For Years, NVIDIA Hid That Nobody Actually Wanted Its First AI Server

Fresh materials from the legal battle between Elon Musk and Sam Altman have unexpectedly revealed how NVIDIA chief Jensen Huang landed a key customer for his first dedicated AI server, the DGX-1. Back in 2016, he assured Musk there was colossal demand for the device, while today he admits: nobody except Musk himself actually wanted it at the time.

In the disclosed correspondence from April 2016, Huang actively lobbies Musk to purchase the DGX-1—a deep learning system packing eight Tesla P100 accelerators, HBM2 memory, and an NVLink interface. The device had been unveiled just a week earlier at the GTC conference, and the NVIDIA chief wrote that the DGX-1 was "selling itself online," with orders pouring in "from everywhere," but he was ready to set aside one of the first units for OpenAI. Musk agreed, and soon after, Huang personally delivered the server to the nonprofit’s small office in San Francisco.

Only now has it become clear that the hype was, in part, staged. Ten years later, in 2026, Jensen Huang described the DGX-1 launch in completely different terms: "No one in the world had any interest in buying it. I had zero orders—not a single one. Nobody wanted to get involved in this except Elon." By his account, the situation looked like a flop, and only a chance conversation with Musk saved the project. The contrast with the tone of his 2016 business emails is stark—back then, the situation was presented as a triumphant launch, not a near-hopeless search for a first buyer. In the end, that very deal set off a chain of events that turned NVIDIA into the dominant hardware supplier for AI and transformed the DGX-1 itself into a museum piece and a symbol of the dawning era.

Curiously, the DGX-1 story surfaced in the public sphere almost in sync with another speech by Huang. On May 10, 2026, he delivered a commencement address to graduates of Carnegie Mellon University. The NVIDIA chief spoke of the dawn of a new industrial era, of how the current generation is starting out at the forefront of the AI revolution, and urged that artificial intelligence be developed responsibly, with transparency and a focus on safety. He stressed that scientists and engineers must champion not only capabilities but also the safety of systems, while policymakers must build sensible regulatory frameworks. Yet juxtaposing the two events inevitably raises questions about the flip side of major technological breakthroughs. On one hand, you have inspiring slogans about building the future honestly and openly. On the other, a documented fact that one of NVIDIA’s most famous products came into being largely thanks to a skillfully orchestrated hype campaign.

Today, this deal is interesting not just as a historical curiosity. It vividly shows just how fragile the launch of a technology can be—one without which we now can’t imagine the operation of data centers, language models, and autonomous systems. And it serves as a reminder that behind the flashy product launches, there sometimes aren’t queues of customers at all—just one determined buyer and a seller willing to embellish the picture a bit.

What do you think: are entrepreneurial tactics like this justified if they ultimately accelerate progress? Or is transparency more important in the long run than any marketing success? Share your thoughts in the comments.