人生不如意

I Made This Entire Investment Thesis Video About Applied Digital Using AI — Here’s Why That Matters

How AI created this video, this analysis, and why Applied Digital is building the infrastructure to scale it all.

James Soldinger

Why I Made This Video — and Why I’m Sharing Everything

https://www.youtube.com/embed/Hj_DS5CAr0Q

This analysis of Applied Digital Corp (APLD) isn’t just about a stock — it’s about what’s coming. I created the video at the top of this article using only AI: the slides were built with Gamma, the script and visuals were generated with ChatGPT, the voiceover was cloned with ElevenLabs, and the final video was produced without ever stepping in front of a camera or microphone.

Now — is it perfect? No. The audio could absolutely be better. But that’s not the point. The point is that this entire artifact — this research, the voiceover, the design, the delivery — was created in a few hours, by one person, using tools that are getting exponentially better every month. That speed and accessibility are changing how knowledge is created and shared. And it’s exactly that kind of transformation that Applied Digital is helping enable behind the scenes.

I’m also sharing my full research report, executive summary, and scenario models — not because this information is proprietary, but because it isn’t. It’s all out there. The only difference is how quickly you can get to it, how deeply you can analyze it, and how powerfully you can synthesize and share it — if you know how to use AI. That’s the edge. And the infrastructure behind that edge? That’s the real story. That’s why I’m long APLD.

🚨 Bottom Line Up Front (BLUF)

Applied Digital (APLD) is doubling down on its bet to become a leading AI-focused data center infrastructure provider. Recent quarterly results disappointed Wall Street, and their unexpected decision to offload the high-growth Cloud Services segment caused a knee-jerk sell-off. But beneath the surface, the core investment thesis hasn’t changed — in fact, it’s stronger than ever. With major backing from Nvidia and Macquarie, fully-funded expansion at Ellendale, and explosive industry demand for AI compute capacity, Applied Digital is exactly where you want it to be: undervalued and misunderstood by the market. The recent stock plunge isn’t a warning; it’s a gift.

📝 Executive Summary

Applied Digital Corporation (NASDAQ: APLD) is having an identity crisis — or so it might appear. Quarterly results came in soft, triggering a sharp sell-off. More significantly, the company announced a pivot away from its fast-growing GPU Cloud Services business, refocusing purely on high-performance computing (HPC) data centers designed specifically for AI workloads.

Wall Street panicked, seeing short-term disruption and uncertainty. But smart investors should zoom out. This strategic pivot positions Applied Digital for clearer execution, stronger margins, and, critically, opens the door for future REIT conversion — potentially unlocking major valuation upside.

The heart of the story is the 400 MW Ellendale HPC campus, which is fully financed and nearing completion. Multiple hyperscalers — think tech giants urgently needing GPU-heavy infrastructure — are currently in lease negotiations. Big-name strategic backers like Nvidia (who recently took a meaningful equity stake) and Macquarie (committing nearly a billion dollars in project-level capital) aren’t going anywhere; they’re doubling down.

Here’s how this could play out over the next 12–24 months:

🚀 Bull Case ($12–15/share): Quick and substantial leasing announcements; strong divestiture proceeds; rapid execution leading to major re-rating of shares.

🎯 Base Case ($6–9/share): Moderate leasing progress, successful sale of Cloud Services, steady financial improvement — stock gradually moves higher.

🐻 Bear Case ($2–4/share): Leasing delays, underwhelming sale price, sustained execution issues causing persistent stock pressure.

Applied Digital isn’t without risk — execution must improve, and the near-term will remain volatile. But at current prices, the risk-reward is remarkably attractive. Investors bullish on the continued explosive demand for AI computing capacity should take advantage of recent weakness and buy before the market catches on.

Applied Digital (APLD) at a Crossroads: Post-Earnings Deep Dive

Overview: Turning Point After a Disappointing Quarter

Applied Digital Corp. (NASDAQ: APLD) — a builder and operator of next-gen data centers for high-performance computing — just released fiscal Q3 2025 results (for the quarter ended Feb 28, 2025) that underscore the company’s transitional moment. Revenue came in at $52.9 million, up 22% year-on-year, but ~18% below Wall Street expectations of ~$64.5M (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). The miss was driven by an unexpected 36% sequential drop in Cloud Services revenue, due to a shift in business model and technical hiccups (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). GAAP net loss widened to $36.1M (−$0.16/share) (8k.pdf), missing the Street’s EPS estimate (–$0.11 (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq)), though the non-GAAP loss of $0.08/share beat by 2¢ (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) after adjusting for large depreciation and stock comp. Adjusted EBITDA was $10M, also well below the ~$17M consensus (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints), reflecting margin pressure from heavy infrastructure investments.

Investors reacted harshly. The stock plunged ~30–40% the next day to around $3.90 (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga) — now roughly 70% off its 52-week highs (it had traded above $10 earlier on AI hype). The market’s alarm stemmed not only from the revenue shortfall, but from uncertainty around a major strategic shift: Applied’s board approved a plan on April 10 to sell off its Cloud Services segment entirely (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). This effectively unwinds the company’s foray into providing GPU-based cloud computing, which had been a high-growth business line. Management says the divestiture will sharpen focus on core data-center hosting and set the stage for a potential conversion into a data center REIT (real estate investment trust) down the road (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). But in the near term, investors are left with questions: Why exit a fast-growing segment? Can Applied fill the gap by scaling its Ellendale data center campus for AI customers? How will this pivot impact financials, valuation, and risk/reward for the company?

In this report, we analyze Applied Digital’s latest developments — including the Cloud Services spin-off decision, progress at Ellendale, booming AI infrastructure demand (and Nvidia’s involvement), and what the bull, bear, and base cases look like from here. We’ll also examine market/analyst sentiment, and the company’s financial health, to assess whether APLD’s post-drop price represents a bargain opportunity or a value trap in the making.

Q3 Earnings Highlights and Key Developments

Cloud Services whiplash: The quarter exposed the volatility of Applied’s Cloud Services unit. This segment, launched in mid-2024, rents out high-end GPU clusters (largely Nvidia-based) to AI and machine-learning customers. In Q3 it generated $17.8M revenue, up 220% year-over-year — impressive growth — but down sharply from $27.7M in Q2 (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). Management explained that they shifted some GPU capacity from single-tenant “reserved” contracts to a multi-tenant, on-demand cloud model, causing near-term disruption (technical integration issues delayed revenue recognition). The issues have since been resolved, but the quarter’s dip was stark. Meanwhile, the legacy Data Center Hosting business (mainly crypto mining customers co-locating at Applied’s facilities) contributed $35.2M, down 7% YoY. Hosting revenue has been relatively flat to declining, as it depends on stable fees for power/space (and was already near full capacity).

Total Q3 revenue of $52.9M missed consensus by a wide margin, which “negatively impacted investor sentiment” (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq) (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq). On the earnings call, CEO Wes Cummins emphasized that Cloud Services was not a core long-term focus and revealed the board’s decision (just days before the report) to divest this unit (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). “We believe separating the Cloud Services business from our data center operations better serves the long-term interests of our shareholders,” Cummins said (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). In other words, management sees more value in being a pure-play HPC data center operator — potentially even a REIT in the future — than in running a cloud services platform. The Cloud segment’s rapid growth had been a double-edged sword: it drove >50% of company revenue in recent quarters, but also introduced heavy capex (purchasing pricey Nvidia GPUs) and operational complexity outside Applied’s traditional expertise. By selling it, they can refocus on what they view as a higher-margin, lower-competition opportunity: providing infrastructure for the AI boom, rather than competing in cloud services against larger players.

Financial and personnel moves: In conjunction with the strategic refocus, Applied has shored up its finances and team:

Balance sheet check: As of Feb 28, Applied had $261.2M in cash (including restricted) against $689.1M in total debt (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq). Net debt was ~$428M, reflecting the significant borrowing to fund construction and GPU purchases. The Macquarie and SMBC deals, however, improve the picture moving forward: the SMBC loan refinanced short-term debt and funds expansion, while the Macquarie facility (once active) will effectively inject equity capital per MW leased, reducing the need for further debt. Still, interest costs have spiked — Q3 net interest expense was $8.9M, +87% YoY — and will remain a drag until the new data centers start generating cash. Liquidity appears sufficient for now (especially if the Cloud unit sale brings in additional cash), but execution is critical to avoid any funding gaps. We’ll discuss this more under Risks/Financial Health.

Shedding the GPU Cloud: Why Divest the Cloud Services Segment?

The planned sale of Cloud Services marks a pivotal strategy shift. This segment was only launched in mid-2024 but ramped extremely fast, leveraging Applied’s data center space and power to host Nvidia GPU clusters that customers could rent for AI workloads. By Q2 FY2025, Cloud Services had hit $27.7M in revenue — roughly 43% of total company revenue — and likely would have been even higher in Q3 absent the model change. Why would Applied jettison this business, just as AI demand is exploding?

There are a few likely reasons:

  • Capital intensity and margin profile: Running a cloud service means buying cutting-edge GPUs (like Nvidia A100s or H100s) and maintaining a cloud software platform — essentially stepping into the shoes of a mini-AWS or a specialized competitor like CoreWeave. This is expensive. Indeed, industry-wide, GPU cloud providers have been raising huge sums: e.g. CoreWeave raised $1.1B (plus $7.5B debt) in 2023 to expand its fleet (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register), and even giants like CyrusOne took on $7.9B debt to finance AI hardware (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register). Applied itself took on lease obligations for GPUs (note: $97M in new finance leases in the six months to Nov 2024 (10-Q — 01/14/2025 — Applied Digital Corporation)) and issued equity — including a $160M private placement in Sept 2024 involving Nvidia and others — to fund its GPU cloud build-out (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register). While Cloud Services yields high revenue per MW, it also carries high depreciation, R&D, and maintenance costs. In Q3, cost of revenue climbed to $49.1M (93% of sales) (8k.pdf) (8k.pdf), in part due to operating those GPU clusters. By contrast, pure hosting or leasing of data center capacity can be more predictable and potentially higher-margin once built out. Management hinted at this, noting that as GPU units shifted to revenue-generating mode, depreciation moved from SG&A to cost of sales — implying the Cloud business was a major driver of overhead.
  • Strategic focus and valuation clarity: Applied’s leadership believes the company’s long-term value will be maximized by focusing on data center infrastructure, not competing in cloud services. They even floated the idea of becoming a REIT (real estate investment trust) to emphasize stable, lease-based income (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints). Operating a cloud services arm alongside a data center landlord business muddles the story and likely the valuation. Separating them could unlock value — either by selling the Cloud unit to a strategic/financial buyer who values it on its own, or by enabling Applied (the remaining company) to potentially get REIT-like multiples on its steady hosting revenues. “If we were to transition into a data center REIT in the future, we need to focus… separating Cloud Services better serves shareholders,” the company said (8k.pdf) (8k.pdf). In essence, Cloud Services might be worth more to someone else (who is focused on cloud), and Applied can then redeploy the sale proceeds to its core projects.
  • Execution risk and customer overlap: The Cloud business, while fast-growing, had a setback this quarter (the on-demand transition issues). It also faces heavy competition from the likes of AWS, Azure, Google Cloud, and specialized providers (CoreWeave, Lambda Labs, etc.) chasing the AI gold rush. There’s risk in scaling it — both technical and sales-wise — that management may not be comfortable taking on, given their DNA is more in infrastructure. Additionally, there may be overlap or conflict between Cloud customers and hosting customers. Interestingly, Applied has been in negotiations to lease large chunks of capacity to “multiple U.S.-based hyperscalers”. It’s conceivable that some hyperscalers (or large cloud providers) might be uncomfortable leasing from Applied if it’s also running its own cloud service (i.e. a potential competitor). By exiting Cloud Services, Applied can position itself as a neutral provider of capacity to all big players, rather than a competitor to them.

Overall, while the Cloud unit sale will remove a high-growth revenue stream (estimated ~$100M annualized run-rate) in the short term, it could de-risk the company and improve its financial flexibility. We will have to watch what price and terms Applied can get for this business — that’s a key catalyst ahead. No guidance was given on the expected sale value or timing, which adds uncertainty (Why Applied Digital (APLD) Shares Are Sliding Today). Management only provided “visibility” that they have interest and are moving forward. If they fetch a strong price, it could inject cash to pay down debt (improving net leverage) or invest in more data centers. A low price or protracted sale process, however, would raise questions. For now, the market seems to be taking a “show me” attitude, which contributed to the stock’s selloff (uncertainty around losing the Cloud segment without clear recompense) (Why Applied Digital (APLD) Shares Are Sliding Today).

Ellendale HPC Campus: Build-Out Progress and Leasing Prospects

At the heart of Applied Digital’s bull thesis is its Ellendale campus in North Dakota — envisioned as a flagship high-performance computing (HPC) data center campus tailored for AI infrastructure. Here’s the latest on Ellendale:

So far, no lease wins have been officially announced — and this remains the make-or-break catalyst for Applied Digital in the coming months. The market likely will reward the stock if any sizable contract is inked (validating the demand) or punish it further if the Ellendale capacity sits uncontracted for too long. It’s worth noting that Nvidia’s own CEO highlighted data center bottlenecks as a constraint on AI growth: “AI has made GPUs one of the hottest commodities… but without data centers, the chip powerhouse and its customers have nowhere to put all that tech. With capacity in short supply, VC and chipmakers alike are pumping billions into datacenters to keep the AI hype train from stalling,” wrote The Register (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register) (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register). This context explains why Macquarie and SMBC are backing Applied — and it underpins the bull case that “if you build it, they will come.” But until we see actual leases, some skepticism will linger.

Beyond Ellendale, Applied is already eyeing further expansion: they have a planned 200 MW HPC facility in Iowa on the drawing board (Applied Digital closes $375m financing for Ellendale campus in North Dakota — DCD), and mention a pipeline of over 1 GW of additional capacity opportunities (Applied Digital closes $375m financing for Ellendale campus in North Dakota — DCD). The Macquarie partnership’s 30-month ROFR (right of first refusal) on up to $4.1B for future projects (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD)) (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD)) suggests both parties anticipate aggressive growth beyond North Dakota. If Ellendale proves out with marquee tenants, Applied could replicate the formula in other power-rich locales (Iowa being one). Essentially, management is positioning Applied to become a top-tier specialist in AI-focused data centers — potentially a valuable platform in an industry where M&A is common (large REITs or infra funds might eventually acquire successful niche operators).

Industry Tailwinds: Soaring AI Infrastructure Demand and Nvidia’s Role

The macro environment for AI infrastructure is extraordinarily robust. The rise of large-scale AI models (GPT-4 and beyond) has triggered a gold rush for GPUs and high-performance data center space:

  • Unprecedented GPU demand: Nvidia — the leading GPU supplier — has reported record data center revenues (>$30B in a quarter recently) and still cannot meet all the demand for its AI chips (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register). Enterprises and AI startups alike are racing to secure compute capacity. However, simply obtaining GPUs isn’t enough — you need somewhere to put them. That’s created a capacity crunch in data centers that can handle the power density and cooling needs of these GPU pods. Traditional cloud regions are often at capacity, and building new ones takes time. This dynamic hugely benefits specialist players like Applied (and competitors such as CoreWeave or Lambda). As The Register quipped, “Nvidia and chums [investors] inject $160M into Applied Digital to keep GPU sales rolling… datacenters are the lifeline for its $30B ML boom” (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register) (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register). In September 2024, Nvidia itself participated in Applied’s $160M equity round (alongside venture firms and real estate player Related Co.) (Applied Digital Announces $160 Million Strategic Financing, Fueling Transformative Accelerated Compute and AI Infrastructure :: Applied Digital Corporation (APLD)), showing Nvidia’s interest in ensuring ample third-party capacity for its hardware. According to regulatory filings, Nvidia held ~7.72M shares of APLD (about 3% stake) as of Q3 2024 (Nvidia Discloses Stake In Applied Digital, Keeps Arm, SoundHound …). This is a strong vote of confidence and aligns with Nvidia’s strategy of investing in its ecosystem (they’ve also invested in CoreWeave). Applied is officially a Preferred NVIDIA Cloud Service Provider (Applied Digital Announces $160 Million Strategic Financing, Fueling Transformative Accelerated Compute and AI Infrastructure :: Applied Digital Corporation (APLD)), which likely gives it early access to new GPU tech (the company even announced plans to host Nvidia’s next-gen Blackwell GPUs in the future (Applied Digital closes $375m financing for Ellendale campus in North Dakota — DCD)). Nvidia’s involvement not only brings credibility but could hint at customer referrals or collaborative go-to-market efforts.
  • AI data centers = the new growth frontier: The broader data center industry is undergoing a shift — beyond just hyperscale cloud for general workloads, there’s now a huge wave of specialized AI computing needs. Companies like Microsoft, Google, and Amazon are investing tens of billions in their own AI supercomputers. But not every company can build like the hyperscalers; many will look to outsource or lease. Even the big players may augment capacity via partners. For example, Microsoft has reportedly contracted with third-party HPC infrastructure providers to support OpenAI’s needs. Cloudless AI startups (those not tied to a major cloud provider) also need places to train their models — that’s why venture capital has poured into independent GPU cloud firms (as noted, CoreWeave’s multibillion financing, etc.).
  • The addressable market for AI-focused hosting is expanding rapidly. Applied’s move to pivot fully into this space is timed well: it coincides with the early phase of what many call an “AI arms race” in computing power. If the overall demand for AI compute continues its exponential trajectory, high-performance data center capacity could remain supply-constrained for several years, supporting high utilization and pricing for providers like Applied. On the Q3 call, management noted that lease rate expectations have improved since last year (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga), implying that customers are willing to pay a premium to secure capacity now (especially given the costs of AI downtime or model training delays).
  • Competitive landscape: It’s important to acknowledge that Applied is not alone in chasing this opportunity. Aside from big cloud providers and internal enterprise data centers, a handful of private competitors exist. The most notable is CoreWeave, a New York-based GPU cloud provider that has snagged marquee clients (like being a cloud partner for OpenAI) and, as mentioned, raised enormous capital. CoreWeave is building out multiple facilities (e.g. in Texas, North Carolina). There’s also Lambda Labs, Cerebras (via partner colos), and others specializing in AI hosting. Furthermore, traditional colocation giants (Equinix, Digital Realty) are also starting to tailor offerings for high-density AI racks, and hyperscalers might eventually lease spare capacity to third parties. Applied will need to leverage its first-mover advantage in ND and its financing partnerships to establish a strong foothold. The good news: the pie is growing so fast that multiple winners can thrive, at least in the near-to-medium term. And Applied’s blend of real estate acumen (via Related Co. and Macquarie), power access, and Nvidia’s blessing gives it a solid foundation.
  • Nvidia’s impact on sentiment: For many investors, the Nvidia connection was a key reason to pay attention to APLD in 2023. The stock ran up significantly after Nvidia’s stake was disclosed — in fact, APLD stock nearly doubled in the six months after that deal, reaching highs around $10–11 by late 2024 (Nvidia Discloses Stake In Applied Digital, Keeps Arm, SoundHound …) (Applied Digital (APLD) Valuation in Focus After Nvidia … — tastylive). Even after the recent plunge, analysts remain bullish largely because of these AI tailwinds. The consensus analyst rating is Strong Buy, and prior to Q3 the average price target was about $12 (APLD Stock Forecast: Analyst Ratings, Predictions & Price Target …). This week, Needham’s analyst reiterated his Buy rating, assigning a $10 target (trimmed from $11) despite the earnings miss (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga). He highlighted that the “megatrend in AI infrastructure” remains intact and that Applied’s management is taking the right steps (monetizing the cloud segment and focusing on leasing) to capitalize on it. As long as Nvidia’s growth (and by extension, demand for external GPU capacity) stays on fire, Applied Digital will have a powerful sector tailwind at its back.

In summary, the industry setup for Applied is highly favorable: rare are the cases where demand so clearly outstrips supply and investors are willing to fund capacity expansion in advance. But translating that into shareholder value will hinge on execution — which we explore via scenarios next.

Outlook: Bull, Bear, and Base Case Scenarios

Given the transformative changes underway, the range of outcomes for Applied Digital is wide. Below we outline three scenarios (bullish, bearish, and baseline) with key assumptions, potential financial outcomes, and valuation implications:

🔼 Bull Case: “Hyperscaler Home Run” — Applied rapidly leases out Ellendale and beyond, becoming a leading AI data center provider.

Assumptions: The company signs full 400 MW of leases by 2026, locking in major hyperscalers for multi-year terms. The Cloud Services sale is executed at an attractive price (perhaps ~$100M+), which is used to pay down debt and fund expansions. With Macquarie’s $900M equity infusion triggered by the leases, Applied builds out the entire campus by 2026 and possibly starts its Iowa 200MW project by 2027. Lease economics are strong — e.g. each 1 MW yields >$1 million in annual EBITDA (as hinted by management (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga)】 thanks to high utilization and favorable rates. By FY2027, Applied could be hosting ~600+ MW of capacity (400 Ellendale HPC + 180MW crypto + initial Iowa), with annual revenues perhaps in the $500M+ range (for illustration: if 400MW HPC generates even $1.5M revenue/MW/year, that’s $600M, plus ~$150M from legacy hosting). EBITDA margins might expand to ~40–50%, given scale and the shift to pure hosting. Importantly, with stable lease income, the company could convert to a REIT structure, attracting income-oriented investors. In this bull case, we’d expect a higher valuation multiple — data center REITs often trade at 20–25x EBITDA or 15–20x FFO. If Applied were doing, say, $200M EBITDA by 2027, a 20x multiple would imply a ~$4B enterprise value. Even discounting for the 15% minority stake of Macquarie in the projects, equity value could be in the mid-$3B range. With ~250M shares (assuming some dilution but also buybacks with cloud sale cash), that’s ~$12–15 per share potential, roughly 3–4x the current price. In a takeover scenario (not unlikely if it’s successful), larger infrastructure players might pay a premium on top. Catalysts in this scenario: announcement of big leases (100MW+ deals), completion of cloud sale at a good price, achieving positive net income (or FFO) by FY2026, and possibly initiation of dividends (if REIT). The bull case is essentially that Applied becomes the “Equinix of AI data centers”, riding a multi-year secular growth wave in AI compute.

🔻 Bear Case: “Stumble and Stall” — Projects face delays or under-utilization, and the cloud sale provides little relief.

Assumptions: Despite the industry buzz, Applied struggles to finalize hyperscaler leases. Perhaps negotiations drag on, or a prospective tenant opts to build in-house or use a competitor. Without signed leases, Macquarie’s $5B partnership might not close (since it’s conditional on at least one lease for the first 100M (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD))9】. That could leave Applied overleveraged after building Ellendale Phase 1. The 100MW facility might sit partially empty through late 2025/2026, forcing the company to rely on smaller AI customers or crypto miners to fill some capacity at lower rates. Meanwhile, the Cloud Services business is sold but maybe at a fire-sale valuation (e.g. just covering the cost of its GPUs). Without that segment’s revenue, and with crypto hosting flat or declining (if some miners go bust or renew at lower rates), total revenues might stagnate or even drop in FY2026 ($150M-$200M/year range). High interest costs ($40M+ annual) and ongoing operating expenses could keep net losses high. In a stress scenario, net debt could remain ~$400M+ while EBITDA stays modest (or even negative if a lot of capacity is idle). That raises the specter of needing to raise equity again (diluting shareholders) or take on costly mezzanine financing to service debt. The Macquarie deal’s preferred equity has a hefty 12.75% PIK dividend and a **1.8x liquidation preference (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD)) (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD))0】 — if Applied cannot grow into that capital, it could end up very expensive. In a bear case, one can imagine the stock languishing in the low-single-digits (or worse). For instance, at $3/share and ~250M shares, the market cap would be ~$750M; if net debt is similar, the enterprise value ~$1.2B might equate to a high multiple of whatever small EBITDA they generate (i.e. the market doubting they’ll ever scale). The downside could even involve restructuring if debt covenants tighten and cash flow isn’t enough. However, the company likely has enough runway (cash + remaining equity backing) that bankruptcy is a low-probability scenario in the next year or two. The more realistic bear outcome is a “show-me story” that never materializes: Applied’s assets (land, partially built centers, etc.) could then be sold off to recover value, but equity holders might see little gain. Catalysts in a bear scenario: continued quarterly losses without clear improvement, failure to announce leases by late 2025, any cancellation of the Macquarie facility, or a downturn in AI investment reducing demand (currently not expected, but if the AI boom cooled off, it’d hurt everyone in this niche).

➡️ Base Case: “Steady Ramp” — Applied executes moderately well: leases a good portion of Ellendale, but growth is stepwise.

Assumptions: The most likely path is somewhere in between. Perhaps Applied secures one major hyperscaler lease (~100–150 MW) by mid-2025, enabling the Macquarie funding to close (partially) and ensuring ELN02 is fully utilized by 2026. Additional leases for the next 100MW or so might come but spread over 2026–2027 as the company negotiates with multiple tenants (instead of all 400MW at once). So by FY2027, maybe ~200–250 MW of HPC capacity is earning revenue (plus the still-running 180MW crypto hosting). This would give total annual revenue in the ballpark of $300M-$400M by 2027 (e.g. 200MW HPC * $1.5–2M revenue/MW + ~$120–150M legacy = ~$300–450M). Cloud Services is gone, but the proceeds (say $50–75M net) help lighten debt a bit. Profitability improves steadily: EBITDA turns positive and grows to perhaps $100M+ by FY2027 (with margins ~30% as the mix shifts to more AI hosting). The company remains leveraged but manageable — net debt might actually increase initially (as they invest in finishing Ellendale) but then level off as cash flows from leases come in. In this base scenario, Applied transitions into a true data center operator with a solid niche. It might not get REIT status by then, but investors could start valuing it on EBITDA/AFFO and growth prospects. A reasonable valuation could be something like 10x forward EBITDA (given some execution risk still, and the pref equity overhead). If FY2027 EBITDA were ~$100M, 10x would be a $1B EV; if growth prospects remain strong, perhaps more like 12–15x, implying $1.2–1.5B EV. Deduct Macquarie’s piece and net debt, equity value might be on the order of $800M–$1B. With share count ~250M, that’s around $4–$6 per share — notably above the current ~$4 level, suggesting modest upside. However, a base case investor would be counting on the company proving itself over the next 1–2 years and likely tolerate volatility along the way. Catalysts and milestones for the base case: signing that first anchor tenant (which would likely bump shares toward the upper end of this range initially), hitting positive adjusted EBITDA consistently (the company did post +$10M this quart (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints)5】 but needs more to cover interest), and successfully refinancing any remaining high-cost debt as creditworthiness improves.

The swing factor across these scenarios is clearly the success of the HPC hosting pivot. One metric to watch will be EBITDA per MW of deployed capacity — management’s commentary that $1M EBITDA/MW is conservati (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga)6】 suggests a lot of potential operating leverage in the model if they can fill the sites. Another factor is what happens to the crypto hosting side: currently 286MW (Jamestown + Ellendale) fully used by mining clien7】. Crypto markets have been soft; if any of those clients reduce operations, that could free capacity (which could then possibly be repurposed to AI, but not guaranteed if design differences). Or conversely, if Bitcoin prices surge again, mining could remain a steady cash cow (the hosting contracts are likely take-or-pay, providing stable income as long as clients stay solvent). It’s somewhat ironic that a year ago, Applied was seen as a crypto proxy (it was even named Applied Blockchain before rebranding); now it’s all about AI — a testament to management’s nimbleness, but also something of a reinvention.

Market Sentiment and Analyst Views

After the post-earnings tumble, sentiment in the near term is cautious, but analysts covering APLD are generally still positive (with tempered enthusiasm). As mentioned, **Needham’s John Todaro reiterated a Buy with a $10 PT (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga)9】, viewing the sell-off as overdone given the long-term opportunity. He acknowledged the quarter’s miss but highlighted that management is taking corrective actions (exiting cloud) and that guidance on 200MW in 2026 was above his prior expectati (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga)9】. The Bulls (including apparently Needham and possibly HC Wainwright or others) argue that Applied is one of the few pure plays on AI infrastructure, and that its partnership with blue-chip investors (Macquarie, Nvidia) provides validation. They see the current price as a deep value if the company even partially executes — essentially buying “dollars of future AI capacity for 50 cents today.” Some point to how Nvidia’s stake of ~$63M as of last fa (Nvidia Discloses Stake In Applied Digital, Keeps Arm, SoundHound …)9】 signaled confidence, and that insiders and institutions hold over 75% of the sto ( Applied Digital (APLD) Stock Price, News & Analysis )9】, implying a strong alignment and support.

The Bears or skeptics, on the other hand, focus on execution risks and the financial strain. One recent Seeking Alpha article labeled Applied “a high-risk lottery ticket in the AI infrastructure gold rush,” noting that despite Nvidia’s involvement, the stock had slid 23% from November to January (even before this latest drop) and that *“fundamentals [were] still concerning. ([PDF] 01/14/2025 — Applied Digital Corporation)4】. Bears highlight that Applied has yet to prove it can attract non-crypto customers at scale; until a major lease is signed, there’s a risk that this AI pivot could be a “build it and hope they come” scenario. They also note that dilution has been significant — shares outstanding have increased substantially with the capital raises (the $160M equity in Sept 2024 issued ~49.4M shares at $3. (Nvidia joins $160M investment into data center builder Applied Digital)3】, plus prior warrants/convertible notes conversions). More could come if additional financing is needed. With a **Zacks Rank of Hold (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq)3】 and no profitability yet, some generalist investors may avoid APLD until it actually generates positive earnings.

Market reaction to recent news encapsulates this mixed view: the stock’s 30% one-day drop on earnings reflects disappointment and uncertainty, especially because management did not offer forward revenue guidan (Why Applied Digital (APLD) Shares Are Sliding Today)7】. The decision to remove guidance was likely due to the pending Cloud sale and timing of leases — but without it, the market is “flying blind” near-term. “Looking ahead, the company offered no definitive sales guidance and revealed plans to divest its Cloud Services unit, raising uncertainty about future revenue streams,” noted one Yahoo Finance bri (Why Applied Digital (APLD) Shares Are Sliding Today)9】. In other words, investors have to trust management’s long-term vision without much near-term predictability.

That said, any concrete positive development could swiftly change sentiment. For example, if tomorrow Applied announced a 5-year agreement with, say, Microsoft or Google for 100MW at Ellendale, the stock would likely rebound sharply — not only for the revenue implications but also because it would validate the business model and perhaps remove the overhang around the Macquarie deal closing. Conversely, continued silence on new contracts by next earnings could further erode confidence.

Trading at around $4 (market cap ~$950M) post-drop, Applied now sits roughly at an EV/Sales of ~5x (TTM) and an EV/EBITDA (forward) that’s hard to pin down but elevated due to heavy investment. These multiples are much lower than many high-growth AI stocks, but higher than established data center peers — reflecting its unique hybrid status. It’s almost a story stock at this point: valued on future potential more than current numbers. This is why the news flow in the coming months — cloud sale outcome, Ellendale tenant wins, etc. — will likely dictate whether APLD stock languishes or finds a new uptrend.

Risks and Challenges

While the growth narrative is compelling, investors should weigh several key risks:

  • Execution and Development Risk: Building large data centers on spec (without committed tenants) is inherently risky. Any construction delays, cost overruns, or technical issues in making Ellendale operational for AI clients could impair Applied’s first-mover advantage. Thus far construction seems on tra (8k.pdf)7】, but integration of advanced cooling and provisioning huge power loads leave little room for error. The technical glitch in Q3 on the Cloud side (when moving to multi-tenant) shows even the best-laid plans hit bum7】. Successfully deploying liquid-cooled GPU infrastructure at scale for the first time will be a major execution milestone.
  • Customer Concentration & Adoption Risk: When/if Applied lands large hyperscaler customers, those deals will bring concentration risk — e.g. one tenant might occupy an entire 100MW building. If that customer has financial issues or decides to exit at contract end, Applied would need to find a replacement in a niche market. Also, AI infrastructure needs could evolve; today it’s GPUs, but future AI could use different architectures (e.g. TPUs, custom ASICs). Applied must stay flexible to host whatever hardware the market demands. There’s also the scenario that hyperscalers manage to build more of their own capacity than expected, reducing need for external leases. For instance, if the supply chain for data center builds eases, companies like Microsoft might not outsource as much — leaving third parties vying mostly for smaller clients.
  • Financial Leverage and Cost of Capital: With $689M debt outstanding as of (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq)0】, Applied is carrying a heavy debt load relative to its current earnings profile. Interest rates are high (the **Macquarie preferred has a 12.75% yield (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD))9】, and any floating-rate debt is expensive in today’s environment). While the company has a substantial cash cushion now, that will be drawn down by construction and any interim losses. If the timeline to fill capacity stretches, Applied may need to service debt longer from its cash reserves or raise additional funds. Refinancing risk could emerge if credit markets tighten. The Macquarie facility, while large, is not free money — its 1.80x liquidation preference means Macquarie expects a significant return before common shareholders see residual val (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD)) (Applied Digital Agrees to Build a Partnership with Macquarie Asset Management for Funding of up to $5.0 Billion to Drive HPC Growth :: Applied Digital Corporation (APLD))0】. This preferred equity acts somewhat like debt in disguise. Failure to eventually generate enough cash to take it out (redeem after 5+ years) could dilute common holders’ value in the subsidiary. In short, the capital structure is complex, and while it enables growth, it also caps some upside (since 15% of the HPC venture’s equity and a chunk of its cash flows will go to Macquarie).
  • Cloud Services Transition: Until the Cloud segment is sold, it remains part of Applied and carries its own risks. The sequential decline showed the model shift issues, and although that was reportedly fixed, we don’t know how Q4 Cloud revenue will look. If AI cloud competition intensifies, the value of this segment could drop. There’s also risk the sale takes longer or fetches less than anticipated. Notably, **the company’s promise to host Nvidia’s upcoming Blackwell GPUs (Applied Digital closes $375m financing for Ellendale campus in North Dakota — DCD)7】 — which was part of its Cloud strategy — might need to be transferred to whoever buys the Cloud business, or perhaps canceled if not applicable. Applied will want to ensure key customers using its GPU cloud are smoothly handed off to a new owner to avoid burning bridges (some Cloud customers might also be lease prospects or vice versa).
  • Regulatory/Geo Risk: Operating large data centers can sometimes attract regulatory attention — for instance, energy usage, environmental impact, or local opposition (though ND is business-friendly and sparsely populated). One fringe risk: if the crypto mining industry faces new regulations or curbs (e.g. high energy taxes or bans in certain areas), it could indirectly affect Applied’s legacy hosting revenue (though presumably they could re-use that capacity for HPC if miners left). Geopolitical tensions could also affect the supply of GPUs (export restrictions, etc.), which might slow down clients’ ability to utilize the data center even if space is ready.
  • Market Volatility: APLD has been and likely will continue to be a volatile stock. It moves not just on its own news, but on sentiment around AI (and previously, crypto). A downturn in the tech sector or a shift in investor focus could hurt the stock regardless of fundamentals. Conversely, exuberant sentiment can swing it up quickly. This volatility is a risk for investors with shorter horizons or lower risk tolerance.

Catalysts and Potential Upside Drivers

On the flip side, several catalysts could unlock upside in the coming quarters:

  • Successful Cloud Sale: Announcement of a definitive agreement to sell the Cloud Services business (and the price obtained) will be a major event. A price that exceeds the book value of its assets would signal a win. It would also simplify Applied’s story and potentially provide a cash windfall. Investors will watch how those proceeds are used — e.g. debt reduction (improving net leverage) or reinvestment into HPC build-outs. A quick close of this deal would remove uncertainty.
  • Major Lease Deal Announcements: As emphasized, any hyperscaler or large-scale lease signing for Ellendale will likely be the pivotal catalyst. Even a deal for a portion (say 50MW) could be incremental good news, but a full 100MW lease to a blue-chip customer would be transformational. It would not only project strong future revenue (a 100MW lease could be worth tens of millions annually) but also trigger Macquarie’s funding to flow. Essentially, it would derisk the remaining development. Applied has promised more color on customer pipeline as things firm up; these announcements can drop anytime (they aren’t tied to earnings schedules necessarily).
  • Ellendale Build Completion and Ribbon-Cutting: In the second half of 2025, as the 100MW facility comes online, Applied might host events or demonstrations. Showing a live, operational AI data center with cutting-edge hardware could attract positive media and investor attention — especially if they can showcase clients running AI workloads there. It’s one thing to talk about plans; it’s another to see racks of Nvidia GPUs humming in a brand-new facility. This completion will also start the clock for revenue generation from those assets (if leased).
  • Improving Financial Metrics: We could see rapid financial improvement once pieces fall into place. For example, if by late 2025 the Cloud sale is done (removing its costs) and at least part of Ellendale is leased, Applied’s quarterly numbers might start to show higher margins and maybe even net profits by 2026. Any quarter where they surprise to the upside (the opposite of this Q3 miss) could shift sentiment positively. They will also likely present backlog or booking figures once leases are signed, which can help analysts model future revenues more confidently.
  • Industry Developments: Given how tied Applied is to Nvidia and the AI ecosystem, news like Nvidia’s earnings beats, new AI chip releases (e.g. H200, Blackwell), or major AI funding rounds can indirectly boost APLD as a “derivative” play. Additionally, if a competitor were to be acquired at a rich valuation, it could read-through to APLD. For instance, if a large REIT or tech company decided to buy an AI infrastructure startup, it would highlight the value of what Applied is building. Speculatively, even Applied itself could become an M&A target if it demonstrates valuable assets — companies like Digital Realty, Equinix, or infrastructure funds might consider buying once the heavy lifting is done.
  • REIT Conversion or Alternative Structures: Management has floated the REIT idea; achieving that would require meeting certain criteria (income from leasing real property, etc.) and likely a couple years of stable operations. If it were to happen, it could significantly broaden the investor base (as REITs attract income investors) and potentially lower the corporate tax (REITs pass income to shareholders without corporate tax). Even if not a formal REIT, any steps toward a more predictable, yield-generating model (like initiating a dividend once cash flows stabilize) could re-rate the stock.

In essence, there are many “proof points” ahead for Applied. The company has put in place the pieces — capital, partnerships, facilities — to potentially ride the AI wave, but now must execute and deliver tangible results.

Conclusion

Applied Digital is navigating a high-stakes transformation. Once a crypto-focused hoster, it’s reinventing itself as a critical infrastructure provider for the AI era. The Q3 earnings debacle and Cloud Services retrenchment illustrate the growing pains of that pivot. In the short term, the company faces skepticism as it sheds a big revenue driver and asks investors to trust in a longer-term vision of AI data center dominance. However, the building blocks of that vision are falling into place: a massive pipeline of projects funded by deep-pocketed partners (SMBC, Macquarie), strong industry demand signals (GPU shortages, hyperscalers seeking capacity), and even a stamp of approval from Nvidia, the central player in the AI hardware world.

For tech infrastructure investors and AI sector watchers, APLD offers a unique, albeit speculative, exposure. Few public companies its size are so directly leveraged to AI compute growth. This means the stock can be volatile and is not without significant risk — but it also means success could yield outsized rewards. One might compare Applied Digital’s story to a smaller Digital Realty or CyrusOne in their early days, or even to an oil wildcatter — drilling (building) in a region where signs point to a gusher of demand below the surface.

As of now, market sentiment is cautious, essentially taking a “show me” stance. Key things to watch in the coming months include: the fate of the Cloud business sale, any announcements of anchor tenants at Ellendale (or elsewhere), and the company’s ability to manage its balance sheet through this expansion. Investors will also want to see improving profitability metrics once the dust settles on these strategic changes.

In summary, Applied Digital’s latest quarter was a setback that lays the groundwork for a potentially exciting future. The company is betting big on the AI infrastructure boom — and thanks to its partners, it has the resources to place that bet. If you believe the demand for AI computing power will continue to explode, then Applied’s data centers — purpose-built for GPUs and HPC — could become very valuable assets in the next 1–2 years. But until leases are signed and revenue starts flowing, a degree of skepticism is warranted. This is a high-risk, high-reward scenario where execution and timing will make all the difference.

For investors with conviction in the AI mega-trend and patience for a volatile ride, APLD is worth watching (or nibbling on) at these beaten-down levels. Just remember: even data center revolutions don’t happen overnight — but when they do happen, the payoff can be substantial. Applied Digital now needs to prove that its strategic moves in April 2025 set the stage for sustainable, long-term value creation at the forefront of the AI revolution.

Sources: Recent company filings and earnings relea (8k.pdf)2】; management commentary from Q3 2025 call and press relea (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) (8k.pdf)3】; industry news on Macquarie and SMBC dea (Macquarie invests up to $5 billion in Applied Digital’s AI data center business — DCD) (8k.pdf)5】; analysis from CoinDesk, Nasdaq/Zacks, and Benzinga (Needha (Crypto Miner and Data Center Applied Digital (APLD) Tumbles 30% as Earnings Report Disappoints) (Applied Digital’s Q3 Earnings & Revenues Miss Estimates, Stock Down | Nasdaq) (Applied Digital Shares Tank Almost 40% On 3Q, Why This Analyst Remains Bullish — Applied Digital (NASDAQ:APLD) — Benzinga)9】; Data Center Dynamics reports on Ellendale progre (Applied Digital closes $375m financing for Ellendale campus in North Dakota — DCD) (Applied Digital closes $375m financing for Ellendale campus in North Dakota — DCD)4】; The Register on Nvidia’s investme (Nvidia and co inject $160M into Applied Digital’s GPU cloud • The Register)2】; Yahoo Finance and others on market reacti (Why Applied Digital (APLD) Shares Are Sliding Today)7】. All financial figures are as reported for fiscal Q3 2025 or from cited sources.

Financial Advice Disclaimer: This article is provided for informational purposes only and is not intended as financial advice. The content reflects the personal opinions of the author and should not be construed as specific investment advice. Investors are advised to conduct their own research or consult with a qualified financial advisor before making any investment decisions. The author and publisher are not liable for any financial losses or damages resulting from decisions made based on this article. Investments in the stock market involve risk, including the potential loss of principal.

Disclosure: The author of this article holds shares in the company mentioned in this article. This disclosure is to inform readers of potential biases in the opinions expressed herein.

Leave a Reply

Your email address will not be published. Required fields are marked *