The Neocloud Supercycle Is Here. This Growth Stock Could Be Its Biggest Winner.

Applied Digital is constructing two North Dakota campuses totaling 700 MW and has already leased 600 MW of that capacity (CoreWeave 400 MW on a 15-year lease), representing about $16 billion in potential lease revenue. The company has begun a new southern U.S. facility that can initially support 430 MW, holds land and power agreements to pursue a 4.3 GW roadmap, and management expects to surpass $1 billion in net operating income within five years; Deloitte projects U.S. AI data-center power demand could reach 123 GW by 2035, underpinning long-term customer demand. Shares have rallied ~350% over the past year, and analysts highlight a robust lease pipeline driving elevated growth expectations.

Analysis

Winners extend beyond the obvious landlord. Scale owners that control land + long-term power rights (in particular those that can stagger build schedules to avoid peak EPC congestion) will capture outsized returns because transformer/switchgear lead times (commonly 12–24 months) and skilled EPC bandwidth create artificial scarcity in delivery capability. Equipment and services exposed to liquid cooling, medium-voltage substation builds, and long-haul fiber splicing are asymmetric beneficiaries — they can pick up 30–50% margin expansion as unit volumes normalize and retrofit demand grows.

The single largest macro sensitivity here is power cost and timing: a $10/MWh move equals ~87,600 USD/MW/year (≈$88m per GW), so a 1 GW rollout facing a $20/MWh step-up swings NOI by almost $175m annually — that’s enough to change multi-year IRR math and covenant headroom. Execution timing matters more than headline bookings; 12–36 month construction slippages compress IRR and force incremental working capital or equity raises which would be dilutive in a higher-rate environment.

Consensus assumes smooth GPU availability and contract roll-through; that’s the contrarian lever. An acceleration in GPU availability would lift tenants’ throughput and shorten payback on colocated capacity, but GPU export controls, priority allocation to hyperscalers, or a new generation SKU with materially different power density could re-rate winners and losers. Watch near-term catalysts (power-hedge announcements, completed campus energizations, and major GPU allocation statements) as binary events that will reprice forward multiples within quarters rather than years.

Key Decisions for Investors

• Initiate a long APLD position (equity OR Jan-2028 LEAP calls ATM) sized 1–1.5% NAV; time horizon 12–36 months. Target 50–100% upside if electrification and campus energizations hit schedule; hard stop/hedge if project-level delays force >25% equity dilution.
• Pair trade: Long APLD / Short CRWV (ratio 1:0.6) over 9–18 months to capture infrastructure owner re-levering vs tenant pricing compression as commissioned capacity alleviates tightness. Expect 30–60% relative return if builds hit on time; downside if GPU scarcity persists or CoreWeave wins more favorable commercial terms.
• Smaller tactical long on NVDA exposure via Sep-2026 calls (size 0.25–0.5% NAV) as a convexity play: if GPU allocation remains the binding constraint, NVDA re-prices on higher ASPs and order volumes. This trade is high-volatility/high-reward — cap exposure and treat as a directional option, not core position.