Altman-Huang Talks Forged $100B OpenAI-Nvidia Partnership

ABILENE, Texas – Sam Altman, CEO of OpenAI, was operating on a tight schedule. He was en route to Texas to unveil a significant infrastructure initiative, and Nvidia CEO Jensen Huang was keen to be a part of it.

Following a flurry of negotiations, late-night discussions, and last-minute contract revisions, OpenAI and Nvidia finalized a monumental $100 billion partnership on Monday, mere hours before Altman’s departure for Abilene, a city of approximately 130,000 residents situated about 180 miles west of Dallas.

Adding a layer of political context, Huang and Altman’s participation in President Donald Trump’s state visit to the U.K. the previous week provided an opportunity to brief the President on the agreement in advance.

Huang characterized the deal to CNBC as “monumental in size,” underscoring its significance as a pivotal moment in the technology sector, where capital and influence are increasingly concentrated within the leading players of the artificial intelligence revolution.

Huang now leads Nvidia, the world’s most valuable publicly traded company, boasting a market capitalization of nearly $4.5 trillion following a $170 billion surge after Monday’s announcement. Altman, on the other hand, heads OpenAI, the most prominent startup globally, estimated to be worth half a trillion dollars.

OpenAI’s rise to prominence in the generative AI space is deeply intertwined with Nvidia’s high-performance graphics processing units (GPUs). Now, their collaboration is even more profound as they embark on a joint venture to develop the next generation of AI supercomputing facilities.

“You should expect a lot from us in the coming months,” Altman stated in an interview with CNBC’s Jon Fortt at Nvidia’s Silicon Valley headquarters on Monday. “There are three core areas where OpenAI needs to excel: groundbreaking AI research, the creation of user-centric products, and addressing the unprecedented infrastructure challenges this entails.”

According to sources familiar with the discussions, Altman and Huang primarily conducted the negotiations through virtual channels and private meetings in London, San Francisco, and Washington, D.C., bypassing involvement from traditional investment banks.

The agreement stipulates that Nvidia will invest $10 billion at a time in OpenAI, the company renowned for ChatGPT. Nvidia will also supply state-of-the-art processors to power a series of new data centers as part of the infrastructure buildout.

While solidifying its relationship with Nvidia, OpenAI faces the challenge of maintaining critical partnerships with other key players.

Sources indicated that OpenAI only informed Microsoft, its major shareholder and primary cloud provider, one day prior to signing the deal. Earlier this year, Microsoft relinquished its status as OpenAI’s exclusive provider of computing resources.

Furthermore, the agreement comes shortly after Oracle revealed an agreement under which OpenAI agreed to allocate $300 billion towards computing resources over a five-year period, commencing in 2027. At the beginning of the year, OpenAI became involved with Stargate, a multibillion-dollar project promoted by President Trump and supported by Oracle and SoftBank, dedicated to developing next-generation AI infrastructure. This strategic move suggests a diversification of OpenAI’s compute sourcing beyond Microsoft Azure, a critical step in mitigating potential vendor lock-in and securing long-term scalability.

Looking ahead, all of OpenAI’s infrastructure projects will be integrated under the Stargate umbrella.

Representatives from Microsoft, Oracle and SoftBank didn’t immediately respond to requests for comment.

Nvidia and OpenAI have disclosed limited details regarding the timelines and locations for the infrastructure buildout, except to note that the first of the 10-gigawatt sites will be operational in the second half of next year. The choice of location is crucial, involving a complex balance between factors like access to renewable energy sources (to minimize environmental impact and operating costs), favorable regulatory environments, and proximity to skilled labor.

Executives have stated that they have evaluated between 700 and 800 potential locations since introducing Stargate in January. Since then, they have received numerous proposals from developers throughout North America, presenting land, power, and facilities. OpenAI has narrowed down its selection based on energy availability, permitting timelines, and funding alternatives, according to the company.

OpenAI has characterized Nvidia as a “preferred supplier.” However, executives told CNBC that this is not an exclusive relationship. OpenAI is proactively working with large cloud providers and other chipmakers in order to avoid dependence on a single vendor. This multi-sourcing strategy reflects a growing trend among AI companies who are seeking to diversify their supply chains and negotiate more favorable pricing.

While the investment in OpenAI represents a substantial undertaking for Nvidia, it is part of a rapidly growing investment portfolio.

Last week, Nvidia invested $5 billion into Intel through a joint venture focused on co-developing data center and PC chips with the semiconductor manufacturer. Additionally, Nvidia invested approximately $700 million in U.K.-based data center startup Nscale, mirroring Nvidia’s support for U.S. AI infrastructure provider CoreWeave, which held its IPO in March.

The financial structure of the OpenAI transaction is designed to mitigate substantial share dilution. Sources familiar with the matter said that the initial $10 billion tranche is secured at a pre-determined valuation of $500 billion and is anticipated to close within approximately one month once the transaction is finalized. There are nine further tranches of $10 billion planned, each to be priced at the company’s current valuation as new capacity becomes operational.

The relationship between Nvidia and OpenAI significantly predates the launch of ChatGPT in 2022.

When OpenAI was still a relatively small nonprofit research lab, and Nvidia was most known for graphics chips used in video games, Huang personally delivered his company’s first DGX supercomputer to OpenAI’s office in 2016. At the time, OpenAI operated from a building located in San Francisco’s Mission District, which now serves as the headquarters for Elon Musk’s xAI.

Nearly a decade and trillions of dollars in value later, Huang and Altman stand as influential figures in the tech industry.

Nvidia officially established its financial stake in OpenAI in October of the previous year, participating in a $6.6 billion funding round that valued the company at $157 billion. A month later, in Tokyo, OpenAI executives met with SoftBank CEO Masayoshi Son to brainstorm a name for their next growth phase. The outcome of this session was “Stargate,” a code name that has since become a shorthand for OpenAI’s foremost aggressive construction plans.

Stargate now encompasses every major contract for compute capacity, including this week’s partnership with Nvidia. While obtaining rights to the name required careful negotiation, OpenAI has adopted it as the label for its long-term infrastructure strategy. The Stargate initiative highlights a broader industry trend of companies investing heavily in proprietary infrastructure to support their AI initiatives.

Nvidia’s $100 billion commitment is a portion of what’s required for the proposed 10-gigawatt buildout. Although OpenAI will lease Nvidia’s chips for deployment, the startup plans to source additional capital to finance the long-term expansion.

As OpenAI’s compute needs increase, a major question is where the company will host its workloads, which have so far predominantly been housed in Microsoft Azure. Moving the activities in-house would put OpenAI closer to working as a first-party cloud provider, a market led by Amazon Web Services, followed by Azure, Google, and Oracle. This strategic decision will have significant implications for the cloud computing landscape, potentially disrupting existing market dynamics.

Executives have openly discussed the idea, indicating it might not be distant. Some told CNBC that a commercial cloud offering could appear in a year or two, once OpenAI has secured enough compute to service its own demands. Currently, the need for training advanced models leaves little capacity to spare, but OpenAI continues to seek new opportunities. Successfully transitioning to a commercial cloud provider would require significant investment in infrastructure, security, and talent.

As Altman and Huang discussed details of the announced agreement, OpenAI’s infrastructure team met in Tokyo with SoftBank’s Son to review broader financing and manufacturing assistance. These parallel discussions underline the scope of Altman’s vision, and the network of global parties now involved in realizing it.