IBM's $240M Together AI Deal For Nvidia Systems Puts It In The Neocloud Business

The cluster will use Nvidia HGX B300 systems and Spectrum-X Ethernet networking and is expected to come online in Q1 2027.

• Together AI is a neocloud that provides GPU infrastructure and AI services to developers and enterprises.
• The neocloud plans to use IBM’s Cloud capacity to support its open-source AI model inference business for enterprise customers.
• The company recently raised $800 million at an $8.3 billion valuation and says its platform processes 400 trillion tokens each month.

International Business Machines (IBM) is expanding its push into the fast-growing neocloud market with a $240 million deal to provide Together AI with dedicated AI computing capacity on IBM Cloud, using Nvidia (NVDA) systems.

Together AI is a neocloud, a category of GPU-focused infrastructure providers that includes publicly traded companies such as CoreWeave (CRWV) and Nebius (NBIS). Unlike traditional enterprise customers, these companies buy large amounts of computing capacity and resell access to AI developers and businesses through their own platforms.

The agreement gives IBM a foothold in that market by making Together AI a major cloud tenant rather than a direct competitor. Together AI says its platform now processes about 400 trillion tokens a month. The deal could therefore give IBM exposure to AI infrastructure demand beyond its traditional enterprise software and consulting customers.

IBM stock edged 0.3% higher in pre-market trade. On Stocktwits, retail sentiment around IBM remained in ‘bearish’ territory over the past day. Meanwhile, NVDA stock gained as much as 1.5% in pre-market trade, with retail sentiment trending in the ‘bullish’ zone. 

IBM Says Inference Cluster Will Use NVDA HGX B300 Systems

IBM said the inference cluster that it will build with Together AI will use Nvidia HGX B300 systems with Spectrum-X Ethernet networking and is expected to come online in the first quarter of 2027.

The infrastructure will support Together AI's open-source model inference business for enterprise customers. Nvidia has said its B300 systems can deliver up to 30 times the AI factory output of prior-generation hardware.

For IBM, the deal provides a concrete deployment tied to its broader Nvidia partnership. The companies announced plans in March to bring Nvidia Blackwell Ultra GPUs to IBM Cloud and integrate Nvidia's data and inference tools into IBM's Red Hat AI Factory stack.

For updates and corrections, email newsroom[at]stocktwits[dot]com.<