SpaceX's Default Insurance Hit a Record on a $40 Billion Chip Loan That Nobody Has Agreed. What a 195 bp CDS Reading Can and Cannot Show

On Wednesday 7 October 2026, the price of insuring SpaceX’s debt against default reached its highest level since the contracts began trading actively in June. According to Bloomberg, citing ICE Data Services, five-year credit default swaps on SpaceX rose as much as 14.5 basis points to about 195.4 basis points a year. The same day, Bloomberg reported that SpaceX is seeking about $40 billion of financing, and the Financial Times reported that the money would buy Nvidia chips, with Apollo Global Management expected to lead. Both reports rely on unnamed sources, and Bloomberg said the talks are at an early stage and could end without a deal. A record reading on a financing nobody has agreed is easy to over-read. This piece explains what a CDS price measures, why new borrowing pushes up the cost of old debt, and how little the reports establish.

What was reported

• Size and split: about $40 billion in total, roughly $10 billion of bank loans and $30 billion of investment-grade debt, per the Financial Times as summarised by investingLive. Whether ‘investment-grade’ means debt that would be rated that way or debt sold to investment-grade buyers is not clear in the reports.
• Purpose: to pay for a large order of Nvidia AI chips for SpaceX’s computing needs, including its Grok models running in Colossus data centres. Elon Musk has said those facilities will use only Nvidia hardware (investingLive).
• Lenders and timing: Apollo is expected to lead and place the debt with a broad group of investors, with Pimco among a small group of lenders in talks. Closing is not expected until 2027. Pimco declined to comment; SpaceX, Apollo and Nvidia did not immediately comment.
• Caveat from Bloomberg: the talks are early and could end without a deal.

Nothing in these accounts says the financing has been agreed, or what security, if any, lenders would hold. SpaceX shares fell about 1% in after-hours trading and Nvidia rose about 0.5%, according to investingLive, hardly a dramatic response.

How to read a credit default swap

A CDS is a contract in which one party pays a regular fee in exchange for compensation if a borrower defaults. The fee is quoted in basis points of the insured amount per year. At 195.4 basis points, insuring $10 million of SpaceX debt for five years costs about $195,400 a year, or 1.954%. A 14.5 basis point rise means the same cover costs $14,500 a year more than the day before. Those are our conversions of the quoted figures.

Two things limit what the number can tell you. First, the SpaceX CDS market is young. The contracts have traded actively only since June, so ‘highest since active trading began’ describes roughly four months of history, not a long record. Second, a thin market can move on small flows. A single large buyer of protection, for example a bank hedging an expected loan commitment, can lift the price without any change in the borrower’s likelihood of default. That is a plausible explanation for the move, not one the reports state.

For scale, spreads of this size are more typical of companies at or below the boundary of investment grade than of the strongest borrowers. That is general market background rather than anything in the sources, and it does not tell us how rating agencies view SpaceX. It sits a little awkwardly beside a report that $30 billion of the new money would be investment-grade debt, which is another reason to treat the details as unsettled.

What new supply does to old bonds

The bond market shows the same thing from the other side. SpaceX sold a 6.65% bond due 2056 in June as part of a $25 billion debt deal, at a spread of 175 basis points, per Trace data cited by Bloomberg. On 7 October, at 9:40 a.m. New York time, that spread was 238 basis points, up 12 basis points on the day. The 63-basis-point difference (238 − 175) is our calculation; the reports give both numbers but not the gap. The reports also do not say what the spread is measured against, which for a corporate bond is normally a government benchmark.

Why would a plan to borrow $40 billion more make existing bonds cheaper? There are three mechanisms, and the reports do not tell us which dominates. The first is supply: investors who hold a borrower’s bonds must absorb more of its paper, so they demand a higher yield to take on additional exposure to the same name. The second is seniority: if the new loans are secured, or rank ahead of existing bonds, the old bonds’ claim on assets is diluted. No collateral terms have been reported, which is the biggest unknown. The third is leverage: a company borrowing $40 billion on top of $25 billion raised in June is becoming more indebted, and the market prices that in. For a 30-year bond a 63-basis-point spread move is material. Using an illustrative duration of about 12 years, typical for a long bond with this coupon, that widening alone would cut the bond’s price by about 7.5%. We have not seen the bond’s actual price, so treat that as an order of magnitude.

The same pressure is visible beyond SpaceX. Sal Naro, chief investment officer at Coherence Credit Strategies, told Bloomberg: ‘This is unprecedented debt supply with no real ending in sight.’ The investingLive summary cites Morgan Stanley’s estimate that AI infrastructure needs $1.5 trillion of external financing by 2028, and notes Goldman Sachs has linked AI-related corporate issuance to higher US Treasury yields. Those are estimates and views, not measurements, but they frame why one name’s spreads might respond to a single headline.

The counter-argument: this is a modest move

The sceptical reading is that nothing very alarming has happened. A 14.5 basis point change on a 195 basis point base is a rise of about 8%, which a thin market can deliver in a day, and the bond spread widened by 12 basis points, less than the CDS. If the borrower is raising $40 billion of debt that lenders are willing to buy, a modest widening is what one would expect as the market digests supply, and the financing itself would be a sign that lenders are willing to fund the build-out. The investingLive account notes that Nvidia’s muted share reaction may mean the demand boost is already priced in. On this reading, the record CDS is a statement about the size of the borrowing, not a distress signal.

The opposing reading is that spreads are the market’s verdict on circular financing: debt raised to buy chips from a supplier, to run models whose revenue is not yet proven. The reports we reviewed do not support either reading, because they contain no collateral terms, no revenue data and no pricing for the loan itself.

What the evidence does not establish

• Whether any financing is agreed. Bloomberg says talks could end without a deal; the Financial Times says closing is not expected before 2027.
• The terms: interest rates, collateral, covenants, and the identity of the lenders beyond Apollo and Pimco. None has been reported.
• What drove the CDS move. The contracts are thinly traded and the reports do not say who bought protection.
• A causal link between the financing reports and the spread moves. They happened on the same day, which is suggestive, not proof.
• Figures from other outlets. A newsletter printed a CDS level of 194 basis points; we use 195.4 as reported by Bloomberg from ICE Data Services, and note it is an intraday high rather than a close. A figure of 244 basis points circulated for Oracle’s five-year CDS, which we could not confirm and do not use.