Reported $40 Billion Debt Plan Would Fund SpaceX's Nvidia Buildout
The financing — split between roughly $10 billion in bank loans and $30 billion in bond debt, with Apollo expected to lead — is reported by secondary sources with inconsistent attribution to Bloomberg and the Financial Times. It remains at the discussion stage, and the companies are staying silent. Still, the story is a case study in how a newly public, loss-making AI infrastructure company finances its capital buildout: in Q2 2026, revenue was $7.81 billion against capital expenditures of $18.37 billion.
What has been reported — and what is not confirmed
On October 7, 2026, multiple outlets — Yahoo Finance citing Bloomberg, Semafor, 24/7 Wall St., and The Motley Fool — reported that SpaceX is seeking roughly $40 billion in financing to pay for an order of AI chips from Nvidia. According to 24/7 Wall St., the plan is split in two: about $10 billion in bank loans and around $30 billion in investment-grade bond debt. Apollo Global Management is expected to lead the loan package, and Pimco (Pacific Investment Management Company) is reportedly in negotiations among the lenders.
It is important to stress what this actually is: reporting based on anonymous sources, at an early stage. 24/7 Wall St. writes that the financing "remains at the discussion stage" and that it is expected to close in 2027. Neither SpaceX, Apollo, nor Nvidia responded to requests for comment. There is no primary documentation — no press release, no regulatory filing, no company statement — in the available source material.
The attribution is also inconsistent. The Motley Fool (in Yahoo syndication) attributes the original reporting to the Financial Times, while Yahoo Finance and other Motley Fool articles cite Bloomberg. Neither of the original articles is in the source set. What we can say with certainty is that several independent news channels reported the same story the same day with largely matching details — and that the market reacted: the stock fell around 2% in early trading on Wednesday, October 7, the same day the reports emerged, according to Yahoo Finance (source, c57faa99).
The funding gap in numbers
Why does a company with growing AI contracts need to borrow $40 billion? The Q2 2026 figures explain it. SpaceX reported revenue of $7.81 billion against capital expenditures of $18.37 billion — of which $15.83 billion went to AI. The company booked a net loss of $541 million (24/7 Wall St., e6539c5f).
In other words: capital expenditures are more than double revenue, and the gap is not filled by operations. SpaceX has real cash flows from Starlink and its launch business, but Semafor points out that Starlink is no "cash cow" in the class of its rivals' advertising businesses. The Semafor analysis, written by Liz Hoffman, claims that SpaceX carries five times more debt than Alphabet and three times more than Meta, relative to their respective earnings (Semafor, e9a47ea2). These comparison figures should be read as analytical judgments — Semafor does not show the methodology behind them — but the point they support is confirmed by the published quarterly numbers: the buildout is running faster than the company's own earnings allow.
Management signaled at the August 4 earnings presentation that the next two quarters would look "very similar" on the spending side, according to 24/7 Wall St., citing the presentation. The spending pressure is thus not planned to abate.
The revenue side: the contracts meant to pay the bill
The counterpart to the spending is a rapidly growing portfolio of AI compute contracts. According to The Motley Fool's recap of company disclosures and reported deals (source, db22b473), the customer list includes:
- Anthropic: $1.25 billion per month for access to roughly 325,000 Nvidia GPUs.
- Google: $920 million per month for around 110,000 GPUs.
- Reflection AI, a startup backed by Nvidia: a deal worth $150 million per month for GB300 chips in SpaceX's Colossus 2 datacenter.
These figures are not independently verified in this material — they come from company disclosures and press as relayed by The Motley Fool — but they sketch the business model: SpaceX builds datacenters, fills them with Nvidia's fastest chips, and leases the capacity to some of the world's largest AI players on monthly contracts.
CFO Bret Johnsen added a new deal in September. "We're on our way, or we believe we're on our way, to reaching $100 billion in ARR," Johnsen said of the company's target for annual recurring revenue, according to Yahoo Finance. He added: "An update is that earlier this month we closed a new hosting deal, which equates to roughly $1.11 billion per month starting December 1 of this year, so about an additional $13 billion in ARR" (Yahoo Finance, c57faa99; quotes translated to Norwegian in the original edition).
The arithmetic reveals the paradox: the revenue contracts are large and growing, but they pay out over time — while the GPUs and datacenters must be paid for now. Hence the debt.
Nvidia exclusivity as a deliberate choice
The financing need is tied to a strategic decision Elon Musk explained at the August 4 earnings presentation: "And going forward, we have decided to build exclusively on Nvidia, because we believe the Vera Rubin architecture is the best architecture. We believe it is the best AI computer, and we greatly value our close collaboration and partnership on many levels with Nvidia. So we are exclusive to Nvidia" (The Motley Fool via Yahoo Finance, c0784bd9; quote translated to Norwegian in the original edition).
This means the entire capital base — and thus the entire debt burden — is tied to one chip vendor and one generation of architecture. It offers operational simplicity and a close partnership with Nvidia, but also concentration risk: $40 billion in new debt is financing equipment whose value depends on Vera Rubin remaining competitive through the depreciation cycle.
The credit mechanics: why Apollo and Pimco
The loan structure itself reveals a great deal about how this kind of financing works in practice. SpaceX secured a BBB credit rating shortly after going public earlier this year, according to The Motley Fool (c0784bd9). BBB sits in the lower tier of the investment-grade category — and, per Semafor, two notches below Meta and Alphabet, making the debt "a harder sell for blue-chip bond funds." Their conclusion: "enter Apollo, which has a pool of money to match apparently every risk out there" (e9a47ea2, translated).
That explains the lender mix: a large share is raised as classic bank loans (the $10 billion), while the bond portion (the $30 billion) must largely be matched with private credit funds and specialized institutional investors like Apollo and Pimco — capital that can price BBB risk at a four-month-old, listed company running a deficit.
The analyst Dan Ives, a prominent SpaceX bull on Wall Street, described the deal as "firepower" — "firepower for [SpaceX's] AI buildout" — even as the debt picture grows, according to Yahoo Finance (c57faa99). That is the correct reading in both directions: the financing provides headroom, but each new loan increases the financial leverage a company with negative cash flows is already balancing on.
A four-month-old public company squeezed between two stories
The context is remarkable. SpaceX listed on June 11, 2026 at $135 per share, peaked at $225.64, and closed at $171.92 on October 6 — still well above the introduction price, but far below the peak (24/7 Wall St., e6539c5f). The stock fell around 2% in early trading on October 7, the day after that close and the same day the reports of the debt plan emerged.
At the same time, the lockup periods are beginning to expire: 328.4 million eligible shares are released on October 9, another 328.4 million on October 24, and up to 1.3 billion shares after the third-quarter report (e6539c5f). That means the company is simultaneously managing a significant supply shock in its stock and a massive debt raise — two financing channels that both depend on market confidence.
What we don't know — and what to watch
Let's be honest about the uncertainty. None of the companies has confirmed anything. Every detail — the size, the split between bank loans and bonds, Apollo's lead role, Pimco's involvement, the 2027 timeline — comes from anonymous sources relayed through secondary reporting. The terms could change, shrink, or disappear entirely before an expected close in 2027. Nor are the Anthropic, Google, and Reflection AI contract figures independently verified in this material, and Semafor's debt comparisons are analytical claims without shown methodology.
The concrete markers to follow:
- The lender group. Are Apollo's lead role and Pimco's participation confirmed through official mandates or ratings correspondence?
- Final size and terms. Does the reported 10/30 split change? What coupon will a BBB-rated company at this growth pace have to pay?
- Ratings commentary. One or more ratings agencies' assessment of the debt raise will say more about sustainability than any analyst note.
- The third quarter. The earnings report, which also triggers the largest lockup release of up to 1.3 billion shares, will show whether the ARR curve is actually stretching toward Johnsen's $100 billion target.
The big picture is clear regardless: AI infrastructure building has become a debt-financed industry. A company that listed four months ago is seeking $40 billion to buy chips from a single vendor, leased out to customers who pay over time — with a credit rating two notches below the tech giants and lenders willing to take the risk. Whether this specific deal closes in the reported form or not, the model it is about is here to stay.