Applied Digital: A Buy Despite Risks and Uncertainties.

Applied Digital Corporation (NASDAQ:APLD) has seen significant gains since its tariff sell-off lows, with investors now sitting on around 200% gains. The company's focus on selective growth narratives has been rewarded despite multiple pressures and uncertainties.

Applied Digital Corporation (NASDAQ:APLD) has seen significant gains since its tariff sell-off lows, with investors now sitting on around 200% gains. The company's focus on selective growth narratives has been rewarded despite multiple pressures and uncertainties. APLD is a digital infrastructure provider for high-performance computing applications, offering physical hyperscale data centers tailored for workloads that demand high computing power like AI, ML, and blockchain. It also provides enterprises access to NVIDIA (NVDA) H100 GPU clusters through its AI cloud services. Initially heavily reliant on crypto-based revenues, APLD has now moved into more recurring revenue from enterprise AI clients, leading to its re-rating in the market.

The recent surge in APLD share prices was largely due to material long-term leases secured from CoreWeave. Two 15-year leases from the AI hyperscaler for APLD's Ellendale facility (currently 250MW and expandable to 400MW) and $7 billion expected revenue is not only a boost for APLD's AI revenue stream but also validates its move away from legacy crypto businesses towards its AI-focused positioning. This has helped APLD surge in an environment of massive data center investing from big tech firms—a strong tailwind for AI infra stocks. APLD has made significant progress in its infrastructure cost advantage, too. A recently published whitepaper shows how its North Dakota Polaris Forge campus is leveraging cheap power and free cooling to deliver industry-leading efficiencies. A potential $2.7 billion cost savings reinforces its "AI Factory" credentials.

APLD's AI compute segment (Cloud Services) saw a topline of ~$17.8 million in Q3 FY 2025. While that is ~220% up YoY, it has dropped by ~$10 million sequentially. This sequential drop is attributed to technical issues when shifting from a single tenant reserved GPU structure to a multi-tenant on-demand model. The execution is still clearly early stage, and while the demand surge is verified from the continued YoY topline growth, teething issues will exist till operational fine-tuning matures over live demand cycles. I am not looking at numbers so much in the early phase deployments, as monetization will expectedly be in flux. Broader trends like new signings and operational developments through setbacks are enough to support APLD's AI tailwind narrative. APLD has already shown actual revenue, real contracts, and a tangible buildout compared to some other AI plays built on narratives and potential only at this point.

APLD posted a GAAP net loss of $36.1 million, significantly lower from $62.8 million a year before. Adjusted net loss narrowed to $17.8 million and adjusted EBITDA turned positive at $10 million. Costs rose only modestly despite increased deployment activity, signaling improving operational efficiency. The profitability trajectory is encouraging and points to efficiencies and scale payoffs, but that is only a supporting part to APLD's thesis, which is more about the potential and advancing proof of concept. APLD has $261 million in cash and $689 million of debt. It secured $375 million funding from SMBC, using part of that to repay a prior Macquarie note. More importantly, it has a $5 billion commitment from Macquarie to support its HPC buildout. Although APLD's model is capex heavy, I think institutional backing is strong and markets should be now looking away from funding risk and focus on execution and lease conversions.

APLD's proof of concept is very encouraging, and the long-term tailwinds are strong. There is even real AI monetization in there—gold dust in today's AI hype narratives. The recent CoreWeave contract has even added contractual visibility—many GPU rental and cloud native startups lack that level of progress and risk off the table. The recent rally is partly attributable to an increase in forward revenue estimates, but it is also a huge rerating due to the strengthening narrative. In comparison with CORZ, the valuations look benchmarked. CORZ itself has been pivoting from legacy crypto infra offerings into AI businesses. Both CORZ and APLD therefore reflect a common market scarcity premium on players with tangible infra and early revenues, in addition to the AI tailwinds. That could justify the apparently rich valuations even after a marked expansion.

APLD's reliance on CoreWeave is not free from risks. CoreWeave is not a mature hyperscaler like Google or Microsoft. It is going through its own cycle of capital requirements and AI demand. APLD's concentration on CoreWeave therefore translates to a double bet investors are taking on both firms when buying APLD today. CoreWeave should be able to scale sustainably and APLD should be able to execute flawlessly. In comparison, CORZ has a longer history, is closer to profitability, and less concentrated in its demand pipeline. So, at similar valuations, CORZ appears to be offering a cleaner balance sheet trajectory and lower demand risk profile.

APLD's AI revenues are already significant as a proportion of its overall revenues—showing focused AI traction compared to CORZ (not numbers but AI focus as % of revenue). APLD's infra is also looking more purpose-built for HPC and AI workloads, with liquid cooling and energy optimization—data center design around GPUs. CORZ's approach is more about using its legacy infra to support AI workload. That could have limitations in a scaled demand scenario. Is APLD Ready to Invest? Overall, I believe despite higher valuations now, APLD's thesis has a stronger AI monetization story relative to a close business model like CORZ's. As highlighted earlier, CoreWeave's own capital cycle presents a risk to the long-term contracts. Recently, CoreWeave exited its equity stake in APLD. However, I believe that is more about financial management than strategic disengagement, as CoreWeave was granted warrants to buy an equivalent number of shares at a premium ($7.19). However, it goes to show that the CoreWeave contract is only to be used as a proxy for validation of APLD's capability. The concentration risk is high for it to be counted as a basis for top-line growth potential. I approach APLD as still a conviction play backed by proof of concept validation, not numbers alone. Valuations look overheated, but I believe in APLD's fundamental narrative. I think that even in today's markets, there is scope for growth bets with proof of concept (and revenue). Investors who are comfortable riding short-term volatility can consider small starter positions in APLD even today (down ~25% from June highs) and play longer term. There could be further short-term corrections after a larger rally of this magnitude, and that could be used to accumulate more. Waiting for perfection may not work, so APLD is a Buy for Execution at today's prices.

References:
[1] https://seekingalpha.com/article/4799783-applied-digital-stock-real-ai-revenues-real-risks-but-still-buy
[2] https://www.dmagazine.com/business-economy/2025/07/conversation-with-applied-digital-ceo-wes-cummins-on-doing-business-with-nvidia/