Is Applied Digital Your Ticket to Becoming a Millionaire?

Applied Digital is rapidly expanding AI-focused data-center capacity in North Dakota and South Dakota, citing excess energy and cooler climates, with roughly 700 MW under construction and an additional 4.3 GW in development. The stock has jumped ~50% year-to-date through Feb. 3 and ~400% since the start of 2025, but the company is unprofitable, trades at about 32x sales, and has not given profit guidance; Wall Street projects revenue growth of ~61% for FY2026 and ~55% for FY2027. The author cautions that much of this growth appears priced in and prefers lower-risk exposure to the AI buildout via established hardware leaders such as Nvidia.

Analysis

Market structure: APLD’s 700 MW under construction plus 4.3 GW planned (multi-year) directly benefits GPU vendors (NVDA), container/infrastructure vendors, local utilities and renewable generators in ND/SD while pressuring traditional colocation providers (EQIX-type) and compressing margin capture for asset-light hyperscalers. The market signal is acute demand for AI compute capacity; near-term GPU/accelerator tightness will support NVDA pricing while APLD’s 32x sales multiple prices multi-year revenue realization rather than current cashflows.

Risk assessment: Key tail risks are permitting/grid-interconnect failures, adverse PPA/electricity price moves, hardware supply shocks (GPU shortages or HBM bottlenecks), and rising rates that increase funding costs for capital-intensive builds. Immediate (days) risk is headline-driven volatility in APLD; short-term (weeks–months) risk centers on quarterly commissioning milestones; long-term (2–5 years) risk is oversupply of wholesale compute that could strand assets.

Trade implications: Favor semiconductor/accelerator exposure (NVDA) and power-infrastructure plays over pure-play colo operators. Implement relative-value trades (long NVDA, short APLD), use 6–12 month option structures to limit capital, and add utility/renewables exposure (Midwest PPAs). Monitor GPU inventory, announced customer contracts, and commissioning notices as catalysts to scale positions.

Contrarian angles: Consensus underestimates customer concentration and margin squeeze for APLD — floor space is not the binding constraint, GPUs and networking are. The 400% move since 2025 suggests momentum over fundamentals; historical parallels (colo cycles 2010–2013) show hyperscalers can flip between outsourcing and insourcing, creating reversal risk and potential wholesale price deflation 3–5 years out.

Key Decisions for Investors

• Establish a 2–3% portfolio long in NVDA via a 6–12 month call spread (buy ATM call, sell ~25% OTM call) to capture GPU tailwinds while capping premium; target 40–60% gross upside in 6–12 months, reduce position if NVDA falls >20% from entry or misses two consecutive quarters of revenue guidance.
• Initiate a 1.0–1.5% notional short position in APLD (short equity or buy 3-month ATM puts) sized to limit drawdown; set hard stop-loss at a 25% adverse move and take profits on a 30–50% decline or if APLD announces multi-GW contracted revenue that materially derisks guidance.
• Put on a pair trade: long NVDA (2 units) / short APLD (1 unit) as a relative-value hedge; rebalance monthly and trim shorts by 50% if NVDA outperforms APLD by >50% in any 90-day period or if APLD secures >50% of its planned GW as firm, multi-year customer contracts.
• Reduce exposure to pure-play colo/REITs (e.g., EQIX) by 1–2% and redeploy into Midwestern power/renewables exposure (increase XEL by ~1% or equivalent utility/renewable operator) for 6–24 months to capture PPA and grid-integration upside; exit if regional wholesale power prices decline >20% or XEL guidance deteriorates.