Rackspace and Together AI have signed a multiyear agreement under which Rackspace will provide and operate dedicated NVIDIA GB300 NVL72 infrastructure to power Together AI's artificial intelligence inference workloads. The deal, announced on 10 November, carries an anticipated total contract value of $380 million over several years, Rackspace said on its third-quarter earnings call the same day.

The agreement covers the full stack around the hardware. Rackspace will supply and operate the dedicated NVIDIA Blackwell Ultra systems, including power, liquid cooling, networking and storage, in a purpose-built liquid-cooled facility, connected with NVIDIA Spectrum-X Ethernet networking. Together AI, which describes itself as the AI Native Cloud, will run its managed platform on top, giving customers access to open-source models, inference optimisation and developer tools.

Deployment planned for 2027

Deployment is planned for 2027, subject to certain conditions being met, and Rackspace said it expects the infrastructure to be in service by mid-2027. The company told investors it now expects to be operating about 20 to 22 megawatts of cumulative high-performance computing capacity for AI applications by the end of 2027, including the Together AI deployment.

There is an important caveat. Rackspace's obligations under the agreement depend on it securing financing for the GB300 systems on acceptable terms, and no definitive GPU financing commitment had been obtained when the deal was announced. The company said it was in active discussions with potential lenders. On 6 November, Rackspace entered into a share purchase agreement with an investor group led by Voss Capital for a $66 million private placement of common stock.

A wider partnership in view

The two companies also signed a separate, non-binding letter of intent to explore a joint go-to-market approach for enterprise customers, with Together AI providing inference capabilities and Rackspace contributing infrastructure, platforms and forward-deployed engineering support. Rackspace told investors the letter of intent also covers potential future capacity expansion that it believes could represent a multibillion-dollar long-term opportunity, although that part of the arrangement is not a definitive agreement and remains subject to future deals and conditions.

The announcement, carried by GlobeNewswire, said Together AI chose Rackspace for its engineering depth, operational capability and ability to support demanding enterprise AI workloads. Rackspace, an NVIDIA Cloud Partner, will operate the deployment with dedicated GB300 NVL72 systems, the rack-scale platform built around 72 Blackwell Ultra GPUs and 36 Grace CPUs in a single liquid-cooled rack.

The deal reflects the shift in the AI infrastructure market from training to inference, the work of running trained models to answer questions, generate text and serve applications. Inference demand is rising as businesses put AI agents and assistants into everyday use, and providers are racing to offer dedicated, high-performance capacity that enterprises can rent rather than build themselves.

Rackspace announced the agreement alongside its third-quarter results. On the earnings call, management said the Together AI contract and a separate $230 million agreement with Blacksmith for AMD EPYC bare-metal computing brought $610 million in aggregate anticipated contract value from new definitive agreements. The company also raised its full-year 2026 revenue guidance to between $2.5 billion and $2.6 billion.