Applied Digital Corp. (APLD) stock intrinsic value fair value estimate & owner earnings analysis

Applied Digital is a U.S. developer and operator of large-scale data centers, pivoting from its original crypto-mining hosting business to leasing purpose-built, high-power campuses to AI/HPC compute companies on very long-term contracts. The company has signed 15-year, take-or-pay, non-cancellable leases across five campuses (Polaris Forge 1, 2 and 3, and Delta Forge 1 and 2) totaling roughly 1,410 megawatts of committed capacity and about $36.2 billion of contracted revenue over the initial lease terms. The largest is a 400-megawatt, roughly $11 billion lease with CoreWeave at Polaris Forge 1 in North Dakota, delivering in phases between late 2025 and mid-2027; the other four campuses are leased to unnamed "investment-grade" hyperscale customers, with delivery dates running from mid-2026 through the second half of 2028. Alongside this, the company still runs a much smaller, legacy business renting powered space to a single crypto-mining customer in North Dakota (about $154 million of revenue in the year ended May 2026). Management has said this segment will shrink as a share of the business, and the customer's remaining contract term was roughly a year and a half as of mid-2026, with no replacement customer named. In May 2026 the company also spun its cloud-computing business into a separately-traded, majority-owned (about 97%) subsidiary, ChronoScale, which still consolidates into these results but is no longer central to the story. This is a business being built ahead of its earnings. In the year ended May 2026 the company spent about $2.9 billion on construction - more than four times that year's total revenue of $611 million - financed with roughly $5 billion of new debt, including $2.35 billion of senior secured notes carrying a 9.25% interest rate, plus new preferred stock and a joint-venture partner's capital. Cash flow from operations, after removing the large non-cash stock compensation the company pays employees and normal working-capital swings, has been negative every year we have full records for: roughly $0.80 per share of cash loss in the year ended May 2024, $0.44 in the year ended May 2025, and $0.68 in the year ended May 2026. The worsening in the most recent year mainly reflects a sharp increase in stock-based pay tied to headcount growth and performance milestones on the buildout, not a deteriorating core business - so this pattern is not new to the latest year, and we do not expect it to reverse on its own. The story only turns because capacity that isn't generating cash yet starts to. A meaningful share of the interest on the new construction debt is currently capitalized rather than run through the income statement, which is standard while a project is under construction, but that interest shows up in cash flow in full once each campus is delivered and the debt has been outstanding a full year. Because virtually the entire value of this business sits in leases that either haven't started yet or have barely started, we based our estimate of value almost entirely on the year just completed and on the contracted lease schedule itself, rather than on a longer earnings history that describes what was, in effect, a different company. Base case: the five campuses deliver roughly on the schedule disclosed in the filing; the legacy crypto-hosting revenue ends when its contract does, with no renewal assumed since none is disclosed; and the company needs to raise meaningfully more capital - a mix of additional project-level debt and some further stock issuance - to finish construction, moderately diluting today's shareholders. Cash losses persist through the year ending May 2027 before turning modestly cash-generative as more leases go live, reaching a mature run rate only once all five campuses are fully delivered around 2029. On this basis we estimate the shares are worth on the order of $6-7 today. Bull case: the company continues to execute on schedule, as it has so far (the first phase of Polaris Forge 1 was delivered on time in October 2025); finances the rest of the buildout mainly with additional secured project debt and joint-venture capital rather than heavy new stock issuance; and the business reaches the higher end of the margins its older, similar hosting operation has actually produced in past years. The renewal options that exist on two of the five campuses' leases are exercised as their initial 15-year terms approach expiration. On this basis we estimate the shares could be worth on the order of $13. Bear case: financing the roughly $10 billion-plus of construction still needed for the three not-yet-delivered campuses proves harder or costlier than the company's own recent 9.25% borrowing rate would suggest, pushing delivery out by a year or more and forcing considerably more stock issuance to bridge the gap - we assume roughly double the dilution of the base case. The legacy crypto-hosting revenue ends on schedule with nothing to replace it. We also note that in March 2026 the company amended its main lease with CoreWeave, shifting part of the leased capacity to a CoreWeave financing subsidiary backed by a guarantee rather than leaving it with the original tenant directly - a reminder that even a "non-cancellable" contract can be restructured, and that CoreWeave itself is a young, fast-growing, heavily-leveraged company whose ability to keep paying under a 15-year, multi-billion-dollar commitment hasn't been tested through a full economic cycle. Under this combination of delay, dilution and financing stress, we estimate the shares are worth on the order of $2-3 today - still positive, because most of the contracted revenue sits with several different large customers rather than one, and the leases are structured so that walking away requires paying the full remaining contract value. Putting these together, the margin-of-safety price sits at a meaningful discount to our base-case estimate - close to our bear-case value - to allow for the real risk that construction, financing or a major tenant relationship doesn't go as planned. One item we could not fully reconcile from the filings: disclosed segment results do not separately break out corporate overhead or interest expense by segment, so our estimate of owner earnings once all five campuses are delivered required judgment about the margin and interest cost the business will carry at full scale, rather than a figure taken directly from the filing.