What Is Nebius Group and Why Did Its Biggest Customer Just Become Its Rival?

Most investors had never heard of Nebius Group eighteen months ago. Today NBIS stock is a $60 billion company that joined the Nasdaq-100 in June, counts NVIDIA as a direct investor, and has a five-year contract with Meta worth up to $27 billion. It is also down approximately 15% today because that same Meta just announced it is entering Nebius's market.

The Nebius story is unusual enough that the background matters as much as today's news, and understanding one requires understanding the other.

Where Nebius Came From

Nebius Group is not a startup. It is a company with roots that trace back to one of the most dominant technology businesses in Russian history.

Yandex was Russia's Google — a search engine that became the country's most valuable technology company and expanded into ride-sharing, cloud services, food delivery, and dozens of other digital businesses over two decades. When Russia invaded Ukraine in 2022, the business became untenable in its original form. International investors could not hold Russian assets. The company faced sanctions exposure. And the people running its technology businesses largely wanted to operate outside Russia.

The restructuring that followed was complex. The Russian operations were separated and sold to local buyers. The international technology businesses, primarily the cloud and AI infrastructure operations, were reorganized under a new holding company and listed on Nasdaq under the name Nebius Group. The CEO is Arkady Volozh, Yandex's founder, who had renounced his Russian citizenship and relocated to Europe.

What Nebius inherited from Yandex was not just people and technology. It inherited a genuine engineering culture that had spent years solving large-scale distributed computing problems at a level that most Western startups have not had to tackle. The team that built Yandex's infrastructure brought that experience into building GPU clusters and AI cloud platforms for a completely different market.

What Nebius Actually Builds

The business Nebius operates today is fundamentally about one thing: giving AI companies access to GPU computing power without requiring them to build their own data centers.

Training large language models, running inference at scale, and developing AI applications all require vast amounts of GPU compute. The companies building those models and applications — from frontier AI labs to enterprise software companies adding AI capabilities — have three basic options: build their own infrastructure, rent from the hyperscalers like AWS, Azure, and Google Cloud, or use specialized AI cloud providers like Nebius.

Nebius competes in the third category, which the industry calls neocloud. The neocloud value proposition is specific. The hyperscalers offer broad product suites covering storage, databases, networking, and dozens of services alongside compute. Nebius offers a focused, GPU-optimized environment that can be faster to provision, more flexible in configuration, and purpose-built for AI workloads rather than general computing.

The company operates GPU clusters built on NVIDIA hardware, offers cloud platforms where developers can deploy AI models and applications, and provides tooling for the specific workflows that AI development requires. It is not trying to be AWS. It is trying to be the best possible option for the specific use case of AI infrastructure.

The Numbers That Made NBIS a $60 Billion Company

The speed of Nebius's ascent from obscure spinoff to major index constituent is explained entirely by one thing: the numbers.

Q1 2026 revenue came in at $399 million, up 684% year over year. AI cloud revenue within that grew 841% year over year to $389.7 million. Core AI cloud annualized recurring revenue reached $1.92 billion exiting Q1, up 54% quarter over quarter. Management guided contracted capacity to 4 gigawatts for the full year.

Those growth rates are not normal. They reflect a company that found itself in the right position at the exact moment the AI infrastructure buildout accelerated beyond what the existing hyperscalers could supply quickly enough. When demand for GPU capacity exploded faster than AWS, Azure, and Google Cloud could provision new data centers, companies like Nebius filled the gap.

CEO Arkady Volozh consistently described the constraint as supply rather than demand. Four or more customers were competing for every GPU cluster Nebius brought online. The company was not selling in a competitive market where it had to fight for customers. It was allocating scarce capacity to the customers with the best contracts and the longest commitments.

That supply-constrained dynamic is what made the business model extraordinary and also what made the Meta relationship so central to everything.

The Three Deals That Built the Current Nebius

Three specific transactions in the first half of 2026 transformed Nebius from a fast-growing AI infrastructure company into a genuinely large-cap name with Nasdaq-100 status.

In March, Nebius announced the Meta deal: a five-year agreement valued at up to $27 billion, including $12 billion in dedicated NVIDIA Vera Rubin compute and $15 billion in committed capacity purchases. At the time, this was the largest customer agreement in the company's history by an enormous margin and immediately established Meta as Nebius's defining customer relationship.

Also in March, NVIDIA announced a $2 billion direct investment in Nebius. This was not a routine financial investment. NVIDIA investing in an AI cloud provider is a supply chain move. It gives Nebius priority access to the next-generation GPU platforms that customers will be competing for. The NVIDIA investment validated Nebius's technical capabilities and supply chain position in a way that no analyst note could replicate.

In June, Nebius closed its acquisition of Eigen AI, adding inference optimization capabilities to the platform. And on June 22, Nebius joined the Nasdaq-100, pulling in passive index buying and broadening its institutional shareholder base significantly.

Within roughly four months, Nebius went from a fast-growing neocloud to a company with a $27 billion customer contract, a $2 billion strategic investment from the most important hardware company in AI, an inference acquisition, and a major index inclusion. The stock reflected all of that in its price.

Why the Meta Announcement Is a Different Kind of Problem

Most competitive threats to a business come from rivals. The threat Nebius faces today is more unusual and in some ways more structurally challenging: it comes from its biggest customer.

When Bloomberg reported that Meta is organizing Meta Compute to sell AI computing capacity to outside customers, the competitive implications for Nebius were immediate and specific. Meta is not some distant new entrant that might compete with Nebius in three years. Meta is the company whose $27 billion contract represents the backbone of Nebius's contracted revenue and the primary validation of its market position.

The dynamic this creates is one that corporate strategists have a name for: a customer-competitor relationship, sometimes called a coopetitor situation. Nebius and Meta will simultaneously be executing the existing five-year capacity agreement while Meta builds infrastructure that could eventually make it less dependent on Nebius's services.

This is different from Meta simply being a tough negotiator or reducing its orders in a given quarter. It is Meta developing an internal capability that, if successful, changes the nature of the relationship from dependency to optionality over a multi-year horizon.

The Structural Vulnerability It Revealed

The Meta announcement did not create a new weakness in Nebius's business so much as it illuminated a structural vulnerability that existed before today.

Any business with a single customer representing a disproportionate share of contracted revenue carries concentration risk. The $27 billion Meta deal is extraordinary in absolute dollar terms, but it also means that Nebius's long-term trajectory is unusually dependent on Meta's continued appetite for third-party AI infrastructure capacity.

In a world where Meta never builds its own cloud business, that concentration is a strength — a deep, multi-year partnership with the company building some of the world's largest AI systems. In a world where Meta is simultaneously developing internal capacity to serve outside customers, that concentration becomes a vulnerability, because the very scale of the Meta relationship means any reduction in Meta's dependence on Nebius has an outsized impact on the business.

The question Nebius now needs to answer, and that investors need to watch for over the coming quarters, is whether the company can build a customer base diverse enough that no single relationship, however large, determines the long-term outcome. The supply-constrained dynamic that Volozh has described, with multiple customers competing for every cluster, suggests the demand base is there. Whether translating that demand into a genuinely diversified contracted revenue base happens fast enough to reduce the Meta concentration risk is the question today's selloff is really asking.

What Remains True About the Business

Today's news changes the competitive landscape for Nebius. It does not change several things that made the business compelling in the first place.

NVIDIA's $2 billion direct investment is not contingent on Meta's competitive decisions. The supply chain priority that investment secured remains intact. The engineering team that built Yandex's infrastructure and has been building Nebius's GPU clusters for the past two years has not changed. The 841% year-over-year growth in AI cloud revenue reflects genuine customer demand that exists across the market, not just from Meta.

The Eigen AI acquisition's inference optimization capabilities become more rather than less valuable in a competitive environment where customers have more choices and will favor providers with better performance and efficiency. And the Nasdaq-100 inclusion brought in institutional investors whose ownership of the stock is not conditional on any single customer relationship.

The business Nebius has built is real. What has changed is the price the market is willing to pay for the uncertainty that now surrounds its largest customer relationship.

For investors tracking stock, WEEX provides access to stock trading products, including the First Stock Trade Protected campaign offering eligible users additional protection on their first stock trade.

Conclusion

Nebius Group's story is one of the more unusual in recent technology history: a spinoff from a sanctioned Russian internet giant that rebuilt itself as one of the fastest-growing AI infrastructure companies in the world, landed a $27 billion customer contract, secured a $2 billion NVIDIA investment, joined a major index, and then watched its largest customer announce plans to enter its market in the span of a few months.

The 15% selloff today reflects the market repricing the concentration risk that the Meta deal always represented but that nobody was focused on while the growth numbers were extraordinary and the competitive dynamics were simple.

Whether Nebius emerges from this competitive shift as a company that successfully diversified its customer base while executing the existing Meta contract, or as a company whose growth narrative was too dependent on a single relationship, is the question that the next several quarters of earnings will answer. Today's selloff opened that question. The business's response to it will determine how it closes.

FAQ

1. What is Nebius Group?
Nebius Group is an AI cloud infrastructure company that grew out of the international operations of the former Russian technology company Yandex. It operates GPU clusters and cloud platforms specifically designed for AI workloads, serving companies that need computing power without building their own infrastructure.

2. Why did Nebius stock fall 15% on July 1?
Bloomberg reported that Meta, Nebius's largest customer with a $27 billion contracted agreement, is building a competing cloud business called Meta Compute to sell AI computing capacity to outside customers, creating a customer-competitor dynamic that the market priced as a significant risk.

3. How did Nebius grow so fast?
Q1 2026 AI cloud revenue grew 841% year over year, driven by explosive demand for GPU computing capacity during the AI infrastructure buildout. CEO Arkady Volozh described the constraint as supply rather than demand, with multiple customers competing for every cluster the company brought online.

4. What is Nebius's connection to Yandex?
Nebius was formed from the international technology operations of Yandex, Russia's dominant search and technology company, after Yandex restructured following Russia's invasion of Ukraine. Yandex founder Arkady Volozh leads Nebius and relocated to Europe.

5. Does the Meta competitive threat immediately hurt Nebius revenue?
Not immediately. The existing five-year contract with committed capacity purchases remains in place. The risk materializes over time as Meta potentially reduces its dependence on third-party providers like Nebius rather than as an immediate revenue decline.

Disclaimer

 This content is provided for general informational and educational purposes only and should not be considered financial, investment, legal, or tax advice. Nothing in this article constitutes an offer, recommendation, solicitation, or invitation to buy, sell, or trade any crypto asset or use any specific service. Crypto assets are highly volatile and involve a high degree of risk. You may lose some or all of the value of your investment and should not invest funds you cannot afford to lose. WEEX services may not be available in all regions and are subject to applicable laws, regulations, and user eligibility requirements. Please carefully assess risks and confirm local requirements before making any financial decisions.