SpaceX pledges Nvidia exclusivity

Elon Musk said on SpaceX’s August 5 earnings call that the company will buy GPUs only from Nvidia, giving the chipmaker an exclusive position in SpaceX’s future AI infrastructure purchases. Musk said SpaceX had decided to “build exclusively on Nvidia” because it viewed Nvidia’s Vera Rubin architecture as the best option for its compute needs. Business Insider reported the comments from the call, and a Yahoo syndication of the report carried the same remarks. The statement lands as AMD is trying to frame the next phase of AI infrastructure competition around openness rather than lock-in. On AMD’s August 4 earnings call, executives said the company’s open approach to software was a competitive advantage against Nvidia, according to Business Insider. DigiTimes reported on August 5 that AMD and Nvidia are pursuing different paths into the same spending cycle, with AMD emphasizing rack-scale systems for a smaller set of hyperscalers and frontier AI labs while Nvidia broadens its platform reach. (businessinsider.com) ### What exactly did Musk say about SpaceX’s GPU purchases? Musk said SpaceX had “decided to build exclusively on Nvidia,” according to Business Insider’s account of the August 5 earnings call. He said the company believed Nvidia’s Vera Rubin architecture was “the best architecture” and called it “the best AI computer,” adding that SpaceX valued its cooperation and partnership with Nvidia. (businessinsider.com) The wording matters because it goes beyond a preference for one generation of chips. Musk described an ongoing procurement stance, not a one-off order, and Business Insider characterized the relationship as effectively exclusive. ### Why does AMD keep talking about “open” software? AMD executives said on their August 4 earnings call that openness is one of the company’s clearest arguments against Nvidia’s more integrated stack, Business Insider reported. (businessinsider.com) The company’s pitch is that open-source software gives customers more flexibility in how they build and run AI systems. DigiTimes said AMD is pairing that software message with a rack-scale systems strategy aimed at large buyers such as hyperscalers and frontier labs. Nvidia, by contrast, is extending its position beyond chips into a broader platform, DigiTimes reported. ### Why is this bigger than one SpaceX supplier decision? SpaceX is one buyer, but it is a visible one, and Musk’s comments show that some large AI customers are still willing to deepen dependence on Nvidia even as rivals argue for more optionality. (businessinsider.com) That matters because the more compute, networking and software are tied to one vendor, the harder it can be for customers to switch parts of the stack later. That reading is an inference from the procurement stance Musk described and AMD’s competing argument about openness. (digitimes.com) Nvidia’s advantage in these accounts is not presented only as raw chip performance. Musk pointed to architecture and partnership, while DigiTimes described Nvidia as expanding a broader platform moat. ### What should readers watch next? AMD reported second-quarter 2026 revenue of $11.5 billion on August 4, according to its earnings release carried by Markets Insider, giving investors a near-term benchmark for whether its AI strategy is translating into growth. (businessinsider.com) The next test for this story is whether more large buyers publicly choose Nvidia-only deployments or whether AMD wins additional rack-scale system deals with hyperscalers and frontier labs. (finance.yahoo.com) (markets.businessinsider.com)