Skip to main content

SpaceX discusses $40 billion financing for Nvidia chips

8 OCTOBER 2026·2 MIN READ·3 SOURCES

SpaceX is discussing financing for a reported $40 billion purchase of Nvidia GPUs, CNBC reported, citing an earlier Financial Times report. The talks are preliminary, and key details have not been settled.

SpaceX discusses $40 billion financing for Nvidia chips

Key takeaways · 4

  • 01

    CNBC reported that Apollo Capital Management and several banks were discussing financing a $40 billion Nvidia GPU purchase by SpaceX.

  • 02

    The financing discussions are preliminary, and the lenders and other key details have not been finalized.

  • 03

    CNBC said the plan would rely primarily on investment-grade debt; The Inference reported a possible $10 billion in bank loans and $30 billion in investment-grade debt.

  • 04

    The Inference reported that the proposed financing would not close until 2027.

Financing talks remain preliminary

CNBC reported that Apollo Capital Management and several banks were discussing financing SpaceX’s planned $40 billion Nvidia GPU purchase.[1] CNBC said the discussions were preliminary and that the lenders and other key details had not been finalized.[1] It reported that Apollo was playing a leading role in facilitating the financing, while Apollo declined to comment.[1] Nvidia did not immediately respond to a request for comment, CNBC reported.[1] CNBC said the Financial Times was the first to report the talks.[1]

Debt structure and timing are unsettled

CNBC reported that the financing would rely primarily on investment-grade debt and that Nvidia GPUs would likely serve as collateral.[1] The Inference reported a possible package of about $10 billion in bank loans and $30 billion in investment-grade debt.[3] That breakdown is a reported possibility, not a finalized structure: CNBC said key details had yet to be settled.[1] The Inference also reported that the financing would not close until 2027.[3] For teams assessing the proposal, the reported timing and structure should therefore be treated as provisional.

A different debt-market backdrop

SpaceX raised $25 billion in bonds two weeks after its mid-June IPO, and CNBC reported that the sale drew very high demand from fixed-income investors.[1] Since that sale, AI-related bonds have sold off and credit spreads have widened, CNBC reported.[1] Fixed-income investors have been demanding higher yields for AI-related debt, a sign of greater selectivity about AI exposure, according to CNBC.[1] SpaceX had reported nearly $40 billion of debt in its second-quarter financials, and CNBC said its existing debt was trading near junk-bond levels.[1] Those conditions make the proposed financing’s final terms worth watching.

The chips support a growing compute buildout

Elon Musk said SpaceX’s Colossus data center in Memphis was running 230,000 Nvidia AI chips, while Colossus 2 had 550,000.[1] He said SpaceX planned to more than double its chip count to 1.21 million by late December, if it got lucky.[1] CNBC reported that SpaceX had compute deals with Google, Anthropic, Reflection AI and Cursor, which SpaceX later bought.[1] The market expected GPUs to retain their value for about seven years, and CNBC reported that scarce computing power supported the value of high-performing GPUs.[1] These factors may inform how lenders assess the proposed collateral, but the reported financing terms remain unsettled.[1]

For finance and infrastructure teams, the reported deal links a large AI hardware purchase to debt markets that CNBC says have become more selective about AI exposure. The eventual financing structure, collateral and timing could matter when evaluating the cost and risk of compute expansion.

Why it matters
Daily session

Put this to work — one session a day, built for your industry.

Create a free account for a daily session — eight questions and one real-work challenge, on the news that affects your role.

Start free

How this developed

  1. 8 October 2026

    SpaceX discusses $40 billion financing for Nvidia chips

Sources

AI fluency, one session a day, built for your work.