Iren vs. Applied Digital: What's the Better Long-Term Play?

Iren (formerly Iris Energy) signed an approximately $9.7 billion deal with Microsoft and is presented as a vertically integrated data‑center operator owning land, electrical infrastructure and GPUs; analysts project a 175% jump in EPS to $1.07 in fiscal 2026. Applied Digital signed a $7 billion, 15‑year capacity deal with CoreWeave but is expected to report a $0.45 per‑share loss this year; hyperscalers spent an estimated $350 billion on data centers last year, driving 2025 share rallies (Iren +285%, Applied Digital +221%) and implying that Iren's asset control could provide stronger cost and scaling advantages for long‑term earnings visibility.

Analysis

Market structure: Hyperscalers (MSFT, AMZN, GOOGL) and vertically integrated builders (IREN) are the primary winners as multibillion, multi‑year contracts lock in steady demand; smaller colo operators and power‑dependent builders (APLD) are exposed to margin compression because they lack substation access and guaranteed power. The $350B hyperscaler data‑center spend last year implies multi‑year demand growth, but supply is constrained by land, power hookups, and GPUs, concentrating pricing power in firms that control those inputs.

Risk assessment: Key tail risks are regulatory limits on grid connections or ESG permitting delays, a renewed GPU shortage or price spike, and contract termination or renegotiation by hyperscalers; any of these could wipe out near‑term IRR on projects. Immediate moves (days) will be earnings/stock reactions; in 1–6 months watch GPU supply and financing needs; over 2–5 years the thesis depends on utilization ramp and contracted revenue realization. Hidden dependencies include long‑dated power purchase agreements, capital intensity (need to raise equity/debt) and single‑customer concentration (MSFT, CoreWeave).

Trade implications: Favor capital‑light, vertically integrated exposures (establish IREN and selective NVDA/MSFT exposure) and de‑rate pure‑play colo names without guaranteed power contracts. Use relative positions (long IREN, short APLD) to express the vertical‑integration premium; size to idiosyncratic risk and prefer options to cap downside when entering during high vol windows. Monitor quarterlies for utilization %, GPU inventory, and announced PPA terms as primary catalysts.

Contrarian angles: Consensus underprices dilution risk and financing needs—IREN’s forecasted +175% EPS (to $1.07 in FY26) assumes smooth builds; a single major project delay would compress returns materially. Conversely APLD’s market pessimism may already price in losses (consensus ‑$0.45), so catalyst risk exists if CoreWeave ramps ahead of schedule. Historical parallels to past data‑center booms warn of overbuild and rate sensitivity; stress‑test models for 10–20% utilization misses.

Key Decisions for Investors

• Establish a 2–3% long position in IREN (NASDAQ:IREN) over the next 2–6 weeks ahead of FY26 results; add on intraday pullbacks >15% and set a hard stop at -30% from entry, target holding 12–24 months to capture contracted revenue ramp (EPS catalyst: consensus $1.07 for FY26).
• Implement a pair trade: go long IREN and short Applied Digital (NASDAQ:APLD) equal dollar notional (1–2% portfolio exposure) to isolate integration vs. supplier risk; close or rebalance if APLD outperforms IREN by >20% or if IREN misses utilization guidance by >10 percentage points in a quarter.
• Use options to lever exposure: buy IREN Jan 2027 LEAP calls (~delta 0.30–0.40) sized to 0.5–1% portfolio risk and fund with 3–6 month covered calls or sell APLD 3–6 month puts to express short bias; limit theta bleed by scaling into positions over 4–8 weeks around earnings and GPU supply updates.
• Overweight infrastructure/AI supply chain (NVDA, MSFT) by +3–5% combined versus benchmark and trim consumer cyclical positions by -3% to fund; review quarterly GPU spot prices, hyperscaler capex guidance, and announced PPA/power‑permit outcomes within the next 90 days as go/no‑go triggers for further capital deployment.