Applied Digital is moving from concept to monetized AI infrastructure, with revenue surging and major campuses now online or under construction. The stock offers real growth, but heavy capex, losses, and a rich valuation keep execution risk high.
Applied Digital(APLD) is a high-upside, high-volatility AI infrastructure developer whose core investment case rests on one simple fact: the business has already moved from concept to monetized scale. Fiscal Q3 2026 revenue reached $126.6M, up 139% YoY, adjusted EBITDA rose to $44.1M from $6.3M a year earlier, and the company reported that its first 100 MW HPC building is online while additional large campuses remain under construction. That combination matters. APLD is no longer selling a blueprint alone. It is selling operating capacity, contracted capacity, and a financing model built around hyperscale demand.
The bull case is driven by contracted infrastructure and visible buildout. Management said it has secured approximately $16B in contracted lease revenue, including $11B tied to CoreWeave and $5B tied to an investment-grade hyperscaler, while the investor presentation showed Polaris Forge 1 at 400 MW contracted critical IT load and Polaris Forge 2 at 200 MW contracted. On May 20, 2026, the company also said it surpassed 1 GW of contracted capacity with about $7.5B in total contracted value over an estimated 15-year take-or-pay term. Those are not small numbers for a company with trailing annual revenue of $319.3M.
The bear case is just as real. APLD remains GAAP unprofitable, with a trailing net margin of -59.5%, operating margin of -20.5%, and FY2025 operating cash flow of -$115.4M. Capital intensity is extreme. Annual capex reached $681.6M in FY2025, quarterly capex reached $775.2M in the February 2026 quarter, and debt stood at $2.7B against $2.1B in cash, cash equivalents, and restricted cash as of February 28, 2026. This is not a sleepy data center landlord. It is a project-finance-heavy builder trying to outrun dilution, execution slippage, and customer concentration.
For a balanced, moderate-risk investor with a medium-term horizon, APLD looks most attractive as a selective accumulation story rather than a blind momentum chase. The company has real assets, real tenants, and real operating progress. It also has a valuation that already prices in a lot of future success, with EV/Revenue at 38.9x and forward P/E at 526.3x. That leaves little room for construction mistakes or financing friction. The right stance is constructive but disciplined: respect the platform, respect the execution, and do not ignore the price being paid for that promise.
Get AI-powered research reports, daily market intelligence, and a personal analyst in your pocket.
Get Full Access →Applied Digital(APLD), based in Dallas, Texas, designs, develops, and operates digital infrastructure for high-performance computing and AI workloads in North America. The company has 205 employees and operates through two core segments: Data Center Hosting and HPC Hosting. It was formerly known as Applied Blockchain and changed its name to Applied Digital in November 2022, a useful clue that management has been repositioning the business away from a crypto-only identity and toward broader digital infrastructure.
That repositioning is now visible in the numbers. Annual revenue rose from $8.5M in FY2022 to $55.4M in FY2023, $136.6M in FY2024, and $215.5M in FY2025. Through the first nine months of fiscal 2026, revenue reached $352.6M versus $177.5M in the comparable prior-year period. The business is scaling fast, but it is doing so while carrying the bruises of a build-first model, including recurring net losses and heavy capex.
Management is led by Chairman and CEO Wes Cummins, CFO Mohammad Saidal Mohmand, COO Laura Laltrello, and President Jason Zhang. In the April 8, 2026 earnings call, Cummins framed the company as an early mover in high-density AI infrastructure, saying APLD now operates one of the only 100 MW direct-to-chip liquid cooled data centers in the world online today. That statement is central to the company’s identity. APLD wants the market to see it less as a commodity host and more as a specialist builder for hyperscale AI demand.
The company’s current strategic map is concentrated in North Dakota and expanding into the Southern U.S. Polaris Forge 1 in Ellendale, North Dakota is the flagship CoreWeave campus. Polaris Forge 2 in Harwood, North Dakota is a 200 MW campus tied to an investment-grade hyperscaler. Delta Forge 1 is a 300 MW AI factory campus in a strategic Southern U.S. market with initial operations expected in mid-2027. That campus language matters because APLD is not pitching one-off buildings. It is pitching repeatable infrastructure clusters with long-duration lease streams.
APLD’s business is best understood through its two operating engines: HPC Hosting and Data Center Hosting. The HPC Hosting business is the growth engine tied to AI and hyperscale demand. The Data Center Hosting business is the steadier legacy engine tied largely to crypto-related hosting assets. A third activity, cloud services, is being separated through the proposed combination with EKSO Bionics to form ChronoScale Corporation, with Applied Digital expected to own about 97% of the combined company at closing in fiscal Q4 2026.
In fiscal Q3 2026, the HPC Hosting business generated $71M of revenue. CFO Mohmand broke that into $44.1M of base rents, $18.9M of tenant fit-out services, and $8.1M of power pass-through and other ancillary revenue. Segment operating profit was $17.6M. This is the segment investors care about most because it is where AI infrastructure economics start to show up in reported results.
The Data Center Hosting segment generated $37.5M of revenue in fiscal Q3 2026, up 7% YoY from $35.2M, and produced $13.9M of operating profit on $119.6M of reported assets. Management called it the highest return-on-assets business in the company. In plain English, this is the cash-generating workhorse that helps support the broader buildout, even if it no longer carries the strategic glamour of AI campuses.
Cloud services contributed $18.1M of revenue in the quarter, but the segment also recorded a $59.7M non-cash write-down due to reclassification from held for sale and reported a $52.2M loss. Management excluded the segment from certain non-GAAP results because the company is evolving toward a pure-play data center platform. That makes strategic sense, though it also means investors need to separate core campus economics from accounting noise tied to portfolio reshaping.
The segment mix is improving. In earlier years, APLD looked more like a speculative infrastructure roll-up with uneven economics. In fiscal Q3 2026, the HPC business became large enough to materially reshape the company’s revenue base and margin profile. Gross profit in the February 2026 quarter rose to $53.8M with a 42.5% gross margin, versus $26.0M and 20.6% in the November 2025 quarter. That kind of step-up usually means the newer, higher-value infrastructure is starting to matter.
Get AI research on any stock
Instant reports, daily intelligence, and an AI analyst in your pocket.
APLD’s flagship product is not a software application. It is its Polaris Forge campus platform, especially Polaris Forge 1. The investor presentation described PF1 as a 400 MW contracted critical IT load campus in Ellendale, North Dakota, with 100 MW energized and 300 MW under construction. Management said PF1 has three leases with CoreWeave for a combined 400 MW and about $11B in anticipated aggregated contract value over a 15-year horizon.
That is the heart of the story. PF1 is proof that APLD can source power, build specialized AI-ready capacity, sign a major tenant, and begin monetizing the asset. In the earnings call, Cummins said the first 100 MW building is operating and 1,200 skilled craft professionals are progressing in parallel on two new 150 MW facilities. Mohmand added that the company realized a full quarter of lease revenue from the 100 MW data center in the HPC hosting business and expects revenue to ramp significantly over the next 12 months as the two 150 MW buildings come online.
That quote matters because it captures the product differentiation. PF1 is not just powered shell space. It is purpose-built for high-density AI workloads, where cooling architecture, power distribution, and speed to service are part of the product. In infrastructure, the best product is often the one that disappears into uptime, efficiency, and tenant confidence. That is less flashy than software, but hyperscalers care about it a great deal.
PF2 is the next major product extension. The investor presentation described Polaris Forge 2 as a 200 MW contracted campus in Harwood, North Dakota. Management said both buildings are advancing well, with foundations largely complete and work shifting to precast direction and interior fit-out. The company also disclosed a $2.15B private offering of 6.75% senior secured notes due 2031 to support 200 MW of critical IT load at PF2. Product quality is one thing. Product finance is another. APLD is trying to show it can do both.
APLD’s advantage is an infrastructure moat, not a software moat. The company competes on site control, available grid power, specialized cooling design, execution speed, and financing access. In the April 2026 investor presentation, management highlighted anticipated site NOI margins of 88% ± 3% for PF1, 86% ± 3% for PF2, and 85% ± 3% for Delta Forge 1, with anticipated capex per MW of $11M to $13M. Those are ambitious economics, but they show where management believes the value sits: once a campus is leased and operating, the annuity stream can be very attractive.
The most concrete competitive advantage is early execution in liquid-cooled AI infrastructure. Management said APLD broke ground on its first 100 MW facility more than two years ago and now operates one of the only 100 MW direct-to-chip liquid cooled data centers online. In a market where many players advertise AI readiness, having an operating asset is the difference between a blueprint and a bridge that already carries traffic.
Customer validation is another edge. CoreWeave is a major anchor tenant, and management said the company has approximately $16B of contracted lease revenue split between $11B from CoreWeave and $5B from an investment-grade hyperscaler. On March 30, 2026, APLD executed amendments with CoreWeave that restructured portions of the ELN-02 and ELN-03 leases through a CoreWeave SPV, added an unconditional springing parent guarantee from CoreWeave, and secured a $50M letter of credit. Mohmand said the SPV received an investment-grade A3 rating, an improvement from BB.
That credit upgrade matters because infrastructure returns are often won or lost in the financing stack. Mohmand told investors that single-A spreads are generally in the low- to mid-200 basis point area, while BB spreads can run 350 to 450 basis points. If APLD can refinance future tranches closer to investment-grade economics, the equity value of these campuses improves materially. In project finance, a few hundred basis points is not a rounding error. It is the whole weather system.
A final edge is capital partner support. Management said it has access to $4.1B in preferred equity from Macquarie Asset Management following a mutually agreed executed lease with an investment-grade hyperscaler. That kind of structured capital is important because many smaller AI infrastructure hopefuls can secure land or tell a good story, but far fewer can fund multibillion-dollar campuses without repeatedly tapping public equity.
Execution is the operating test for APLD, and so far the company has tangible progress to point to. Cummins said all buildings under construction at PF1 and PF2 are progressing on time and on budget. He also said that building large-scale data centers through a North Dakota winter is no small task, but the company is executing with years of experience, thousands of skilled professionals, and partners including McGough, ABB, Adolfson and Peterson, and BASX.
The operating footprint is expanding. At PF1, the first 100 MW building is online and two additional 150 MW facilities are under construction. At PF2, both buildings are advancing with foundations largely complete. During the quarter, APLD also broke ground on Delta Forge 1, a 300 MW campus spanning more than 600 acres in a Southern U.S. market, with initial operations expected in mid-2027.
The company is also actively marketing four development sites with approximately 1 GW of total grid power capacity across those locations, subject to approvals. Those include Delta Forge 1, an additional North Dakota site, and two sites in unnamed states. Management said some campuses are in advanced stages of negotiation. That pipeline is encouraging, though it also reinforces that APLD is managing a construction and development machine, not a mature stabilized portfolio.
Power sourcing is central to operations. Cummins repeatedly emphasized that everything APLD is marketing is grid power and that grid power takes priority over behind-the-meter generation. He also said the company looks for pro-business states and markets where labor can be secured for construction. That focus is sensible. In AI infrastructure, power availability and labor execution are the equivalent of inventory and logistics in retail. If either fails, the whole model stalls.
There are still operational frictions. Management said it delayed the South Dakota site after failing to get a tax exemption from the legislature and is evaluating reallocation of the associated power agreements. That is a reminder that permitting, local politics, and tax treatment can alter project economics quickly. Infrastructure investing always sounds clean in a slide deck. On the ground, it is mud, steel, utility negotiations, and county meetings.
APLD is riding a powerful market wave: the scramble for AI compute capacity and the power infrastructure needed to support it. Gartner forecast worldwide data center spending to rise 31.7% in 2026 and server spending to accelerate 36.9% YoY, driven by hyperscale cloud demand for AI workloads. That backdrop aligns directly with APLD’s strategy as a builder of high-density AI campuses.
Management reinforced that demand picture in the April 2026 call. Cummins said the largest U.S. hyperscalers’ annual capex had increased from about $400B to nearly $700B and described it as one of the largest investment cycles in U.S. history compressed into a short time frame. He added that every hyperscaler APLD targets remains engaged aggressively in the market and that the company sees multiple hyperscalers interested at every location it is marketing.
The demand side looks strong, but supply is constrained by power. That is why APLD’s North Dakota and broader grid-access strategy matters. The company is not competing only on square footage. It is competing on energized, financeable, AI-ready megawatts. In this market, a megawatt with permits, cooling design, and a tenant path is worth much more than a brochure promising future optionality.
The company’s own growth numbers show that market demand is translating into financial results. Trailing revenue growth was 139.3% YoY. Quarterly revenue rose from $52.9M in the February 2025 quarter to $126.6M in the February 2026 quarter. Nine-month revenue nearly doubled to $352.6M. Those figures support the idea that APLD is participating in a real infrastructure build cycle rather than a purely speculative narrative.
Still, investors should separate market demand from guaranteed shareholder returns. AI infrastructure demand can be enormous while individual developers still overbuild, overpay for capital, or dilute equity holders. APLD has the tailwind. The question is whether it can convert that tailwind into durable per-share value rather than just bigger campuses and bigger financing packages.
Like what you're reading?
Get full access to AI-powered research reports, market analysis, and portfolio tools.
APLD’s customer base is increasingly defined by large-scale infrastructure users rather than retail-facing or small enterprise accounts. The most important customer disclosed in the materials is CoreWeave, which anchors Polaris Forge 1. Management said the company has $11B of contracted revenue tied to CoreWeave and $5B tied to an investment-grade hyperscaler, for a total of about $16B in contracted lease revenue.
That customer profile has advantages. Hyperscale and AI tenants sign large, long-duration contracts and can support project-level financing. Management described each new campus as intended to create a 15- to 30-year revenue stream backed by strong credits. The company’s May 2026 update also cited about $7.5B in total contracted value over an estimated 15-year take-or-pay term for more than 1 GW of contracted capacity. Those are the kinds of contracts that can turn a developer into an annuity platform.
The drawback is concentration. The business context explicitly identifies customer concentration as the single biggest business risk, and the facts support that view. A large share of near-term value is tied to one tenant and one platform. Management is aware of this. Cummins said he prefers to diversify customers at existing campuses rather than simply signing more capacity with current customers, and he set a goal of getting total contracted revenue to 70% investment grade.
The legacy Data Center Hosting segment serves crypto mining customers and remains profitable, but the strategic center of gravity has shifted. The future customer profile APLD wants is clear: hyperscalers, AI infrastructure buyers, and large enterprise-grade counterparties that can support long-term lease economics and lower-cost financing. That is a better customer mix than speculative spot demand, but it also means each contract negotiation carries outsized importance.
APLD competes in overlapping markets that include AI/HPC data center operators and larger colocation platforms. The industry context identifies Equinix(EQIX), Digital Realty(DLR), NTT, Switch, and Core Scientific(CORZ) among the relevant competitors or adjacent alternatives. The competitive challenge is broad: power access, land, permitting, financing, speed of deployment, and tenant credibility.
Against incumbents like Equinix and Digital Realty, APLD is far smaller and less diversified. Those larger operators have broader customer bases, established balance sheets, and more mature portfolios. APLD’s answer is specialization. It is targeting purpose-built AI capacity, liquid cooling, and rapid deployment in markets where power can be secured. That is a narrower lane, but it can be a lucrative one if executed well.
Against converted crypto infrastructure players and smaller private operators, APLD’s edge is stronger customer validation and financing structure. The CoreWeave relationship, the investment-grade hyperscaler relationship, and the Macquarie preferred equity access all help distinguish it from operators that have sites but lack institutional-grade counterparties or capital support. In infrastructure, credibility compounds. One signed hyperscaler tends to attract the next conversation.
The weakness is that APLD does not yet have the diversification or margin stability of mature peers. Its valuation also leaves less room for error. Without a usable peer valuation screen in the provided data, the cleanest conclusion is qualitative: APLD deserves a premium to weaker speculative developers because it has operating assets and contracted capacity, but it also deserves a discount to mature diversified data center leaders because its business remains concentrated, capital hungry, and under construction.
The macro backdrop for APLD is a mix of strong demand and expensive capital. On the positive side, AI infrastructure spending is surging, and management said hyperscalers are as aggressive as they have ever seen them. On the negative side, APLD is a capital-intensive builder, so interest rates, credit spreads, and project finance conditions matter enormously. This is one of those businesses where the macro tide does not just lift or sink the boat. It changes the cost of the boat while it is still being built.
Credit conditions are especially important. Mohmand said the CoreWeave lease restructuring improved the company’s credit position through an A3-rated SPV, a springing parent guarantee, and a $50M letter of credit. He also noted that single-A spreads are materially lower than BB spreads. That means a more favorable financing environment or better tenant credit support can directly improve project economics and equity value.
Energy and power policy are another macro variable. APLD’s campuses depend on grid access, utility coordination, and supportive local policy. The delayed South Dakota site after a failed tax exemption effort shows that state-level politics can shape economics. At the same time, management said the operating Polaris Forge 1 site has saved North Dakota ratepayers about $31M, which helps the company’s case in local development discussions.
APLD is also backing a power strategy through Base Electron, an independent power producer that plans to build roughly 1.2 GW of natural gas-fired generation capacity in the Dakotas region in partnership with regional utilities. Applied Digital is providing limited credit support through a guarantee that terminates if Base Electron raises at least $50M in financing or completes an IPO, and APLD shareholders would own about 10% of the new company. That move is strategically interesting because it links the company more directly to future power availability, though it also adds another layer of execution complexity.
Debt reached $2.7B versus $2.1B in cash, cash equivalents, and restricted cash as of February 28, 2026, leaving Applied Digital with a leveraged balance sheet that must support an aggressive buildout.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →Fiscal Q3 2026 revenue jumped 139% year over year to $126.6M and adjusted EBITDA climbed to $44.1M, but the company still posted a trailing net margin of -59.5%.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →Management says Applied Digital has about $16B in contracted lease revenue and more than 1 GW of contracted capacity, giving the growth outlook unusual visibility for a company with $319.3M of trailing annual revenue.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →EV/Revenue of 38.9x and forward P/E of 526.3x show the market is already pricing in a lot of future success, leaving little margin for construction or financing missteps.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →The report’s price framework points to a $58 fair value, with the stock viewed as a Buy but only for investors willing to tolerate high volatility and project-execution risk.
Unlock the full analysis
Subscribers get the complete breakdown — pick rationale, financial metrics, and recent earnings detail.
Get Full Access →Applied Digital(APLD) is one of the more interesting infrastructure stories in the AI buildout because it has already done the hard part that many smaller players still only describe. It has operating capacity, contracted megawatts, major counterparties, and a financing framework that is becoming more institutional. Fiscal Q3 2026 made that visible with $126.6M of revenue, $44.1M of adjusted EBITDA, and a full quarter of lease revenue from the first 100 MW HPC facility.
The company also has the right strategic instincts. It is pushing toward investment-grade customer mix, using structured capital instead of relying only on public equity, and trying to tie power strategy more directly to long-term campus development. Those are the moves of a management team that understands where value is created in this market.
But none of that makes the stock cheap. The fair value estimate of $58 reflects a business with real momentum and real optionality, offset by leverage, concentration, heavy capex, weak GAAP profitability, and internal-control concerns flagged in the 2025 10-K. APLD is not a stock to own casually. It is a stock to own deliberately.
The bottom line is simple. APLD has built enough to earn investor attention, but not enough to earn blind faith. For medium-term investors who can tolerate volatility and respect valuation discipline, the shares still merit a Buy. Just remember what kind of machine this is: powerful, expensive, and unforgiving of mistakes.
Applied Digital Corporation (APLD) climbs after landing a long-term 210 MW lease for its AI-focused data center and securing major financing support. The move reflects stronger demand, better customer quality, and improving execution as the company scales its high-performance computing infrastructure.
Arm Holdings plc American Depositary Shares (ARM) drops as a broader semiconductor selloff hits AI-linked stocks. The decline appears driven by sector-wide de-risking rather than company-specific news, leaving investors to weigh ARM’s strong long-term AI exposure against its stretched valuation and high volatility.
GE Vernova Inc. (GEV) drops 6.6% as investors take profits after a strong run, even though the latest company news was positive. Elevated volume suggests active repositioning, while the long-term business case remains supported by strong orders, raised guidance, and bullish analyst targets.