NVIDIA CEO Jensen Huang wrote in an official blog post today (the 17th) that OpenAI has committed to deploying NVIDIA AI infrastructure at scale by 2030.
OpenAI’s existing and planned deployment commitments total about 12GW (gigawatts) of NVIDIA compute capacity. If NVIDIA expands the PORTS-Pike partnership beyond the initial 4.25GW phase, the total could rise further to around 16GW.
Based on this scale, the business opportunity associated with NVIDIA compute capacity could reach approximately $600 billion by 2030 (approximately RMB 4.05 trillion at current exchange rates).
Huang said that building AI factories requires a complete system of critical resources, including advanced chips, packaging, memory and networking, as well as land, power and building shells. NVIDIA has historically leveraged its scale, long-term demand predictability and supply-chain partnerships to secure critical semiconductor resources. It now plans to apply the same approach to land, power and shells (LPS), specifically securing the capacity needed for AI factories powered by NVIDIA compute.
At the same time, NVIDIA announced a partnership with SB Energy to secure LPS capacity at the PORTS-Pike technology campus in Portsmouth, Ohio, for the deployment of NVIDIA compute, with OpenAI serving as the tenant.
Huang revealed that OpenAI will build and operate a world-class AI factory at PORTS-Pike. The facility will use NVIDIA’s complete DSX AI factory platform, including GPUs, CPUs, networking and infrastructure software.
The initial deployment is expected to provide 4.25GW of AI factory capacity. Each generation of NVIDIA AI factory systems deployed at PORTS-Pike is expected to correspond to approximately 1.5 million NVIDIA GPUs, generating an estimated $150 billion to $200 billion in revenue for NVIDIA (approximately RMB 1.01 trillion to RMB 1.35 trillion at current exchange rates). Over a 20-year period, the campus could support multiple rounds of system upgrades.
The core economic rationale is that the LPS partnership secures a long-term site for AI factories, while the NVIDIA computing equipment inside can be continuously upgraded. Each new generation of systems can deliver greater output, more intelligence and better economics.
NVIDIA may further expand the PORTS-Pike partnership beyond the initial 4.25GW phase in the future, securing the remaining 3.75GW of capacity.