together AI announced a multi-year $240 million collaboration to deploy a large NVIDIA HGX B300 cluster on IBM Cloud for open-source AI inference, with availability expected in Q1 2027.
- The deployment will use NVIDIA HGX B300 systems and Spectrum-X Ethernet networking, and IBM says it is the first dedicated large-scale inference cluster of its kind on IBM Cloud.
- Together AI will use the cluster to improve enterprise inference performance and token economics as it expands its open-source AI platform.
- The deal builds on Together AI’s recent $800 million Series C at an $8.3 billion valuation and its reported scale of 400 trillion tokens served monthly.
- The collaboration extends IBM and NVIDIA’s broader enterprise AI push, combining cloud infrastructure, GPU capacity, and open-source model deployment for production workloads.
Impact
The deal reinforces a broader shift toward open-source AI inference as enterprises look for lower-cost alternatives to closed models. It also signals continued demand for large-scale GPU infrastructure, which should support investment in cloud-native AI capacity and enterprise deployment tools over the next 12–24 months.
