APLD
Five-Pillar
Analysis

Frequently asked questions

Which pillars does Applied Digital pass, and which fail?

Only the sunrise sector passes: AI compute infrastructure is a multi-decade demand wave, with about $36 billion of contracted 15-year take-or-pay leases to prove it. Leadership and moat are partial. The iron-fortress balance sheet and free cash flow fail: about $5 billion of long-term debt against about $1.7 billion of equity, and free cash flow deeply negative.

What would change the grades?

The balance sheet upgrades on sustained FCF-positive quarters with net debt falling; leadership upgrades on two more years of clean delivery with no governance flags. Downgrade triggers include a governance flag, a major construction delay, a guidance cut, or lease cancellations and renegotiations.

How the engine judges

The five-pillar engine asks five questions. Each pillar passes, partially passes, or fails on evidence, not narrative. A partial grade means the evidence is real but incomplete. A fail means the current state does not meet the standard.

Pillar 1: Sunrise sector

Pass

The question: is the industry growing for a decade or more?

AI compute infrastructure is a multi-decade demand wave, and Applied Digital has the contracts to prove it. The company designs, builds, and operates purpose-built AI data centers, called AI Factories. As of its fiscal 2026 results, it reports about 1.4 gigawatts of contracted critical IT load under 15-year take-or-pay leases worth about $36 billion in base-term revenue. During the year it signed three 15-year take-or-pay leases totaling 810 MW and about $20.2 billion with investment-grade hyperscalers. Take-or-pay means the tenant pays whether it uses the capacity or not: this is contracted revenue, not forecast demand.

The backdrop supports the long runway. Hyperscalers are spending hundreds of billions on AI capacity, and the announced lease roster shows tenants committing for 15 years: CoreWeave holds 400 MW at Polaris Forge 1 (Ellendale, North Dakota) under a contract expanded to about $11 billion; a 200 MW Polaris Forge 2 (Harwood, North Dakota) lease worth about $5 billion went to an investment-grade hyperscaler; and a US investment-grade hyperscaler took 300 MW at Delta Forge 1 (Boyce, Louisiana) plus 300 MW at Polaris Forge 3, each worth about $7.5 billion. The legacy bitcoin-mining hosting business (286 MW, $37.3M of Q4 revenue at full utilization) is now a sideshow.

What would change the grade: a sustained collapse in AI compute demand, a wave of lease cancellations or renegotiations, or chronic overbuild in AI data centers. None is visible today.

Pillar 2: Leadership

Partial

The question: has the leader scaled before, and done what they said?

Wes Cummins co-founded the company in 2021 with Jason Zhang as Applied Blockchain, a bitcoin-miner hosting business, and renamed it Applied Digital in November 2022. By mid-2022 he was exploring AI workloads for the company's large power sites, ahead of the industry-wide AI pivot. The GPU cloud launched in 2023. Cummins is a technology investor by background: he joined B. Riley in 2002 as a publishing equity analyst covering tech hardware, became its Director of Research, then President of the firm across a nine-year tenure; he founded 272 Capital in 2020 and sold it to B. Riley in 2021; and he spent eight years before that at Nokomis Capital focused on TMT special situations. He holds roughly 4.2 million shares directly, so his financial outcome is tied to the stock.

For a pass: he called the AI pivot early, signed about $36 billion of contracted revenue, and delivered construction on schedule: the first 100 MW building at Polaris Forge 1 reached Ready for Service in November 2025, and Polaris Forge 2 broke ground in September 2025.

Against: his background is investing, not operating companies at scale; the company has a heavy dilution record (including a 49.4-million-share $160 million placement in September 2024); and the governance record carries a real blemish. In June 2023 the company disclosed in an 8-K that its Audit Committee had investigated a threatened, never formally asserted, allegation of sexual harassment by then-CMO Regina Ingel arising from a personal relationship with Cummins. The committee determined the relationship was consensual and the allegations unfounded, and the board reaffirmed Cummins as CEO. The stock fell about 16% to $8.09 that day, and short-seller reports and securities class actions followed.

What would change the grade: upgrade on two more years of clean delivery with no governance flags. Downgrade on any governance flag, a major construction delay, or a guidance cut.

Pillar 3: Moat

Partial

The question: can competitors take the customers?

The commercial traction is genuine: about $36 billion of 15-year take-or-pay contracted revenue with investment-grade counterparties is the kind of backlog most developers never assemble. The build spec is competitive too: closed-loop liquid cooling, a PUE of about 1.18, and 12-to-14-month construction timelines, all aimed at AI density.

But there is no structural barrier. Powered land, buildings, and liquid cooling are replicable, and others are replicating them: IREN signed a $9.7 billion five-year GPU cloud deal with Microsoft, and TeraWulf and Cipher Mining are pursuing the same hyperscaler lease model while trading at very high sales multiples. Customer concentration is real: the CoreWeave contract alone is about $11 billion of the roughly $36 billion backlog. And Nvidia, the most visible validator of the strategy, ended its equity endorsement: it participated in the $160 million placement in September 2024, then sold its entire 7.72-million-share stake in Q4 2025 (disclosed February 2026), sending the stock down about 8.4% after hours. APLD remains an Nvidia Cloud Partner and builds on Nvidia GPUs, so the equity exit is a portfolio rebalance, not a commercial divorce, but it removes a layer of third-party validation.

What would change the grade: upgrade on backlog growth and re-leasing at higher rates as the first campuses prove out. Downgrade on lease cancellations or renegotiations.

Pillar 4: Iron-fortress balance sheet

Fail

The question: can the company survive a bad two years without help?

No. Recent company disclosures show about $4.97 billion of long-term debt against about $4.15 billion of cash and about $1.72 billion of common equity: a debt-to-equity ratio near 2.9 and debt to capital near 0.74. The cash is construction funding, not a fortress: it arrived through financing, not through the business. The financing stack is heavy by design: $2.15 billion and $1.59 billion of 6.75% senior secured notes due 2031, a $430 million revolving credit facility, a $5 billion perpetual preferred equity facility with Macquarie, and a $300 million Goldman bridge facility that has been repaid.

Watch item: the $5 billion Macquarie perpetual preferred facility and repeated capital-market raises mean further dilution or subordination is always one funding need away. Any covenant breach or distressed raise would be the next shoe dropping.

What would change the grade: upgrade only on sustained free-cash-flow-positive quarters with net debt materially reduced. Interest coverage from operations, not from the cash pile, is the threshold.

Pillar 5: Free cash flow

Fail

The question: does the business generate real cash, repeatedly?

Not yet. Free cash flow is deeply negative, roughly negative $2.8 billion, against capital spending of roughly $2.9 billion: the campuses are being built with financed money, not with cash the business throws off. Progress is visible but partial: operating cash flow reached about $89.7 million in the latest quarter, adjusted EBITDA was $107.2 million for fiscal 2026 ($42.4 million in Q4), adjusted net income was $36.1 million, and Net Operating Income was $90.4 million. The GAAP picture is darker: a $249.2 million net loss for fiscal 2026, driven by heavy stock-based compensation and depreciation.

The path is clear but unproven. Fifteen-year take-or-pay leases convert to cash as buildings reach Ready for Service through 2027, starting with the first 100 MW at Polaris Forge 1 in November 2025. Until that conversion shows up in free cash flow for more than a quarter or two, the pillar fails.

Watch item: the GPU cloud was separated into ChronoScale (Nasdaq: CHRN) in May 2026, with APLD retaining about 96%. The non-GAAP measures exclude ChronoScale, which flatters the trend; judge the consolidated cash.

What would change the grade: upgrade on two consecutive free-cash-flow-positive quarters. A return to financing-dependent quarters keeps the fail.

What would change each grade

The backlog, the Nvidia exit, and the comps

The company reports total contracted critical IT load of about 1.4 GW and about $36 billion of base-term lease revenue. The table above lists the disclosed anchor leases; three leases totaling 810 MW and about $20.2 billion were signed with investment-grade hyperscalers during fiscal 2026 alone.

The Nvidia exit, framed factually. Nvidia participated in the $160 million private placement in September 2024 and disclosed 7.72 million APLD shares in its Q3 2024 13F. Its Q4 2025 13F showed the entire stake, worth about $177 million, sold; the stock fell about 8.4% after hours. The same filing exited Arm and WeRide while adding Intel (about $7.9 billion), Nokia, and Synopsys: a portfolio rebalance, not a verdict on APLD specifically. In January 2026 Nvidia deepened its CoreWeave relationship with a $2 billion equity investment. APLD remains an Nvidia Cloud Partner and builds on Nvidia GPUs. The equity endorsement ended; the commercial relationship continues.

How the group is priced. CoreWeave carries about a $58 billion market cap at roughly 5x forward sales with a $99.4 billion backlog. Nebius is near $58 billion at about 17x forward sales. IREN is about $16.5 billion with a $9.7 billion five-year Microsoft GPU cloud deal. TeraWulf and Cipher Mining trade at very high sales multiples while loss-making. At about $7.9 billion, APLD is the cheapest in the group against contracted revenue, roughly 0.2x the $36 billion backlog. That discount is the market pricing the leverage and the cash burn, which is exactly what the failed pillars measure.

Sources and methodology

Method. Each pillar is graded on evidence available through September 23, 2026. Financial figures come from company press releases and SEC filings. Grades are strict by design: a partial marks exactly what evidence is missing, and a fail marks what the standard requires but the company does not yet show. Where the company's stated totals differ from the sum of disclosed anchor leases, the company's totals are used and the difference is noted.

Disclaimer: This analysis 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. This analysis discusses the ticker only and contains no portfolio or position information.