저자: OneCk채널진행자 · 외국인
요약 보기
Nebius(NBIS)는 Nvidia 9.3% 지분, 374억달러 계약잔고, 매출 454% 성장 등 펀더멘털이 강해 AI 인프라 성장 수혜주로 평가. 가격 인상(17~21%)과 높은 수익성으로 회수기간 단축. 반면 Michael Burry는 GPU 감가상각(5년 vs 2~3년)과 대규모 자금조달 격차를 이유로 풋옵션 베팅. 진행자는 공매도 잔고 18.49%와 레버리지 ETF 수급으로 숏스퀴즈 가능성을 언급하며 단기 $254.50 돌파, 장기 $320~399 목표 제시. 단, 진행자는 투자 권유가 아니라고 명시.
다듬은 글
Okay, I want to start with a question, because honestly, this might be the strangest situation in the entire AI market right now. Did you know that a company that most people have n't even heard of has grown by more than 180 % this year, has a portfolio of signed contracts worth about $37 billion, and received a direct investment from Nvidia itself?
And yet, at the same time, Wall Street can't agree on how much it's actually worth. Price targets range from a low of $84 to as high as $410 per share . This company is Navitas Group , ticker NVTS. It currently trades somewhere between $242 and $248 per share with a market capitalization in the range of $61 billion to $65 billion.
Its 52-week range ranges from $73.52 all the way up to $299.86. And if you've been following the news headlines, you already know why she's so controversial. On one hand, you have retail interest and large institutions investing massively. On the other hand, you have legendary shortstops, and yes, I mean Michael Burry, the one from the " Downside Game," who actively plays against it.
So, the whole point of this video really boils down to one thing. Is Navitas the perfect bet on AI hypergrowth, or is it an over- leveraged bubble about to burst? And I'll be honest with you, when I started this research, I thought it was probably a bubble. But I finished it, to be honest, not so sure, which in itself makes the action worth discussing.
Before we go any further, a quick request . If you are one of those who likes to understand what is really going on "under the hood" and not just chase green candles on the chart, do me a favor — like and subscribe. This is genuinely helpful, and tells me you want more of these kinds of breakdowns here on 1CK.
Okay, let's begin. And now it gets interesting, because I want to give you my verdict right away, instead of making you wait 15 minutes for it. In my opinion, Navitas is a sure buy on pullbacks, but only if you are disciplined in your entries and setting stop losses. And I want to emphasize: these are my own considerations, not a recommendation.
More about this at the end. That's why I think so. Nebius has quietly transformed from what was essentially a “ GPU rental company” into something much more—a powerful full-cycle company called Neocloud. And when I analyze this thesis, it rests on three pillars. First, the strategic moat, which is really difficult to overcome, is entrenched by Nvidia.
Second, a revenue driver supported by some of the biggest names in the tech industry signing multi-year deals. And third, a price catalyst that I think most investors completely missed, and it happened just a few weeks ago. And I want to say right away: these three pillars are exactly why I didn't dismiss this company as just another AI-hyped stock.
But you have to evaluate them for yourself, so let's really test each one for strength, instead of taking everything on faith. To understand what Nebius really does, we have to look at who holds the keys behind the scenes, and that's Nvidia. So Nvidia doesn't just supply Nebius chips, they literally own 9.3% of the company , that's 22.25 million shares.
I want you to think about this for a moment because this is not a typical supplier relationship. When your supplier owns a portion of your capital, the interests become very, very aligned. This ownership effectively buys Nebius VIP access to Nvidia's next-generation Vera Rubin NVL 72 racks deployed in their flagship data center in Finland.
And if you follow AI hardware even a little bit, you know how crazy the waiting lines are right now. Companies wait months, sometimes longer , just to get their hands on current-generation equipment. Nebius skips this turn entirely. Their competitors are literally stuck in line while Nebius plugs in next- generation racks.
And this goes beyond Nvidia. Palantir has officially selected Nebius as its preferred sovereign AI infrastructure partner. This is a big deal because sovereign AI, the desire of governments and states to have their own AI infrastructure, is one of the fastest growing and least competitive segments in this field .
On top of that, Nebius was recently added to the FTSE All-World index, meaning index funds are now forced to automatically buy millions of shares. They do n't even have a choice. The money just keeps coming. So when I say "moat ", I don't mean a fancy brand name. I mean advantages that a competitor literally can't buy, because you can't just buy Nvidia owning your stock.
This is not a partnership that can be negotiated. This is a signal. So, the moat is real. But having Nvidia on your side is a huge advantage that is worthless if the financials don't match. So let's look at the real numbers. And if we dig deeper, we find, frankly , one of the most absurd earnings reports I've seen in the tech sector.
The group's revenue grew by 454% year-on-year to $582.3 million . This crushed Wall Street's estimates. Their specialized AI cloud technology segment grew 514% to $574.9 million, now accounting for 98% of total revenue. So, this is not a company with a bunch of random lines of business. This is , in essence, one focused growth engine.
At first glance, it looks great, but I've learned to be careful when a company grows so quickly, because hypergrowth can hide many problems. So let's look at this in more detail. That's what really surprised me. Adjusted EBITDA went from a $21 million loss last year to a $236.2 million profit. This is a profit margin of 41 to 50%.
A company that grows so quickly usually loses money, but Nebius does the opposite. And there is also an indicator that made me stand up straight. Annual revenue at the current rate has already reached $3 billion , and management is forecasting $7 billion to $9 billion by the end of the year. So the trajectory here is not linear.
It is vertical. If you think $3 billion in annual revenue sounds impressive, wait until you see their confirmed order book, because this is where it stops being a story about growth and becomes a story about unfulfilled commitments. Nebius has a huge contract backlog of $37.5 billion in performance obligations.
This is not a list of possible deals. These are signed contractual obligations. And the names of the companies on the other side of these contracts are simply amazing. $27 billion commitment from Meta and $17.4 to $19.4 billion from Microsoft. We are talking about multi-year mega-deals with the two largest AI vendors on the planet.
But here's a hidden metric that I think protects Nebius from what kills most infrastructure companies— capital shortages. 70% of their second- quarter contracts include upfront customer payments, covering 50 to 60% of the total capital expenditure for the equipment. Customer down payments are projected to exceed $9 billion this year alone.
Read this again. Hyperscalers are effectively paying to build Nebius infrastructure for them. This is a completely different risk profile than a company that borrows billions and prays for demand to emerge. Okay, huge contracts—that's great. But what happens when demand for computing power grows so rapidly that a company can simply set any price it wants?
That's exactly what happened, and I don't think enough people are talking about it. On October 1, while traditional cloud providers were cutting prices to attract customers, as they always do, Nebius did the opposite. They increased GPU cloud prices by a total of 17–21% . Let me give you specific numbers, because details matter.
The hourly rate for the Nvidia H100 increased from $3.85 to $4.50. Their flagship B300 instances have increased in price from $7.85 to $9.50 per hour. The cost of services based on AMD EPYC processors has increased by 25%. Managed token endpoints are up 18.3%, with their latest Blackwell capacity auction closing 15% above previous peak levels.
Think about what this number actually means at auction. They literally put the capacities up for auction, and they sold for more than ever before. In a market so desperately in need of computing resources, Nebius is not a price-sensitive player . They have an impact on pricing, a real impact, and that's why it's important to the business model.
When you can raise prices and customers still line up , you are no longer competing on cost, but on availability. This is a completely different game. And when you combine price increases with high demand, the math of return on capital changes dramatically. So, let's do these calculations, because that's what convinced me.
Standard long-term contracts generate $20 to $25 million in annual contract value per megawatt. This is the basic level. But for short-term contracts of 3-6 months, desperate AI labs are paying $40 to $50 million per megawatt, almost twice as much. And thanks to the October price increase, Nebius reduced the hardware payback period from 2-3 years to just 1 year and 10 months.
Honestly , I didn't expect this. Think about this for a moment. They are recovering their multi-million dollar investment in servers in less than 22 months. Each subsequent month is, in essence, a net cash flow. This is the difference between a capital-intensive business and a capital-efficient one, and it is the difference that most analysts fail to take into account when modeling these companies.
Getting more out of your hardware is critical right now, but Nebius has just solved an even bigger software problem that plagues the entire GPU industry, and it's a really smart solution. Just a few days ago, Nebius acquired an inference software specialist called Inferize, and they are using it to power their own Nebius Token Factory.
Here's the problem this solves. In traditional data centers, GPUs sit idle between complex workloads. This is a huge loss of profit, and in the industry it is called a “cold start” of an idle GPU. Every second the chip is doing nothing is money you've already spent and will never get back. The Inferize software dynamically packages and distributes inference tokens between underloaded chips.
Simply put, it makes the equipment work . This increases the profitability of each server without spending a single additional dollar on hardware. This is margin growth that does not require any new purchase orders. And in a business where the largest expense item is equipment, this is huge. So, with all this going well, big institutional money has quietly started pouring into the company.
But that's only half the story. Leading institutions have been aggressively increasing their stakes. BlackRock increased its position to 3.93%. Fred Alger holds 3.79%, along with Susquehanna, Jane Street and Citadel. And here is the statistic that caught my attention. Seven of the eight well-known market gurus have recently increased their stakes.
This is not retail FOMO. This is some of the smartest money in the world, quietly building up positions. And when such names appear on the list of shareholders, it changes the view on price movements, because large funds do not chase the market, they accumulate. And accumulation looks exactly like what we observe.
Now, headline watchers can point to $179.9 million in insider selling last year and say, "Yeah, insiders are dumping stocks." At first glance, it looks terrible. But if you consider the context, the story changes completely. These were pre- planned automatic trades under the 10 b 5-1 rule, scheduled months after the stock's massive 180% rally .
And the chairman of the board of directors still owns $85 million worth of stock. If he really thought it was a peak, he wouldn't have kept them. Of course. So, with big buying and insiders effectively just rebalancing portfolios, why is Wall Street so radically divided on the future of these stocks?
This is the question that led me into the jungle of analytics. The answer is that analysts are fighting a fierce battle, creating a staggering price target range of $326. I have honestly never seen such a wide range for a company of this size. On the optimistic side, Northland Securities sets the highest price on the market at $410.
BNP Paribas Exane raised its valuation to $399, predicting annual revenue will reach $22 billion by 2027. Truist predicts $355. Bank of America expects revenue to reach $27 billion by 2028. The consensus forecast is in the range of $283–289, which implies an upside potential of about 20%. But at the very bottom of this range, we have Rothschild Redburn with a "sell" rating at $84.
$84—the company says this asset is worth about a third of its current price. And this cannot be simply ignored because someone with serious research analyzed the same company and came to the opposite conclusion. So, all of this is irrelevant unless we consider the risks that inspired this pessimistic $84 forecast.
And this is where the story gets really awkward. Let me think about this, because the person who is playing against these stocks is not some random social media pessimist. Legendary investor Michael Bury, known for his big shorts, purchased put options on Neebius with an expiration date of June 2027.
And I want to be fair to him, because his argument is not absurd. In fact, this is the most sensible " bearish" forecast I've read about this entire sector. Bury's main argument is this : Neo clouds overstate their true profits by amortizing GPUs over 5 years, while AI hardware obsolescence occurs in 2-3 years.
That's the whole point. If you amortize a chip for 5 years but have to replace it after two, your reported profit is a fiction. You are effectively borrowing from the future and calling it income. And there is also a balance sheet. Neebius plans capital expenditures of $20 billion to $25 billion in 2026 on revenue of just $3 billion .
This leaves a colossal $200 billion funding gap for their long-term 5 gigawatt roadmap. Meanwhile, total debt rose to $10.06 billion . So yes, when I put it this way, the "bear" scenario has merit. The question is whether growth will outpace the need for financing, and that's really hard to know. Anyone who tells you they know for sure is just guessing.
But while the risks of debt and equity dilution are real, something is brewing at a technical and structural level that could trigger a massive short squeeze. And this is the part where market mechanics get really interesting. Short positions (shorts) quietly increased to 46.78 million shares. This represents 18.49% of the total free float and takes only 3.6 days to cover.
To make matters even crazier, the Leveraged Shares 2X Long NBISE ETF holds over 50% of its portfolio in Nabia shares. Half of an entire ETF is one stock. This means that every time money comes into this fund, it is forced to buy more Nabia shares. This is a mechanical amplifier. And with a massive open interest in calls at the $ 250, $260, and $280 strike levels , any positive catalyst could trigger a spiral of forced buying through a gamma squeeze and a short squeeze.
Simply put: if the stock breaks these levels, market makers will be forced to buy shares to hedge, which will push the price higher and force even more buying. This is a feedback loop. And with 18.49% of shares in shorts, there is plenty of fuel here. So, when we look at the actual price chart, does the technical situation confirm that a big move is coming?
Honestly, there is something like that. Let me show you what I mean. The chart is currently in what technical analysts would call a textbook “ bullish” consolidation wedge. The stock is trading well above its 20-day moving average at $221 and its 50-day at $214, comfortably holding 50% above its 200- day trend line at $162.
This is a stock in a clean, established uptrend, with no broken structure or lower highs. The 14-day RSI is at a healthy 55.0 , which is not an overbought zone, and this is important because it means there is potential for growth without overheating. Momentum indicators are 97 out of 100, and a close of the day above key resistance at $254.50 paves the way for a quick technical test of the all-time high at $299.86.
So, the situation is this: a strong trend, healthy momentum, a tight consolidation, and a huge short position underneath it. This is a powder keg, and frankly, technical analysis is not magic. It does not predict the future. But it shows where pressure is building up, and right now that pressure is pointing upwards.
So how exactly do you trade this asset for maximum profit with minimal risk? Of course, this is just educational material. This is my view and it is not financial advice. But let me show you the levels I will be monitoring. For short-term traders, the approach would be to buy on intraday pullbacks to the $235– $240 zone or enter on a high-volume close above $254.50.
Profit targets will be at $260, $ 280, and ultimately $298 with a strict stop loss at $224.50. This gives a risk-to-reward ratio of about 3.2 to 1, which is exactly the asymmetry you need. For long-term investors, the strategy is different. Form the main position in three tranches. The first tranche is currently priced at $238 to $244.
The second tranche is on a pullback to $216–$221, and the third is at $200. The target is $320 for 12 months and $399 for 18 months. And the thesis loses relevance only when the stock closes below $ 185. Below this level, the entire structure becomes invalid, and the situation needs to be reassessed , which brings us to the main question.
Why am I still calling Nvidia an attractive buy right now, even after showing you the bearish scenario? Here is the main argument, and it is simple. Nvidia isn't just riding the wave of AI hype , they're supplying the engines for it. With Nvidia's 9.3% ownership stake, a $37.5 billion order book , a profit margin of over 50%, and an undeniable price advantage, the hyper-growth momentum outweighs the short-term capital risk.
In my opinion, the bearish scenario is real, but it is a timing risk, not a collapse of the entire thesis. As long as demand holds, growth is outpacing debt, and I want to be upfront about how I assess that. I'm not saying there aren't risks. I'm saying that the market is pricing these risks as if growth will stop tomorrow, while the contracts say otherwise.
Before we finish, I have something to say . This is not financial advice. I am not a financial advisor. I am not your financial advisor. And nothing in this video is a recommendation to buy or sell anything. Everything here is my own opinion and research, shared for educational and entertainment purposes.
Do your own analysis and never invest money you cannot afford to lose. Markets are unpredictable, and any stock, especially one that moves so quickly, can depreciate to zero. Now I want to hear from you because I'm really interested in this topic. If Michael Burry is right about GPU obsolescence, and these neo- clouds are really overstating their profits, what does this mean for the entire field of AI infrastructure?
Is Burry seeing something that the entire market is missing, or is he just jumping to conclusions? Share your thoughts in the comments below. I read them, and the best ideas really influence what I explore next. And I'm not asking this for a fancy word. I hesitated about this for weeks . So, if you see something I missed, I sincerely want to hear about it.
And if you've watched this far , thank you, seriously. Hit the "like" button if this breakdown gave you food for thought, and subscribe to 1CK if you want more in-depth analyses like this , where we get to the heart of the matter rather than just reacting to headlines. That's all for today. Take care of yourself, stay disciplined, and see you in the next video.
Comments 0