원문 보기
Two years ago, Nibbius was the leftover piece of Russia's biggest tech company. It ended its first month back on NASDAQ at about $22 a share. Friday, it closed at 221. It's now worth about $60 billion. In August, Michael Bur shorted it at $2177, calling his new AI shorts a bit like shooting fish in a barrel.
Days later, Nibbius reported earnings and the stock jumped 34% in a day. Bur shorted more at 247. His words, "Nebius is what the top of a boom looks like." Then in late September, he closed the stock short and switched to puts expiring next June with strikes below $100. Those only pay off big if the stock more than halves.
Meanwhile, 22 analysts have an average target of about 284. So, is Nbius the next great cloud company or the textbook bubble Bur says it is? Let's break down Nbius. are case, and what to watch next. Nbius is what's called a Neocloud. It buys Nvidia's most powerful chips, packs them into huge data centers, and rents that computing power to companies building AI.
Its biggest customers are tech giants. Microsoft signed a deal worth up to $19.4 billion. Meta signed two deals worth up to $30 billion combined. AI labs like Coher rent from it, too. The number that matters most is power. Nibbius is aiming for 5 gawatts of contracted power by the end of this year. That's roughly the output of five large nuclear reactors and Nvidia is both its supplier and one of its investors.
The story starts in Russia. Nibbius used to be Yandex, often called Russia's Google, co-founded by Ardi Valage in the 1990s. By the end of 2021, it was worth about $31 billion. Then Russia invaded Ukraine. NASDAQ halted the stock and Valage stepped down after the EU put him under sanctions. In July 2024, the Dutch parent company sold all of its Russian businesses for about $5.4 billion and soon renamed itself Nibbius.
What was left, a data center in Finland, a few side businesses, and about $2 billion of cash. Valage, now off the sanctions list, came back as CEO. That October, the stock returned to NASDAQ. Two months later, Nvidia joined a $700 million investment round. Here's the stock since it came back. In early 2025, it more than doubled, then lost almost 2/3, bottoming near $18 in April.
In September 2025, Microsoft signed its deal and the stock jumped more than 40% in one day. By October, it hit $141. Then came an AI bubble scare. By February, it had dropped to about $74, almost half off the peak. In March, Nvidia invested $2 billion, and Meta signed a deal worth up to 27 billion. The stock went nearly vertical to a record near $300 in late June.
On July 1st, a report that Meta might start renting out its own spare computing power knocked Nebius down 17% in a day. By late July, it had dipped below 150. In August came Bur short and then that 34% earnings jump. Friday, it closed at 22107. That's about 10 times its first month back on NASDAQ, but still about a quarter below the peak.
Now, the latest quarter reported August 12th. Revenue $582 million, up 454% from a year ago. Annual run rate revenue hit $3 billion at the end of June, up from $1.9 billion in March. adjusted EBITDA was $236 million, a 41% margin. But under standard accounting, it lost $190 million. Depreciation on all those chips was bigger than that EBITDA and interest costs added more.
Nibbia stuck to its targets for this year. 3 to3.4 billion of revenue and 7 to9 billion of run rate revenue by December. Now, the part that matters most for Bur's argument. In that one quarter, Nibbia spent $5.7 billion on chips and data centers. That's almost 10 times its revenue. For the full year, it plans to spend 20 to 25 billion.
So, where does the money come from? about $8 billion of cash on hand, about 8.5 billion of debt, mostly convertible bonds, customer prepayments, which it expects to top $9 billion this year, and new shares through June. It sold about $2.8 billion of stock. So, the bullcase five points. One, explosive growth.
Revenue up 454% and run rate revenue expected to more than double again by December. Two, the demand is signed. More than $40 billion of customer commitments, much of it from Microsoft and Meta, two of the richest companies on Earth. Three, Nvidia is in its corner. Nvidia invested $2 billion and Nibbius gets early access to new chips like Vera Rubin four strong unit economics the AI cloud business already runs at about a 50% adjusted Ebit Dah margin bulls argue that once the building phase slows that turns into serious cash flow five Wall Street likes it 22 analysts a buy consensus and an average target about 28% of of Friday's close and the day after Bur switched to puts BNP Paribas upgraded it to outperform with a 399 target the bull view in one line is a pure play on AI computing and demand is still running ahead of what it can build now the bare case and Bur's version of it one the 1960s leasing trap in the late 1960s companies borrowed to buy IBM computers and rent them out.
They assumed the machines would last 8 years. IBM assumed four. When IBM launched a better machine, rents for the old ones collapsed. And by 1970, leasing stocks had fallen about 80%. Bur says Neoclouds look the same. And this year, Nibbia stretched the assumed life of its servers from 4 years to 5.
Two, the price of urgency. Bur points out that short-term contracts at Nibbius go for $40 to $50 million per megawatt while longer ones go for 20 to 25. Power doesn't wear out, he says. So if customers pay double to get it now something else is losing value fast. The chips, the customers or both. Three big customers can become rivals.
Microsoft and Meta make up a big share of the contracted business. And in July, a report that Meta might rent out its own spare computing power was enough to knock the stock down 17% in a day. For Michael Bur, he shorted at 212, added at 247, and has now moved to puts that need a crash by next June.
He wrote that he's moving his timelines up and wants more leverage. He's not alone. About a fifth of the shares available to trade are sold short. Five, the valuation at $60 billion. Nibbius trades at about 18 times this year's expected revenue. Analysts expect a loss per share this year and next, and every new share sale dilutes existing owners.
The bare view in one line. NBIS is a real business growing fast, but it's built on borrowed money, fastaging chips, and a few giant customers. So, what decides who's right? First, the big tech companies report earnings in the coming weeks. Watch whether Microsoft and Meta keep raising their AI spending.
Then, Nebius's own report expected around mid November. Watch run rate revenue against that 7 to9 billion target. How much power is actually online and how the spending gets paid for? Debt, prepayments, or more new shares. On the chart, the record near 300 is the ceiling. The July low near 146 is the floor.
The average analyst target is about 284. Bur strikes sit below 100. For options traders, Nbius is a wild one. It fell 17% in a day in July and jumped 34% in a day in August. That makes its options expensive. Back in August, Bur said the puts were exorbitantly priced with implied volatility above 100%.
So, he shorted the stock instead. Defined risk spreads let you trade it without betting the account. And Bur's new structure puts out to next June struck below $100. If he's wrong, he loses only the premium. If there's a crash, the payout is big. So, who's right? 22 analysts at 284 or bur betting at more than halves. .
Comments 0