Applied Digital: A Gigawatt-Scale Bet on the AI Infrastructure S-Curve

The investment case for Applied Digital hinges on a single, undeniable fact: the exponential growth of AI is creating a fundamental shift in infrastructure. We are moving from a world of individual data centers to one defined by gigawatt-scale campuses. This isn't just a scaling up; it's a paradigm shift in how compute power is delivered, and it creates a new kind of moat-one built on sheer scale and execution discipline.

The numbers paint the picture. By late 2027, U.S. AI data centers alone will need between 20 to 30 gigawatts (GW) of combined power. That's a colossal demand, equivalent to 5% of the nation's current power generation. To meet this, the industry is consolidating into megascale campuses. As developers and industry groups now describe it, we are entering an era of gigawatt-scale data center campuses, where a single site can support 1 gigawatt-equal to 1,000 megawatts or 1 billion watts. This move is driven by the need for renewable energy integration, grid responsiveness, and the sheer physical footprint required for next-generation AI models.

Applied Digital's Delta Forge 1 campus is a disciplined, repeatable step into this new S-curve. It is designed from the ground up to support an initial 430 MW of total utility power, enabling up to 300 MW of critical IT load. The first phase will be two 150-MW buildings on over 500 acres. The company's plan is clear: build a foundation that can scale considerably, with initial operations expected in mid-2027. This is not a speculative bet on a single facility. It is a blueprint for deploying the fundamental rails of the AI infrastructure layer, one campus at a time.

The bottom line is that the race is now for scale. The largest upcoming facilities, like Meta Hyperion and Microsoft Fairwater, are projected to reach capacities of five million H100-equivalents each by late 2027. To support that kind of compute, you need power in the gigawatt range. Applied Digital is positioning itself not as a builder of isolated data centers, but as a developer of the next-generation campuses that will house the AI factories of the future. Its strategy is to be a reliable, scalable partner in that build-out, executing a repeatable model as the demand curve accelerates.

Scale Comparison: The Power Race to Gigawatts

The race to power the AI paradigm is a race for scale, and the strategies are diverging. While some players are building massive portfolios, others are making single, colossal bets. Applied Digital's path is one of the most aggressive, but it must be measured against the established giants and the emerging multi-provider models.

Digital Realty Trust sets the benchmark for scale and financial heft. The company has more than 5,000 customers globally, spanning more than 300 data centers and operates with a market cap near $55 billion. Its strategic focus is on securing power through partnerships, not just owning it. This allows it to leverage its vast real estate footprint to deliver capacity where needed, a model that has served it well for over two decades. Its scale is undeniable, but its approach is one of integration and risk mitigation through collaboration.

Vantage Data Centers is taking a different tack, showcasing a multi-provider, flexible strategy. Its partnership with Liberty Energy is designed to deliver up to 1 GW of power capacity over the next five years. This isn't a single plant; it's a planned, phased build-out that spreads risk and ensures a steady pipeline of power. The model is built for resilience in power-constrained markets, using on-site generation to supplement the grid. This approach offers agility and cost control, but it requires deep coordination and a long-term commitment to multiple partners.

Applied Digital's strategy is a stark contrast: a massive, single-source bet. The company has secured a $1.5 billion contract for a 1 GW power plant with Babcock & Wilcox, with operations targeted for 2028. This is a direct, capital-intensive play on a single technology and partner. The stated advantage is speed-to-market, with the steam-generation design promising faster deployment than traditional plants. It represents a vertical integration of power and data center development, aiming for a distinct advantage in bringing capacity online.

The bottom line is that Applied Digital is betting on a single, monumental execution. It is not building a portfolio like Digital Realty, nor is it spreading its power risk like Vantage. Its entire Delta Forge 1 campus is built around this one power source. The success of its S-curve bet hinges entirely on the flawless delivery of this $1.5 billion plant. For now, it is the most aggressive scale play in the field, but it is also the most exposed.

Execution on the Exponential Curve: Power, Partnerships, and Timeline

Applied Digital's growth narrative is a single, massive execution play. The company's entire S-curve bet depends on the successful and timely commissioning of a 1 GW power plant by 2028. This is not a portfolio of smaller, de-risked projects; it is a monumental, single-source dependency that introduces significant operational and timing risk.

The core vulnerability is its reliance on one partner: Babcock & Wilcox. The company has secured a $1.5 billion contract with the power giant for the steam-generation plant. While B&W brings deep expertise and a proven track record, the partnership concentrates all execution risk in one vendor. Any delay, cost overrun, or technical snag in this single $1.5 billion project directly threatens the timeline for the Delta Forge 1 campus. The company's speed-to-market advantage, a key part of its pitch, is entirely contingent on B&W delivering on its targeted 2028 operation date.

This contrasts sharply with the multi-provider strategies emerging among competitors. Vantage Data Centers, for instance, is pursuing a partnership with Liberty Energy to deliver up to 1 GW of power capacity over the next five years. More importantly, it has secured a reserved 400MW of generation capacity scheduled for 2027. This approach spreads risk across multiple providers and timelines, allowing for a more rapid and flexible build-out of power capacity. Applied Digital's model, by contrast, is a binary outcome: the 1 GW plant either comes online on schedule, or the entire growth trajectory for its flagship campus is delayed.

The bottom line is that Applied Digital is betting its exponential growth on a single, high-stakes execution. The key catalyst is the successful and timely commissioning of the 1 GW plant by 2028. For now, its strategy is the most aggressive in the field, but it is also the most exposed. The company must navigate the complex logistics of a gigawatt-scale power project with a single partner, while competitors are securing capacity through diversified, multi-year plans. The path to the next paradigm is paved with power, but Applied Digital is laying its entire foundation on one, very long, single track.

The Infrastructure Layer Play: Dual Revenue Streams and Long-Term Levers

Applied Digital's financial model is built for the long haul, operating on a dual-revenue engine that positions it squarely as a pick-and-shovel provider for the AI paradigm. The company has two distinct segments: next-generation datacenter colocation services and its own AI GPU cloud platform, Sai Computing. This creates a hybrid model. The colocation arm provides a steady, recurring revenue stream from hyperscalers leasing space and power for their own equipment. Simultaneously, Sai Computing allows Applied Digital to capture upside from the AI compute demand it is helping to enable, selling GPU hours directly. This dual stream is the essence of the infrastructure play-building the rails while also running the first trains.

The strategy is a classic capital-intensive, high-margin bet on exponential adoption. By focusing on the fundamental layers of the stack-power, cooling, and physical space-Applied Digital aims to be the essential vendor for the AI factories of the future. This is the "pick-and-shovel" model in action. As BlackRock's CEO noted, investing in AI means investing in HVAC and IT, investing in power grids and power supplies. Applied Digital is building those grids and supplying those shovels. The model promises high margins because it captures value at the infrastructure layer, away from the volatile chip pricing wars. However, it demands massive upfront capital, as seen in its $1.5 billion contract for a 1 GW power plant.

Valuing this company requires looking past short-term earnings. The returns are tied directly to the long-term AI adoption curve. The company's initial Delta Forge 1 campus, designed for 430 MW of total utility power, is a foundational asset. Its value accrues over decades as it scales and as AI workloads grow. The $1.5 billion power plant contract is not an expense; it is a capital investment to secure the fuel for that growth. The financial model is one of deferred gratification: heavy spending now to secure a dominant position in the gigawatt-scale infrastructure layer, with profits materializing as the AI S-curve accelerates and customers fill the colocation space and use the Sai Computing cloud.

The bottom line is that Applied Digital is trading near-term profit for long-term control of a critical resource. Its dual revenue streams provide a path to profitability, but the valuation must account for the significant capital expenditure required to build and power these campuses. The payoff is a high-margin, essential role in the AI supply chain, but it is a bet that the exponential adoption of AI will continue to justify the massive upfront investment. For investors, this is a play on the infrastructure layer, where the returns are tied to the slope of the adoption curve, not the quarterly earnings report.

Catalysts and Watchpoints: Validating the Exponential Thesis

The investment thesis for Applied Digital rests on a single, monumental execution. The path from its current position to a gigawatt-scale infrastructure leader is paved with specific milestones that will either validate its exponential bet or expose its vulnerabilities. Investors must watch three key catalysts.

First, the company needs to demonstrate demand for its foundational asset. The successful commissioning of the 1 GW power plant is the ultimate operational milestone, but near-term validation comes from securing hyperscaler leases. Applied Digital has already announced a $5 billion AI factory lease with a U.S. investment-grade hyperscaler, a powerful signal. The next watchpoint is for additional announcements of new leases for the Delta Forge 1 campus. These would provide concrete revenue visibility and confirm that the market sees value in its purpose-built, power-optimized infrastructure. Without these commitments, the thesis of exponential adoption remains unproven.

Second, the broader trend of power partnerships in the sector is a critical competitive signal. Applied Digital's strategy relies on a single, massive contract with Babcock & Wilcox. This contrasts with the multi-provider, flexible approach gaining traction elsewhere. For instance, Vantage Data Centers has partnered with Liberty Energy to deliver up to 1 GW of power capacity over the next five years, including a reserved 400MW of generation scheduled for 2027. If other developers secure capacity faster through diversified partnerships, it could highlight a competitive disadvantage for Applied Digital's single-source model. The watchpoint here is the pace and scale of new power deals announced by peers; a trend toward multi-year, multi-partner plans could pressure Applied Digital's speed-to-market narrative.

Finally, the primary operational milestone remains the successful commissioning of the 1 GW power plant by 2028. This is the non-negotiable foundation for the entire Delta Forge 1 campus. The company has secured a $1.5 billion contract with Babcock & Wilcox, with operations targeted for 2028. Any delay or cost overrun in this project would directly threaten the timeline for the campus's initial operations, expected in mid-2027. This is the binary event that will determine the thesis. For now, the company is on track to release the full contract in the first quarter of 2026, a near-term checkpoint that will provide more detail on the project's scope and schedule.

The bottom line is that Applied Digital's exponential growth is a high-stakes, binary play. The watchpoints are clear: new hyperscaler leases to prove demand, the pace of competitor power deals to gauge its strategic position, and the flawless execution of its single, massive power plant contract. Success on all fronts would validate its bet on the AI infrastructure S-curve. A stumble on any one could derail the entire build-out.