Revenue Analysis

The Quarter in Context

For months, the APLD thesis has been built on contracts and commitments. Billions in prospective revenue, megawatts under agreement, buildings under construction. All of it pointed in the right direction, but it was still forward-looking. The Q2 FY26 10-Q changes that. Now there's actual lease revenue hitting the income statement, and we can reverse-engineer what the economics actually look like.

Start with the timeline. Fiscal Q2 ran September 1 through November 30, 2025. Ninety-one days. The company announced Phase I Ready for Service on October 27, meaning the first 50 MW came online with thirty-five days remaining in the quarter. Phase II followed on November 24, bringing the second 50 MW online with just seven days left. So when management reported $12 million in HPC lease revenue for the quarter, that wasn't from 100 MW running for three months. It was from a fractional deployment. Most of the capacity online for barely a month, and half of it for barely a week.

Capacity Utilization

The math gets interesting when you quantify exactly how fractional. Multiply 50 MW by thirty-five days and you get 1,750 MW-days from Phase I. Multiply 50 MW by seven days and you get 350 MW-days from Phase II. Combined, that's 2,100 MW-days of capacity delivered during the quarter. Now compare that to what a full quarter would look like: 100 MW multiplied by ninety-one days equals 9,100 MW-days. The quarter operated at roughly 23% of its theoretical full-quarter capacity. A little less than a quarter of a quarter, if you want to think about it that way.

Implied Run Rate

This reframes the $12 million entirely. That revenue wasn't generated by a stabilized asset running at full capacity. It was generated by an asset that spent most of the quarter either under construction or ramping. If 23% of capacity produced $12 million, then 100% of capacity for a full quarter implies somewhere around $52 million. Annualize that and you get approximately $208 million per year from 100 MW, which works out to roughly $2.08 million per megawatt annually.

Contract Comparison

Here's where it gets interesting. The simple contract math tells a different story. CoreWeave signed for 400 MW at Polaris Forge 1 with approximately $11 billion in total lease revenue over fifteen years. Divide that out and you get about $1.83 million per MW per year. The second hyperscaler at Polaris Forge 2 signed for 200 MW at approximately $5 billion over fifteen years, implying roughly $1.67 million per MW annually. But the actual realized rate from Q2 comes in at $2.08 million, meaningfully higher than either contract would suggest on a straight-line basis.

A few mechanisms could explain the gap. Infrastructure leases often front-load payments, with higher rates in early years that taper over time to account for asset depreciation and inflation dynamics. There may also be built-in escalators, annual increases of two or three percent that compound over a fifteen-year term, which would push the average rate higher than the simple division implies. And there's likely a pass-through component for power and utilities that sits on top of the base lease rate. Whatever the mix, the important observation is that realized economics are running at or above the headline contract math, not below it. That's the direction you want the surprise to go.

Forward Projections

The implications for the next several quarters become fairly mechanical from here. Q3 FY26, December through February, will be the first full quarter with 100 MW online. If the $2.08 million run rate holds, that should produce somewhere around $52 million in HPC lease revenue, roughly four times what Q2 delivered.

Building 2, the 150 MW facility at Polaris Forge 1, is expected to come online mid-2026. That would start contributing partial-quarter revenue in Q4 FY26 or Q1 FY27 depending on exact timing. Building 3, another 150 MW, follows in 2027. Meanwhile, Polaris Forge 2, the 200 MW campus for the second hyperscaler, is targeting initial capacity in 2026 with full delivery by early 2027.

When you stack all of that together, the trajectory becomes clear. By the time the company reaches full delivery across both campuses, 600 MW total, the quarterly HPC lease revenue run rate should approach $310 million using the $2.08 million per MW figure. That's roughly $1.25 billion annually from HPC leases, which is the business that matters for the thesis.

Margin Validation

The margin structure matters too. Management has consistently pointed to 85-90% NOI margins for the HPC colocation model, and the Q2 numbers suggest that guidance is credible. Once construction activity winds down and the buildings are simply collecting lease payments, the operating leverage should be substantial.

Consider what the 10-Q actually shows. The $12 million in lease revenue was recognized alongside operating costs that were dominated by fit-out pass-throughs, not recurring facility expenses. Strip away the one-time construction revenue and its associated costs, and the underlying lease economics look clean. Power in North Dakota runs cheap. The facilities use proprietary waterless cooling that eliminates major operating expense lines. The buildings are new, so maintenance capex stays minimal for years. When you remove the noise of the construction phase, the path to 85%+ NOI margins becomes visible in the structure of the business itself.

Valuation Context

If APLD hits 85% NOI on $1.25 billion in stabilized HPC lease revenue, that's over $1 billion in annual NOI from the HPC segment alone. Against a current enterprise value in the $10-11 billion range, that implies the market is pricing the company at roughly 10x forward NOI on contracted, investment-grade-backed revenue. For context, stabilized data center REITs typically trade at 15-25x NOI. The gap has narrowed since initial publication but still implies significant upside if the contracts convert as expected. The Q2 10-Q makes that math harder to ignore.

None of this requires heroic assumptions. The contracts are signed. The counterparties are investment-grade. The buildings are under construction with financing in place. The first one just delivered on time and started generating revenue exactly as expected. What remains is execution. More buildings, more megawatts, more quarters of lease revenue hitting the income statement. Each quarter that passes converts another piece of the thesis from projection to proof.

The 10-Q gave us the first real data point. The math works. The NOI guidance checks out. Now we watch it compound.