TradingKey - On June 12 ET, SpaceX (SPCX) officially debuted on the Nasdaq. It closed up 19% on its first day of trading, with a closing market capitalization reaching $2.11 trillion, ranking it sixth in the U.S. stock market and making Elon Musk the first person in history to surpass a $1 trillion net worth.
On the same day, Nvidia (NVDA) posted on X to congratulate SpaceX on its "historic IPO debut," with Musk responding that he hopes to take their "exciting collaboration" with Nvidia to new levels. Nvidia stated that its partnership with SpaceX has spanned nearly a decade; from delivering the world's first NVIDIA DGX-1 supercomputer in 2016 to today's custom DGX Spark, Nvidia remains committed to pushing the limits of accelerated computing to power the future of space exploration.
According to NVIDIA, the delivery of the world’s first NVIDIA DGX-1 supercomputer to Elon Musk in 2016 marked the inception of its collaboration with SpaceX. At the time, however, the machine was nominally delivered to OpenAI—then a non-profit organization that Musk had not yet exited and where he oversaw research operations.
Building on this foundation, the two companies subsequently expanded their partnership, with NVIDIA GPUs now serving as the physical backbone of SpaceX’s data centers. This is underscored by recent computing power agreements between SpaceX and its partners: according to a cloud services agreement signed with Google on the eve of SpaceX’s IPO, the former will provide the latter with computing power equivalent to "110,000 NVIDIA GPUs, CPUs, memory, and related components"; similarly, SpaceX’s agreement with Anthropic mandates that the former provide access to "220,000 NVIDIA GPUs."
According to reports from tech outlet Wccftech, the direction in which Musk seeks to deepen cooperation with NVIDIA may be linked to his recently disclosed "AI1" satellite venture—SpaceX’s first satellite design specifically dedicated to AI computing.
In short, AI1 differs from SpaceX’s previous core businesses; it is not a standard Starlink network device. According to Musk’s descriptions, the design is even “simpler” than a Starlink satellite, eschewing complex broadband phased-array communication antennas in favor of larger solar arrays, thermal management panels, and modular computing payload bays. Each AI1 satellite supports a sustained computing load of 120 kilowatts, with peak capacity reaching 150 kilowatts. Functionally, the satellite effectively serves as a miniaturized AI data center. Given NVIDIA’s position as the world’s preeminent AI chip supplier, SpaceX is likely to pursue deeper collaboration with the company in this domain.
As a partner of SpaceX for a decade, SpaceX's public listing will undoubtedly benefit NVIDIA. Although NVIDIA does not hold a significant stake in SpaceX, their business relationship is expected to provide NVIDIA with a new growth engine—the computing power foundation for space data centers.
While Terafab, the self-developed chip project between SpaceX and Tesla, is accelerating, as Jensen Huang noted, Elon Musk may have underestimated the barriers to entry in chip manufacturing; SpaceX will remain unable to move away from NVIDIA chips for the foreseeable future. As SpaceX expands its space data center operations, its demand for NVIDIA chips is expected to grow significantly.
Following the successful launch of its IPO, SpaceX will gain access to ultra-low-cost financing channels. Although it is still burning through cash, SpaceX will no longer need to worry about cash flow and can more aggressively increase capital expenditure, purchasing NVIDIA chips on a larger scale.
Furthermore, as NVIDIA ties its chip R&D to SpaceX and captures market share in space infrastructure, it will establish a monopolistic advantage over other chip companies. Consequently, the space chip businesses of AMD and Google will be further squeezed, widening the gap with NVIDIA.