OpenAI has secured 457MW of dedicated AI data centre capacity under a six-year agreement with TensorWave across several facilities in the US.
The contract represents TensorWave’s largest infrastructure project to date and will support the computing requirements needed for OpenAI to build and operate advanced AI models.
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Under the arrangement, the provider will install AMD Instinct MI455X graphics processing units (GPUs) inside data centre facilities tailored specifically for AMD Helios rack-scale architecture.
Initial data centre capacity is scheduled to go live in the first half of 2027, with the entire rollout expected to conclude in early 2028.
The dedicated, single-tenant data centre environments are designed to meet escalating computational needs for complex AI workloads.
TensorWave will connect AMD Helios racks using high-performance Ethernet networking to handle both large-scale training and inference operations.
Fulfilling the deployment requires integrated management across electrical power, data centre cooling, networking architecture, and processor clusters spanning hundreds of megawatts across multiple locations.
TensorWave president and co-founder Piotr Tomasik said: “AI infrastructure is entering a new phase, where the ability to deliver hundreds of megawatts of advanced compute reliably and at speed will separate the companies that can support frontier AI from those that cannot.
“TensorWave was built to prove that AMD infrastructure could operate at the scale and reliability that the world’s most demanding AI companies require.
“This deployment takes that work into an entirely new level, with TensorWave helping deliver a meaningful portion of the infrastructure behind OpenAI and AMD’s next generation of AI compute.”
The development expands upon an earlier strategic collaboration established between OpenAI and AMD. Through this project, TensorWave becomes the first neocloud infrastructure provider publicly confirmed to support the joint venture.
TensorWave was previously the earliest public cloud provider to deploy AMD Instinct MI300X processors, subsequently broadening its systems across MI325X and MI355X hardware generations.
The contract marks a substantial addition to TensorWave’s American operations, building on over 2GW of long-term data centre capacity secured to strengthen future AI workloads.
Prior to this, Firmus signed a multi-year deal to supply dedicated computing capacity from two data centres in Malaysia to OpenAI, taking its total contracted capacity beyond 900MW.