APLD · Applied Digital Corporation · Stock research

Coverage note: this advances coverage of Applied Digital first published 2026-04-19; previous update 2026-08-24.

Current Thesis

The reclaim did not come. Applied Digital closed at $26.89 on 2026-08-24, beneath the $27.21 close of 2026-08-21 and still beneath the $28.23 low of 2026-08-19 that had framed the summer range. RSI(14) printed 30.6 on the same adjusted daily series, against 41.0 three sessions earlier, with the conventional 30 line immediately below. The shares sit 45.8% under the $49.65 high of 2026-05-28, and the three-month price change is -40.4%.

The prior update set a five-session condition: reclaim $28.23 or the range resolves lower. That window runs through 2026-08-28. Two sessions are spent and price moved away from the level in both.

What preceded the last session makes it informative. B. Riley maintained a $75 target after a Polaris Forge 2 site visit, reported 2026-08-22, citing the first phase of Building 2 at Polaris Forge 1 delivered 2026-06-30 — 75 MW added on schedule, taking live capacity at the campus to 175 MW — completed in 12 months against the 24 months the first building took. The next session closed lower. A maintained top-of-range sell-side target, struck on fresh primary fieldwork, drew no marginal bid. That is an observation about who is left to buy, and it is the single most useful datapoint added since the last update.

Own-name silence is now 20 trading sessions. The investor-relations press-release archive, re-checked 2026-08-25, still ends at the 2026-07-27 FY26 results; the prior items are dated 2026-07-15 and 2026-07-01. Fiscal Q1 2027 does not close until 2026-08-31, and third-party earnings calendars place the print at 2026-10-14 — a date the company's own events page does not yet carry. Roughly seven weeks pass before Applied Digital files anything that tests the thesis.

The leg an investor is buying has not changed: a signed lease book converting into collected rent. The FY26 Form 10-K, filed 2026-07-29, reports approximately 1,410 MW of contracted critical IT load across five campuses and about $36.2B over initial 15-year base terms, against roughly 175 MW energized. Q4 FY26 revenue of $258.748M cleared a $94.836M consensus, but $152.4M of that was one-time tenant fit-out fees; the recurring data-centre rental line was $50.6M and quarterly NOI was $39.9M at a 91% margin. Management said on the 2026-07-27 call that it expects roughly $1B of annualized NOI within about a year. The distance between $50.6M of quarterly rent and a $1B annual run-rate is the whole equity argument, and nothing the company files measures it before 2026-10-14.

The narrative is maturing. The theme keeps drawing vehicles and holders: Roundhill listed its Neocloud ETF (NCLD) on 2026-08-06, and Benzinga reported on 2026-08-14 that Tiger Global disclosed a new 885,000-share stake in the Q2 2026 13F cycle, positions as of 2026-06-30. Coverage is intact but dispersed — B. Riley $75 (2026-08-03, maintained 2026-08-22), Morgan Stanley $37.50 at Equal-Weight (2026-08-06), Needham Buy $83 (2026-08-14), an average of $74.23 across 11 analysts as of 2026-08-17 with a $37.50–$109 spread struck off identical filings. What has thinned is APLD's share of the flow. The cohort's attention items on 2026-08-18 and 2026-08-19 were a Futurum Group ledger ranking IREN first on contracted revenue per megawatt at $40M/MW ahead of CoreWeave; Applied Digital appeared in neither write-up. The 2026-08-12 cohort rally on CoreWeave and Nebius results named IREN, HUT and GLXY. The name is being deselected inside a theme that still works — a different condition from the theme rolling over, and the one that would flip if peer prints stop lifting the group.

Bullish and bearish views on Applied Digital Corporation

The model's bull view on Applied Digital Corporation (APLD), in brief: Contracted book, FY26 Form 10-K filed 2026-07-29: approximately 1,410 MW of critical IT load across five campuses, about $36.2B over initial 15-year base terms, roughly 175 MW energized. The bear view: Twenty trading sessions from 2026-07-28 through 2026-08-24 with no company disclosure of any kind, confirmed against the IR archive on 2026-08-25. Both cases follow in full.

Bull Case

• Contracted book, FY26 Form 10-K filed 2026-07-29: approximately 1,410 MW of critical IT load across five campuses, about $36.2B over initial 15-year base terms, roughly 175 MW energized. The book grew through the drawdown.
• Build cadence is measurably faster: the first phase of Building 2 at Polaris Forge 1 delivered 2026-06-30 in 12 months against 24 months for Building 1, per the B. Riley note reported 2026-08-22, with management attributing the improvement to construction sequencing.
• Financing has cleared in debt, not common equity: the last two construction tranches priced at 6.750% and 7.000%. Benzinga's 2026-07-28 item attributes the lower cost of borrowing to the unnamed investment-grade hyperscaler counterparty.
• Unit economics on live capacity: Q4 FY26 NOI of $39.9M at a 91% margin, against management's stated path to roughly $1B annualized NOI within about a year (2026-07-27 call).

• Mechanical fuel on any own-name positive: short interest of 69.87M shares, 26.13% of float and 23.97% of shares outstanding, 3.65 days to cover, float 267.44M, as of 2026-08-04.

Bear Case

• Twenty trading sessions from 2026-07-28 through 2026-08-24 with no company disclosure of any kind, confirmed against the IR archive on 2026-08-25. A narrative with no new supply of information depends entirely on someone else's headlines.
• Recurring revenue is small relative to the headline: $50.6M of data-centre rent inside a $258.748M quarter that carried $152.4M of non-repeating tenant fit-out fees.
• Leverage against a non-GAAP conversion story: FY26 NOI of $90.4M and a GAAP net loss of $249.2M sit against roughly $5,100M of total debt at 2026-05-31.
• Counterparty concentration: 810 MW and about $20.2B of base-term revenue rest with one unnamed "high investment-grade hyperscaler" the company has not identified.
• Insider mechanics into a falling tape: four Form 4s filed 2026-08-04 report 3,045,000 performance stock units vesting 2026-07-31 to the CEO, President, COO and CFO, with 1,222,748 shares withheld at $27.39 to cover tax, and a Form 144 the same day noticing resale of 75,000 shares with an aggregate market value of $2,336,107.50.
• Residual crypto exposure inside an AI story: 286 MW of hosting capacity across two North Dakota sites. On 2026-08-21 bitcoin traded above $77,000 while APLD closed lower, so the beta is not currently paying.
• Consensus carries little information: an $37.50–$109 target spread on identical filings, average $74.23 as of 2026-08-17.

Setup & Price Structure

Reference levels, from the adjusted daily series: last close $26.89 on 2026-08-24; $27.21 on 2026-08-21; $28.64 close on 2026-08-20; $28.23 low on 2026-08-19, the shelf that gave way; $25.79 the summer low; $25.56 the April base low; $49.65 the 2026-05-28 high. RSI(14) at 30.6.

Structure reads as a failed range rather than a base. The $28.23 shelf held through mid-August and broke on 2026-08-21; the five-session reclaim test expires 2026-08-28. Below the last close, the only prior demand shelf on the record is the $25.56–$25.79 zone, which is where every leg since April has stopped. A weekly close through it removes the last reference point above the $20 area.

Crowding and positioning observables, stated as observables:

• Retail-facing coverage density has decayed. Benzinga carried own-name explainers on 2026-07-29, 2026-07-30 and 2026-08-03 ("Why Is Applied Digital Stock Surging Monday?"), then one 13F item on 2026-08-14, then nothing own-name through 2026-08-21.
• Short interest of 26.13% of float and a five-year beta of 5.77 (both as of 2026-08-04) mean cohort headlines move this name harder than the cohort in both directions.
• No earnings date sits inside the next 30 days. The compression that usually precedes a print is absent; the next own-name catalyst is roughly seven weeks out.
• No ATM reactivation or registered equity offering has been disclosed since the FY26 results. Financing to date has arrived as debt at 6.750% and 7.000%.
• The 2026-07-31 PSU vest and the 2026-08-04 Form 144 are share supply arriving without a matching demand event.

Catalyst Calendar (next 30 days)

• 2026-08-26 — Nvidia Q2 fiscal 2027 results, quarter ended 2026-07-26; release approximately 1:20 p.m. PT, call 2 p.m. PT. Nvidia has guided to approximately $91.0B of revenue for the quarter. The only scheduled event in the window capable of repricing APLD without any Applied Digital disclosure — the mechanism behind the 2026-07-30 and 2026-08-12 advances.
• 2026-08-28 — Structural marker, not an event: expiry of the five-session window to reclaim the $28.23 shelf lost on 2026-08-21.
• 2026-08-31 — Fiscal Q1 2027 quarter end (fiscal year ends May 31). First full period after the $152.4M of Q4 tenant fit-out fees; sets the starting point for the recurring rental line.
• ~2026-09 (est.) — Construction financing for Delta Forge 1 and Polaris Forge 3, 300 MW each. Debt priced inside the 6.750%/7.000% precedent versus common equity settles the dilution question that has capped the shares since spring.
• ~2026-09 to 2026-12 (est.) — North Dakota Public Service Commission ruling on the Montana-Dakota electric service agreement for Polaris Forge 3, which gates power for the 430 MW campus targeted for initial operations in August 2027.
• 2026-10-14 — Outside the window, and the next own-name print: Q1 FY27 results before the open per third-party earnings calendars. The company's IR events page carries no scheduled date as of 2026-08-25.

What Would Change Our Mind

The reclaim test is the near-term arbiter. If $28.23 is not recovered by the 2026-08-28 close, the range that held from mid-July is resolved downward and the $25.56–$25.79 zone becomes the only structure left. Below that, a weekly close below $25.56 breaks the April base low that has floored every leg this year and leaves no reference above the $20 area — that is the gradeable break of the thesis.

Three non-price conditions would do the same work. First, construction financing for Delta Forge 1 or Polaris Forge 3 arriving as an ATM reactivation, a registered equity offering or a convertible instead of debt near 6.750%–7.000%: the debt-not-equity precedent is load-bearing for the whole conversion argument. Second, an 8-K disclosing an amendment, deferral or termination on the 810 MW single-counterparty block, or any change in the "high investment-grade" characterisation. Third, the 2026-10-14 print showing the recurring rental line flat or down sequentially from $50.6M, which would put management's roughly $1B annualized NOI statement out of reach on the stated timetable.

The bear read fails if the theme re-selects the name: an own-name lease, financing or delivery release ending the disclosure gap, or APLD leading the cohort on a peer up-session rather than lagging it, would show marginal buyers returning ahead of the print.

Correlation Notes

• Cohort: IREN, CoreWeave, Nebius, HUT and GLXY. The 2026-08-12 rally on CoreWeave and Nebius results lifted the group; APLD was not named in the Benzinga write-up, nor in the 2026-08-18 and 2026-08-19 Futurum ledger items ranking IREN at $40M/MW.
• Nvidia is the sector's repricing instrument. The 2026-07-30 advance followed Microsoft's Q4 report, not company news; the 2026-08-26 print is the same mechanism.
• Bitcoin beta persists through 286 MW of hosting across two North Dakota sites, but it disconnected on 2026-08-21, when bitcoin cleared $77,000 and gold hit three-month highs while APLD closed down.
• Rate sensitivity runs through construction financing rather than the income statement in the near term: the 6.750% and 7.000% tranches are the benchmark the next two 300 MW facilities are measured against.
• ChronoScale (Nasdaq: CHRN) is roughly 96–97% retained and consolidated after the 2026-05-05 cloud separation, so CHRN moves are not independent of the APLD income statement.
• Vehicle flow: Roundhill's Neocloud ETF (NCLD) listed 2026-08-06, adding a passive bid to the theme whose constituent weights are worth watching for whether APLD is included and at what weight.