IBM Teams with Together AI on $240M Nvidia-Powered AI Cluster
IBM has signed a $240 million deal with startup Together AI to build an AI cluster powered by Nvidia systems. The proposed infrastructure will be hosted on IBM Cloud and use Nvidia's HGX B300 systems, incorporating its latest Blackwell processors.
The cluster is designed for inference tasks, running open-source AI models such as DeepSeek, MiniMax, and Kimi. According to Nvidia, the infrastructure can deliver 30 times more AI factory output than previous generations.
Initially, about 2,000 Nvidia Blackwell chips will be deployed in the U.S., with the cluster expected to become available in the first quarter of 2027. This deal marks a significant collaboration between IBM and Nvidia, building on their partnership to expand AI infrastructure for enterprise customers.
Enterprises are increasingly looking for cost-effective alternatives to closed-source AI models, which can be expensive. Vipul Ved Prakash, CEO of Together AI, noted that 'enterprises want the performance of the best frontier models without the closed-model price tag, and that only works if the infrastructure underneath is fast and reliable at scale.'
This deal comes amid growing attention to the capabilities and security risks of autonomous AI systems. As interest in open-source AI grows, Together AI aims to make it an 'obvious choice for enterprises', with this cluster being a key step forward.