SpaceX picks Nvidia over AMD, but AMD's data center revenue surges 107%

SpaceX's choice of Nvidia represents a real competitive vulnerability for AMD: Nvidia's control over chip allocation capacity gives it the ability to convert scarcity into exclusivity with the largest AI spenders, a mechanism AMD cannot yet replicate. This is especially stinging for Lisa Su as she attempts to convince Wall Street that her Instinct roadmap can rival Jensen Huang's platform. The timing compounds the pain—the deal lands exactly when AMD is trying to establish momentum in the hyperscale AI market.

Yet the framing of SpaceX as evidence of AMD's collapse misreads the underlying data. AMD's Q2 results show the company won the broader cycle: data center revenue jumped 107% year over year to $6.72 billion, and the segment swung to $2.1 billion in operating income from a loss a year ago. Crucially, AMD has already locked in multi-gigawatt commitments from the companies actually building frontier models—OpenAI, Meta, Anthropic, and Microsoft—before SpaceX became an issue. SpaceX, by contrast, is a single-tenant buyer with a specific alignment to Musk's xAI, a profile that does not generalize to the rest of the market.

Management's confidence in the forward path reinforces this reading: Su told analysts the company enters the second half with strong momentum, and rather than trimming guidance, AMD raised its long-term targets. Data Center is expected to more than double year over year in 2027. Nvidia retains a material structural advantage in allocation politics, and AMD's forward P/E of 65 reflects the assumption that advantage will narrow. But the loss of one customer—however high-profile—inside a market AMD believes will approach $2 trillion by 2030 is a setback, not a collapse.