IBM and Together AI sign a multi-year $240 million agreement for AI inference, according to multiple reports. The deal centers on deploying Nvidia HGX B300 systems on IBM Cloud as part of Together AI’s inference offering.
Quartz reports that the arrangement is structured as an open-source AI inference deal, with availability expected in the first quarter of 2027. Channel NewsAsia similarly describes the agreement as a $240 million Nvidia-powered inference cluster deal tied to IBM Cloud, emphasizing the use of the same Nvidia HGX B300 platform.
While both sources converge on the core commercial figure and the hardware and cloud context, they differ mainly in emphasis: Quartz foregrounds the open-source nature of the inference arrangement, while CNA focuses more directly on the Nvidia-powered “inference cluster” framing. Neither report, based on the provided excerpts, includes further details on pricing structure, deployment scope beyond the systems and cloud venue, or specific performance or workload targets.
