Applied Digital just landed a massive $5 billion AI infrastructure lease with an unnamed U.S. hyperscaler, marking one of the largest data center deals of the year. But investors aren't celebrating - the stock tumbled over 7% on the news, extending a brutal week that's seen shares drop more than 20%. The deal highlights the complex dynamics in AI infrastructure investing, where even billion-dollar contracts can't guarantee immediate market rewards.
The AI infrastructure gold rush just got more complicated. Applied Digital's announcement of a $5 billion lease agreement with an unnamed U.S. hyperscaler should have been cause for celebration, but the market had other plans. Shares plunged more than 7% in Wednesday trading, adding to a punishing week that's wiped out over 20% of the company's value.
The deal itself is massive by any measure. The 15-year agreement will deliver 200 megawatts of capacity at Applied Digital's Polaris Forge 2 campus in North Dakota, bringing the company's total leased capacity to 600 megawatts across its two facilities. For context, that's enough power to run a small city, all dedicated to AI workloads.
"We started down this path a couple years ago and we stubbed our toe a few times, but I think we've really dialed in the process of the ability to build at scale," CEO Wes Cummins told CNBC's Squawk on the Street. The admission reveals just how challenging it's been for smaller players to crack into the hyperscaler ecosystem.
While Applied Digital won't name its new tenant, the company did reveal it's an "investment grade hyperscaler." Cummins narrowed down the field during his CNBC interview, identifying the five major U.S. hyperscalers as Microsoft, Meta, Oracle, Amazon, and Google. He also disclosed that CoreWeave was the tenant for their first major lease agreement.
The timing couldn't be more critical for AI infrastructure. Major cloud providers and tech giants are racing to secure data center capacity as AI workloads explode. But the market's reaction to Applied Digital's announcement suggests investors are growing more cautious about the sector's frothy valuations, even as deals reach record sizes.
This latest agreement builds on Applied Digital's momentum from earlier this year. In June, the company announced two long-term lease agreements with CoreWeave for 250 megawatts of capacity, expecting $7 billion in rental revenue over 15 years. That news sent shares soaring 48%, but Wednesday's muted response shows how quickly sentiment can shift.
The company also recently secured $5 billion in infrastructure funding from Macquarie Asset Management, giving it the capital firepower to build out its ambitious expansion plans. Cummins mentioned the company has a 4-gigawatt "active pipeline" - a massive scale that would make Applied Digital a major player in AI infrastructure.
Despite Wednesday's sell-off, Applied Digital shares have still nearly quadrupled this year, reflecting the broader AI infrastructure boom. The stock's volatile reaction to what should be unambiguously good news highlights the challenges facing investors trying to value companies in this rapidly evolving market.
"We believe Polaris Forge 2 builds on that momentum, reflecting the strength of our partnerships and the speed at which we're reshaping the AI infrastructure landscape," Cummins said in the company's announcement. But reshaping entire industries, as Applied Digital is discovering, doesn't always translate to immediate stock gains.
The Applied Digital story captures the strange economics of AI infrastructure investing, where billion-dollar deals can still send stocks tumbling. While the company has clearly positioned itself as a serious player in the AI data center game with over $10 billion in recent agreements, investors seem increasingly focused on execution risk and market saturation. The unnamed hyperscaler partnership adds another major validation to Applied Digital's approach, but the market's reaction suggests that in today's environment, even massive deals need to prove their profitability before investors celebrate.
