Applied Digital (APLD 3.75%) is in an enviable position among companies in the artificial intelligence space. The data center developer has $16 billion in contracted lease revenue.
To the company's bulls, that number is key to its investment thesis. It suggests years of locked-in cash flow and a level of security that helps investors sleep at night.
But dig a little deeper, and the cracks emerge. There is a major weakness at its heart: counterparty risk -- that is, the money is only "guaranteed" if the company on the other side of the deal can pay the rent.
But first, the leases themselves may not be quite as ironclad as many investors believe.
Image source: Getty Images.
There's an exit for Applied Digital's future tenants
These contracts come with termination provisions that shouldn't be taken for granted. If Applied Digital falls far enough behind its construction targets, its lessees can walk away without incurring any kind of penalties, and with no requirement of partial payments.
Now, Applied Digital has stayed on schedule so far. The company just completed the second phase of its Polaris 1 campus, which now boasts 100 megawatts (MW) of capacity ready for service. That's significant.
But the company has another 600 MW that it will need to deliver without delays. Supply chain disruptions, permitting delays, equipment shortages, labor constraints -- any of these could make completing the remaining capacity a real challenge.
Of course, it's more than possible -- likely even -- that its tenants wouldn't choose the nuclear option if Applied Digital does miss its targets. But the threat of cancellation gives them enormous leverage to renegotiate terms. If construction slips, the neocloud company could find itself agreeing to lower lease rates or more favorable terms just to keep them from exercising their exit rights.
A Russian nesting doll of leverage
The contract risk is a concern, but the real weakness, in my opinion, lies in layers of leverage and unprofitable AI businesses.
Let me explain. Applied Digital's growth is fueled almost entirely by debt. In November 2024, the company carried roughly $468 million in debt. As of the end of its last quarter, that figure had ballooned to $2.6 billion -- more than a fivefold increase in just one year. This is happening all on the basis of future earnings; the company lost $125 million over the trailing 12 months.
NASDAQ: APLD
Key Data Points
This alone wouldn't be too much of a concern if it weren't for the fact that a large fraction of those future revenues -- $11 billion out of $16 billion -- are slated to come from CoreWeave, which is itself a highly leveraged, currently unprofitable company.
CoreWeave's debt surged from $5.5 billion in December 2024 to $10.6 billion by November 2025, while posting a net loss of $766 million over the last 12 months. This is the counterparty that Applied Digital's entire growth thesis hinges on.
CoreWeave, in turn, is heavily reliant on OpenAI -- the creator of ChatGPT -- as its primary customer. It's still privately held, so OpenAI's financials aren't public, but the company itself carries little debt -- key investors carry a significant amount -- and it, too, is burning through cash. But it's doing so at a pace and on a scale that simply has not been seen before.
From 2026 through 2030, OpenAI expects to lose $211 billion. Uber famously operated at a loss for years and, until now, held the title for the company with the biggest cash burn until it started to turn a profit. It lost just $18.2 billion from 2016 to 2022.
What you end up with is a Russian nesting doll of counterparty risk. Applied Digital depends on CoreWeave, which depends on OpenAI, which depends on the continued willingness of investors to fund massive AI spending before the revenue materializes.
If any layer of this structure cracks -- if OpenAI's fundraising stalls, if CoreWeave can't service its debt -- the consequences will flow downhill directly to Applied Digital's shareholders.
The bottom line
Applied Digital's $16 billion pipeline is real, but it's not quite guaranteed in the ways that investors hope. Am I being too cautious or too pessimistic about the future of AI? Possibly.
But for my money, there are layers of risk here that make me uncomfortable.
