HomeNews
Share

SpaceX has issued $32 billion in debt to purchase AI chips. A portion of the debt carries a yield of over 8%.

Yana Zakomoldina

Yana Zakomoldina

Reporter
SpaceX issued $32 billion in debt securities to purchase AI chips / Photo: Wirestock Creators/Shutterstock

SpaceX issued $32 billion in debt securities to purchase AI chips / Photo: Wirestock Creators/Shutterstock

SpaceX, Elon Musk’s aerospace company, has issued $32 billion in debt securities as part of a total financing package worth over $40 billion to purchase Nvidia AI chips, Bloomberg reports, citing unnamed sources. The chips will formally be purchased by a specially created entity, Saturn Funding Trust, managed by the Apollo division, and SpaceX will lease them for approximately five years. This arrangement is necessary to ensure that the debt does not appear on SpaceX’s balance sheet, the agency notes.

In premarket trading on November 20, the company's shares are down about 6%.

What is known about the deal

According to sources who spoke with Bloomberg, the debt securities were sold on November 19 in three tranches, with Goldman Sachs and Apollo Global Management acting as lead managers for the transaction. Representatives from SpaceX, Goldman Sachs, and Apollo did not respond to the agency’s inquiries.

The largest tranche—the so-called Class A3 securities, totaling approximately $25.2 billion—was sold at a yield of 8.259% per annum. This is significantly higher than the average yield on high-quality U.S. corporate bonds (5.95%, according to Bloomberg indices). In addition, the new bonds offer a significant yield premium over SpaceX’s previous bonds, which were sold in June. The prices of those bonds subsequently plummeted, sending shockwaves through the credit market, Bloomberg notes.

The funds raised—including bonds and bank loans—will be used to pay for Nvidia’s latest Rubin-architecture chips. According to previously published documents, the equipment in question is valued at approximately $47.3 billion, and deliveries are set to begin as early as January. The formal purchaser of the chips will not be SpaceX, but rather Saturn Funding Trust—a special-purpose entity created for the transaction and managed by one of Apollo’s subsidiaries. Musk’s company itself will lease these chips for approximately five years. In doing so, SpaceX is taking on strict financial obligations: lease payments cannot be suspended even in the event of equipment failure or force majeure.

This off-balance-sheet structure should help SpaceX maintain its investment-grade credit rating: it allows the company to borrow more and faster, which is critical in the AI race.

News that SpaceX was seeking to raise $40 billion to purchase Nvidia chips broke in early October. According to the Financial Times, the deal was intended to strengthen ties between SpaceX and Nvidia after Musk pledged to increase his investment in the company’s technology for his AI initiatives.

What Analysts Are Saying

Leading players in the AI industry have already raised hundreds of billions of dollars in the U.S. investment-grade bond market this year, according to Bloomberg. Barclays analysts predict that the credit boom will continue into 2027, but the composition of the loans will change. The share of shorter-term debt tied to chips and data centers will grow, as the unsecured debt market for hyperscalers—the largest owners of cloud data centers—reaches its capacity limits. “The unsecured debt market is close to the limit of what it can absorb in a year,” Barclays analysts write.

Companies will need about $1.5 trillion to purchase AI chips, and deals to finance such purchases are likely to become more common, said Scott Goodwin, co-founder of the hedge fund Diameter Capital Partners: “This is a new asset class,” he noted.

However, after investors lost money on SpaceX’s June bonds, some clients have become extremely skeptical and are viewing the new deal with caution, notes Davis Herbert, head of the media and telecommunications technology division Davis Herbert, head of the technology, media, and telecommunications division at the research firm CreditSights. According to him, investors had to offer higher yields to attract more traditional bond buyers to the deal.

In addition, Bloomberg notes that it is still difficult to analyze this type of arrangement involving a specially created company for buyers of investment-grade bonds. Financing deals for the purchase of graphics processing units (GPUs) for AI are a relatively new phenomenon, and the risks associated with the chips’ lifespan and cost are still being assessed.

This article was AI-translated and verified by a human editor

Share

Trending

Stock Screener
Buy
Sell
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
‌
Small Caps
Investment and Finance News