APLD · Applied Digital Corporation

Current Thesis

The BTC-miner→AI-infrastructure pivot is no longer a hope trade; it is a contracted-backlog trade. After Polaris Forge 3 ($7.5B / 15-year take-or-pay, 300MW), total contracted lease revenue stands near $31B and contracted capacity has pushed past 1.2GW. That sits on top of CoreWeave's 400MW at Polaris Forge 1 (~$11B, now amended to an investment-grade SPV) and the $5B Polaris Forge 2 lease with a U.S. investment-grade hyperscaler. The neocloud theme (gpu-cloud-neoclouds) is ACCELERATING and the cohort is bid. What changed since the late-May refresh: the tape cooled. Shares pulled back roughly 12% from the ~$48–50 high into ~$42 (June 5 range $39.92–$43.75), landing in the ~$40–44 20-EMA retest zone a cleaner fresh-entry structure than the extended late-May chase. The counterweight: analyst coverage is saturated (26 Buy / 0 Hold / 0 Sell), the price-target cluster that fired the narrative-acceleration signal in May is spent, and an ATM/equity raise remains the dominant structural overhang. Net: constructive pullback in an accelerating theme, but crowding plus dilution risk keep a fresh entry at MEDIUM, not HIGH. No earnings until ~mid-August removes near-term binary risk.

Bull Case

• Polaris Forge 3 ($7.5B / 15-year take-or-pay, 300MW) lifted contracted capacity past 1.2GW and total contracted backlog to ~$31B (announced May 2026; Compass Point raised PT to $70 from $45, 2026-05-22).
• CoreWeave at Polaris Forge 1: 400MW across ~15-year leases, ~$11B total contracted; tenant converted to an investment-grade-rated SPV, cutting financing cost and counterparty risk.
• Polaris Forge 2: $5B AI-factory lease with a U.S.-based investment-grade hyperscaler a second distinct anchor, proving Forge 1 was not a one-off.
• Q3 FY26 (printed 2026-04-08): revenue $126.64M, +139% YoY; adjusted EPS flipped positive live-billing GPU/hosting revenue, not pipeline.
• Analyst stance is uniformly bullish: 26 Buy / 0 Hold / 0 Sell, average PT ~$61 (range $40–$106), Compass Point $70; ~$42 spot leaves ~45% to the average target.
• Theme tailwind intact: IREN +4% premarket 2026-05-26; IREN CEO's supply-scarcity frame ("an AI factory started today may not go live until 2030", 2026-05-25) keeps the neocloud cohort bid, and APLD is the higher-beta second-derivative.

Bear Case

• Analyst saturation: 26 Buy / 0 Hold / 0 Sell means the upgrade-catalyst well is dry. The May 21–22 PT-cluster acceleration signal has already fired; there is little fresh sell-side fuel left to lift the tape.
• 2026-05-22: shares slumped on the company's own landmark lease news fast money selling strength is how a crowded late-stage move behaves.
• ATM/equity dilution is the structural overhang: capex is debt- and equity-funded, and a single raise 8-K can erase 10–15% in a session regardless of fundamentals.
• Valuation stretch: after a +540% 12-month run, the minimum analyst PT ($40) sits right at the current price some desks see fair value already here.
• Residual BTC beta: tourist money still conflates the cohort with bitcoin on BTC-volatile days, adding noise unrelated to the AI thesis.
• Margin risk into the Q4/FY print (~mid-August): North Dakota utility costs load as capacity energizes, and the consensus slack that made the 2026-04-08 beat easy is gone.

Setup & Price Structure

• Spot ~$42 (2026-06-05 range $39.92–$43.75), down ~12% from the ~$48–50 late-May high.
• 20-EMA retest zone ~$40–44 price is sitting on it now; this is the pullback-to-support that the late-May extended levels lacked.
• May lease-breakout shelf ~$40: a daily close below it on volume marks the first structural crack.
• April pivot-confirmation base ~$34: the line that confirmed the miner→infra re-rate breakout; a daily close below it kills the structure.
• Momentum reset: RSI cooled off the late-May high-60s back toward neutral on the pullback a healthy reset, not a blowoff top.
• Theme state: gpu-cloud-neoclouds ACCELERATING; cohort (IREN et al.) still leading, which favors buying the dip over fading it.

Catalyst Calendar (next 30 days)

• No earnings in the window. Q4/FY26 print estimated ~2026-08-10 to ~2026-08-20 (fiscal year ends May 31) well outside 30 days, so no binary blackout near-term. (Corrects the prior "mid-July" estimate.)
• Unscheduled but live: additional Polaris Forge lease announcements and ready-for-service milestones. Lease cadence has run roughly monthly; watch the 8-K feed.
• Risk-side watch: any ATM/equity-raise 8-K (no scheduled date; dilution event).
• Macro: BTC tape (still material beta on volatile days).

What Would Change Our Mind

• Weekly close below the rising 20-EMA (~$44) → first warning that the pullback is becoming a top.
• Daily close below the ~$40 May lease-breakout shelf on volume → structural crack.
• Daily close below the ~$34 April pivot base → the BTC→AI re-rate breakout has failed; thesis dead.
• An 8-K disclosing an ATM or equity raise → dilution event; de-risk immediately regardless of price.
• Q4/FY print (~mid-August) margin miss as North Dakota utility costs load at scale → reopens the "thin GPU-hosting margin" bear case.
• gpu-cloud-neoclouds theme flips ACCELERATING → SATURATED/DEAD (IREN and peers roll over) → the cohort tailwind that carries the higher-beta name is gone.

Correlation Notes

• Pair vs IREN: APLD is the higher-beta, less-proven expression of the same neocloud / BTC→AI build-out thesis it leads on the way up and bleeds faster on the way down.
• Cohort correlation with broader neocloud and AI-datacenter-buildout names; theme-level moves dominate single-name news on most days.
• Residual BTC beta despite the AI-revenue mix the cohort gets mispriced on BTC-volatile sessions; that dislocation is an entry window, not an exit signal.
• Counterparty concentration: a large share of contracted revenue is tied to CoreWeave's health, partly mitigated by the conversion to an investment-grade SPV and the addition of new hyperscaler tenants at Forge 2 and Forge 3.