Applied Digital adds 75 MW of AI capacity in North Dakota

Applied Digital brought an additional 75 MW of AI infrastructure online at its Polaris Forge 1 campus in Ellendale, North Dakota. The expansion raises operational critical IT load to 250 MW at the fully leased facility, advancing the campus toward its full contracted capacity and supporting the company’s AI data-center growth strategy.

Analysis

The key investable question is not incremental MW commissioned but whether APLD can convert contracted capacity into durable, investment-grade cash flow without further equity dilution. At this stage, execution reduces construction and customer-onboarding risk, but it also shifts investor focus toward lease economics: rent escalators, tenant credit quality, power pass-throughs, uptime guarantees, and the gap between project-level EBITDA and corporate interest expense. APLD remains structurally more sensitive to financing costs and a single-campus operational disruption than scaled data-center peers such as CoreWeave (CRWV), Equinix (EQIX), or Digital Realty (DLR).

Near term, the announcement can support momentum as the market extrapolates remaining capacity activation, but a 1-3 month rerating requires independently verifiable evidence of revenue recognition and stable gross margins in reported results. The more consequential 6-18 month effect is that successful delivery strengthens APLD's credibility in the scarce power-and-land pipeline for AI compute; that could lower its cost of capital and improve negotiations for future pre-leases. Conversely, a delay in customer acceptance, power-curtailment event, or higher-than-expected financing need would likely compress the equity multiple disproportionately because the valuation rests on future contracted EBITDA rather than mature FCF.

Consensus may overvalue the headline MW figure: capacity is only scarce if it is deliverable with reliable power, network connectivity, and economically viable GPU tenant demand. North Dakota's power economics are an advantage, but geographic concentration creates weather, transmission, and tenant-concentration exposure. Watch whether APLD discloses lease duration, contracted annual revenue per MW, and customer credit support; absent those data, the operational milestone alone is insufficient to underwrite a new fundamental long.

Key Decisions for Investors

• Maintain APLD as a tactical long only into the next earnings/update window if price action remains constructive; target a 10-15% upside from confirmation of revenue ramp and no incremental equity issuance, with a 7-10% stop or exit on evidence that commissioning has not translated into recognized hosting revenue.
• Do not add a core APLD position until management provides contracted revenue/MW, lease tenor, customer concentration, and project-level EBITDA-to-corporate-FCF bridge. Treat any new capital raise or material increase in net-debt guidance as thesis falsification.
• For lower-volatility AI-infrastructure exposure over 6-18 months, pair a modest long APLD position with larger exposure to DLR or EQIX; this retains upside to AI capacity scarcity while reducing single-site execution and refinancing risk.
• Set alerts around the next quarterly filing for utilization, realized revenue per operational MW, adjusted EBITDA margin, interest expense, and capex guidance. A revenue ramp materially below the implied pace of newly operational capacity would warrant closing the tactical long rather than averaging down.