In short: Asked whether the lab slowdown means pulling back "on semis, pull back on like the CoreWeaves of the world": "Yeah, you should have been pulling back a few weeks ago," once Texas and Pennsylvania put data-center connections on hold. As a neocloud it is in the bucket he expects to "get destroyed" when the bubble breaks — overflow capacity, debt-funded, no cash flow.
CoreWeave is the best-known neocloud — the same business model as Nebius: borrowed money, AI chips, rented capacity. When the host asked if investors should pull back on semiconductors and "the CoreWeaves of the world," Niles said they should have done so weeks ago, once states like Texas and Pennsylvania started pausing power hookups for new data centers. It carries all the neocloud risks: first to lose customers in a slowdown, heavily indebted, and no cash flow cushion.
10:17Plus, now you've got these comments from OpenAI, Anthropic, Elon Musk is supportive of that, too, of slowing the pace of improvement in AI which means slowing the spending if you're not going out and training new models. — So does that mean you pull back on semis, pull back on like the CoreWeaves of the world? — Yeah, you should have been pulling back a few weeks ago.
In short: "The unsecured CoreWeaves are trading around 12% yield… That is meaningfully distressed," yet street models have it tripling its debt over two years: "You are not going to raise that debt because the cost of capital is too high." Its 12% cost of capital sits inside his ~$1T/yr carrying cost for AI compute.
CoreWeave is a "neocloud": it borrows heavily to buy NVIDIA chips and rents the computing power to AI labs. Taylor's key clue is its bonds. Its unsecured debt (debt not backed by specific assets) trades at about a 12% yield, a level lenders demand from companies they worry may not pay them back, even though the stock market treats CoreWeave as a secure growth story.
Wall Street models assume CoreWeave triples its debt in two years. Taylor says that can't happen at a 12% cost of borrowing, and that the high interest bill is part of why AI computing costs about a trillion dollars a year to carry. When bond investors and stock investors disagree this sharply, he trusts the bonds, as in 2006 before the housing crash.
1:56Stocks were good. Everything's great. The numbers are awesome. The earnings are fantastic. And the bond said, no. And then CoreWeave, Oracle, and that's what really piqued my interest. I said, what are the bond guys seeing? Currently, the unsecured CoreWeaves are trading around 12% yield. And for our listeners, it's kind of a crazy number given where we're at right now for something that is so secure and so wonderful and such incredible growth.
In short: Harrington's name for where the AI froth actually sits — used explicitly as the contrast to Oracle. "I don't think you need AI to come crashing down to have a problem… the challenge right now is that you could have AI contract at the margin. And there's a lot of fluff and a lot of froth and a lot of stocks out there. And I think of like the CoreWeaves more than I think of the Oracles." Her discriminator is balance-sheet experience, not valuation: "given how big their debt load is and how they frankly don't have experience managing a debt load, that's different. Whereas the energy companies and the real estate companies, they've lived through huge debt loads, super long dated, laddered out forever. They know how to deal with it."
CoreWeave is a "neo-cloud" — it buys Nvidia chips with borrowed money and rents them out to AI companies. Harrington uses it as the counter-example to Oracle when she explains where the real risk sits.
Her point is subtle and worth keeping: AI does not have to collapse for these businesses to break. It only has to "contract at the margin" — a modest slowdown in demand — because the debt payments do not slow down with it. And her screen is not the valuation but the management: energy and property companies have run big, long-dated debt loads through cycles and know how to handle them, while the newer AI borrowers have never done it once.
In short: Named as one of the AI credits he monitors daily, with no stock view: "one of the charts I look at every morning is the credit default swaps for the big guys, Microsoft and CoreWeave and all the rest of them, Nvidia, to see what they're doing. What's the cost to insuring that debt?"
CoreWeave rents out AI computing capacity, and it appears here for one reason: it is one of the credits on the chart Niles checks every morning.
That chart shows credit default swap prices — effectively the annual insurance premium a lender pays to be protected if the company fails to repay. Rising premiums mean professional lenders are getting nervous, usually well before equity investors do.
He offers no view on the stock. The point is the screen: a heavily capital-intensive AI business funded increasingly with debt is exactly where he expects the first sign of strain to appear, and the debt market is where he expects to see it first.
40:48And so all of those things together bother me because to some degree, and you can't separate out what's what very easily, right? I mean, I look at one of the charts I look at every morning is the credit default swaps for the big guys, Microsoft and CoreWeave and all the rest of them, Nvidia, to see what they're doing.
In short: Asked where to be in AI he "started with a negative": the risk of extrapolating a very short-term dynamic — "the pricing of compute, certainly maybe with neoclouds" — while Elon's entry may hasten the glut. Named among the nine or ten scaled compute providers that turned a three-vendor market into a crowded one with fewer customers.
CoreWeave is a "neocloud" — a company that buys enormous quantities of AI chips and rents the computing power out by the hour. Its economics depend heavily on the price of compute, which is extremely high right now because demand far exceeds supply.
Asked where investors should be in AI, Keller deliberately opened with a negative, and this is it: the danger of assuming today's very high compute prices persist. Elon Musk is entering the same business at scale, which he thinks could "move us quicker to the glut." And the wider structural point — the market for renting compute has gone from three serious providers to nine or ten (the hyperscalers, Oracle, SpaceX, the labs building their own, plus the neoclouds), while the pool of customers is small and concentrated. More sellers and fewer buyers is not a combination that protects pricing.
29:01You include Oracle and SpaceX and the labs who are procuring directly and the neoclouds and CoreWeave and Nebius at least. So you have a market that went from maybe three to 10. They still have scale advantages, but you have fewer customers. So I understand why those stocks are not trading at higher multiples given their growth.
In short: Named by Harrington as one of the names for which NVIDIA's guidance actually is a live catalyst, even though she thinks it is a non-event for NVIDIA itself: "it might matter more to CoreWeave or Microsoft or OpenAI." The neo-cloud whose economics depend directly on GPU supply, pricing and demand commentary. No stance on the stock.
In short: Appears in the second tier of the AI broadening — "Intel, Astera Labs, Qualcomm, Nebius, and CoreWeave also made appearances" among top buys. The other half of the neocloud pair with NBIS.
In short: Bought for a trade, not for long-term, alongside Nebius off the GPU-lease-rate signal, and it "also had some strong results" — but it carries the debt Nebius doesn't, and "we're selling some of the names that rallied."
The same business as Nebius — renting AI computing capacity — bought off the same rising-rental-rate signal, and it also reported strong results. The difference he flags is the balance sheet: CoreWeave carries a lot of debt where Nebius doesn't.
He is explicit that this was bought "for a trade, not for long-term," and that he is selling down the names that have already rallied.
7:23It's a well-followed name. We're selling some of the names that rallied. On the software side, one of the big shorts for hedge funds were the software names because, as we know, we'll talk about situational awareness in a minute, but a lot of these tech names had to short software as a hedge to their long AI stocks.
In short: Owned into the print and it paid: "we own CoreWeave ahead of the earnings. Shares were up about 17% pre-market… they're up 26% ahead of that." The quarter: revenue beat at $2.6B, EBITDA $1.5B vs $1.43B at a 59% margin, backlog $104B (+46%), active power more than tripled to 1.5 GW, FY26 guidance raised to $12.4-13.2B of revenue and $960M-1.15B of adjusted operating income, the active-power target lifted above 1.85 GW, and a year-end run rate of $18.5-19.5B. Deck page 37.
CoreWeave rents AI computing capacity by the hour. Singh owned it into the results and it rose 17% before the market opened. The numbers that matter are not the quarter but the two forward-looking ones: signed future business of $104 billion, and the amount of electricity actually running through its data centres — which more than tripled to 1.5 gigawatts, with a target now above 1.85.
Power is the real constraint in this industry, not chips. A company that can point to gigawatts already energised and a hundred billion dollars of contracts against them is selling something scarce. Management raised guidance and pointed to an $18.5-19.5 billion annualized revenue rate by year end.
Full passage: premium transcript (PDF).
In short: "The real king of debt I would say is CoreWeave, that just reported. I've never seen a bigger gap between EBITDA and net income loss. It is quite extreme and it's a little railway-like." Its "giant backlog" is really the hyperscalers' off-balance-sheet commitments; its delayed-draw term loans are a credit line dressed as prudence. The killer question is the fallback: people still use YouTube even if Google wastes a few hundred billion — "with a CoreWeave, do they have that backup? Maybe crypto mining, I don't know."
CoreWeave is a "neocloud" — it buys GPUs, builds data centres and rents the computing power to the big technology companies. Its headline selling point to investors is a huge contracted backlog, which Wigglesworth points out is largely the other side of the hyperscalers' off-balance-sheet commitments: the same promises, counted once as a liability nobody labels debt and once as a company's future revenue.
His verdict is unusually blunt: "the real king of debt I would say is CoreWeave… I've never seen a bigger gap between EBITDA and net income loss." EBITDA is profit before interest, tax and the cost of writing down equipment; net income is what is actually left. A record gap between them means the depreciation and interest — the two costs a chip-buying, debt-funded business cannot escape — are eating everything the operating business earns. He calls the whole thing "a little railway-like," which in this conversation is not a compliment: the railways were real, transformative, and bankrupted their financiers. Its delayed-draw term loans (borrow only when you need it) are a credit line dressed up as prudence.
The test he applies is the fallback question. If AI disappoints, "people are still going to be going on YouTube even if Google wastes a few hundred billion dollars." A neocloud has no second product: "with a CoreWeave, do they have that backup? Maybe crypto mining, I don't know, but I'd worry about those essentially more than I do the big hyperscalers."
48:48I think one thing that CoreWeave, the neocloud, is doing is delayed draw term loans. So, oh, you don't have to actually borrow the money until you need it. So it's like a credit line. Yeah, the real king of debt I would say is CoreWeave, that just reported. I've never seen a bigger gap between EBITDA and net income loss.
In short: "CoreWeave is Sold Out" — the purest version of the neocloud model: buy NVIDIA GPUs, install them, rent the compute to OpenAI, Microsoft and Meta. Revenue +112% Y/Y to $2.6B with committed contracts generating 98% of it and on-demand just 2%; gross margin contracted 8 points to 66% as data-center rent, power and ramp costs grew faster than revenue. A $49M operating loss became a $626M net loss, "heavily weighed down by $640 million in interest expense tied directly to its GPU-collateralized debt facilities" — the financing cost, not operations, is what makes the loss. And the P&L understates the spend: Q2 capex $9.4B, more than three times quarterly revenue. What's improving: backlog reached $104B (+246% Y/Y) before another $25B of commitments in early Q3, so near-term capacity is "effectively sold out"; adjusted operating margin rose sequentially 1% → 5%, new Q2 contracts should carry contribution margins 5–10 points above recent deals, and that is before July's roughly 25% price increase. Mix is broadening too — storage, CPU, networking and software now exceed $400M of ARR, and managed inference went from $1M to more than $100M of booked ARR in a single quarter. FY26 capex outlook raised to $35–39B chasing >1.85 GW of active power by year-end. Bottom Line: the $104B backlog "sounds almost absurd next to a targeted 2026 exit revenue run rate of $19 billion," but conversion is capped by physical capacity — the case "comes down to how quickly it can turn power and GPUs into revenue without letting financing costs overwhelm the margin gains."
CoreWeave does one thing: it buys NVIDIA chips, plugs them into data centres, and rents the computing power to companies like OpenAI, Microsoft and Meta. It is the cleanest way to own the AI compute shortage — and the most financially exposed.
The demand side is extraordinary. 98% of its revenue this quarter came from contracts signed in advance, and the order book — work already sold but not yet delivered — reached $104 billion, up 246%, with another $25 billion signed weeks later. Revenue more than doubled to $2.6 billion. Management says near-term capacity is effectively sold out, and it pushed prices up about 25% in July.
The problem is the bill. Building capacity costs money years before customers pay for it: CoreWeave spent $9.4 billion on construction and chips in a single quarter — more than three times its revenue — and plans $35–39 billion for the year. To fund that it borrows against the GPUs themselves, and this quarter the interest on those loans was $640 million. That single line is why a business with a $49 million operating loss reported a $626 million net loss. Operationally it is close to break-even; financially it is deep in the red.
Two things pull in the right direction: profit per contract is improving (adjusted operating margin went from 1% to 5%, with new deals expected to be 5–10 points better), and the mix is broadening beyond raw chip rental — storage, networking and software passed $400 million of annual recurring revenue, and managed inference went from $1 million to over $100 million of booked recurring revenue in one quarter.
Why Neutral: the enormous backlog cannot be converted faster than CoreWeave can physically energize data centres — it is targeting more than 1.85 gigawatts of live power by year-end — and every month of delay is another month of interest. The whole case is a race between rising margins and a rising interest bill. Analysis, not a recommendation.
In short: "CoreWeave beat on both the top and bottom line. Revenue climbed 112% versus last year" — still unprofitable (EPS a loss of about a dollar), but "the market focused on the revenue and the stock was up 19% on Wednesday," and on the call "management stated that demand and pricing remain very strong." His framing: "as long as the hyperscalers keep spending, AI infrastructure players like CoreWeave and Super Micro can't help but benefit."
CoreWeave rents out AI computing power — essentially a specialist data-centre landlord for the AI boom. Its quarter beat on both revenue and profit, with revenue up 112% versus a year earlier, and the stock rose 19% in a day even though the company still loses money (about $1 per share).
What Eisman took from it wasn't the profit line but the pricing commentary: management said demand and pricing "remain very strong," which is direct evidence that the AI build-out has not cooled. His framing is mechanical rather than promotional — as long as the giants keep spending, the infrastructure suppliers "can't help but benefit." The corollary is that this is a derivative position: it works exactly as long as the spending does.
14:38Both companies had good results. Coreweave beat on both the top and bottom line. Revenue climbed 112% versus last year. However, Coree is still an unprofitable company. EPS was a loss of about a dollar. The market focused on the revenue and the stock was up 19% on Wednesday. Also, on the conference call, management stated that demand and pricing remain very strong.
In short: Its strong earnings are the standard objection to his sell call — and his answer is that a discounting market has already priced it: "that's known." Cited as evidence of the recency bias driving retail back in, not as a view on the company.
12:19seller of this move when you just saw CoreWeave's earnings and you saw Nebius's earnings and you saw Taiwan semiconductor saying that demand is insatiable and Micron and all these guys and the answer is because that's known. So what you had, you have not seen revenue growth were earnings expectations at levels this high, 15% revenue growth since Q4 of 2021.
In short: Reports Tuesday after two deals in a week — the $335M five-year storage agreement with Backblaze and a multi-year agreement with Solidigm (an SK Hynix subsidiary). Same overhang as Nebius: Meta's ~400,000 spare GPUs could depress GPU rental yields, the one thing that breaks Morgan Stanley's "Neoclouds win in every scenario" case.
CoreWeave is the largest of the AI compute landlords, and it signed two deals in a week: a $335 million five-year storage agreement with Backblaze and a multi-year deal with Solidigm, SK Hynix's memory-storage arm. Both signal it is building out capacity for committed demand.
It carries the same single risk as Nebius: if Meta dumps its roughly 400,000 spare graphics chips onto the rental market, the price CoreWeave can charge falls. That is the one scenario that breaks the otherwise favourable outlook for these operators.
Full passage: premium transcript (PDF).
In short: A neo-cloud winner on two counts: "we said that they were quite interesting as AI lease rates were bouncing" (it rallied hard last week), and Meta's admission that it is being offered a significant premium for its existing compute — and is itself still buying from neo-clouds — is a direct read-through on lease pricing.
CoreWeave is a "neo-cloud" — it buys AI chips and rents computing time to companies that need it. Two things this week confirmed the rental market is tight: prices for AI capacity started bouncing (Singh had flagged these names as interesting on that basis, and they rallied hard), and Meta disclosed it is being offered a meaningful premium for its own spare computing power while still buying from neo-clouds. When the largest buyers are also would-be sellers at a premium, existing capacity is scarce and valuable.
Full passage: premium transcript (PDF).
In short: The credit-market read on the AI trade: "CoreWeave, the AI data center company, is in the process of raising debt of 2.6 billion to fund additional computing capacity. The loan is being priced with a yield of more than, get this, 9%. That is expensive debt. There is something of a credit cycle here as fixed income investors are discriminating between the large companies like Google that they know can pay the money back and smaller newer companies like CoreWeave that are more risky."
CoreWeave rents out AI computing capacity, and it has to borrow to buy the chips and build the buildings. It is raising $2.6 billion of debt at a yield of more than 9% — very expensive money in a market where the largest technology companies borrow at a small fraction of that.
Eisman's interest is in what the price says rather than the company itself: "there is something of a credit cycle here as fixed income investors are discriminating between the large companies like Google that they know can pay the money back and smaller newer companies like CoreWeave that are more risky."
That is a meaningful change. When credit markets stop treating an entire theme as one risk and begin charging very different prices within it, the era of easy financing for the weaker participants is ending — and bond investors usually notice before equity investors do.
9:28That is why NASDAQ is down 7% from its all-time high on January 2nd. Also, the Socks Index, which is the EyesShares semiconductor ETF, is down 23% from its peak on June 22nd and is down 4% this week. The change in the AI debate is also impacting fixed income markets. Coreweave, the AI data center company, is in the process of raising debt of 2.
In short: Q&A (why Nebius over CoreWeave): CRWV "is also likely a buy" and looks cheap, but it carries ~$40B of debt, burns a lot of cash and is raising more — whereas Nebius grows ~3-4× faster with no net debt. "You could buy both" on an AI recovery; he prefers NBIS.
Asked why he prefers Nebius to CoreWeave (the other big AI-cloud rental company), Singh says CoreWeave is "also likely a buy" and looks cheap — but it carries about $40 billion of debt, burns a lot of cash, and keeps borrowing more. Nebius grows 3-4× faster with essentially no net debt, so he'd rather own Nebius; you could hold both if you believe in an AI recovery.
Full passage: premium transcript (PDF).
In short: "Allegedly Coreweave has been looking at a way to hedge downside in asset prices" — a tell that contradicts the "assets only go up in value" narrative the neo-clouds sell.
CoreWeave is another neo-cloud renting out AI computing power. Chanos notes it has reportedly been looking for financial instruments to hedge against a drop in the price/value of its assets.
The point: if the whole story is "these assets only go up in value," why would a leader quietly be buying protection against them going down? It's a tell that insiders don't fully believe the narrative they're selling.
27:20And so therefore these chips are going to last forever. — Right. — And of course there's a fallacy again in using spot prices to do that. — Well, it's interesting, and I saw one neo cloud. It was an article, I can't remember where I saw it today. They were talking about coming up with some sort of derivatives to hedge out pricing.
In short: Collateral of the Meta cloud news — fell 13%. Meta spent the past year as one of CoreWeave's biggest AI-compute customers; a Meta cloud offering "could turn into a massive new rival." (Recap, not a stance call.)
CoreWeave is a "neocloud" — a company whose whole business is renting out banks of AI chips (mostly NVIDIA GPUs) to others who need computing power. Its stock fell 13% on the Meta news for a simple reason: Meta had been one of its biggest customers, and a customer that decides to rent out its own spare chips instead becomes a competitor. Losing a big buyer and gaining a big rival at the same time is a double blow. The article flags it as collateral damage from Meta's move, not a judgment on CoreWeave's own business — a recap, not a call.
In short: Data-center-delay beneficiary: with operational capacity and secured power it (with DGXX, Nebius) earns premium lease rates until new builds catch up — the AI bubble has "at least a year" before the delays resolve, despite Chanos's bearishness.
Full passage: premium transcript (PDF).
In short: "They just added CoreWeave to the Nasdaq… companies that are really financially a mess" — his exhibit for index committees adding fragile, late-cycle names that passive money is then forced to buy.
CoreWeave rents out AI computing power. His interest is not the company but the fact that it was "just added to the Nasdaq" — an index add for a business he calls "really financially a mess." Index membership forces every tracker fund to buy it, which is how fragile, late-cycle names end up in ordinary retirement accounts.
51:16Remember, if they brought SpaceX public in 2019, it would have been a $30 billion deal, right? And now it's a two trillion-dollar deal, right? And so over time, look at the companies that they're adding to the S&P. They just added CoreWeave to the Nasdaq. They're adding CoreWeave into the Nasdaq. Companies that are really financially a mess. They put Lululemon in the S&P at all-time highs. The stock's down 60%.
In short: "Effectively just a conduit that buys Nvidia GPUs and builds its own data centers" — a pass-through in the circular-financing web.
39:46OpenAI's data centers and AMD is part of this circular funding as well and you have Micron and Nebius and Cororeweave which effectively is just a conduit that buys Nvidia GPUs and builds its own data centers., their names like Figure AAI which are private where you don't even know what the financials are and Microsoft and Honey AI.
In short: Cited as one of NVIDIA's large publicly disclosed equity holdings contributing to the ~$16B of Q1 equity-investment gains.
In short: The neocloud category is "absolutely durable." Their GPUs are utilized 2–3× more per hour than a bottom-of-the-barrel provider's, so they can charge a real premium. Regrets being conflicted out of investing at a $1.1B valuation.
CoreWeave is a "neocloud" — a company that rents out GPU computing power to AI firms, instead of being a giant like Amazon or Microsoft. The common skepticism is that this is a commodity business with no lasting advantage.
Baker disagrees. He says running a GPU cluster well is "like driving a Formula 1 car" — it looks easy but is extraordinarily hard, and the best operators get their GPUs used 2–3× more productively per hour than weak rivals. That efficiency lets them charge a premium, which makes the advantage durable. He regrets being legally blocked from investing back when CoreWeave was valued at just ~$1.1 billion.
19:37And I was conflicted out by Crusoe, and I love Crusoe, and I think Crusoe's going to work, and we have a large position in Crusoe. But every time I think of being able to put 50 into CoreWeave at a billion dollars, I get a little sad. But I'm very happy you put 7 and 1/2 billion in. Yes.
In short: Named as the archetypal "neocloud" in the revenue-circularity chart — NVDA "sells GPUs to neoclouds like Coreweave" that are funded by NVDA equity stakes or hyperscaler-anchored rounds. A structural example of the loop, not a stance.
In short: Data-center developer: bonds fell 104→low-90s (~12% yield, single/double-C-like) with ~$15B negative FCF next year — "in flames."
CoreWeave rents out AI computing power and is building data centers. He points to its bonds: they fell from 104 to the low-90s, pushing the yield to about 12% — the kind of return investors demand only from very risky borrowers — even though it's officially a higher-rated credit. On top of that it's projected to burn ~$15 billion of cash next year. To him it's a poster child of the credit crisis already starting.
9:22Right now the S&P is almost 30% AI stocks, which is insane. And the street's expecting 820 data centers over the next 5 years. But there are three big hurdles — NIMBY (not in my backyard), power, and infrastructure equipment. A lot of data centers are probably going to get delayed, and the credit markets are already sniffing this out. Just look at Blue Owl. Look at Jefferies equity versus Goldman Sachs. But the CoreWeave bonds are trading with almost 12% yield.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.