The AI Infrastructure Stock That Escaped Russia
From $30 billion Russian tech giant to delisted and sanctioned, to 500% gains in three months. The unlikely story of Nebius and why it matters for AI investors.
Disclaimer: This article is for educational and informational purposes only. The author is not SEBI registered and this does not constitute investment advice or recommendations of any kind. Do your own research before making any investment decisions.
Most investors have never heard of Nebius. They know Nvidia. They know the hyperscalers. They’ve probably heard someone at a dinner party mention CoreWeave. But Nebius? The company that went from being worth $30 billion to getting delisted, sanctioned, and left for dead, only to crawl back onto the Nasdaq and rip 500% in three months?
That story doesn’t get told enough. And the setup right now is worth paying attention to.
The Origin Story: From Russia’s Google to AI Infrastructure Play
To understand Nebius, you need to understand Arkady Volozh. He’s the guy who built Russia’s answer to Google, and now he’s building something entirely different in the West.
Volozh started working on search technology in 1989. Let that sink in. While most of the world was still figuring out email, this guy was building search engines. He co-founded Yandex in 1997 and spent the next two decades turning it into Russia’s everything company. Search. Maps. Ride-hailing. Payments. Cloud computing. By 2021, Yandex was worth $30 billion and traded on the Nasdaq under the ticker YNDX.
Then Russia invaded Ukraine. Everything changed.
Nasdaq suspended trading in February 2022. The stock was frozen. Volozh himself got hit with EU sanctions. The company most investors had known was effectively dead. But Volozh had been living in Tel Aviv since 2014. He saw what was coming. And he started planning the escape.
Over the next two years, Yandex executed the largest corporate exit from Russia in history. They sold all Russian assets for $5.2 billion to a consortium of Russian investors. They moved hundreds of engineers out of the country. Volozh publicly disavowed the invasion, which got his sanctions lifted in March 2024.
What emerged from the wreckage was Nebius: a company stripped of its Russian operations but loaded with something valuable. Two decades of infrastructure expertise. Relationships with Nvidia that made Yandex one of their largest customers outside the US and China. Data centers in Finland, France, and Iceland. And a team that had been building high-performance computing systems since before most AI startups existed.
The company rebranded, changed its ticker to NBIS, and returned to the Nasdaq in October 2024 at around $17 per share. As of late January 2026, it trades above $85. That’s roughly 500% in fifteen months.
What Does Nebius Actually Do?
Here’s the simple version: Nebius rents out GPUs to companies building AI.
But that undersells it. What they’ve actually built is full-stack AI infrastructure. They don’t just buy Nvidia chips and rack them up. They design custom servers. They build their own data centers. They operate proprietary cooling systems. They’ve created an entire cloud platform optimized specifically for AI workloads.
Think of it this way. If you’re an AI company and you need to train a large model, you have three options. You can go to the hyperscalers, where you’ll compete with everyone else for limited GPU capacity and pay premium prices. You can try to build your own infrastructure, which takes years and billions in capital. Or you can go to a specialized provider like Nebius that focuses exclusively on AI infrastructure.
Nebius calls itself a “neocloud.” The positioning is deliberate. They’re not trying to compete with AWS on general cloud computing. They’re not selling storage and databases and serverless functions. They do one thing: provide the compute infrastructure AI companies need to train and run models. Everything in their stack is optimized for that single use case.
The secret sauce is their vertical integration combined with their Nvidia relationship. They’re not just a customer buying chips at retail. They’ve been partners since the Yandex days. Nvidia participated in Nebius’s private placement as an investor. That relationship gives them priority access to hardware that others can’t get.
The Infrastructure Moat: Why This Business Is Hard to Replicate
Building AI infrastructure at scale isn’t like spinning up a SaaS company. You can’t do it in a garage with a laptop and a dream. The barriers to entry are enormous.
Capital intensity is brutal. A single data center with modern GPU clusters costs hundreds of millions of dollars. Nebius has facilities in Finland, France, Iceland, Missouri, and a new one under construction in New Jersey. That’s billions in physical infrastructure that took years to build.
The supply chain is constrained. Nvidia GPUs remain in short supply. Companies that established relationships early, like Nebius through its Yandex history, get priority allocation. New entrants go to the back of the line.
The technical expertise runs deep. Running GPU clusters efficiently isn’t just about plugging in hardware. It’s about cooling systems, networking, software optimization, and operational expertise built over years. Nebius has been doing high-performance computing since Yandex was building search infrastructure. That operational knowledge is difficult to acquire quickly.
The contracts create stickiness. Once an AI company is training models on your infrastructure, with their data and workflows integrated into your platform, they don’t switch easily. The relationships tend to be multi-year and expanding.
Here’s a stat that illustrates the competitive position: Nebius operates the ISEG supercomputer, which ranks 13th globally in computing power. That’s not something you can replicate by raising a Series B.
The Customer Base: Microsoft, Meta, and the Hyperscaler Irony
The most interesting thing about Nebius’s customer list is who’s on it. Microsoft. Meta. The hyperscalers who supposedly dominate cloud computing.
Think about what that means. These companies have more compute infrastructure than almost anyone on Earth. They could build their own GPU clusters. They have the capital. They have the engineering talent. And yet they’re signing multi-billion dollar contracts with Nebius.
In September 2025, Nebius announced a $17.4 billion deal with Microsoft to supply AI infrastructure over five years, with potential expansion to $19.4 billion. A month later, they signed a $5 billion three-year deal with Meta.
Why would the hyperscalers outsource this? Because demand for AI compute has outpaced even their ability to build. Every AI company on Earth wants GPU capacity right now. Microsoft is racing to support OpenAI and their own Copilot products. Meta is building Llama and needs infrastructure for training runs that consume enormous resources. Building new data centers takes years. Nebius has capacity available today.
These contracts also validate Nebius’s technical capabilities. Microsoft and Meta don’t sign billion-dollar infrastructure deals with vendors who can’t deliver. The fact that they chose Nebius over alternatives tells you something about the quality of what Nebius has built.
The Financial Picture: Explosive Growth, Premium Valuation
Let’s look at the numbers because they tell an unusual story.
Revenue Growth: The company generated $146 million in revenue last quarter, up 355% year-over-year. That’s not a typo. Triple-digit growth from a company that already has substantial infrastructure.
Forward Guidance: Management is targeting annualized revenue between $7 billion and $9 billion by end of 2026. Against expected 2025 revenue of around $550 million, that implies they expect revenue to grow roughly 13-16x in a single year. Those Microsoft and Meta contracts are obviously driving this projection.
Current Valuation: The stock trades at roughly $85-90 per share with a market cap around $22 billion. On trailing revenue, that’s expensive: about 40x sales. But against the 2026 guidance, the multiple looks more reasonable if they execute.
Profitability: The company is not yet profitable on a GAAP basis and has negative free cash flow. This is the nature of infrastructure businesses in growth mode. You have to spend billions building capacity before you can generate the revenue. The Microsoft and Meta contracts should dramatically improve the picture as that infrastructure comes online.
Balance Sheet: Nvidia led a $700 million investment round in 2024, which strengthened the balance sheet. The company has the capital to fund expansion without dilutive equity raises in the near term.
The bull case on valuation is straightforward: if they execute on the $7-9 billion revenue guidance, the current market cap is a bargain. The bear case is equally clear: if something goes wrong with those contracts, the valuation looks extremely stretched on current fundamentals.
The Competitive Position: Where Nebius Fits
AI infrastructure is a growing market with several players. Understanding where Nebius sits helps evaluate the investment case.
vs. The Hyperscalers (AWS, Azure, Google Cloud): These are the 800-pound gorillas. They have massive infrastructure, established customer relationships, and unlimited capital. But they also have constraints. They’re serving millions of customers across every workload type. Their GPU allocation gets spread thin. Nebius focuses exclusively on AI infrastructure, which means AI-first optimization and dedicated capacity.
vs. CoreWeave: The most direct competitor and the one investors often compare to Nebius. CoreWeave has raised massive funding and is building GPU cloud capacity rapidly. The difference: Nebius has two decades of operational history and existing Nvidia relationships, while CoreWeave is newer but growing fast. Both can probably win in a market this large.
vs. Lambda Labs / Together AI / Specialized Providers: These companies also provide AI compute, but typically at smaller scale or focused on specific use cases. Nebius operates at enterprise scale with the contracts to prove it.
vs. On-Premise Build-Out: Some large enterprises try to build their own GPU clusters. This makes sense for a few massive players but is impractical for most. The capital requirements, operational complexity, and supply chain challenges make Nebius’s rental model attractive for everyone except the largest tech companies.
The positioning is: full-stack AI infrastructure at hyperscaler scale, with the operational expertise of a company that’s been running high-performance computing for two decades. Not many companies can make that claim.
The Nvidia Factor: Why the Partnership Matters
The Nvidia relationship deserves its own section because it’s central to the thesis.
During the Yandex era, the company was one of Nvidia’s largest customers outside the US and China. They ran massive GPU clusters for search, machine learning, and autonomous driving research. That relationship survived the restructuring.
When Nebius raised capital in 2024, Nvidia didn’t just participate; they led the round. Having Nvidia as an investor aligns incentives. Nebius gets priority allocation of the most sought-after chips in the world. Nvidia gets a customer committed to deploying their hardware at scale.
This matters because GPU supply remains constrained. Companies are waiting months for hardware. The ability to get chips faster than competitors is a significant advantage in a market where customers need capacity immediately.
Nebius has also announced plans to be an early adopter of Nvidia’s next-generation Rubin platform. Getting early access to new architectures before competitors can be the difference between winning and losing major contracts. Nebius is definitely among the leaders in this race.
Beyond the Core: The Other Nebius Businesses
Nebius Group isn’t just cloud infrastructure. They have several other businesses that came out of the Yandex restructuring:
Toloka: A data labeling platform for AI. Training models requires enormous amounts of labeled data. Toloka provides the workforce and platform to create it. This is a picks-and-shovels play on the same AI trend.
Avride: An autonomous driving company developing robotaxis and delivery robots. This was Yandex’s self-driving division. It’s still early, but the technology has logged millions of test miles.
TripleTen: An ed-tech business focused on reskilling people for tech careers. Different from the core thesis but generates revenue.
ClickHouse Stake: Nebius founded ClickHouse, the popular open-source analytics database, and retains a stake. ClickHouse has become widely adopted and could be valuable as an independent company.
These businesses are optionality. The core thesis is AI infrastructure, but the other segments could become meaningful over time.
The Technical Setup
Let’s look at what the chart says. As of late January 2026, shares trade around $85-90 after pulling back from highs above $140.
The stock ran hard after the Microsoft deal announcement in September 2025, nearly doubling in a few weeks. That kind of move needs to consolidate. The pullback from $141 to current levels is roughly 35%, which shakes out momentum traders and creates a more sustainable base.
Volume patterns suggest institutional accumulation during the pullback. Large buyers tend to come in when stocks pull back to areas of previous support. The $80-85 zone appears to be where they’re comfortable adding.
The stock has earnings coming up on February 12, which creates a near-term catalyst. Beat expectations, and the stock likely rips. Disappoint, and it could retest lower levels. The setup is binary around that event.
The Numbers to Watch
| Level | Price | What It Means |
|---|---|---|
| Current Price | ~$87 | Where we are now |
| Near-term Support | $80-82 | Area of recent buying interest |
| Deeper Support | $65-70 | Previous consolidation zone |
| Previous High | $141 | Major resistance on any rally |
| Downside Risk | ~$65 | Roughly 25% below current |
The risk/reward depends heavily on your view of the contracts. If Nebius executes on the Microsoft and Meta deals, this stock could trade significantly higher. If something goes wrong, the valuation on current revenue doesn’t provide much cushion.
Key Metrics at a Glance
| Metric | Value |
|---|---|
| Market Cap | ~$22B |
| Revenue (TTM) | ~$550M |
| Revenue Growth | 355% YoY (last quarter) |
| 2026 Revenue Target | $7-9B (annualized) |
| Key Contracts | Microsoft ($17.4B), Meta ($5B) |
| Nvidia Relationship | Strategic investor, priority allocation |
| Data Centers | Finland, France, Iceland, Missouri, New Jersey (building) |
| Founded | 1997 (as Yandex), 2024 (as Nebius) |
| Stock Performance | +500% since Oct 2024 |
The Bull Case
Here’s how this plays out optimistically:
The Microsoft and Meta contracts ramp as expected. Revenue goes from $550 million to the $7-9 billion guidance in 2026. Suddenly the stock looks cheap on forward revenue. More hyperscaler deals get announced. The pipeline is full because AI demand isn’t slowing down.
Margins improve as infrastructure scales. The fixed costs of data centers get spread across more revenue. Operating leverage kicks in. The company moves toward profitability.
The Nvidia relationship continues producing advantages. Early access to new chips lets Nebius win deals competitors can’t. The GPU supply shortage persists, making their allocation advantage more valuable, not less.
Multiple expansion happens as Wall Street recognizes the growth and competitive position. Analyst coverage increases. Institutional ownership grows. The stock gets rerated from “speculative” to “high-growth infrastructure.”
From here, a move back to previous highs of $141 would be roughly 60% upside. If the 2026 guidance materializes and the multiple holds, significantly higher prices are possible.
There’s also M&A optionality. A company with Nebius’s infrastructure, contracts, and Nvidia relationship would be attractive to someone looking to enter the AI infrastructure market or expand their position. Any acquisition speculation would lift the stock.
The Bear Case
Nothing goes up in a straight line. Here’s what could go wrong:
Contract Execution Risk: The revenue guidance depends entirely on the Microsoft and Meta deals ramping as expected. These are large, complex infrastructure projects. Delays happen. Requirements change. If the ramp is slower than expected, the stock could fall hard.
Valuation Without Contracts: Strip out the forward revenue assumptions and look at trailing numbers. The company trades at roughly 40x trailing revenue. If something causes those contracts to stumble, there’s significant downside to a more normalized valuation.
Competition Intensifies: CoreWeave and others are building capacity rapidly. The hyperscalers may decide to serve their own AI infrastructure needs more aggressively. A price war could compress margins even if revenue grows.
GPU Supply Normalizes: Part of Nebius’s advantage is constrained supply giving them allocation priority. If Nvidia eventually floods the market with chips and supply exceeds demand, that advantage disappears.
Geopolitical Risk: The company’s origin story involves Russia, sanctions, and restructuring. While they’ve successfully exited Russia, there’s always headline risk. A new geopolitical development could create uncertainty even if it doesn’t directly impact operations.
Customer Concentration: Microsoft and Meta represent a huge portion of expected future revenue. Losing either relationship, or having them scale back, would be devastating.
Macro Headwinds: If the broader market sells off, high-multiple growth stocks get hit hardest. In a risk-off environment, Nebius is vulnerable despite strong fundamentals.
The Bottom Line
Nebius is one of the stranger stories in the market. A company that was essentially dead two years ago. That escaped from Russia and rebuilt itself in the West. That now has contracts with Microsoft and Meta worth over $22 billion combined.
The founder, Arkady Volozh, has been building technology companies for 35 years. He navigated Yandex through everything from the 2008 financial crisis to Russian sanctions to complete corporate restructuring. Whatever you think about the stock, underestimating his ability to execute would be a mistake.
The fundamental question is simple: do you believe the contracts will ramp as guided? If yes, the stock is probably undervalued. If you’re skeptical, the trailing valuation looks stretched.
For investors looking for AI infrastructure exposure beyond Nvidia, Nebius offers a direct play on the picks-and-shovels thesis. They’re not building models. They’re not betting on which AI application wins. They’re providing the compute capacity that everyone needs to build AI. As long as the AI investment cycle continues, companies like Nebius should benefit.
The technical setup supports accumulation on pullbacks to support levels. The earnings date creates a near-term catalyst. The contract book provides visibility that most growth companies don’t have.
This isn’t a safe, boring investment. It’s volatile, concentrated, and dependent on management execution. But for investors who can handle the risk, the potential upside if everything works is substantial. The AI infrastructure buildout is a multi-year trend. Nebius is positioned to be one of the companies that enables it.
This analysis is for educational and informational purposes only. It does not constitute financial advice, investment recommendations, or an offer to buy or sell any securities. The author is not a SEBI-registered financial advisor or licensed investment professional. The rankings and assessments reflect one analytical framework among many possible approaches and should not be interpreted as buy, sell, or hold signals. Always consult a qualified financial professional before making investment decisions. Past performance does not guarantee future results.