Rackspace Technology and Together AI announced on November 10 a multiyear strategic agreement under which Rackspace will provide and operate dedicated Nvidia GB300 NVL72 infrastructure, connected by Nvidia Spectrum-X Ethernet, to support Together AI’s inference workloads. ConvergeDigest put the anticipated total contract value at $380 million, a figure the companies describe as expected contract value rather than recognised revenue. Deployment is planned for 2027 in a purpose-built, liquid-cooled facility; neither company disclosed the site, the rack count or the total GPU count.
Each GB300 NVL72 system integrates 72 Blackwell Ultra GPUs and 36 Grace CPUs in a single liquid-cooled rack. Rackspace will run the physical layer, including power, liquid cooling, networking and storage, while Together AI supplies the managed AI cloud platform, inference optimisations, access to open models and developer tools on top. The companies also signed a separate, non-binding letter of intent to explore a joint go-to-market offering for enterprise customers.
The binding constraint is money. Rackspace’s obligations are conditional on obtaining financing for the GB300 purchase on acceptable terms, and the company said no definitive GPU financing commitment had been obtained at announcement, though it is in active discussions with lenders. Separately, on November 6, Rackspace entered a share purchase agreement with an investor group led by Voss Capital for a $66 million private placement, proceeds earmarked for general corporate purposes and GPU-related initiatives, a raise this site covered on November 10.
A second, smaller agreement came with the same announcement cycle. Rackspace will provide Blacksmith, an infrastructure platform for building and testing code, with dedicated compute powered by AMD EPYC processors plus storage, totalling approximately 2 megawatts in Northern Virginia, with deployment beginning in early 2027. Rackspace already has an agreement with AMD to deploy 30 megawatts of dedicated AMD compute across its data centres through 2028, and its third-quarter investor summary targets 20 to 22 megawatts of high-performance AI compute across its planned AMD and Nvidia deployments by the end of 2027, with $420 million to $460 million of annualised revenue, targets that depend on the same financing and deployment conditions.