SpaceX CEO Elon Musk unveiled a detailed mock‑up of an AI data‑center satellite, called AI1, during a 30‑minute video on X. The design features 230‑foot solar arrays and an average compute capacity of 120 kW, peaking at 150 kW. Musk tied the project to a planned 1 million satellites network for future AI in orbit operations soon to scale to launch.

SpaceX will seed the first nodes with Nvidia chips, CFO Bret Johnsen said. Musk also highlighted a 11‑million‑square‑foot Gigasat complex in Bastrop, Texas, to fabricate the massive panels, and a forthcoming 100‑million‑square‑foot Terafab that would eclipse Tesla’s Austin gigafactory. These facilities aim to support the satellite data‑center rollout for the company’s IPO push and AI.

SpaceX’s push into orbital compute feeds its broader IPO strategy, positioning the firm as a supplier for AI workloads. The company has signed deals with Anthropic and Google to lease its chips and data centers, while keeping its own chatbot, Grok, in development. Investors view the satellite network as a new revenue stream that could boost valuation for the market.

With the satellite architecture outlined, SpaceX is tightening its supply chain by building Gigasat and Terafab on U.S. soil, reducing reliance on foreign manufacturing. This vertical integration could lower costs for the 1 million‑satellite goal and strengthen the company’s bargaining power with chip suppliers, solidifying its edge in the emerging AI infrastructure race for the future.