NBIS: The Cleanest Pure Play On AI Cloud Most People Do Not Know Exists

If you have heard of CoreWeave you have probably had the conversation about whether neoclouds, the new generation of GPU-first cloud providers, are a real category or a temporary bubble that flames out when Nvidia capacity normalizes. I think the category is real. The bigger question is which name in the category you want to own.

CoreWeave gets all the attention. Nebius gets very little. That is the opportunity.

Nebius is a fortress-balance-sheet, Nvidia-backed AI cloud pure play with a fast-growing GPU capacity footprint and a publicly listed US ticker, trading at a meaningful valuation discount to peers because most retail investors still associate it with its Yandex origin.

Nebius is the surviving entity after the unwinding of Yandex NV, the parent of the Russian search and tech business. The Russian operations were divested. What remained is a Netherlands-based holding company holding a portfolio of non-Russian assets including Nebius AI infrastructure, the AI cloud business, Toloka, the AI data labeling company, ClickHouse, the open source database investment, and Avride, an autonomous mobility company.

The market is still mentally pricing the company like it is Russian-exposed. It is not. The post-restructuring entity has clean European operations, US-listed shares on Nasdaq, and a board and management team focused on AI infrastructure.

The balance sheet that emerged from the restructuring was unusually clean. Billions of dollars in cash with relatively limited debt. That balance sheet has been the fuel for one of the fastest AI capacity build-outs in Europe.

Nebius operates AI-first cloud infrastructure. GPU clusters. High-speed interconnects. Custom orchestration. Multi-region deployment. The product is most similar to CoreWeave or Lambda. The target customer is AI labs, foundation model trainers, and enterprises building large-scale AI workloads.

The differentiator is integrated stack quality. Years of engineering experience running large-scale search and machine learning infrastructure at Yandex translates directly into deeply optimized cloud infrastructure. The team is not new at this.

Capacity is being expanded aggressively across Europe and the US. Nvidia GB200 deployments are coming online. Capacity bookings have been growing rapidly. The company has been signing multi-year contracts with foundation model customers.

Pillar one: Nvidia partnership. Nvidia is a strategic investor in Nebius. That is not a courtesy. It signals priority allocation in a capacity-constrained GPU market. When Nvidia is allocating B200 and B300 chips, partners get treated differently than non-partners. That access alone is worth a premium.

Pillar two: balance sheet. Most AI cloud upstarts are scaling on debt or burning cash. Nebius is scaling from a position of strength. They can sign multi-billion-dollar capacity expansion deals without bet-the-company financing risk. That is a strategic advantage that compounds.

Pillar three: portfolio optionality. ClickHouse, Toloka, and Avride are each interesting independent businesses. ClickHouse alone is a real-time analytics database used by major tech companies. Toloka is a data labeling provider in an AI-hungry market. Avride is autonomous mobility. Any one of these monetizing meaningfully is upside on top of the core cloud thesis.

Pillar four: discounted valuation versus comps. Nebius trades at a meaningful EV-to-forward-revenue discount to CoreWeave and to the implied valuations of comparable private neoclouds. Some of that gap is justified by smaller scale. A lot of it is sentiment overhang from the Yandex origin. Sentiment gaps close as the new operating story plays out.

Capacity in megawatts. GPU count by generation. Contracted backlog. Revenue growth quarter over quarter. Customer concentration. EBITDA trend. Cash position. Capex versus capacity additions.

I track each one quarterly. The metric that matters most for the next 12 to 24 months is contracted backlog versus capacity coming online. The more booked the new capacity, the cleaner the unit economics.

The first risk is hyperscaler price competition. AWS, Azure, GCP, and Oracle are all expanding GPU cloud offerings aggressively. Neoclouds compete on access, performance, and price. If the hyperscalers decide to crush pricing, smaller players take pain.

The second risk is customer concentration. If a small number of large foundation model customers represent most of the bookings, losing one is a meaningful event.

The third risk is capital intensity. Every gigawatt of new capacity costs billions. Nebius has the balance sheet today. They need to keep raising or rotating capital efficiently as they scale.

The fourth risk is sentiment overhang. As long as some investors still treat this as Russian-adjacent, the multiple stays compressed. That is also part of the opportunity, but it requires patience.

The fifth risk is regulatory. European data sovereignty and AI rules are evolving. Compliance complexity could become a drag.

NBIS is a higher-volatility position than ARM or NVDA. The right size is meaningful enough to matter but small enough that a 40% drawdown is survivable. The asymmetry comes from the valuation discount, not from leverage on the operating story.

If the bull case plays out, the multiple expansion alone is a meaningful return. If the operating story also delivers, the upside is much bigger.

The combination of US listing, Nvidia partnership, fortress balance sheet, multiple independent business lines, and discounted valuation is genuinely unusual. Most AI cloud upstarts have at most two of those. Nebius has all five.

That is the case in one paragraph.

Read Nebius's most recent quarterly update. Look at the capacity additions chart. Look at the bookings line. Look at the cash position. Now pull CoreWeave's most recent filings and compare. The valuation gap will jump off the page.

Read next: APLD: The Gigawatt Bet | CEG: Why Nuclear Is The Trade Of The Decade

*⚠️ Important Disclaimer: MentorSurge is not a financial advisor. This post is for educational and entertainment purposes only. Nothing on this site constitutes financial, investment, or trading advice. AI cloud stocks are volatile and dependent on capital intensive build-outs and customer demand cycles. Always do your own research and consult a licensed professional.*