Power On: Applied Digital's First AI Data Center Goes Live

Applied Digital shares jumped ~12.8% on Nov. 24 after its first 100 MW building at the Polaris Forge 1 campus reached Ready for Service, a key step toward generating lease revenue. The 400 MW Polaris Forge 1 campus is fully leased to CoreWeave under an agreement projected to produce ~ $11 billion over 15 years, Polaris Forge 2’s initial 200 MW is pre-leased under a ~$5 billion deal, and the company has secured a $2.35 billion senior note facility plus project-level equity from Macquarie—providing funding while limiting common-stock dilution; note that short interest remains elevated at >31% amid analyst price-target upgrades to ~$39–$41.

Analysis

Market structure: Applied Digital (APLD) converting 100MW RFS into revenue materially tightens available AI-ready power capacity and benefits land/power-rich infrastructure owners (APLD, CRWV, select utility partners) while pressuring legacy colocation REITs (DLR, EQIX) that lack liquid cooling/power. The pre-sold 400MW → ~$11bn over 15 years implies high revenue visibility but only if tenant uptime and power delivery meet SLAs; hyperscaler capex (~$350bn in 2025) sustains demand but creates concentrated counterparty risk. Expect upward pressure on specialty infrastructure pricing; new builds that can deliver 100s of MW with liquid cooling will command premium rents and longer terms than general-purpose colo.

Risk assessment: Major tail risks include tenant default or GPU demand collapse (negative revenue shock >30%), sudden grid curtailments or sharp wholesale power price spikes (+50% power costs would stress margins), and a credit market shock that widens APLD senior note spreads by >300bp. Near-term (days–weeks) volatility is driven by >31% short interest and potential squeezes; medium-term (3–12 months) risk centers on completing Polaris Forge 2 funding/capex and Macquarie equity drawdowns; long-term (years) execution/maintenance of liquid-cooling tech and contract renewals determine cashflow longevity. Hidden dependencies: project economics depend on contracted power rates, utility interconnect deliverables, and GPU lifecycle demand elasticity.

Trade implications: For event-driven desks, APLD is a high-conviction tactical long but position-size constrained by leverage and short interest; volatility will remain elevated—expect IV spikes around earnings/RFS milestones. Use option structures to cap downside (calendar/LEAPS spreads) and exploit short-covering in the next 2–6 weeks after any positive catalysts (additional RFS announcements or revenue recognition). Rotate capital away from single-tenant legacy colo REITs (DLR, EQIX) into AI-infrastructure and select hyperscalers/GPU exposure (CRWV, NVDA) where contract duration and pricing power are improving.

Contrarian angles: The market may be overstating perpetual $11bn revenue headline without discounting tenant concentration, capex curves, and financing covenants—NPV at 10% discount yields materially less attractive equity value. The positive reaction could be overdone near-term given 31% short interest; a forced deleveraging or slower-than-expected GPU buildout would quickly reverse gains. Historical parallel: telecom tower rollouts—early leases looked like annuities but were sensitive to technology cycles and power costs; expect similar asymmetric risks here.

Key Decisions for Investors

• Establish a 2–3% long position in APLD equity now (size 2.5% portfolio), set hard stop at $21 (≈22% below current price) and plan to take 50% profits at $34 and trim remaining at $41 within 6–12 months if execution continues; be prepared to reduce to 1% if senior note spreads widen >200bp.
• Construct a capped upside options trade: buy APLD Jan 17 2026 25–45 call debit spread (or nearest strikes), allocate 0.5–1.0% notional; max loss = premium, target 2–4x return if RFS revenue recognition or Polaris Forge 2 RFS/lease occurs within 6–12 months.
• Enter a relative-value pair: long APLD (1.5–2.0%) vs short DLR (Digital Realty) (1.0–1.5%) to capture AI-infrastructure premium vs legacy colo; unwind if spread narrows by >30% or if APLD misses next revenue milestone within 90 days.
• Reduce discretionary exposure to EQIX and DLR by ~50% over next 3 months and redeploy proceeds: 1% into CRWV (CoreWeave) and 0.5–1% into NVDA to express GPU demand tailwinds; rebalance if wholesale power prices rise >25% y/y or GPU capex guidance falls by >15%.
• Do not add incremental APLD exposure >1% until one of these catalysts occurs: (a) company reports recognized lease revenue from Polaris Forge 1 in next quarterly filing (within 60 days) or (b) Polaris Forge 2 RFS/firm financing announcement within 120 days.