In short: "I don't like the neoclouds at all because when this bubble breaks, they're going to get destroyed. They are in my mind overflow capacity." If spending or model advancement slows, "the neoclouds are going to be the first ones to feel that pain," because their biggest customers are the platforms themselves — "if you're Microsoft, are you going to say, 'I don't need that neocloud capacity anymore.'" They are debt-funded and not generating cash ("cash is king. Everything else is an opinion"), so they fail his screen even though they "can absolutely scream higher."
Nebius is a "neocloud" — a company that borrows money to buy AI chips and rents that computing power out. Niles calls these firms "overflow capacity": they do great while the big clouds (Microsoft, Amazon, Google) can't build fast enough, but when spending slows, those same big clouds will stop renting from them first rather than idle their own data centers.
They also fail his basic safety test: they are funded with debt and don't generate cash. In the 2001–02 bust, the companies that went bankrupt were the ones with weak balance sheets burning cash. The shares can "scream higher" in good times, but he expects them to "get destroyed" when the bubble breaks.
46:14$61 billion market cap company. What does Nebius do? Is it on your radar? — Yeah, I don't like the neoclouds at all because when this bubble breaks, they're going to get destroyed. They are in my mind overflow capacity. When you don't have enough capacity, the neoclouds do great. When these companies have, if spend, you and I have just talked about, well, does spending slow down? Does model advancement slow down? The neoclouds are going to be the first ones to feel that pain because a lot of the cloud providers themselves, the big
In short: Her career case study — "I probably will never see anything like this again in my lifetime." "This is clearly going to be one of the biggest providers of cloud compute in the United States," yet holders who got it from frozen Russian positions "suddenly just had this thing show up in their brokerage account one day and it's just clicking the sell button." The lesson: "whenever you're in a situation in public markets where you know who your counterparty is, you're probably going to make a lot of money… and you have a materially different view… you can make phenomenal life changing money." Sourced from her network, which "was so close to getting burned" betting the Russian market would reopen. (Current holding status not stated.)
Nebius rents AI computing power, and it emerged from the Western part of Russia's Yandex. Many investors who had money trapped in Russian holdings suddenly received Nebius shares, and they just wanted out — they sold without caring what the business was worth.
Wilde's lesson is about the "counterparty," the person on the other side of your trade. If you know the seller is dumping for reasons that have nothing to do with value, and you believe the business sits in a long-lasting trend (data centres), you can make "life changing money." She calls it the best setup she will probably ever see.
In short: The other named neocloud in the same caution: today's very high compute pricing is the short-term dynamic he'd be most careful extrapolating, and the supply set (Oracle, SpaceX, the labs, the neoclouds) keeps widening.
Nebius is the other neocloud he names, and it falls under exactly the same caution: its value rests on renting out AI computing power at today's very high prices, which is precisely the short-term dynamic he'd be most careful about extrapolating. New supply keeps arriving — from Musk, from the labs building their own data centers, from Oracle and SpaceX — into a market with relatively few, very large customers.
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: Two offsetting events in one week (the transcript garbles the name as "Nuvea"). The good: it "got approval for its Vineland, New Jersey data center expansion, a major hurdle tied to its $17.4 billion Microsoft cloud deal, which is insane." The bad, mechanically: it "announced a proposed private offering of 4.5 billion of convertible senior notes, which put pressure on the name given the convertible hedging — when you issue a big convert, there's usually convertible arbitrages that buy it and short some of the stock." A supply-and-hedging drawdown, not a thesis change.
Nebius rents out AI computing capacity — it builds data centres full of Nvidia chips and lets other companies use them. Its biggest contract is a roughly $17.4 billion agreement with Microsoft, and this week it cleared a real obstacle: local approval for the expansion of its Vineland, New Jersey site, without which the contract could not be delivered.
The share price nonetheless came under pressure, for a purely mechanical reason. The company is raising $4.5 billion through convertible bonds — debt that can later turn into shares. The funds that buy these bonds are not making a bet on the company; they buy the bond and simultaneously sell the shares short to neutralise the equity risk, harvesting the difference. That short selling pushes the stock down at issuance regardless of how the business is doing.
So the week contained one genuinely good event and one temporary technical drag, which is why the stance is neutral rather than a change of view.
Full passage: premium transcript (PDF).
In short: Named twice: in the opening summary of where new money went — "newer infrastructure names like Cerebras and Nebius featured prominently among top buys" — and again in the second tier of the AI-broadening list. The neocloud leg of the supply-chain rotation.
In short: Bought for a trade off rising GPU lease rates and it crushed: Q2-26 revenue $582M vs $573M (+454% y/y), adjusted EBITDA $236M vs $175M (against −$20M a year ago), AI cloud revenue +514% to $575M, operating cash flow +$2.2B, $8B cash, ~$9B of customer prepayments, contracted capacity raised 4 → 5 GW; management says it could sell all 2027 capacity today. "Stock was up 34% on the day that it reported and we trimmed."
Nebius rents out AI computing power — you pay for time on its graphics chips rather than buying them. Singh's entry signal was concrete and trackable: during the AI sell-off, the rental rates for those chips were rising on Bloomberg, meaning demand was strengthening while the stocks fell.
The quarter validated it — revenue up 454%, profitability flipping from a loss to a $236 million profit, $2.2 billion of operating cash flow, $8 billion of cash, and customers pre-paying about $9 billion. It also has far less debt than CoreWeave. He trimmed into the 34% one-day pop, which is the point: this was sized and treated as a trade.
6:22Both had earnings; both of them crushed. I think Nebius reported Q2 26 revenue of 582 million, which beat estimates of 573. It was up 454% year-over-year. Adjusted EBITDA came in at 236 versus 175, compared to negative 20 million of EBITDA last year and AI cloud revenue was up 514% to 575 million and operating cash flow actually reached positive 2.
In short: The neo-cloud print that punished the shorts: revenue $582M vs $573M expected, +454% y/y, adjusted EBITDA $236M vs $175M, cloud revenue +514%, operating cash flow $2.25B on prepayments and $8B of cash. "So Nebius absolutely crushed it. It was up 34% on the day." He uses it as the evidence for Huang's case: "you saw a number of investors short Nebius ahead of earnings, including Michael Burry, and he got completely smoked… if you look at the secondary markets, Jensen's right — the demand and the lease rates for GPUs are still quite high and they're growing."
Nebius is a European competitor to CoreWeave, renting AI computing capacity. Revenue grew 454% in a year, profits came in far ahead of forecasts, and it holds $8 billion of cash. The shares rose 34% in a day.
Singh uses it to make a point about betting against a shortage. Several investors, including Michael Burry, had shorted Nebius going into the results and "got completely smoked." His reasoning is that anyone who checked the secondhand market for GPU rentals would have seen prices rising, not falling — the demand was observable before the earnings. Shorting a company because its valuation looks absurd, when the underlying product is still getting more expensive, is a bet on sentiment rather than on facts.
Full passage: premium transcript (PDF).
In short: "Nebius Finds Pricing Power" — the one section whose Bottom Line leans. Revenue +454% Y/Y to $582M, of which AI Cloud is $575M (98%); gross margin +6 points to 77% and adjusted EBITDA $236M at a 41% margin. The $176M operating loss is smaller than the $260M of D&A now landing on the income statement as "billions of dollars of new infrastructure started hitting" it. Unusual origin: Nebius is the international remnant of the 2024 Yandex breakup — the Nasdaq-listed Dutch holding company sold the Russian operations for $5.4B, rebranded, and put co-founder Arkady Volozh back in charge, then used the engineering talent and capital base to build a purpose-built AI cloud rather than rebuild an internet conglomerate. The unit economics are the story: four new deals averaged more than $1B of contract value at $20–25M per megawatt, with shorter-term capacity fetching $40–50M/MW; management now estimates Q2 contracts repay their capex and operating costs in about 22 months, down from a two-to-three-year range. The build is enormous — Q2 capex $5.7B, almost 10x quarterly revenue, against $20–25B for the year — but customers are financing it: more than $9B of prepayments in 2026, covering roughly 50–60% of the associated capex. Bottom Line: "rising prices, faster paybacks, and customer prepayments are making each new megawatt more attractive. That capital efficiency may ultimately matter more than its 454% revenue growth."
Nebius rents out AI computing power, the same basic business as CoreWeave. What makes this quarter interesting is not that revenue grew five-fold to $582 million — it is how the growth is being paid for.
Building an AI data centre is like building a toll road: you spend billions up front and collect fees for years afterwards. The question is always who lends you the money for the gap. CoreWeave borrows it against its chips and pays heavy interest. Nebius is increasingly getting its customers to pay in advance — it expects more than $9 billion of prepayments this year, covering 50–60% of what it spends on new capacity. Money from a customer costs no interest and cannot be margin-called.
The price it charges is also going up, and the newsletter gives the metric to watch: dollars per megawatt of power sold. Nebius's four newest deals — each averaging over $1 billion — were struck at $20–25 million per megawatt, and shorter contracts are going for $40–50 million per megawatt. In other words, customers pay a large premium for flexibility, which is what a genuine shortage looks like.
Put those together and the "payback period" — how long a new site takes to earn back everything it cost to build and run — has fallen to about 22 months, from two-to-three years previously. That is fast for infrastructure this expensive.
It is also an unusual company. Nebius is what was left of Russia's Yandex after the 2024 breakup: the Amsterdam-based, Nasdaq-listed parent sold the Russian business for $5.4 billion, kept the international operations and the cash, and re-pointed the whole thing at AI under founder Arkady Volozh. It started this race already funded.
The caution: it is still spending roughly ten times its quarterly revenue on construction, and its reported operating loss ($176M) exists because the new equipment is being written down ($260M of depreciation). Everything depends on the contracts being real and the customers staying solvent. Analysis, not a recommendation.
In short: Same bucket as CoreWeave: a strong AI-infrastructure print being used to justify chasing the bounce, which he dismisses as already discounted.
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: "We did well buying during the AI sell-off and it rallied, but there was an insider at Nebius… talking about how Meta selling excess compute could put pressure on some of these guys, Nebius and CoreWeave. So it gave back some of the earnings." On top of that, Michael Burry has targeted Nebius as an AI beta short to hedge his book. Reports Wednesday.
Nebius rents out AI computing power — it buys graphics chips, installs them in data centres, and charges others to use them. Singh bought it during the AI sell-off and it rallied, but it has given some of that back on a specific fear.
An insider note circulated saying Meta has excess computing capacity it may sell into the same rental market. If a company that large starts renting out spare capacity, rental prices fall — and rental prices are the entire business model for Nebius and CoreWeave. On top of that, Michael Burry has picked Nebius as a way to bet against the AI trade broadly, which adds selling pressure independent of results.
Full passage: premium transcript (PDF).
In short: Profit-taking, not a view change: trimmed 50% on July 31 after a +27% single-day pop. Still named (with IREN and CoreWeave) as a beneficiary of Meta's compute-resale disclosure and of bouncing AI lease rates.
Nebius rents out AI computing capacity like CoreWeave and IREN, and benefits from the same tight market. Singh sold half his position after it popped 27% in a single day — a discipline call on the price move, not a downgrade of the story, which he still cites approvingly.
Full passage: premium transcript (PDF).
In short: Follow-up to last week's featured NeoCloud deep-dive: Nebius has deployed its first full Vera Rubin Ultra ("Varin Rubin" in the transcript) in a finished facility — "so maybe we see a bounce there."
A short follow-up to last week's deep dive on Nebius, the European company that rents out AI computing power. It has now switched on its first complete Vera Rubin Ultra installation — Nvidia's newest generation of AI systems — in a finished facility, meaning the capacity is live and can start earning. After the stock's beating, "maybe we see a bounce there."
Full passage: premium transcript (PDF).
In short: The featured NeoCloud deep-dive. A pure-play GPU-as-a-service provider spun from Yandex's divested international assets, net-debt-neutral (far less debt/EBITDA than CoreWeave), Amsterdam-HQ'd (a sovereign-AI European alternative). Q1 rev jumped to $399M; de-risked via a $775M asset-backed facility (less dilution); targeting >4 GW contracted power by 2026 with a Vineland NJ build online Q4. "Venture company inside a large-cap wrapper" — EBITDA seen +5×. Consensus buy $244-265 vs the $177 close (+41%; beta >2 — high-risk). The kicker: ~$8B of private stakes the Street under-counts — ClickHouse (28%, ~$4.2B), AV-Ride (83%, ~$3.3B), Toloka (50%, ~$250M), TripleTen (100%) — on top of ~$46B of signed Microsoft/Meta deals (Wolfe: $34B rev / $21B EBITDA by 2030).
Nebius rents out AI computing power — it buys huge clusters of Nvidia chips and lets AI companies pay to use them (a "NeoCloud"). It was carved out of Yandex, the Russian internet company, and is now run from Amsterdam, which it pitches as a selling point: European governments and firms increasingly want their AI kept on European soil under European law. The demand is real because the big clouds (AWS, Azure) are sold out, so AI startups rent from nimbler providers like this.
Singh's edge is the balance sheet and the hidden assets. Unlike rival CoreWeave — which is drowning in ~$40B of debt — Nebius has essentially no net debt, so it can keep building without constantly selling new shares. And it owns pieces of four fast-growing private companies (a database firm ClickHouse, self-driving AV-Ride, data-labeler Toloka, and ed-tech TripleTen) that he tallies at roughly $8 billion — value he says Wall Street analysts basically ignore. Wall Street's average price target is $244-265 vs a $177 stock, but it's a high-risk name (it swings twice as hard as the market), so size it carefully.
Full passage: premium transcript (PDF).
In short: "The most asset-heavy neo cloud" ($4.40 of capital per $1 of revenue) suddenly unveiling an asset-light model — "I think that was a tremendous admission" the capital-intensive story doesn't work.
Nebius is a "neo-cloud" — a company that buys GPUs and rents out computing power. It was the most capital-hungry of them, spending $4.40 to generate every $1 of revenue. For two years these firms told investors that owning the physical hardware was the goldmine.
Now Nebius has flipped to an "asset-light" model — it will run other people's data centers for a fee instead of owning them (like a Marriott franchising hotels). Chanos calls that "a tremendous admission" that owning the expensive hardware was never the money-maker they claimed.
27:38— So allegedly Coreweave has been looking at a way to hedge downside in asset prices, which I thought was interesting given the story is — Well they should be, right? — Theoretically yes, but the narrative is this stuff is going up in value not down in value. — what I thought. — The bigger development was by Nebius today which came out and unveiled a new business model that's asset-light.
In short: Baruch's final trade: an AI-infrastructure "ultra-growth story" that's sold off 40% from its recent high into big support — "I think it snaps back quickly." Notes the super-volatile area.
Nebius is a fast-growing AI-infrastructure company (it rents out AI computing capacity). Bill Baruch's final-trade pick: it's an "ultra-growth" story that has sold off 40% from its recent high into what he sees as strong technical support, and he expects it to "snap back quickly." He flags it's a very volatile part of the market.
In short: Collateral of the Meta cloud news — dropped 15%, the hardest hit. Like CoreWeave, a neocloud that had counted Meta as a customer and now faces it as a potential rival for AI-compute rental. (Recap, not a stance call.)
Nebius is another AI-compute rental company in the same boat as CoreWeave, and it fell even harder — down 15%. The logic is identical: Meta was a customer, and a Meta cloud offering would turn that customer into a rival competing for the exact same AI-rental dollars. It was the hardest-hit name on the news. Referenced as collateral of Meta's strategy shift, not a stance on Nebius itself.
In short: Used as the example "neo-cloud" counterparty that could take Comstock's secured gas/power capacity and bring the turbine/fuel-cell capex — with the land-vs-infrastructure economics split TBD.
In short: Grouped with CoreWeave/DGXX as a neocloud with live capacity that benefits from the power/transformer-driven data-center delays (premium lease rates).
Full passage: premium transcript (PDF).
In short: Listed among the circular-funding/neocloud cohort alongside Micron and CoreWeave.
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: Named among the "high-quality neoclouds" (with CoreWeave and Crusoe) whose superior GPU utilization is a durable, underappreciated competitive advantage.
Nebius is another neocloud, which Baker groups with CoreWeave and Crusoe as "high-quality" operators.
Same thesis as CoreWeave: the durable edge is operational excellence — getting GPUs utilized far harder than a "bottom of the barrel" provider — which lets the best names charge more and earn a lasting premium, something he thinks the market underappreciates.
21:05And those CoreWeave GPUs are being utilized two to three X more per hour on average than the GPUs from maybe a more bottom of the barrel provider. And by the way, this all goes for Crusoe and Nebius and other high-quality neoclouds. But I think this is very under appreciated. And people think, "Oh, that can't be a durable competitive advantage."
In short: #8, +174.4% — "Nebius Group quietly spent years building AI-focused data centers. In 2025, the market finally noticed." Triple-digit revenue growth as GPU demand exploded, with the trigger named: "Microsoft signed a multi-year AI infrastructure deal worth over $17 billion. Meta soon followed." Slegers' framing: "It's a classic pick-and-shovel play."
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.