THE POST
Pretty much. As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers. SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue. That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
REPLIED TO
Chamath Palihapitiya@chamath
As someone with a portfolio of energized land assets here is what I am seeing: 1. Neoclouds are disorganized and poorly capitalized. Some are definitely better than others however. 2. Hyperscalers are intractable, bureaucratic organizations that take 3 months to do an NDA and want 18mo escrow periods. We built our portfolio by competing against these terms because of our speed and simplicity. The more future assets orgs like ours win, the bigger the problem for the hyperscalers downstream (see 3). 3. There are delays in the existing pipeline that will catch people off guard in 2027. Hyperacalers will (obviously) react by focusing on their own internal compute needs over any third party. If the third party is a model company, they will tell them to go through their model router where they will also take a non-trivial vig. Them telling Ant they have no compute for them (rumored in Ant’s S-1) if true is a huge tell that this is already happening. All of this is to say that it does look like there will be a lot of partially stranded compute and compute sitting in inventory in 2027 than is currently forecasted. This is bought and paid for but effectively underearning (below total cost of capital) or collecting dust. We see this as true for everyone - model companies, hyperscalers, neoclouds except ONE (you can guess which one). This dislocation will benefit two kinds of orgs: the non-bureaucratic big scalers who can build industrial scale quickly and the nimble, well capitalized players feeding them. 2027 will be a turbulent year because it will clear out a bunch of wrong/bad bets. It will also be a consolidation year for a few.
https://x.com/chamath/status/2114389987563381004
ENGAGEMENT
- likes33,071
- reposts3,183
- replies1,611
- quotes1,980
- bookmarks9,674
- views8,681,723
AS RETURNED BY THE X API · FETCHED 2026-10-26 06:19 UTC
WHAT IT SAYS
Agreeing with Chamath Palihapitiya, who expects a lot of partially stranded compute in 2027, Musk says that in the past few weeks debt markets have almost entirely stopped lending to companies whose revenue depends on AI data centres. He says SpaceX closed $40 billion of debt for GPUs because it is investment grade and earns money from Starlink and launches, and predicts that most of the 200-plus neoclouds will fail or be bought. In a follow-up reply he said the money is mostly callable debt for GPUs arriving in the first half of 2027.
CONTEXT
The credit stress at neoclouds is visible in market data. CNBC reported on 21 October that CoreWeave's stock was down 36% for the month, its unsecured debt yielded about 15% and its five-year credit default swap spread stood at 1,145 basis points. Nebius and IREN shares fell 33% and 38% in October, Firmus pulled its IPO, and credit-desk sources said the high-yield market is effectively on hold until after the US midterms because of political pushback against data centres. Goldman, JPMorgan and Barclays analysts still published bullish notes on CoreWeave, pointing to its investment-grade customers.
CNBC confirmed the $40 billion figure from Musk's posts and noted earlier Bloomberg reporting that the deal may sit in a special purpose vehicle to keep it off SpaceX's balance sheet. SpaceX's AI segment brought in $7.56 billion in the third quarter, up from $813 million a year earlier.
On its 20 October earnings call SpaceX guided fourth-quarter capital spending to $35–40 billion, against $19.6 billion in the third quarter. It ended September with 1.7 GW of nameplate compute and expects 2.5 to 3 GW by the end of the year. The CFO said new compute pays back in under a year at current rental prices, but SpaceX rents at spot rates, so a fall in GPU prices would hit revenue quickly.
WHAT IT MEANS FOR K
Money now flows to builders with revenue outside AI, and away from the long tail of GPU renters. That can concentrate compute into fewer, larger operators that can afford to bring their own power, but it also cancels or delays some capacity. Either way, K only moves when watts are actually energised.
CLAIMS
- Debt markets have almost entirely shut off for companies without substantial revenue independent of AI data centres.
- SpaceX closed $40 billion of debt for GPUs, mostly callable, for chips arriving in Q1 and Q2 2027.
- Most of the 200-plus neoclouds will fail or consolidate.
EVIDENCE
- CoreWeave's unsecured debt yields about 15% and its CDS spread is 1,145 bp; Nebius and IREN shares fell 33% and 38% in October (CNBC).
- Credit-desk sources told CNBC the high-yield market is on hold until after the midterms.
- CNBC reported the $40 billion SpaceX financing from Musk's posts and Bloomberg's earlier reporting.
- SpaceX expects to go from 1.7 GW to 2.5–3 GW of nameplate compute by year end.
- Observatory found no count of neocloud failures so far; the consolidation is a forecast.
REASONING
- The tightening of credit is well supported; the claim that it has 'almost entirely' shut is stronger than the evidence, since banks still lend to investment-grade players and analysts remain split on CoreWeave.
- Consolidation moves capacity between owners rather than adding or removing watts, and big balance sheets build power faster, but cancelled projects are a real drag in the short run.
- The net effect on harnessed power depends on what happens next, so the verdict is speculative.
SEE ON THIS SITE
TAGS
- COMPUTE
- AI FINANCING
- NEOCLOUDS
- DATA CENTRES