NVIDIA’s CEO, Jensen Huang, has revealed a little-known story about the early days of the modern artificial intelligence revolution: according to his own account, Elon Musk was not only one of the first to bet on their AI supercomputers but is also pointed out as the “original founder” of OpenAI and, by extension, the precursor to ChatGPT.
Far from the high-stakes scenes and billion-dollar valuations, the scene unfolds in 2016, in a context where the idea of investing billions in hardware to train AI models still sounded risky even to many tech giants.
An extremely expensive machine… with no customers
Huang recalls announcing a new computing system specifically designed for artificial intelligence: a supercomputer costing approximately $300,000 per unit, stemming from an investment of several billion dollars.
However, the venture did not have a triumphant start.
Nobody wanted to buy it. No orders. Not a single purchase.
Amid this climate of uncertainty, during an event discussing the future of autonomous cars, Elon Musk approached him and said a phrase that would change the direction of that project:
“I have a company that could really use this.”
For Huang, that meant finally landing a first client willing to invest in an architecture designed for large-scale AI model training. But the enthusiasm was short-lived.
“It’s a non-profit organization”
In his account, the NVIDIA CEO describes how, after that initial hopeful comment, he was met with a detail that made him turn pale:
Musk explained that the organization he was referring to was a non-profit dedicated to artificial intelligence, and they needed one of those supercomputers.
Huang admits that at that moment, “all the blood drained from his face.” After spending an astronomical sum developing this machine, the likelihood that a non-profit entity could shoulder that cost seemed minimal.
Nevertheless, he decided to move forward and accept the challenge.
A supercomputer heading to a small office in San Francisco
According to Huang himself, NVIDIA built the first system for internal use and prepared the second for Musk and his team. The gesture was almost artisanal: he put the supercomputer in a box, loaded it into a car, and drove to San Francisco to deliver it personally.
Upon arrival, he went up to the second floor of a building where, as he describes, the entire team was working “crowded into a smaller room” than one might imagine for such an ambitious project. That place turned out to be the initial headquarters of OpenAI.
It is in this context that Huang claims Elon Musk was the “original founder of OpenAI and ChatGPT,” emphasizing his role as a driver of the organization and as the first major client willing to invest in the necessary hardware to train advanced language models when hardly anyone saw the potential of that technology.
When infrastructure was still a gamble, not a certainty
Today, the story seems almost counterintuitive: NVIDIA has become the leading provider of GPUs for data centers and AI projects, and accelerated computing is one of the pillars of the current tech economy. However, in 2016, the landscape was very different:
- The market for AI supercomputers was far from consolidated.
- Many actors saw training giant models as an expensive experiment.
- The social and business phenomenon associated with tools like ChatGPT had not yet emerged.
In this context, Musk’s decision to invest in that hardware was both a visionary move and a key vote of confidence that validated NVIDIA’s strategy.
OpenAI, Musk, and the narrative of origins
OpenAI has publicly presented itself as a co-founded organization by several figures, including Elon Musk, although over time the trajectories of the company and the entrepreneur diverged, with public tensions surrounding its evolution.
What Jensen Huang now contributes to the story is an additional layer: that of infrastructure and initial risk. His account reminds us that, before generative AI became a central topic for governments and markets, someone had to be convinced to trust expensive hardware, a new architecture, and a nonprofit organization dreaming of training large-scale language models.
A different picture in a highly competitive market
Huang’s statements come at a time when the AI market is more fragmented and competitive than ever:
- NVIDIA is a strategic piece of the global AI infrastructure.
- Elon Musk is pushing his own model company, xAI, in direct competition with others.
- OpenAI has become one of the sector’s central names, with a strong commercial focus.
In this context, the story of “that first supercomputer in the trunk” serves as a reminder that many projects that now move billions of dollars started from personal decisions, trust-based relationships, and gambles that seemed almost irrational.
Beyond the numbers and current conflicts, the overarching message is clear: the AI revolution is not only about algorithms but also about hardware, risk, and visionaries willing to buy an incredibly expensive machine when no one else wanted to.