Is Nebius Stock Set to Double in 2026?

Nebius reported Q3 revenue of $146 million, up 355% year-over-year, with a net loss of $120 million, and says it has sold out available capacity; full-year revenue is guided to $500–$550 million while ARR is forecast at $900 million–$1.1 billion. Management raised its 2026 contracted compute target from 1 GW to over 2.5 GW and now projects $7 billion–$9 billion ARR by end-2026, but the company remains unprofitable and trades at roughly 60x price-to-sales, leaving valuation and market sentiment as key risks to the stock’s upside.

Analysis

Market structure: Nebius’ +2.5GW 2026 demand upgrade and Q3 sell-out (Q3 rev $146M, +355% YoY) signals demand for GPU compute is materially outstripping available hyperscaler supply. Immediate beneficiaries: GPU suppliers (NVDA), hyperscalers with balance-sheet scale (MSFT, META) who lock capacity; losers: smaller colo/infra players without direct NVDA supply or capital to pre-buy power. Higher contracted demand implies pricing power for suppliers of chips and power but margin compression risk for middlemen who must scale capex rapidly.

Risk assessment: Key tail risks are tighter GPU export controls or sanction-related governance scrutiny (company spun from Yandex) that could sever western hyperscaler relationships — low probability but value-destructive. Timeline: days–weeks for sentiment-driven volatility and contract announcements; quarters into 2026 for capex execution and ARR to materialize; through 2026–27 for profitable unit economics. Hidden dependencies: concentration of revenues with Meta/MSFT and single-sourced GPU supply (NVDA) create counterparty and supply-chain single points of failure.

Trade implications: Favor cash-flow leaders providing GPUs (NVDA) and hyperscaler counters (MSFT, META). For NBIS, prefer defined-risk options: small debit call spreads or puts depending on catalyst sequencing; avoid uncovered equity sized >1–2% due to 60x PS valuation. Pair idea: long NVDA (2% portfolio) vs short NBIS equity/options (1% notional) to express dispersion between supply (NVDA monetization) and overvalued infra reseller (NBIS).

Contrarian angles: Consensus assumes ARR conversion equals cash revenue — that may be overstated if contracts are capacity rights, not billed consumption. The market may be under-pricing regulatory/spend-through delays while over-pricing growth certainty (NBIS at ~60x PS). Historical parallels: infrastructure vendors in prior AI cycles (early cloud GPU rent models) showed revenue recognition and capex lag; unintended consequence is that a rapid capacity buildout can dilute ARR economics and give NVDA more leverage to capture rent.

Key Decisions for Investors

• Establish a 1.5–2.0% portfolio long position in NVDA over 6–12 months (either buy equity or 9–15 month LEAP calls) to capture continued GPU pricing power; trim/stop-loss 15% if NVDA underperforms market for 30 trading days.
• Avoid buying NBIS equity outright; instead size a defined-risk bullish play as a 0.5% portfolio debit call spread on NBIS with 6–9 month expiry (max loss = premium). Close if NBIS fails to deliver sequential revenue growth or guidance by two consecutive quarters.
• Implement a pair trade: long NVDA (2% notional) and short NBIS (1% notional) via stock or liquid options to express capture of supplier economics vs reseller valuation risk; rebalance if spread narrows to <30 PS multiple differential or widens beyond 80 PS.
• Reduce aggregate exposure to unprofitable AI-infra resellers with PS >30x by 25–50% within 30 days; redeploy proceeds to NVDA/MSFT/META or to cash. If within 60 days export-control guidance tightens (US/EU announcements limiting GPU exports), increase NVDA exposure by 50% and close NBIS shorts.