1 hour ago
25
SpaceX just held its first earnings call as a public company, and Elon Musk used the occasion to draw a line in the semiconductor sand. The company will build its entire AI infrastructure exclusively on Nvidia hardware, Musk declared, citing the superiority of Nvidia’s Vera Rubin architecture for processing AI workloads.
AMD investors felt the impact immediately. Shares dropped roughly 9% following the announcement, even though AMD had just reported record quarterly revenues of $11.5 billion, up 50% year-over-year. Nvidia, meanwhile, ticked up between 2% and 3.4% on the news.
SpaceX’s AI business is growing absurdly fast
Q2 2026 revenue hit $7.81 billion, a 92% year-over-year increase. The AI segment is where the numbers get genuinely startling. AI-related revenue reached $2.56 billion in the quarter, representing a 247% jump compared to previous figures. That means roughly one out of every three dollars SpaceX earned came from AI, not rockets or satellite internet.
The company has already installed 1.4 gigawatts of compute capacity as of June 2026. Its target: surpass 2 gigawatts by the end of this year and scale to a staggering 10 gigawatts by the end of 2027.
The Starmind program takes AI off-planet
Perhaps the most unusual detail from the call was the announcement of Starmind, a collaboration between SpaceX and Nvidia to deploy Vera Rubin-based compute payloads aboard satellites in low Earth orbit. The program is scheduled to begin launches in 2027.
Nvidia’s role in the partnership extends beyond simply selling chips. The company is also a significant shareholder in SpaceX, making this less of a simple vendor relationship and more of a strategic alliance with aligned financial incentives.
What this means for the chip wars
AMD’s position is particularly awkward. The company posted genuinely strong Q2 numbers, growing revenue by 50% year-over-year to $11.5 billion. Instead, the market punished the stock because a high-profile customer publicly chose the other side.
For Nvidia, SpaceX represents a customer that is growing its AI spending at triple-digit percentages annually. Guaranteed orders from a company targeting 10 gigawatts of compute capacity provide revenue visibility that most chipmakers would envy. The Starmind collaboration also opens an entirely new market category—space-based AI infrastructure—where Nvidia could establish a first-mover advantage with virtually no competition.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·