IBM and Together AI Sign Multi-Year Agreement to Scale Open-Source AI Inference with NVIDIA AI Infrastructure on IBM Cloud

Details

IBM and Together AI have announced a multi-year $240 million agreement to build what IBM describes as the first large-scale inference cluster on IBM Cloud using NVIDIA HGX B300 systems and Spectrum-X Ethernet networking. Together AI will use the cluster to deliver open-source model inference at enterprise scale, with infrastructure expected available in Q1 2027.\n\nThe scale is notable: Together AI reports serving 400 trillion tokens monthly across its inference, training, fine-tuning, and agentic workflow platform. The company recently closed an $800 million Series C at an $8.3 billion valuation. According to NVIDIA, the HGX B300 systems deliver 30x more AI factory output compared to prior-generation hardware.\n\nTogether AI's pitch is enterprise-focused: open-source model inference at performance and economics competitive with frontier proprietary models. The IBM partnership is positioned as delivering that combination at scale, while giving IBM Cloud a credible marquee AI workload to showcase.\n\nFor IBM, this is a meaningful signal that its cloud platform is being chosen for serious AI workloads. The press release does not compare pricing or SLAs against AWS, Azure, or GCP — those comparisons will matter as enterprises evaluate provider choices.