Applied Digital lands a $5B lease deal with hyperscaler at its planned North Dakota campus


Applied Digital (APLD) said this week that it had entered into a lease agreement with a US-based “investment grade” hyperscaler for its Polaris Forge 2 campus in North Dakota.

The deal is worth about $5 billion in total contracted revenue over an estimated 15-year lease term. It will cover 200 megawatts (MW) of critical IT load that will be purpose-built to support the hyperscaler’s AI and high-performance compute (HPC) infrastructure.

Applied did not name the hyperscaler.

ADVERTISEMENT

The company broke ground last month on Polaris Forge 2, a $3 billion, 280-megawatt (MW) “AI Factory” near Harwood, ND.

The facility is designed to expand beyond its initial 280MW, with the first stage set to be operational in 2026 and full capacity expected by 2027.

Polaris Forge 2 is Applied’s second major site in North Dakota, joining its Polaris Forge 1 campus in Ellendale.

CEO Wes Cummins said in August when announcing the groundbreaking on Polaris Forge 2 that the company was in “advanced negotiations with a US-based investment-grade hyperscaler” on a lease deal for the new facility.

Applied noted that it has now leased a total of 600 MW of capacity with “two of the world’s largest hyperscalers” at its two Polaris Forge facilities.

The company in June signed two massive 15-year lease agreements with AI hyperscaler CoreWeave (CRWV).

Under the deal, Applied will deliver 250 MW of critical IT load to power CoreWeave’s AI and high-performance computing (HPC) infrastructure, all housed at the Ellendale campus.

The deal is expected to generate roughly $7 billion in total revenue over its lifetime.

ADVERTISEMENT

The company said that the latest agreement is “reinforcing the company’s position as one of the fastest-scaling builders of AI infrastructure in the United States.”

Polaris Forge 2 is spread out across 900 acres in North Dakota.

“What sets us apart isn’t just the size of our pipeline - it’s how fast we can deliver," Cummins said in a statement. “The real constraint in this industry is execution, and our team continues to prove that large-scale, next-generation data centers can be designed, financed, and brought online faster and more efficiently than anyone thought possible.”

The question of “execution” has become more pronounced as the money pouring into the construction of AI data centers continues to grow.

Praetorian Capital CIO Harris Kupperman recently wrote about what he called “an insanity bubble,” projecting that AI data centers built this year will suffer $40 billion of annual depreciation, while generating between $15 and $20 billion in yearly revenue.

“Simply put, at the current trajectory, we’re going to hit a wall, and soon,” Kupperman wrote. “There just isn’t enough revenue and there never can be enough revenue. The world just doesn’t have the ability to pay for this much AI.”

Applied’s stock gained 9% on Thursday and has surged 336.9% for the year.

ADVERTISEMENT

ADVERTISEMENT