IBM and Together AI have announced a significant collaboration backed by a multi-year, $240 million agreement to advance open-source artificial intelligence for enterprises. This partnership will see IBM deploy a large-scale NVIDIA-powered AI inference cluster on its cloud platform for Together AI's use. The new infrastructure, expected to be operational in the first quarter of 2027, aims to deliver enhanced performance and efficiency for AI workloads.
A Strategic Partnership for Enterprise AI
Under the terms of the agreement, IBM is set to deploy a substantial cluster of NVIDIA HGX B300 systems specifically for Together AI. This initiative represents the first dedicated, large-scale inference cluster of its kind on IBM Cloud, designed to optimize performance for production AI. The collaboration is structured to improve token economics, enabling businesses to scale their AI deployments more efficiently and cost-effectively.
This strategic move aligns with Together AI's core mission to establish open-source models as a cornerstone of future AI development. The company's growth is bolstered by a recent $800 million Series C funding round, which elevated its valuation to $8.3 billion. This capital injection is intended to expand its AI Native Cloud platform, which has already seen significant momentum, currently serving 400 trillion tokens monthly.
Harnessing Advanced NVIDIA Technology
The technological foundation of this new cluster consists of NVIDIA's state-of-the-art HGX B300 systems and Spectrum-X Ethernet networking. This powerful hardware combination is engineered to provide a massive leap in AI factory output, with NVIDIA reporting a potential 30-fold increase compared to prior generations. Such performance is critical for supporting the demanding, real-time AI services that modern enterprises require for their operations.
Together AI's decision to partner with IBM and NVIDIA was driven by their forward-looking product roadmaps and their capacity to deliver advanced GPU infrastructure at scale. Vipul Ved Prakash, CEO of Together AI, highlighted the need for fast and reliable infrastructure to make open-source AI competitive with closed models. He emphasized that this cluster is a significant step toward making open-source AI the clear choice for enterprises.
Expanding a Broader Tech Alliance
This project is the latest milestone in a broader and deepening alliance between IBM and NVIDIA, which focuses on advancing AI across both hardware and software. Alan Peacock, General Manager of IBM Cloud, remarked that the collaboration provides the scalable and economical infrastructure needed to help enterprises accelerate innovation. This partnership offers a dependable foundation for organizations to build, deploy, and manage their AI systems effectively.
Industry leaders view this development as a move toward establishing AI factories as essential business infrastructure, akin to public utilities. Dion Harris of NVIDIA noted that the platform turns compute and data into intelligence, providing the performance and scale required for enterprise-grade AI services. The broader IBM and NVIDIA collaboration also extends to GPU-native data analytics, unstructured data extraction, and comprehensive consulting services.
The $240 million agreement between IBM and Together AI marks a pivotal investment in the enterprise open-source AI ecosystem. By integrating cutting-edge NVIDIA technology within the robust IBM Cloud environment, this partnership is set to redefine performance and accessibility for AI inference. This strategic alliance is poised to accelerate the adoption of scalable, efficient, and powerful AI solutions for businesses around the world.