Applied Digital Shares Surge on Bright Outlook. Is It Too Late to Buy the Stock?

Applied Digital reported fiscal Q2 revenue up 250% year-over-year to $126.6 million, with HPC revenue of $85 million (including $73 million in tenant fit-out services) and data-center hosting revenue of $41.6 million; adjusted EBITDA rose to $20.2 million from $6.1 million a year ago while adjusted net income was $0.1 million and adjusted EPS was flat. Management said it is in advanced talks with hyperscalers for roughly 900 MW across two to three sites, and the company expects Polaris Forge to ramp to 400 MW for CoreWeave and a 200 MW commitment at Polaris Forge 2 — developments that could drive significant growth but require substantial capital. The balance sheet shows $1.9 billion in cash versus $2.6 billion in debt and negative operating cash flow of $97.9 million and negative free cash flow of $899.4 million through the first half, underscoring execution and return-on-capex risk despite strong revenue momentum.

Analysis

Market structure: Applied Digital (APLD) is a winner if power-constrained AI workloads remain binding — hyperscalers, GPU suppliers (NVDA), and regional power generators/PPAs gain pricing power while commodity colo incumbents (DLR, EQIX) risk share loss in high-density AI builds. The announced ~900 MW talks imply a multi-year squeeze on grid capacity that should keep power prices and contractor demand elevated; financing markets will reprice capital intensity into higher credit spreads for smaller landlord REITs.

Risk assessment: Key tail risks are deal failure (900 MW not contracted), build cost overruns (>20%+), PPA price spikes, permitting delays, and concentration (CoreWeave/CRWV dependency). Timeline: immediate (30–90 days) news sensitivity; short-term (6–18 months) execution and financing risk as campuses are built; long-term (2–5 years) depends on recurring hosting margin conversion and REIT tax/structural outcomes. Hidden dependency: secured long-term PPAs and GPU availability drive true economics.

Trade implications: Tactical play is event-driven — size exposure small given negative FCF (~$899m H1) and leverage (cash $1.9bn vs debt $2.6bn). Use structured exposure: conservative long equity position on contract announcements, paired with short/underweight legacy colos (DLR/EQIX) and energy longs (NEE or gas generators). Options: buy 12–24 month call spreads to limit cash at risk ahead of contract confirmations.

Contrarian angles: The market may be overstating immediate margin conversion — $73m of low-margin fit-out revenues dominated Q2; stock is +250% Y/Y but operating cash flow remains negative, so upside is binary on a few large deals. Historical parallels to crypto-miner capex cycles caution that asset-heavy growth without secured long-term cash flow can rapidly compress returns; REIT conversion could crystallize tax/operational limits and reduce upside optionality.

Key Decisions for Investors

• Establish a tactical long in APLD equal to 1–3% of portfolio on two possible triggers: (A) a signed hyperscaler contract announcement covering >=300 MW, or (B) a PPA securing >=50% of a new site's expected load. Set a hard stop-loss at -30% from entry and take-profit at +50–100% within 12–24 months if recurring hosting revenue ramps.
• Buy an APLD 12–24 month call spread sized to 0.5–1.0% of portfolio notional (buy LEAP calls, sell calls ~40–60% OTM) to capture upside to contract confirmations while capping premium outlay; roll or exit if no material contracts/PPA terms disclosed in 90 days.
• Implement a relative-value pair: long APLD (1%) vs short DLR (0.5%) or EQIX (0.5%) to isolate power-access premium. Close the pair if APLD outperforms by 30% or after 12 months; widen if APLD announces >=900 MW contracts with secured PPAs.
• Reduce traditional colo exposure: trim DLR/EQIX positions by 1–2% and redeploy into utilities/generation names (e.g., NEE or regional gas generators) by 1–2% to capture higher wholesale power demand. Reassess within 180 days or upon APLD REIT conversion filing.