APLDStock Deep DiveNeutral

APLD: Is the 23 billion yuan contract enough to sustain a 6x increase?

Adjusted EBITDA turned positive alongside GAAP losses, with valuation anchors shifting from price-to-sales ratio to contract cash flow

May 8, 2026
Share

NEXT PICK · Market Insights

APLD: Is the 23 billion yuan contract enough to sustain a 6x increase?

Friday, May 8, 2026

Adjusted EBITDA turned positive alongside GAAP losses, with valuation anchors shifting from price-to-sales ratio to contract cash flow

Over the past 12 months, it has risen 665%, with the latest stock price at $41.25 and a market capitalization above $11.79 billion—this is Applied Digital's most glamorous side in the capital markets.

You can flip to the other side of the financial statements

GAAP net loss for Q3 FY2026 was $100.9 million, with a loss per share of $0.36. In the same report, adjusted EBITDA was positive $44.1 million, and adjusted net profit was positive $33.2 million.

More importantly, the company already holds over $23 billion in long-term contract revenue, with more than 50% backed by investment-grade megascale tenants. What makes a company that transitioned from a crypto mining farm to an AI factory support such a valuation structure?

Latest price
$41.25
▼ -0.67%
Q3 adjusted EBITDA
$44.1 million
▲ Promoted to full position
Lock in the total value of the contract
$23 billion+
▲ Investment grade accounts for 50%+
$APLDApplied Digital Corporation $41.25▼ -0.67%

AI industrial real estate developer turning from crypto mining farm

If you treat APLD as an ordinary tech company, you won't understand it; Treating it as an "industrial real estate developer in the AI era" suddenly makes sense.

The company doesn't sell GPUs, doesn't sell clouds, and doesn't write software. It built large data centers on remote land in North Dakota, connected to cheap water and electricity, equipped with liquid-cooled piping and gigabit cables, and then leased the entire campus to massive computing power tenants like CoreWeave for 15 years. Rent is charged per MW, plus an additional electricity surcharge, essentially a "customized industrial park" built for AI training.

Currently, it has three major parks in its hands

Polaris Forge 1 in Ellendale, North Dakota (first phase 100MW online, second phase 150MW under construction), Harwood's Polaris Forge 2 (200MW under construction), and Delta Forge 1 in the south (430MW under construction). The three campuses together have a 4 GW development pipeline, propelling APLD into the ranks of leading AI data center suppliers in North America.

So where does its irreplaceability come from? The answer lies in two details. First, site selection—Polaris Forge 1 uses North Dakota's extremely low-cost renewable energy, with a per-MW combined cost far lower than traditional hubs in Virginia or Texas.

Second, credit binding—two of the three hyperscale clients are investment-grade, and in Q3, CoreWeave's main lease was restructured into an A3-rated SPV subsidiary. With improved credit quality, the discounted anchor of lease cash flow has also stabilized.

Is the 28x price-to-sales ratio a bubble, or just another anchor for the 23 billion contract?

Based on its current market cap of $11.79 billion and revenue of about $400 million over the past four quarters, APLD's price-to-sales (P/S) ratio is about 28-30 times—a figure that would make any value investor hesitate.

But what if you use a different ruler to measure it? The $23 billion locked contract is amortized on a 15-year average, equivalent to about $1.5 billion in annualized contract revenue. Once all three major parks are operational, the price-to-sales ratio will naturally be diluted below 8 times, close to the software industry average, and even lower than many AI concept stocks.

Cash flow has the same dual nature. The Q3 GAAP net loss of $100.9 million includes $59.7 million in non-cash impairment notes from the cloud business divestiture, as well as substantial equity incentive expenses; Excluding these, adjusted net profit turned positive to $33.2 million, and adjusted EBITDA was $44.1 million.

More notably, the financing succession — operating cash outflow for the nine months ending Q3 was only $42.9 million, with ending cash at $1.91 billion; In April, it issued $2.15 billion in 6.75% senior secured bonds, firmly covering the capital expenditure gap by external capital.

So, should it be discounted with a 28x price-to-sales ratio, or at a premium endorsed by 23 billion yuan in contracts and investment-grade endorsements? This is precisely the most dilemma when pricing APLD.

Adjusted EBITDA turns positive—has the break-even turning point really arrived?

The report classifies APLD as the "eve of breakout" phase, but the Q3 financial picture has already rewritten this assessment.

Adjusted EBITDA jumped from $500,000 in Q1 to $44.1 million in Q3, with adjusted net profit turning positive for the first time at $33.2 million. The first 100MW building at Polaris Forge 1 is running at full speed, contributing rental income for the entire quarter; The next 100MW building is also ready. The leverage of economies of scale has already emerged somewhere in the financial statements.

Technically, the latest price of $41.25 is well above the 50-day moving average of 29.56 and the 200-day moving average of 27.02—the market has voted with real money that the "breakout has happened." But be careful: stock prices are moving faster than fundamentals. A 665% increase in one year means the market has already priced in several years of earnings improvements into the stock price. The true "turning point confirmation" can only be stamped on the day GAAP net profit turns positive.

Is CoreWeave both a backer and a potential rival?

The AI data center hosting sector, where APLD operates, is a fierce battle of "two giants leading, multiple companies chasing."

The most direct competitors are the larger IREN (Green Energy-Driven Dual Business—Crypto + HPC) and NBIS (Vertically Integrated AI Cloud). Both are competing for orders for North American customized computing infrastructure. APLD's relative advantage lies in its already deployed capacity and the on-site endorsement of three hyperscale customers.

But the real tricky relationship lies with CoreWeave—it's not only APLD's core customer, but also the largest GPU leasing operator in North America, while massively building its own data centers. Once self-built capacity ramps up, demand for third-party suppliers may decline. Q3 restructured CoreWeave's main lease into an A3-rated SPV, precisely because APLD is tightening credit binding and locking in this layer of risk.

The longer-term threats come from Amazon AWS, Microsoft Azure, and Google Cloud—these hyperscale vendors are building their own AI campuses worth hundreds of billions of dollars. Currently, they are mainly self-built, with overflow demand flowing to third parties. But if one day cloud providers' capacity fully keeps up with demand, the market space for third-party hosting will be squeezed.

APLD's moat is more like a three-piece set of "location + first-mover + customer binding," rather than a technology patent.

4 GW development pipeline: How big can APLD take from the market?

Breaking down the AI data center market into three scales: global TAM is roughly $60-80 billion (AI/HPC accounts for about 25-30% of total data centers), North America's SAM is about $22-33 billion, and the customized HPC hosting segment is only about $2.2-5 billion.

APLD's three contracted campuses together total nearly 700MW of critical IT load, and a 4GW development pipeline means the long-term annualized revenue ceiling could reach several billion dollars. The mid-term target set by management in the earnings call is more specific: to achieve a net operating income of $1 billion within five years, corresponding to a 5-10% share of North American SAM.

This pie is not small, but whether it can be cut depends on two key factors: whether AI computing power demand continues to expand at an annual growth rate of 40%-60%, and whether hyperscale manufacturers continue to outsource overflow demand to third parties. The former appears to be accelerating in 2025-2026, while the latter depends on the details of capital expenditures each quarter.

Three things to do next: financial reports, new leases, and new capacity

The stock price rhythm over the next 6-9 months will be dominated by three types of catalysts: whoever cashes first will have the say in the next phase:

  • Q4 FY2026 Financial Report (expected July-August): Market attention is focused on whether the ramp-up of Polaris Forge 1 Phase II capacity will drive revenue growth quarter-on-quarter, and whether GAAP margin will substantially narrow
  • Delta Forge 1 first capacity delivery window (planned for mid-2027) and announcement of the fourth megascale tenant—once multi-signed, the valuation anchor will shift from price-to-sales to contract cash flow
  • The first phase of Polaris Forge 2, 200MW (expected around late 2026 to early 2027), combined with $2.15 billion in senior secured bond funding in place, marks a key point in testing execution

In terms of bypass risk, any changes in the 97% equity held by APLD after ChronoScale's IPO and any changes in CoreWeave's self-construction progress could act as reverse catalysts.

From $5 to $41, a three-part twist in the story

Breaking down the APLD trend over the past three years into three segments allows us to more clearly see the capital market's process of "re-re-examining" it.

The first segment is the transformation pains of 2022-2023—from crypto mining farm hosting to AI data center operator, with stock prices hovering around $1 and a very low revenue base. The second paragraph is a revaluation of the story from late 2023 to mid-2025—CoreWeave's multiple long-term lease announcements ignited market sentiment, pushing the stock price from $1 to above $10.

The real outburst happened in the third section

From the second half of 2025 to mid-2026. From the slightly better-than-expected Q1 earnings in October, to the issuance of $2.35 billion in guaranteed bonds and the first phase of Polaris Forge 1 energy conversion in November, to a 250% year-on-year surge in Q2 revenue in January and the signing of a $7.5 billion Delta Forge 1 deal in April, each event pushed valuations up a notch.

A 12-month increase of 665%, a 52-week range of $5.13 to $44.25—this is not a uniform growth curve, but multiple leaps in revaluation. Market confidence in AI infrastructure is stacked, but once cracks appear in any link, valuations shrink faster than expansion.

⚠️ Risk Notice

APLD's valuation heavily depends on several assumptions that have yet to be fully realized—the continued expansion of AI computing power demand, deepening customer diversification, and a substantial narrowing of GAAP margins over the next 12-18 months; Any delay in fulfilling at any stage could trigger a significant pullback. Customer concentration remains high, and any changes in CoreWeave's self-built capacity or the pace of capital expenditure by hyperscale vendors will directly affect the influence of lease renewal negotiations. In terms of capital structure, the continued advancement of convertible debt and equity financing means long-term dilution pressure. This article is for research and discussion purposes only and does not constitute any investment advice; Investing carries risks; please make independent judgments and make cautious decisions.

💬 Discussion

A story of a 23 billion contract—would you be willing to pay 28 times the price-to-sales ratio for it?

Data source

  • Applied Digital Q3 FY2026 Earnings Report (2026-04-08): https://ir.applieddigital.com/news-events/press-releases/detail/148/applied-digital-reports-fiscal-third-quarter-2026-results
  • Q2 FY2026 Financial Report (2026-01-07): https://ir.applieddigital.com/news-events/press-releases/detail/142/applied-digital-reports-fiscal-second-quarter-2026-results
  • Q1 FY2026 Financial Report (2025-10-09): https://ir.applieddigital.com/news-events/press-releases/detail/131/applied-digital-reports-fiscal-first-quarter-2026-results
  • Delta Forge $17.5 billion lease announcement (2026-04-23): https://ir.applieddigital.com/news-events/press-releases/detail/149/applied-digital-announces-new-u-s-based-high
  • ChronoScale Stripping Complete (2026-05-05): https://ir.applieddigital.com/news-events/press-releases/detail/151/applied-digital-completes-separation-of-cloud-business
  • Roth Capital raises target to $65 (2026-04-23): TipRanks/CNN Markets report
  • Real-time Stock Price/Market Cap: TradingView, Robinhood (Data snapshot as of 2026-05-09)

Disclaimer: This article is for reference only and does not constitute investment advice. Markets carry risk — invest with caution.