Back to News
Market Impact: 0.45

This Under-the-Radar AI Stock Looks Primed to Skyrocket in 2026

NBISNVDAAMZNNFLXNDAQ
Artificial IntelligenceTechnology & InnovationCompany FundamentalsCorporate Guidance & OutlookCorporate EarningsAnalyst InsightsSanctions & Export Controls
This Under-the-Radar AI Stock Looks Primed to Skyrocket in 2026

Nebius, now focused on AI data-center infrastructure, plans to contract 2.5 GW of power for 2026 and projects an annualized revenue run rate of $7–$9 billion by year-end 2026 versus a $551 million run rate as of Q3. The company rents GPU training clusters (primarily Nvidia-sourced), is not yet profitable while it builds capacity, and trades at ~60x current sales (forward P/S ~6.6); using a midpoint $8B revenue and a 35% operating margin yields ~$2.8B operating profit (≈$2B after a 30% reduction), implying roughly 12x forward earnings on a $23 billion market cap. Management’s aggressive capacity and revenue guidance, if realized, would materially re-rate the stock but execution and continued heavy investment remain key risks.

Analysis

Market structure: Nebius (NBIS) increasing contracted power to 2.5 GW and guiding to a $7–9B annualized run-rate by end-2026 reallocates pricing power to GPU-focused data-center owners and Nvidia (NVDA) as sole GPU supplier. Winners: NVDA, AI hyperscalers, power utilities and copper/transformer suppliers; losers: legacy non-AI-optimized cloud nodes and colos without GPU-ready power/LIQ. Cross-asset: higher risk appetite in equities could steepen yields if growth is validated; electricity and copper prices will be second-order beneficiaries over 6–24 months. Risk assessment: Key tail risks are US/EU export controls on advanced GPUs, Nvidia production shortfalls (>20% supply shock), permitting/grid delays in Netherlands (project slippage >6–12 months), and severe customer concentration (top 3 hyperscalers making >50% revenue). Time horizons: expect headline-driven equity volatility in days/weeks, contract realizations over months, and profit conversion/execution risk into 2026–2027. Hidden dependencies: NBIS’s model hinges on long-term fixed-price power contracts, NVDA supply cadence, and dollar-denominated customer commitments. Trade implications: Establish a staged long NBIS position (2–3% portfolio) now, add to 5% only if NBIS reports >=1.5 GW contracted and run-rate >$2B; use a 30% stop-loss from entry. Hedge NVDA exposure by buying 12–18 month NVDA call spreads (bullish but cap premium) sized 25–50% of NBIS delta. Pair trade: dollar-neutral long NBIS / short Digital Realty (DLR) 1:1 for 6–12 months to express AI infra vs legacy colo dispersion. Options: buy NBIS Jan 2027 LEAP call spread (buy lower strike, sell 30–40% OTM) to capture 2026 conversion while capping premium. Contrarian angles: Consensus underestimates execution risk — a miss to even 40–50% of NBIS’s 2026 midpoint shrinks forward P/S sharply and could re-rate the name by >30% in 3–6 months. Historical parallel: AWS-scale rollouts required multi-year capex and margin compression before monopolistic returns; don’t assume immediate AWS-like margins. Unintended consequences include regional policy pushback, accelerated insourcing by hyperscalers, or utility tariff shocks; require ARR >$2B and signed >1.0 GW customer commitments before moving from speculative to core holding.