CoreWeave will be the moment AI resets.
Blain’s Morning Porridge – October 16th, 2026: CoreWeave will be the moment AI resets.
“Markets are not smart or intelligent, but they are very efficient at converging on the wrong conclusions.”
It’s been a bad week in the AI markets. Suddenly folk are waking up to the growing gap between inflated expectations and disappointing revenues, and the risks which include competition. It’s a pattern that reoccurs repeatedly in markets. The name to watch will be CoreWeave – illustrating what can go wrong when blithe assumptions about fantabulous upside ignore the rules of financial commonsense. A default or troubling rescue is on the cards.
Key Takeaways
- AI is a key technology of the future, but the path to riches seldom runs smooth. The AI Investment Bubble is running well ahead of commercial adoption as businesses figure out how best to harness its power effectively and economically.
- CoreWeave is the potential catalyst for an AI reset. It is the canary-in-the-coal mine showing how revenue growth is failing to cover rising debt costs.
- Strong demand for AI Services in a competitive and price-inelastic market (remember the Chinese do AI too!), does not guarantee success when capex, opex and debt costs are rising – Accounting 101.
- Nvidia’s infamous circularity financing of chip sales to datacentres could magnify losses. Nvidia is down nearkly $300 bln this week! It highlights the risks of being an investor in what you supply – making it a leverage play on GPU demand (which is a function of price and competition)!
- A collapse in Chips could trigger a reset across the private capital markets based on the number of deals secured on chips, compute and infrastructure, now on the books of captive insurance firms.
The AI markets took a punch in the face this week as the scale of the undershoot between sky-high expectations and lacklustre revenues became increasingly apparent. I am a great believer in the potential power of AI, but I reckon there is a long way to go in terms of how economy adopts it. The market is all excited about hyperscalers, build out, and expectations of future profits – that’s the hope. The reality is that while companies embraced AI aggressively as it first emerged, now they are figuring out how to use it effectively while keeping AI costs under control. A slowing, more careful, path towards corporate adoption has been missed by the market which remained wildly frothy about valuations.
And that all means the market is going to correct to recognise the widening expectations/adoption gap. Which means we are going to see some losers emerge.
CoreWeave is way out front in the race to be the first major crash of the AI Debt Bubble. All major financial bubbles need that critical “wake-up-and-smell-the-money” (or lack of it) moment. The moment when the market realises it is not in Kansas anymore. In 2007 it was the run on Northern Rock as short-term funding dried up, and later the failing Bear CDO funds… from there the crisis developed its own momentum, and soon there was the looming inevitability that a bank like Lehman was going to fail…
CoreWeave will be the moment reality hits the AI market with the subtly of flying brick. (Other similar firms, like Nebius and IREN, are available.) It will send investors and financiers scrabbling, desperately seeking to understand the difference between what they thought they were buying and what they actually own. One issue will be the unravelling of the AI-circularity illusion (by which Nvidia funds chip buyers to buy its chips).
Across the private capital markets funds that have lent start-ups money secured on GPUs are going to be checking second-hand chip values. (Clue: not worth nearly as much as they’ve notionally marked them.) Of course, the private capital firms have sold their risk on to their captive insurance companies – where the effect of any losses is only seen when the insurance firms can’t meet their future liabilities… meaning the private capital markets are hoping they will have time to resolve any shortfalls to stop the bottom falling out their risk/insurance model!
The risk is CoreWeave, and the other firms in the space, are likely to trigger a reset of the bubblicious AI multiverse… or something like that. It won’t mean the end of AI – far from it, AI is a critical technology of the future. But it will mean a reset and probably a substantial degree of embarrassment.
In 2008 it felt like the End of the Financial Universe as Lehman Brothers went to the wall. – banks around the globe were scrabbling for bailouts and no one expected to get out alive in a storm of failed banking. Smart money… (including my few remaining pennies after a very expensive divorce) bought up bank stocks, and critically bank debt, at knockdown prices… and made off like bandits as the global authorities rescued the financial economy.
Sure enough, the only thing missing in the (sort of thriving) financial sector today are some name plates like Bear Stearns, HBOS, and Lehman… Thus, it will be in the AI sector – it’s unlikely Meta, Alphabet or Microsoft will collapse, but it would be wonderfully delicious if SpaceX was to come a cropper… (Have you seen how much money Elon Musk has thrown at the US Midterm election in Texas to help out an appalling Republican candidate while using his ownership of X to suggest the Democrat candidate suffers from mental illness…? It’s more than enough to sort global hunger…)
What the imminent CoreWeave quietus will illustrate is simple accounting reality – you can have an apparently successful business, crowds clamouring at the doors to use it, but if it costs more to build, run and finance than it earns… the word you are struggling for is bankrupt. That is what happens when expectations run too far ahead of adoption. The credit derivative market is telling us there is a greater than 60% likelihood CoreWeave will default on its $50 bln debt pile!
This is a second repeating theme in global finance. A fantastical new business opportunity that will make everyone excessively rich from spotting the next new, new thing financial opportunity. Back in 2017 I was having great fun with my contrarian perspective on the then current market darling WeWork. I called it and its founder a sham, arguing the fantabulous expectations of the firm to build out a shared-office business through long-term leases and short-term users would never work.
Sure enough, in 2019 it hit the buffers just ahead of its keenly contested IPO as the scale of its outstanding debt and limited cash flows became apparent. It was frankly oh, so predictable and very, very amusing to watch Softbank’s Masayoshi Son caught up in the mayhem.
Back in 2017 CoreWeave started off as crypto-miner called Atlantic Crypto, hacking Ethereum out of the number-sphere… As crypto waned it bought up distressed miners CPUs and entered the graphics design scrabble around Non-Fungible Tokens (which didn’t last long) before relaunching as a design focused cloud-platform to work its stock of computing power. That’s when things got interesting as the nascent AI market realised some GPUs (especially graphics cards) made great AI processors. The renamed CoreWeave start pulling in funding from tech investors, and in March 2025 it IPO’d at $23 billion, climbed to $90 bln by Jun, but is now sliding though $38 bln.
Harry Hindsight is the greatest market trader I ever met. He spotted a specialist cloud-computing firm that became a compute provider renting out capacity (data centres chock full of very expensive yet swiftly obsolete Nvidia chips) to firms like Microsoft and Open AI to train and test AI models on – looks a way, way too predictable bust. CoreWeave raised around $50 bln of debt to buy chips and compute, but its revenues are well short of what will be required to service the burgeoning debt.
The firm reckons on revenues of around $13 bln this year versus $37 bln of capex. The FT’s Alphaville reckons it’s paying about 5 times more in interest payments than its making in income – the losses are mounting. As interest rates creep up (as will happen following this week’s higher than expected US inflation data), then the scale of CoreWeave’s problem rises. The word “unsustainable” should be flashing up in bright red letters in your mind at this point.
It’s caught between rising rates, rising debt investor concern about its sustainability, and growing equity fears that although there is a bull market narrative about AI and the demand for data centres… a datacentre secured on chips which are repossessed by the lender… is just an empty Amazon Warehouse! CoreWeave’s 5yr Credit Default Swap (CDS) has widened from 370 to 1200 over the past year and is now traded across the Distressed debt desks.
At this stage it’s no longer a question of “if” but “when” the endgame comes. I reckon the kiss-of-death moment was yesterday when Nvidia’s CEO Jensen Huang tried to talk it up: “CoreWeave’s strength is improving every day…. I’m optimistic.” Take the opportunity. Funnily enough, the stock rallied immediately after Huang’s statement – the smart money would have seen it for what it was – the exit opportunity.
Anyone want to give me odds on Nvidia buying it up at a distressed price to prevent a rolling wave of defaults across other firms signed up to its circularity trade?
Out of time, and back to the day job… and have a great weekend.
Bill Blain
Author of the Morning Porridge
CEO Windshift Capital
Advisor – Spitfire Strategic Capital
Meanwhile, don’t forget about my new book: The Battle for Hamble
You can read a review on the Society of Professional Economist’s website here.