Nebius Group is one of the more unusual companies in the current AI infrastructure trade.
It did not begin life as a newly formed GPU cloud startup. The listed company was previously Yandex N.V. After completing the divestment of its Russia-based businesses in July 2024, the Dutch parent company changed its name to Nebius Group N.V. and its Nasdaq ticker from YNDX to NBIS. Trading in its Class A shares resumed on Nasdaq on October 21, 2024.
What emerged from that restructuring is now centered on a very different business: a full-stack AI cloud designed for intensive training and inference workloads.
Nebius Group today consists primarily of:
Inside the core Nebius business, the company is also moving further up the AI software stack through Nebius Token Factory, its managed inference platform.
That distinction matters. Nebius is no longer best understood as simply “a company renting NVIDIA GPUs.” Its strategy increasingly spans the physical data center, GPU infrastructure, networking, cloud software, inference optimization and AI deployment tools.
The business is also scaling extraordinarily quickly. Nebius reported Q2 2026 group revenue of $582.3 million, up 454% year over year, while the Nebius AI Cloud business alone generated $574.9 million, up 514%.
MEXC has already published a detailed breakdown of that earnings reaction, so this article will not repeat the Q2 trading story. Readers looking specifically for the earnings move can see Why Is NBIS Stock Up 34% After Nebius Reported Q2 Earnings?.
Instead, the more useful question here is simpler:
What has Nebius actually become, and what has to go right for this business model to work?
This history is easy to misunderstand.
Nebius Group N.V. is the corporate entity formerly known as Yandex N.V., but investors should not treat today's Nebius as simply a renamed version of the old Russian internet business.
In July 2024, Yandex N.V. completed the sale of its remaining interest in its Russia-based businesses. The company said that transaction fully disposed of its ownership in those operations.
The parent company was subsequently renamed Nebius Group N.V., and its Nasdaq ticker became NBIS in August 2024.
When trading resumed in October, management described the new company as a public pure-play focused on AI infrastructure.
That history explains why Nebius entered the AI infrastructure race with something most startups do not have: an experienced engineering organization, public-market access and businesses retained from the previous corporate structure.
At the center of Nebius is an AI-focused cloud platform.
The company builds large GPU clusters and provides the surrounding infrastructure that allows customers to:
Nebius says it designs substantial parts of its own software and hardware stack, including servers, racks and data-center architecture.
That makes the model different from simply buying GPUs and reselling hourly access.
A useful way to think about it is:
Power and data centers
↓
GPU infrastructure
↓
Networking and storage
↓
Cloud orchestration
↓
Training and inference tools
↓
Customer AI applications
Nebius wants to capture value across more of that chain.
Nebius's expansion is closely tied to NVIDIA hardware.
In March 2026, NVIDIA and Nebius announced a strategic partnership under which NVIDIA agreed to invest $2 billion in Nebius. The companies also said their collaboration could support deployment of more than 5 GW of NVIDIA systems by the end of 2030.
This goes beyond buying GPUs in the open market.
The partnership covers areas including AI-factory architecture, inference and agentic AI, while Nebius also expects access to successive NVIDIA architectures.
For readers looking at the wider AI infrastructure supply chain rather than Nebius alone, MEXC has already covered the sector in AI Infrastructure Stocks 2026: Who Is Actually Benefiting From Big Tech AI Capex?.
One of the clearest examples of Nebius's business model came in March 2026.
Nebius announced a five-year agreement to provide Meta with $12 billion of dedicated AI infrastructure capacity across multiple locations using early large-scale deployments of NVIDIA's Vera Rubin platform.
Meta also committed to purchase additional available capacity from certain upcoming Nebius clusters, potentially bringing that additional amount to up to $15 billion over five years.
That creates a possible relationship worth up to $27 billion under the disclosed arrangements, but the numbers need to be interpreted carefully.
They are not $27 billion of revenue already recognized today.
Capacity must still be financed, constructed, equipped and delivered.
That is the central tension in the Nebius investment story.
Software companies can sometimes add customers without building billions of dollars of physical assets.
Nebius cannot.
To recognize much of its contracted future demand, the company needs:
GPU procurement
land and data centers
power
networking
cooling
financing
before the revenue can fully arrive.
This makes Nebius closer to a hybrid of a cloud company and an infrastructure developer than a conventional software business.
On August 24, 2026, Nebius closed approximately $5.75 billion of convertible senior notes, comprising $3.45 billion due in 2030 and $2.3 billion due in 2034.
The company said the proceeds would help finance data-center construction, expansion of its AI cloud platform and procurement of key components including GPUs.
Reuters placed the deal in the context of the enormous financing requirements behind the AI data-center boom, noting both strong demand for AI capacity and investor concern over how much capital the industry requires.
That tension will probably remain central to NBIS for some time:
Demand can be exceptionally strong while the stock still faces financing, dilution and execution risk.
According to Sarah Chen, MEXC senior crypto industry analyst, the more interesting question after Nebius's 2026 expansion is no longer whether demand for accelerated computing exists. Q2 group revenue grew 454%, the company has disclosed more than $40 billion of customer commitments, NVIDIA has committed $2 billion of strategic capital, and Nebius has signed unusually large infrastructure agreements. The demand signal is already visible. The harder question is how efficiently Nebius can turn contracted demand into operating capacity and then into durable free cash flow. Readers can find more of Sarah's work on her MEXC author profile.
Chen views the August convertible financing as a useful illustration of that shift. Raising $5.75 billion gives Nebius more resources to buy GPUs and build data centers, but investors still need to follow what each new dollar of capital produces. If revenue per deployed megawatt, utilization and customer prepayments remain strong, heavy investment may support value creation. If construction runs late or GPU capacity becomes less scarce, the same capital intensity can become a weakness. That is why headline contract values alone are not enough to evaluate NBIS.
Nebius is also trying to move beyond raw infrastructure.
Nebius Token Factory is its managed AI inference platform.
In May 2026, Nebius agreed to acquire Eigen AI, bringing model-level inference and post-training optimization technology into Token Factory.
Shortly afterward, the core engineering and research team from Clarifai joined Nebius, while Nebius licensed Clarifai's inference and compute-orchestration technology. The company described the combination as an effort to optimize both the model and the underlying system.
This is strategically important.
Raw GPU rental can become competitive.
Inference optimization, orchestration and developer tools create a path toward higher-value services layered on top of infrastructure.
Nebius Group is not entirely an AI cloud company.
Avride develops autonomous cars and delivery robots.
Its technology targets areas including ride-hailing, logistics, e-commerce and last-mile delivery.
Avride therefore provides exposure to physical AI and autonomous systems rather than cloud infrastructure alone.
It is still much smaller than the current AI Cloud story, so investors should avoid valuing NBIS as though every group business contributes equally to current revenue.
TripleTen is the group's education technology business.
It focuses on reskilling and upskilling people for careers in technology through online training programs.
Again, this is a real business, but Nebius AI Cloud now dominates the group's financial profile.
That became particularly obvious in Q2 2026, when AI Cloud generated $574.9 million of the group's $582.3 million of revenue.
Nebius currently describes ClickHouse and Toloka as equity interests rather than core consolidated operating businesses in the same way as Nebius AI Cloud, Avride and TripleTen.
That distinction matters when analyzing the group.
Owning an equity stake can create substantial investment value, but it is not the same as consolidating that company's revenue into Nebius AI Cloud.
At the core AI business, customers pay for access to AI computing infrastructure and related cloud services.
Revenue economics depend on several variables:
| Driver | Why It Matters |
|---|---|
| Deployed GPU capacity | Determines how much compute can be sold |
| Utilization | Idle GPUs generate weak economics |
| Pricing | Determines revenue generated per unit of capacity |
| Power cost | Major ongoing data-center expense |
| Hardware cost | GPUs and networking require substantial upfront capital |
| Customer contracts | Provide demand visibility |
| Prepayments | Can help fund deployment |
| Financing cost | Affects return on infrastructure investment |
| Software/services | Can increase value beyond raw GPU rental |
This is one of the easiest mistakes to make with Nebius.
MEXC's current AI infrastructure research notes that Nebius had more than $40 billion of customer commitments and approximately $37.5 billion of remaining performance obligations, while expecting more than $9 billion in 2026 customer prepayments.
Those are valuable demand indicators.
But:
customer commitment
is not the same as:
revenue already recognized
and neither is automatically the same as:
free cash flow.
Infrastructure has to be delivered before much of that contractual demand becomes reported revenue.
Nebius's risks are unusually physical for a technology company.
Power needs to be available.
Data centers need to be finished.
GPUs need to arrive.
Customers need to use the capacity.
Financing needs to remain available at acceptable terms.
The company also faces competition from hyperscale cloud providers, specialized AI clouds and customers increasingly building their own infrastructure.
For readers focused specifically on NBIS valuation rather than the company structure, MEXC already has a separate NBIS stock price prediction and analyst-target article. This article deliberately does not duplicate that discussion.
Nebius shares trade on Nasdaq under NBIS.
Ondo has created a separate tokenized product called NBISon, displayed as NBISON on MEXC.
MEXC opened NBISON/USDT spot trading on March 23, 2026.
Eligible users can access NBISON/USDT Spot on MEXC.
The relationship is:
Nebius business
↓
NASDAQ: NBIS
↓
Ondo NBISon
↓
MEXC NBISON/USDT
NBISON should therefore be understood as tokenized economic exposure linked to NBIS—not as the native cryptocurrency of Nebius.
Nebius Group N.V. is an Amsterdam-headquartered technology company whose core business is a full-stack AI cloud designed for intensive AI workloads. It also includes Avride and TripleTen and holds equity stakes in businesses including ClickHouse and Toloka.
The listed Dutch company was formerly Yandex N.V. It completed the divestment of its Russia-based businesses in July 2024 and subsequently became Nebius Group N.V.
Trading resumed on October 21, 2024.
It provides GPU-based AI computing infrastructure, storage, cloud software, managed services and tools for AI training, inference and deployment.
Q2 2026 group revenue increased 454% year over year to $582.3 million; AI Cloud revenue increased 514% to $574.9 million.
No. NBIS is the Nasdaq-listed security. NBISON is an Ondo tokenized product linked to NBIS.
This article is for informational and educational purposes only and does not constitute investment, financial, tax or legal advice.
Nebius remains a rapidly expanding and capital-intensive AI infrastructure company. Revenue growth and large customer commitments do not guarantee profitable deployment. Key risks include financing, dilution, data-center construction, GPU availability, power constraints, customer concentration, competition and technology cycles. NBISON adds separate tokenization, backing, liquidity, blockchain, USDT and exchange-custody risks.

Cluster Protocol is a unified AI infrastructure layer built around Base, the Ethereum Layer 2 network. Instead of requiring developers to connect separately to AI models, data providers, GPU

Summary NBISON and NVDAON sit in the same AI investment story, but at different places in the value chain. NVDAON is linked to NVIDIA. NVIDIA designs and sells accelerated computing platforms,

Summary The easiest way to understand Nebius Token Factory is to ask what Nebius does after it has built the GPU cluster. If the answer is only: “rent the GPU” then the business can eventually become

Overview As the computational requirements for frontier artificial intelligence foundation model training and high-throughput inference expand along an exponential trajectory, a multi-tier

Overview Global artificial intelligence compute leader Nvidia has executed one of the most consequential corporate capital allocations in semiconductor history, committing $3.5 billion into overseas

Overview Following more than a year of semiconductor dominance where graphics processing units and physical data center hardware captured the vast majority of artificial intelligence capital flows,

Robinhood Chain's ecosystem surged as daily DEX volume exceeded $560M, while institutional adoption of DeFi and cross-chain infrastructure accelerated. Meanwhile, regulators advanced crypto legislatio

BlackRock Global Head of Digital Assets Robert Mitchnick said Bitcoin’s market sentiment had improved in a “clear but subtle” way as the asset began showing signs of separating from equities. His obse

Summary NBISON and NVDAON sit in the same AI investment story, but at different places in the value chain. NVDAON is linked to NVIDIA. NVIDIA designs and sells accelerated computing platforms, network

Summary “Asset-light” can be a dangerous phrase in an industry filled with billion-dollar data centers. Nebius is not suddenly becoming a software company with no physical infrastructure. What changed

Summary The easiest way to understand Nebius Token Factory is to ask what Nebius does after it has built the GPU cluster. If the answer is only: “rent the GPU” then the business can eventually become

Summary For an AI cloud, GPUs are only useful after electricity arrives. That sounds obvious. In 2026, it has become one of the most important constraints in the entire AI infrastructure market. Nebiu