Elon Musk's AI startup xAI is about to become the anchor tenant of what might be one of the world's most ambitious data center projects. The facility, backed by Saudi Arabia's Public Investment Fund and operated by a company called Humain, is planning to deploy around 600,000 Nvidia (NVDA) graphics processing units. That's not a typo.
When Zero Revenue Meets Massive Infrastructure
At the U.S.-Saudi Investment Forum in Washington, D.C. on Wednesday, Nvidia (NVDA) CEO Jensen Huang couldn't resist pointing out the absurdity of the situation. "Could you imagine, a startup company approximately 0 billion dollars in revenues, now going to build a data center for Elon," Huang said, presumably with a grin.
That's the kind of thing that only happens in the AI gold rush era. A company with essentially no revenue is constructing one of the planet's largest AI data centers because, well, Elon Musk needs somewhere to train his models. Tesla (TSLA) CEO Musk appeared alongside Huang at the event to discuss the project's scope and ambitions.
xAI Gets First Dibs
Musk confirmed during the forum that xAI will be the facility's inaugural customer. The data center will harness 500 megawatts of Nvidia chips to power xAI's computing needs. For context, that's enough electricity to power a decent-sized city, all dedicated to teaching AI models how to generate increasingly convincing chatbot responses.
In classic Musk fashion, he joked about future expansion plans, noting that a facility 1,000 times larger "would be eight bazillion, trillion dollars." It's the kind of humor that reminds you we're living in an era where the numbers have gotten so big that actual figures sound made up anyway.
Not Just An Nvidia Show
While Nvidia is grabbing the headlines with its 600,000-chip deployment, Humain is spreading its bets across multiple chip manufacturers. Advanced Micro Devices (AMD) will supply its Instinct MI450 GPUs to the project, with power consumption potentially hitting 1 gigawatt by 2030. That's twice the power draw of Nvidia's contribution, suggesting AMD's footprint in this facility could grow substantially over time.
Qualcomm (QCOM) is also joining the party with its AI200 and AI250 data center chips, which will account for 200 megawatts of power. Meanwhile, Cisco (CSCO) will provide additional infrastructure support to keep the whole operation running smoothly.
The diversified approach makes sense from a risk management perspective. Relying on a single chip vendor for a project of this magnitude would be like putting all your eggs in one very expensive, power-hungry basket.
Why This Matters
This project represents more than just another big tech deal. It's a signal of how seriously Middle Eastern sovereign wealth funds are taking the AI infrastructure race. Saudi Arabia is clearly betting that owning the picks and shovels of the AI gold rush—the data centers, the power infrastructure, the physical hardware—will pay dividends as AI demand continues to surge.
For xAI, securing first-customer status at a facility this large gives Musk's AI ambitions serious computational firepower to compete with OpenAI, Anthropic, and Google. When you're trying to build AGI or whatever the goal is this week, you need chips. Lots of them.
And for Nvidia, this deal reinforces its position as the dominant player in AI infrastructure, even as competitors like AMD try to chip away at its market share. The company continues to perform exceptionally well, ranking highly for both growth and quality metrics relative to industry peers.