Better Artificial Intelligence Stock: Applied Digital vs. Nebius

Applied Digital (APLD) is repositioning toward data-center real estate after pivoting away from crypto, securing roughly $16 billion in 15-year lease commitments (mostly from CoreWeave) but faces near-term growth headwinds as it spins off its Sai cloud unit into ChronoScale; analysts forecast fiscal 2026 revenue of $297 million (up 38%) and a $91 million net loss, valuing the company at an enterprise value of ~$8 billion (~27x 2026 sales). Nebius (formerly Yandex) has refocused in the West after sanctions, positions as a full‑stack AI infrastructure provider, and is forecast to grow revenue 521% to $3.45 billion in 2026 and to $7.8 billion in 2027, with management projecting an implied annual run-rate of $7–9 billion by year-end; Nebius trades at an EV of ~$24 billion (~7x 2026 sales). Both remain unprofitable, but Nebius's faster growth, clearer contract wins and lower sales multiple make it the preferred pick in this comparison.

Analysis

Market structure: Nebius (NBIS), CoreWeave (CRWV) and Nvidia (NVDA) are the primary beneficiaries as demand for GPU-heavy AI training drives outsized pricing power for managed GPU compute. Applied Digital (APLD) faces near-term headwinds from the Sai/ChronoScale spin-off and tenant concentration — APLD has $16bn of secured lease payments over 15 years but trades at ~27x 2026 sales (EV $8bn vs. est. $297m). Supply/demand remains tight for high-end GPUs and colocated racks, implying sustained revenue per GPU and upward pressure on power/commodity markets (electricity, gas); heavy capex borrowers will pressure credit spreads and issuer supply in bonds.

Risk assessment: Key tail risks are (1) renewed export controls on H100/A100-class GPUs or sanctions (30%+ market-impact probability in stress scenarios), (2) tenant concentration/default at APLD (CoreWeave dependence), and (3) spin-off execution failure for ChronoScale in H1 2026. Near-term (days–weeks) risks: NVDA supply notices, client contract renewals; short-term (months) risks: Q1 prints and spin-off milestones; long-term (years) risk: inability to convert to REIT or achieve >$1bn revenue for sustainable margins. Hidden dependencies include NBIS’s reliance on MSFT/META contracts and Nvidia OEM allocation.

Trade implications: Favor NBIS as the primary long — EV ~$24bn at ~7x 2026 sales with consensus growth of +521% in 2026 implies asymmetric upside; consider a 2–4% portfolio long position, scaling on pullbacks of 10–20% or if EV/sales falls to ≤6x. Establish a tactical short or put hedge on APLD (size 0.5–1.5%) ahead of the ChronoScale spin-off and until losses narrow; pair trade long NBIS / short APLD to capture relative re-rating. Use options: buy 4–6 month NBIS call spreads (20–30% OTM) sized 0.5–1% and buy 3–6 month APLD puts (10–20% OTM) as cheap tail insurance.

Contrarian angles: Consensus likely overweights reported NBIS monthly run-rate (monthly×12 is lumpy) and underweights APLD’s locked $16bn lease backlog that could justify a higher long-term valuation if utilization ramps and REIT conversion becomes viable. History shows infrastructure booms can flip to oversupply — watch GPU allocation and data-center build schedules (6–18 months) for a supply shock. Reassess positions around two objective catalysts: NVDA supply/govt export updates (next 30–90 days) and APLD’s ChronoScale spin-off completion (target H1 2026).

Key Decisions for Investors

• Initiate a 2–4% long position in NBIS (Nebius) sized to portfolio risk tolerance; scale in on any pullback of 10–20% or if EV/sales compresses to ≤6x 2026 consensus (~EV $20bn).
• Establish a 0.5–1.5% short or buy 3–6 month puts on APLD (Applied Digital) to hedge spin-off and tenant-concentration risk; cover or reassess if ChronoScale spin-off delivers clear pro forma cashflow guidance or APLD posts trajectory to >$1bn recurring revenue.
• Place a 0.5–1% notional 4–6 month NBIS bull call spread (20–30% OTM strikes) to capture upside with capped cost; concurrently buy 3–6 month APLD puts 10–20% OTM as inexpensive tail protection against spin-off execution failure.
• Tilt sector exposure: overweight AI infra names (NVDA, CRWV, NBIS) by +3–5% relative to benchmark allocation and underweight speculative infra/REIT-transition names (APLD) by −2–3% until profitability/REIT-path clears (re-evaluate at H1 2026 spin-off completion).
• Monitor three triggers and act within specified windows: NVDA export-control or allocation updates (next 30 days), NBIS monthly revenue cadence and major-client confirmations (review weekly until next release), and APLD ChronoScale spin-off progress (decision point H1 2026).