Applied Digital (APLD) Stock Analysis: Leases Are Not Cash

Executive assessment: a financing-sensitive growth investment

Applied Digital has advanced beyond a collection of proposed AI data centers. Its October operating update put live critical IT capacity at Polaris Forge 1 at 250 megawatts, while leases across five campuses covered approximately 1.41 gigawatts. But the distance between signed capacity and cash available to common shareholders remains the central investment issue. The company reported $341.9 million of quarterly revenue and a $221.0 million loss from continuing operations attributable to common stockholders. Those figures can coexist because construction services, financing costs and corporate expenses do not behave like stabilized rental income. [1]

Sources: November 10 price history [3]; August-quarter earnings and cash-flow statements [1] [2]. Percentages are calculations, not company guidance.

The business: powered buildings, not a pure software multiple

Applied Digital designs, builds and operates infrastructure for high-performance computing. Its HPC hosting business sells access to purpose-built AI facilities; its legacy Data Center Hosting business supplies energized space to cryptocurrency-mining customers. The latter generated $37.8 million of quarterly revenue and $13.3 million of segment operating profit, a materially different economic exposure from constructing the next AI campus. ChronoScale, approximately 96%-owned at the October update, houses the historic cloud business and is consolidated in GAAP results but excluded from the company's adjusted measures. [1]

The competitive asset is deliverable power plus a usable building, not merely land or a large announced pipeline. Management reported $36 billion of base-term contracted revenue, rising to roughly $86 billion only if all renewal options are exercised. Neither figure is present value, annual revenue, current backlog cash nor equity value. The leases extend over years, require delivery, and sit alongside debt and other capital claims. CoreWeave leases Polaris Forge 1; other campuses are leased to hyperscalers described by management as investment grade. Customer quality helps, but does not eliminate completion obligations or concentration risk. [1]

Capacity also needs consistent definitions: 250 MW live at Polaris Forge 1 is an October operating milestone; 1.41 GW is leased capacity across the portfolio as of August 31. These are different dates and scopes, not evidence that the balance is immediately revenue-producing. Management's expectation of 300 MW delivered across its North Dakota campuses by calendar year-end is a forward-looking milestone, not an achieved November 10 observation.

Fundamentals: separate fit-out revenue from recurring rent

Fit-out services were 69.9% of HPC revenue. That distinction explains why rapid reported sales growth should not be treated as an equivalent increase in stabilized rental earning power. Tenant recoveries also bring associated expenses: the release identifies $13.3 million of reimbursable recoveries and corresponding rental costs. The better question is how recurring base rent and property-level cash earnings develop as buildings enter service, after allowing for the capital required to finish them. [1]

The gap between the $221.0 million common-stockholder continuing loss and $4.1 million adjusted net loss is not a small rounding issue. Adjusted figures remove ChronoScale and items including stock compensation and investment or derivative fair-value effects. Those exclusions can help isolate operations, but the preferred claims, dilution and corporate costs that affect common owners do not disappear. The cash-flow statement adds back $66.5 million of stock-based compensation; that supports reported operating cash without making compensation economically free. [1] [2]

The funding test: cash on hand is not surplus capital

At August 31, unrestricted cash and equivalents totaled $2.950 billion, with another $728.0 million of restricted cash across current and noncurrent classifications. The company described approximately $6.4 billion of debt. Restricted funds should not be counted as freely available to common shareholders, and unrestricted cash is supporting a multiyear construction program rather than simply waiting for distribution. These are August balance-sheet amounts, not a November cash balance. [2]

Quarterly purchases of property, equipment and other assets were $2.0747 billion. Subtracting that cash expenditure from $63.921 million of operating cash flow leaves a $2.0108 billion shortfall, before the separate $8.181 million investment in companies. This is a transparent cash comparison, not a standardized free-cash-flow measure or a forecast of every future quarter. Financing activities supplied $1.5435 billion, and total cash, equivalents and restricted cash fell by $475.5 million in the quarter. [2]

Financing has been accessible, but it has a price. A subsidiary's $1.59 billion senior secured notes carry a 7% coupon and mature in 2031; the parent provided a completion guarantee. The quarter also included $275.0 million of gross proceeds from Series G preferred issuance, while conversions of 154,500 preferred shares created approximately 4.5 million common shares. Funding growth with preferred securities and debt can preserve construction momentum while changing how much of the eventual asset value belongs to existing common owners. [2]

Technicals: the bounce has not repaired the trend

The November 10 close was $16.68, versus $16.72 on November 9. The 14-period Kaufman adaptive moving average stood at $19.39 and the 50-session simple moving average at $22.57. Price was 14.0% below the adaptive average and 26.1% below the simple average. Both averages declined from the prior session, so a rebound from the November 5 closing low of $14.17 has not yet reversed the broader downtrend.

A sustained move back above the adaptive average, followed by improvement in the slower average, would be better evidence of stabilization than one strong session. Conversely, momentum indicators cannot verify a construction budget or refinance a loan. The technical and fundamental tests should agree before a falling price is described as an attractive margin of safety.

Valuation: do not capitalize the entire contract headline

StockAnalysis reports $872.25 million of trailing revenue through the August quarter and negative trailing net income. With GAAP losses, a conventional earnings multiple is not a useful anchor. Nor is a simple sales multiple sufficient when a large portion of the latest quarter reflects tenant fit-out rather than recurring rent. A quarter's fit-out work should not be capitalized at the same multiple as an established rental annuity. [4]

A useful valuation sequence is to estimate delivered rental capacity and its cash earnings, subtract remaining build costs and financing obligations, allow for preferred and noncontrolling claims, then divide the residual by an appropriately diluted share count. Management's $36 billion base-term contract total does not supply those inputs by itself. This report therefore assigns neither a point price target nor numerical scenario probabilities. The available operating milestones are more defensible decision tools than a false-precision equity valuation.

Catalysts: follow delivery and funding, not just announcements

1. Year-end North Dakota delivery. Management expected 300 MW of delivered critical IT load by calendar year-end, versus 250 MW live at Polaris Forge 1 in its October update. Watch confirmed readiness and rent commencement, not just construction photographs. [1]
2. The next financial update. Look for the mix of base rent, fit-out revenue and recoveries; cash spending; new financing; and actual shares outstanding. These establish whether scale is improving common-owner economics.
3. Financing terms. Completed financing on acceptable terms matters more than a proposed facility's maximum size. The existing 7% secured note coupon, completion guarantee and preferred conversions show why the capital structure belongs in the investment thesis. [2]
4. Power expansion with discipline. The announced Finland agreement covered up to roughly 1 GW of potential power capacity. It is an option on future development, not energized capacity, a signed tenant lease or current revenue. It should earn valuation credit only as commitments and economics become clearer. [1]

Three paths: what evidence would change the view?

Analytical scenarios, not forecasts. No probabilities or investment returns are assigned.

Risk analysis and bottom line

• Completion and power: a contracted customer cannot pay rent on the intended schedule if a building is not ready; completion guarantees can transmit project risk to the parent.
• Capital structure: secured debt, preferred dividends and conversion features mean enterprise growth need not translate proportionately into value per common share.
• Customer and industry concentration: dependence on large AI infrastructure customers and their deployment plans creates correlated demand and financing risk.
• Accounting comparability: ChronoScale consolidation and its exclusion from adjusted measures require consistent scope; adjusted EBITDA is not common-owner cash flow.
• Expansion discipline: international power options and additional campuses can compete with existing projects for attention and capital before they generate rent.

Bottom line: Applied Digital offers real exposure to scarce, powered AI infrastructure, but the common stock is a claim on what remains after a costly delivery and financing cycle. The decisive evidence is not whether leased megawatts or headline revenue rise. It is whether recurring rental cash grows quickly enough to support the capital structure without repeatedly shrinking existing shareholders' share of the outcome.

Coverage limits: prices and technicals stop at the completed November 10 session; November 11 premarket quotes are excluded. Financials retain their August 31 reporting date. Capacity expectations and power agreements are company statements, not independently certified completions. No live macro assessment, social-sentiment score, comprehensive insider conclusion or unsupported target is used. This is investment research, not personalized investment advice.