Applied Digital
Investment
Report

Frequently asked questions

What is Applied Digital's business?

Applied Digital builds AI Factories: purpose-built data-center campuses leased for fifteen years to AI hyperscalers. About 1.4 gigawatts of capacity is contracted for roughly $36 billion of base-term lease revenue. The buildings are financed with debt, and the stock is priced on delivery: the contracts must convert to buildings on schedule.

What is the verdict on Applied Digital?

DalalBytes rates it Positive, high-risk, scoring 70/100. The $36 billion of contracted lease revenue is real and signed. The question is execution: buildings must be delivered on schedule, and the debt that funds them must be serviced until rents ramp. The 2029 base target is $41.50, from a $27.30 reference price (September 23, 2026 close).

How to read this report

Five things that decide the outcome

The business engine: AI Factories

Applied Digital designs, builds, owns, and operates purpose-built data-center campuses it calls AI Factories. The business model is landlord economics for the AI boom: sign a long-term lease with a tenant that needs massive computing capacity, build the building to that tenant's specifications, then collect rent for about fifteen years.

The leases are take-or-pay: the customer pays the contracted rent whether it fills the building with servers or not. This is the crucial detail. Revenue is not a hope about AI demand; it is a signed obligation from investment-grade hyperscalers and CoreWeave, one of the largest AI cloud companies. The capacity is measured in critical IT load: the megawatts of power delivered directly to the computing equipment, the only capacity that earns rent.

The company began in 2021 as Applied Blockchain, co-founded by Wes Cummins (Chairman and CEO) and Jason Zhang (President). It rebranded as Applied Digital in 2022 and pivoted from cryptocurrency hosting to AI data centers. In May 2026 the historic GPU cloud-services business was separated as ChronoScale (Nasdaq: CHRN), with Applied Digital retaining about 96% ownership.

Revenue inflecting as buildings come online. FY25 is derived from the reported +167%. FY28E is the Motley Fool projection of $1.45B. Source: APLD FY2026 release, July 27, 2026.

The economics are infrastructure economics, not software economics. Margins are decent, with adjusted EBITDA of $107.2M on $611.3M of fiscal 2026 revenue, but the buildings cost billions to construct before the first rent check arrives.

Financial quality: revenue inflecting, profits pending

Fiscal 2026 ended May 31, 2026 with revenue of $611.3M, up 167% from the prior year. The fourth quarter alone delivered $258.7M, up 407%, as the first Polaris Forge 1 building began contributing. Adjusted EBITDA reached $107.2M for the year and $42.4M for the quarter. Adjusted net income was $36.1M. Net operating income, the property-level profit measure, was $90.4M.

The GAAP picture is different: a net loss of $249.2M for the year. The gap between adjusted and GAAP results comes from stock-based compensation, depreciation on the new buildings, and interest on the debt. The non-GAAP figures also exclude ChronoScale, the separated cloud business.

The debt-funded buildout

Long-term debt rose from $678M to $4.96B in one fiscal year; cash rose with it. Source: APLD FY2026 10-K, July 29, 2026.

The balance sheet transformed in fiscal 2026. Long-term debt went from $677.8M to $4,959.5M, roughly $5B. Cash and equivalents rose to about $4B. Total equity stands at about $1.72B, with 287.9M common shares outstanding. Debt to equity is about 2.9x.

The financings came in waves: $2.15B of 6.75% senior secured notes due 2031 issued at 98% of par, another $1.59B of senior secured notes, a revolving credit facility committed at $430M, and a $5B preferred-equity facility from Macquarie Asset Management. The debt is project-level and secured by the campuses and their leases, which is standard for infrastructure, but the quantum is large relative to the equity beneath it.

The honest read: this is how infrastructure gets built. Nobody funds $36B of contracted revenue with retained earnings. But the leverage means there is no cushion. Interest must be paid whether buildings deliver on time or not, and refinancing risk arrives with the 2031 maturities. The balance-sheet grade is D because the equity is thin and the debt is not.

Contracts and backlog: the $36 billion

Contracted capacity vs. campus design targets. Polaris Forge 3 capacity not disclosed. Source: company releases; totals per FY2026 results, July 27, 2026.

The three hyperscaler leases alone total 810 MW and about $20.2B of base-term revenue. CoreWeave also holds an option on another 150 MW at Ellendale, and the Polaris Forge 2 hyperscaler holds a right of first refusal on 800 MW more, the full 1 GW expansion of that campus.

The math that matters: $36B of base-term revenue across 1.4 GW over about 15 years works out to roughly $1.7M of revenue per MW per year. That is the unit economics the whole investment case rests on, and it is consistent with the expanded CoreWeave lease: $11B over 15 years on 400 MW is about $1.8M per MW per year. The base-case FY2029 scenario in this report assumes $1.57M per MW per year, slightly below the contracted rate, which is a deliberate cushion.

Delivery track record: on schedule so far

In data-center development, the only credential that matters is delivering buildings on the promised date. Applied Digital's record, so far, is clean.

• Polaris Forge 1, Building 1. Phase I (50 MW) reached ready-for-service in October 2025. Phase II (50 MW) followed on November 24, 2025, bringing the first building to its full 100 MW of critical IT load, on time. The campus is part of a 400 MW fully contracted deployment for CoreWeave, with a path beyond 1 GW.
• Polaris Forge 2. Broke ground in September 2025 on a 280 MW campus. The first 200 MW is being phased across two buildings, energizing through 2026 with full capacity expected in 2027.
• Delta Forge 1. A 430 MW campus in Louisiana with 300 MW leased; initial operations expected in calendar 2027. Polaris Forge 3 follows the same 2027 timeline.

The engineering pitch: closed-loop liquid cooling, near-zero water consumption, a projected PUE of 1.18, and 12 to 14 month build cycles. PUE, power usage effectiveness, is the ratio of total facility power to the power reaching the computers; 1.0 is perfect, so 1.18 is efficient. Speed is the selling point: in an industry where hyperscalers wait years for capacity, delivering a 100 MW building in about a year is the moat.

Competitive position: the neocloud pack

Applied Digital sits in the neocloud pack: specialist AI infrastructure companies building GPU-ready capacity faster than the big clouds. The pack trades on contracted growth, and the multiples are rich across the board.

Market caps and multiples per Finnhub and industry coverage, June to September 2026. CoreWeave and Nebius figures per ainvest, June 2026. IREN's Microsoft deal per Reuters, November 2025.

Applied Digital's relative position is interesting: it is the cheapest name in the pack on a contracted-revenue basis, roughly $7.9B of market value against $36B of signed base-term revenue. The discount reflects the earlier stage of delivery and the leverage. CoreWeave is both its biggest customer and its biggest comparable: if CoreWeave sneezes, Applied Digital catches it twice, as tenant and as valuation anchor.

NVIDIA: the backer that left

The NVIDIA relationship has two chapters. In September 2024, NVIDIA invested $160M in Applied Digital through a private placement, becoming a visible strategic backer. Then, in its 13F filing released February 18, 2026, NVIDIA disclosed it had sold its entire 7,716,050-share stake during the fourth quarter of 2025, a position worth about $177M at year end. The stock fell 8.4% in after-hours trading and closed down 4.9% the next session.

Three facts keep this in proportion. First, the same filing showed NVIDIA also exiting Arm Holdings and WeRide while opening new positions in Intel, Nokia, and Synopsys: a portfolio rebalancing, not a targeted vote against Applied Digital. Second, NVIDIA deepened its commitment to CoreWeave with a $2B investment, which is the tenant that matters most to Applied Digital's revenue. Third, Applied Digital remains an NVIDIA Cloud Partner and builds its campuses around NVIDIA GPUs; the commercial relationship continues.

Technical analysis: basing under heavy overhead

Data through the September 23, 2026 session. Reference price: $27.30. All indicators below use this single dated series.

Price is below the 200-day average with the 50-day as nearby resistance. RSI is neutral. Data: Finnhub daily candles, September 2025 to September 23, 2026.

Reading the chart

Support and resistance. $23.39 is the near support level, then $19, the 52-week low. Overhead: $29.10 is the recent swing high, then the 50-day at $27.66 as a dynamic ceiling, then the 200-day at $32.27 as the line the bulls must reclaim. $50.73, the 52-week high, is a distant ceiling.

The read. This is a basing chart, not a trending chart. The 34% three-month drawdown flushed weak holders; the 12.8% five-day bounce shows buyers returning. But every rally since spring has failed at the declining 200-day average. The technical grade is C+ because the structure is repairing, not repaired.

What are the risks, honestly stated?

Catalysts: what moves the needle

Valuation: what the price assumes

At $27.30, Applied Digital trades at about 12.9 times trailing-twelve-month revenue of $611.3M, on a market cap near $7.9B. On enterprise value, about $8.7B after adding roughly $0.8B of net debt, the multiple is about 14 times trailing revenue.

Is that expensive? Against the neocloud pack, it is the middle of the range: CoreWeave trades near 5x forward sales with a $99.4B backlog, Nebius near 17x forward sales, and earlier-stage names higher still. The market is not pricing Applied Digital as a software company; it is pricing it as contracted infrastructure trading at a discount for delivery risk and leverage.

The right way to think about the price is as a claim on the $36B. At $7.9B of market value, the market is paying about 22 cents for each contracted dollar of base-term revenue. That looks cheap until you remember the dollars arrive over fifteen years, the buildings are not all built, and the debt gets paid first.

Scenario targets: FY2026 to FY2029

Illustrative scenario price paths from the $27.30 reference price. Scenarios, not forecasts. Formula: (FY29 revenue x EV/Sales - net debt) / diluted shares.

Base case. The contracted 1.4 GW is substantially online by FY2029 at $2.2B of revenue, which is $1.57M per MW per year, slightly below the contracted $1.7M ($36B over 1.4 GW over 15 years), so the revenue carries a cushion. 8x EV/Sales sits between CoreWeave (~5x) and Nebius (~17x) forward multiples. Net debt falls to $3.5B as cash flow funds construction. Check the division: $3.3B / 350M = $9.43; $14.1B / 340M = $41.47; $29.0B / 330M = $87.88. Bear case. Deliveries slip, revenue reaches $1.3B, the market pays 6x for a levered developer that missed schedule, and $9.50 follows. Bull case. The pipeline converts beyond 1.4 GW, revenue hits $3.2B at 10x, and $88.00 follows.

Leadership: builders with a record to defend

Wes Cummins, Chairman and CEO, co-founded the company in 2021 with President Jason Zhang. Cummins led the pivot from cryptocurrency hosting to AI data centers and the rebrand to Applied Digital, then the separation of the cloud business as ChronoScale. The operating record under his leadership includes the CoreWeave leases, the two hyperscaler leases, the Macquarie financing, and the on-time delivery of the first 100 MW building.

The governance record carries one disclosed blemish. In a June 2023 8-K, the company's Audit Committee reported it had investigated a threatened, but never formally asserted, allegation of sexual harassment by Chief Marketing Officer Regina Ingel, arising from a personal relationship with Cummins. The Committee found the relationship consensual and the allegations unfounded, and the board reaffirmed Cummins as CEO. The stock fell about 16% to $8.09 on the disclosure. Short-seller reports and class-action filings followed in July 2023.

The honest assessment: the business execution since 2023 has been strong, and the board's process was disclosed rather than buried. But the episode, the short reports, and the dilution history mean governance deserves a permanent watch item. This is part of what the D balance-sheet grade and the 70/100 score are pricing in.

What to watch

• RFS milestones. Every ready-for-service announcement through 2027 is a checkpoint. On-time delivery is the thesis; delays are the thesis breaking.
• New lease signings. The 150 MW CoreWeave option and the 800 MW Polaris Forge 2 right of first refusal are the visible pipeline. New tenants would reduce concentration.
• ChronoScale. The 96%-owned GPU cloud business (Nasdaq: CHRN) is a separate listed asset. Its valuation is optionality the market may start pricing.
• Operating cash flow. About $90M in the latest quarter. The path from OCF to positive free cash flow is the path to the end of dilution.
• Equity raises. Any new share issuance, especially below book value, is a red flag on the financing plan.
• Contract health. Any cancellation, renegotiation, or tenant distress, at CoreWeave or the hyperscalers, breaks the $36B assumption.

Verdict

The one-paragraph case

Applied Digital has what most AI infrastructure stories lack: signed, take-or-pay contracts. About 1.4 GW of capacity is leased for roughly $36B of base-term revenue from CoreWeave and two investment-grade hyperscalers, and the first 100 MW building was delivered on schedule in November 2025. The stock at $27.30 pays about 22 cents per contracted revenue dollar, a discount that reflects two genuine risks: the buildings must be delivered on borrowed money by 2027, and the balance sheet carries about $5B of debt on thin equity. The base case to $41.50 by FY2029 assumes the contracted capacity comes online at economics slightly below the signed rates. The bear case to $9.50 is what delivery failure plus leverage looks like. This is a positive, high-risk thesis: contracted growth, levered balance sheet, execution as the entire game.

Sources and methodology

Method. This report uses only information available through September 23, 2026. Financial figures come from Applied Digital press releases and SEC filings. Market data comes from Finnhub daily candles through the September 23, 2026 session. Scenario targets are illustrative models, not forecasts: scenario revenue times a scenario EV/Sales multiple, minus scenario net debt, divided by scenario shares.

Disclaimer: This report is independent research for informational purposes only. It is not investment advice, a recommendation, or an offer to buy or sell any security. Investing involves risk, including loss of principal.