By: Alex Mercer – SeaPRwire – SpaceX is raising debt to buy Nvidia silicon. Forty billion dollars, led by Apollo. Ten billion in bank loans, thirty billion in investment-grade paper. Closing targeted for 2027. The money is earmarked for chips, not launches. That shift turns AI hardware into a multi-year balance-sheet commitment.
According to a Financial Times report, SpaceX is seeking to complete a forty-billion-dollar financing package led by Apollo Global Management. The funds are intended to pay for Nvidia chips. The structure is approximately ten billion dollars in bank loans and thirty billion dollars in investment-grade debt. The transaction is expected to close in 2027. After the report circulated, SpaceX shares in New York after-hours trading gave back earlier gains and fell as much as 1.8 percent. Nvidia shares rose as much as 0.6 percent. Nvidia, SpaceX and Apollo declined to comment. Last month Elon Musk stated that the Colossus 2 AI compute cluster operated by xAI could more than double its Nvidia chip count by the end of the year. He provided a detailed timeline for the Memphis expansion. On the social platform X he wrote that Colossus 2 currently holds 110,000 Nvidia GB200 chips and 440,000 GB300 chips. An additional 220,000 GB300 chips were due to come online, followed by another 220,000 in November and, if all goes well, a further 220,000 in December. In August Nvidia reached agreements with several major U.S. investment institutions, including Apollo, to create dedicated financing platforms for its customers. The initiative aims to mobilise more than five hundred billion dollars of third-party capital so that cloud providers, frontier AI laboratories and other enterprises can expand data centres and purchase Nvidia hardware. The stated goal is five hundred billion dollars of financing for AI infrastructure and a potential change in how that infrastructure is funded.
The commercial test is delivery versus debt service. If the chips arrive and are fully utilised before the 2027 obligations begin to amortise, the structure works as intended. If utilisation lags or deliveries slip, the investment-grade paper becomes a fixed cost against idle silicon. Watch the actual chip counts reported at Colossus-scale sites in the months leading up to the expected close. Those numbers will show whether the financing is matched by working hardware.
Author bio: Alex Mercer, senior commentator for an international technology weekly covering AI infrastructure financing and large-scale compute deployment.