Nvidia to invest up to $100 billion in OpenAI as companies move to finalize $100 billion deal; first Nvidia-powered AI datacenters due 2026
Nvidia plans to invest up to $100 billion in OpenAI to fund at least 10 gigawatts of Nvidia-powered AI datacenters — the first gigawatt using the Vera Rubin platform is scheduled for the second half of 2026.
Key takeaways
- Letter of intent: Nvidia has signed a letter of intent to invest up to $100 billion in OpenAI, with deal details expected to be finalized in the coming weeks — see the OpenAI announcement.
- Phased investment: Investment is tied to deployment of at least 10 GW of compute; Nvidia will acquire non-voting shares as each gigawatt is brought online, and OpenAI will use much of the capital to buy Nvidia systems (IT News).
- Scale and timing: The first gigawatt will run on Nvidia’s Vera Rubin platform in H2 2026; full buildout targets at least 10 GW to support training future foundational models (OpenAI announcement).
- Energy footprint: Ten gigawatts is massive — roughly equal to electricity use of more than 8 million U.S. households — raising power, cooling and siting questions (The Star).
Details of the agreement and timeline
Nvidia and OpenAI have signed a letter of intent and say they plan to finalize the partnership in the coming weeks. The deal is structured as a phased investment tied directly to new datacenter capacity: Nvidia will purchase non-voting shares in OpenAI as each gigawatt of compute is deployed, and OpenAI will use much of the funding to buy Nvidia’s advanced chips and systems.
The first gigawatt of systems is scheduled for deployment in the second half of 2026 and will use Nvidia’s Vera Rubin platform. The companies describe a plan for at least 10 gigawatts of Nvidia-powered AI datacenters to support training and running future foundational models and their joint work toward artificial general intelligence (AGI). For more, see the OpenAI announcement and reporting from IT News.
Leadership comments and strategic goals
Jensen Huang, Nvidia’s CEO, framed the agreement as a leap forward to power “the next era of intelligence.” Sam Altman, OpenAI’s CEO, emphasized that compute infrastructure is foundational to the future economy and said the partnership will expand OpenAI’s ability to serve enterprises, developers and the platform’s more than 700 million weekly active users (OpenAI announcement).
The companies plan to co-optimize hardware and software roadmaps, with Nvidia named as OpenAI’s preferred compute and networking partner — a relationship that builds on over a decade of prior collaboration.
Ecosystem effects and competition concerns
Analysts warn the deal could reshape the AI infrastructure market by tying a major AI company closely to the dominant chipmaker. Rivals and cloud providers — including Microsoft, Oracle and SoftBank — will be watching to see how access to advanced hardware and services is affected (IT News).
“Because Nvidia would own non-voting shares and be a preferred supplier, some worry about lock-in effects that could limit other chipmakers and cloud partners from competing on equal terms.”
Critics caution the agreement may increase market consolidation and reduce competition, particularly if non-voting equity and supplier preference create effective lock-in for OpenAI’s compute stack.
Power, scale and environmental questions
Deploying at least 10 gigawatts of compute underscores the energy intensity of cutting-edge AI. To put that in perspective, 10 GW is roughly equivalent to the electricity consumption of more than 8 million U.S. households (The Star).
OpenAI and Nvidia say systems will be co-optimized for efficiency, but the announcement did not disclose site-selection criteria, cooling technologies, or long-term energy contracting details. These omissions leave open questions about grid impacts, water use, and how emissions goals will be met.
Impacts on business, users and the AI workforce
OpenAI’s reported user base of over 700 million weekly active users stands to benefit from more compute capacity, which could accelerate model training and enable new product features for enterprises, developers and consumers (OpenAI announcement).
The partnership also illustrates a growing business model in which hardware makers and AI firms form deep commercial ties: Nvidia’s progressive investment to buy non-voting shares as compute is deployed creates a feedback loop — OpenAI gets capital to buy Nvidia hardware; Nvidia sells systems and gains equity exposure to OpenAI’s growth (IT News).
Implications for The Inland Empire
Economic opportunity and jobs: Riverside and San Bernardino counties could compete to host some of the Nvidia-powered datacenters. The region’s ample industrial land and lower costs compared with coastal metros make it a candidate for large facilities that bring construction and permanent operations jobs.
Energy and grid considerations: The Inland Empire relies on Southern California Edison and CAISO for grid reliability. Even a fraction of the proposed 10 GW would require careful planning — power purchase agreements, new transmission, and storage will likely be necessary to avoid local strain.
Land use, housing and permitting: Datacenter campuses can alter tax bases, traffic patterns and housing demand. Local governments may weigh incentives against long-term revenue and community benefits, and projects will require environmental review under California law.
Workforce development: Community colleges and local programs could partner with companies to train residents for datacenter operations, maintenance and related trades, improving local hire rates for construction and ongoing roles.
Reporting and next steps
Nvidia and OpenAI say they will finalize deal details in the coming weeks and that investment will be phased as gigawatts are deployed. Regions like the Inland Empire should expect outreach from company planners, utilities and regulators if sites in Southern California are considered (OpenAI announcement; IT News).
