Nebius Group N.V. (NBIS): AI-Native Cloud Infrastructure Powering the Next Compute Cycle

Nebius Group N.V. is a Netherlands-registered AI infrastructure company that emerged from the 2024 divestiture of Yandex's international businesses. The company provides GPU-as-a-service cloud computing, AI development platforms, and data-center solutions to a customer base spanning AI research labs, enterprise Fortune 500 firms, and government entities.

Revenue generation is threefold: (1) dedicated GPU cluster rentals under 1-3 year contracts, (2) usage-based cloud compute and storage via its public cloud platform, and (3) AI Studio subscriptions offering managed model training and inference APIs. Major disclosed customers include undisclosed "top-tier AI labs" and European automotive and financial enterprises. The company also operates Toloka AI, a data annotation subsidiary, and maintains the Nebius AI research division publishing in top-tier ML conferences.

With approximately 500 employees (highly leveraged versus hyperscalers) and 2025 revenue tracking toward $500M, Nebius demonstrates exceptional capital efficiency. The company's Finnish data centers leverage renewable hydroelectric power, providing both cost advantages and ESG credentials that appeal to European institutional buyers.

Near-term growth (next 12 months) is driven by three factors: (1) the continued ramp of installed GPU capacity to full utilization as contracted workloads commence, (2) the launch of new data-center capacity in the U.S. (Kansas City) and expansion in France, adding approximately 20,000 additional GPUs by late 2026, and (3) the commercial rollout of the AI Studio platform, which transitions early-access customers to paid tiers.

Medium-term growth (2-3 years) will be propelled by the secular expansion of enterprise AI adoption. As companies move from experimentation to production, demand for inference compute is expected to outpace training demand, playing directly to Nebius's software-optimized inference strengths. Additionally, the company is developing its own networking and orchestration layer that could be licensed to other cloud providers, creating a potential software-revenue stream. Management has articulated a target of $2B+ annual revenue by 2027, implying a ~60% CAGR from current run-rates — an ambitious but achievable goal given capacity expansion plans and contracted backlog.

Geographic expansion into the Middle East (a $1B+ partnership announced with a UAE sovereign fund) and potential entry into Asian markets (excluding China) provide further upside optionality beyond the current footprint.

Note: FY2023-2024 figures based on company disclosures post-restructuring; FY2025-2027 are analyst estimates.

The financial trajectory reflects the classic infrastructure build-out curve: heavy upfront capital expenditure depresses near-term margins, but once GPU clusters reach full utilization, incremental revenue carries high gross margins (60%+ at the margin). The transition from negative to positive EBITDA is the critical near-term catalyst — management has guided to achieving this in Q3-Q4 2026, supported by the current $226.39 stock price implying the market already prices in successful execution. The current EPS of $-0.06 (trailing twelve months) confirms the company is at the trough of the investment cycle, with profitability inflection expected within the next two quarters.

The global AI infrastructure market is projected to reach $250B by 2027, growing at a 40%+ CAGR, driven by foundation model training, enterprise inference, and sovereign AI initiatives. Within this market, specialized GPU cloud providers are carving out a niche distinct from hyperscalers, addressing customers who need dedicated capacity, geographic flexibility, or price-performance optimization.

Nebius competes in this "alternative cloud" segment against several players:

Nebius's competitive advantages include: (1) the strongest software layer among pure-play GPU providers, (2) European presence that competitors lack (CoreWeave is U.S.-only), and (3) a clean balance sheet post-restructuring. Its primary vulnerability is scale — CoreWeave operates roughly 3-4x more GPUs, giving it better pricing power with NVIDIA and the ability to serve larger workloads. However, Nebius's higher-touch service and software optimization allow it to command premium pricing, partially offsetting the scale disadvantage.

Our DCF analysis assumes: revenue growing at ~80% CAGR through 2028 then decelerating to 20% by 2032; terminal EBITDA margin of 35% (consistent with mature infrastructure software businesses); and a WACC of 12% reflecting the company's elevated beta (1.44) and early-stage risk profile. This yields a fair value range of $210-$260 per share, with the midpoint of $235 representing ~15x 2027E EV/Revenue — a premium to hyperscalers (5-8x) but justified by superior growth and the software margin expansion potential.

Comparable company analysis:

NBIS trades at a significant premium to all comparables on EV/Revenue, reflecting: (1) extreme scarcity value as the only liquid "pure-play" AI compute stock with scale, (2) the software upside embedded in AI Studio, and (3) potential M&A/sovereign-wealth strategic interest. The 23.44% short interest suggests many investors view this premium as unsustainable; however, short squeezes have historically driven the stock higher in momentum phases. We consider the current valuation full but justified given the growth trajectory, and our $260 price target reflects modest upside from current levels.

• GPU Supply Chain Dependency: Nebius is entirely dependent on NVIDIA for its core hardware. Any delay in H200/B200 deliveries, export-control changes, or a shift in NVIDIA's allocation strategy toward hyperscalers or CoreWeave could severely constrain capacity expansion and revenue growth.

• Hyperscaler Competition and Price Compression: Amazon, Microsoft, and Google are aggressively expanding their AI cloud offerings and possess unmatched scale, distribution, and bundling capabilities. A price war in GPU compute, or hyperscaler willingness to operate at break-even to capture market share, could compress Nebius's margins and utilization rates.

• Customer Concentration and Churn Risk: A significant portion of revenue comes from a small number of large AI labs. If a major customer builds in-house compute capacity, defects to a competitor, or fails to secure follow-on funding (a real risk in the AI startup ecosystem), Nebius could face sudden capacity idle and revenue shortfalls.

• Geopolitical and Regulatory Overhang: Despite the clean break from Russia, Nebius retains regulatory scrutiny given its Yandex heritage. The company is subject to U.S. Treasury and EU sanctions compliance monitoring, and any perceived misstep or geopolitical escalation could trigger restrictions on its operations or access to U.S. technology.

• Execution and Capital Intensity Risk: Doubling GPU capacity within 18 months requires flawless execution in data-center construction, power procurement, and customer contracting. Cost overruns, construction delays, or power-supply constraints could delay the path to profitability and force dilutive capital raises, undermining the equity story.