Nebius vs CoreWeave: Which is the Better Buy According to Wall Street?

Nebius and CoreWeave are benefiting from surging demand for AI compute capacity, with Nebius reporting capacity sold out and revenue growth of more than 300% in the latest quarter and CoreWeave posting 133% revenue growth. Nebius announced large infrastructure deals — including a Microsoft agreement worth as much as $19.4 billion and a $3 billion deal with Meta — and has raised over $4 billion via convertible notes and equity to fund GPU, land and other asset investments; CoreWeave has first-mover uptake of Nvidia Blackwell products but carries higher debt-to-asset leverage. Wall Street consensus favors buys on both names, implying 12-month upside of roughly 67% for Nebius and 43% for CoreWeave, though both require substantial ongoing capital investment and carry execution and financing risk.

Analysis

Market structure: AI inference demand is moving capacity pricing power toward specialized GPU cloud providers and Nvidia (NVDA). Nebius (NBIS) and CoreWeave (CRWV) directly benefit via triple-digit revenue growth (NBIS +300%, CRWV +133%) and long-term contracts (NBIS–MSFT ~$19.4B, META ~$3B), while hyperscalers could see margin pressure on their lower-end IaaS as customers migrate to purpose-built stacks. Tight capacity and rapid fleet fills imply pricing power for suppliers in the next 6–24 months, but also force heavy capex and financing activity now.

Risk assessment: Key tail risks are (1) a multi-quarter model-training slowdown that reduces GPU utilization, (2) a sudden Nvidia supply shock or prioritization shift, and (3) financing/dilution risk from large equity or convertible raises (NBIS already raised >$4B; CRWV elevated leverage). Near-term (days–months) volatility will hinge on quarterly prints and NVDA product cadence (Rubin launch later this year); medium/long-term (quarters–years) outcomes depend on utilization sustaining >70–80% fleet occupancy and healthy EBITDA/interest coverage.

Trade implications: Favor growth exposure to NBIS with capital light hedges and express NVDA conviction via directional-but-capped options into the Rubin cycle. Avoid unhedged exposure to CRWV equity; if taking position, size small (<=1–2% AUM) and protect with puts or short-duration credit hedges because debt-funded expansion is the main vulnerability. Rotate 3–7% from generalist cloud names into AI-infra specialists and underwrite positions to utilization, contract duration, and covenant milestones.

Contrarian angles: Consensus underestimates financing risk and execution friction — NBIS’s gigantic MSFT/META pipeline is valuable but monetization lags (land, build, power take months). The market may be under-pricing a mid-2026 supply relief if NVDA scales Rubin production, which could compress infra margins; conversely, prolonged power/grid constraints or chip allocation to hyperscalers would widen spreads in favor of specialists. Historical parallel: GPU booms (2016–18) showed winners consolidated after a 12–24 month shakeout; prepare for similar consolidation.

Key Decisions for Investors

• Establish a 2–3% long position in NBIS (NASDAQ:NBIS) sized to growth sleeve with 12-month target +60% (roughly Street-implied); set hard protective stop at -25% and trim half position on +50% gain or if quarterly utilization falls below 70% sequentially.
• Avoid unhedged long exposure to CRWV equity; if expressing negative view, initiate a 1% notional short or buy 6–9 month put options (strike ~10–15% OTM) sizing for max loss = premium, and increase hedge if CRWV net leverage >40% or interest coverage <3x on next report.
• Buy a directional NVDA call spread to play Rubin-driven GPU demand: buy a 9-month at-the-money call and sell a call +15% strike (equal notional), sizing to 0.5–1.0% of portfolio to limit premium while capturing upside into product launches (target >2x payoff if NVDA outperforms on Rubin).
• Rotate 3–7% of cloud exposure from generalists (GOOGL, MSFT) into AI-infra specialists (NBIS, selective private/cloud infra) over the next 3 months, reallocating progressively as utilization and contract-book evidence (monthly/quarterly reports) confirm >70% steady-state demand.