APLD Stock Price Surge Masks a $100 Million Loss — Here’s What Investors Are Actually Betting On

There are 180 megawatts of processing power operating nonstop in a town in North Dakota called Ellendale, which has a population of just 4,000. Nothing that you can hold is produced at the facility. When most people ask a chatbot a question, they never consider the massive, power-hungry computations that are used to train artificial intelligence models. That facility was constructed by Applied Digital Corporation. Additionally, the company’s stock increased by over 10% in a single session on April 9, 2026, briefly reaching $29.39 before settling at $27.79 on the NASDAQ. The day was significant. However, even seasoned investors are forced to reconsider due to the intricacy of the numbers involved.

Applied Digital’s fiscal third-quarter earnings headline figure was truly startling. At $126.64 million, revenue exceeded Wall Street’s expectations by about 68% and increased by 139% from the same quarter last year. It’s not a humble beat. That business is expanding at a rate that indicates actual operational progress is being made. The majority of that increase was driven by the Polaris Forge 1 facility, a 100 megawatt liquid-cooled data center that became fully operational during the quarter and contributed about $71 million in HPC hosting revenue. $44.1 million was the adjusted EBITDA. In one version of the story, everything appears to be going well.

NASDAQ: APLD · AI Infrastructure & HPC Hosting

The net loss comes next. For the quarter, Applied Digital reported a net loss of $100.86 million, which was almost three times greater than the $36.09 million loss for the same period last year. The headline EPS figure, which came in at negative $0.36 against analyst expectations closer to negative $0.10, looked especially bad due to a $59.7 million non-cash impairment charge associated with the Cloud Services segment, which was reclassified for a possible spin-off. That represents a 251% miss. The stock fell more than 5% during extended trading after the market closed as investors took in the real message of the bottom line. Together, the two reactions—the after-hours retreat and the daytime surge—capture the unique tension that currently characterizes Applied Digital.

It’s difficult to ignore what Applied Digital is genuinely trying to accomplish here. It’s not the most obvious origin story for a company that now signs 15-year leases with investment-grade hyperscalers for 200 megawatt AI data centers, but the company began with significant exposure to cryptocurrency mining. Thus far, the pivot has been aggressive and in the right direction. 400 megawatt leases at Polaris Forge 1 were signed by CoreWeave, a GPU cloud provider that has grown to be one of the most well-known brands in AI infrastructure. For Polaris Forge 2, a second unidentified hyperscaler agreed to a long-term contract. Delta Forge 1, a 300 megawatt AI factory campus in the southern United States that is scheduled to start operations in the middle of 2027, has already begun construction. The company is expanding at a rate that suggests a sincere belief in long-term demand, or at the very least, a readiness to place significant bets on it.

It’s important to comprehend the financial framework that underpins all of this. In order to give the company ownership of power generation assets without bearing all of the risk on its balance sheet, Applied Digital established Base Electron as an independent power producer. The goal of the planned merger with EKSO Bionics to create ChronoScale Corporation, a specialized accelerated-compute platform, is to clearly divide the Cloud Services division from the main data center operations. These structural choices are significant. They suggest management is thinking carefully about how to present the business to different categories of investors, and possibly how to unlock value that the current single-entity structure doesn’t fully capture. It’s still genuinely unclear if those actions have the desired effect.

At least on days like April 9, investors appear to think that the revenue trajectory is the most important factor at the moment and that the losses are the inevitable expense of rapidly developing physical infrastructure. In the early years of Amazon Web Services, the market observed AWS consuming capital at frightening rates before the economics finally turned around. This historical parallel is worth taking into account. There are clear limitations to the comparison, and Applied Digital is far from that scale. However, the fundamental reasoning behind this is the same as what the bulls are saying: early infrastructure development, even at a cost, can result in long-term benefits.

The 52-week range has a narrative of its own. Not too long ago, this stock dropped as low as $3.31. It has increased by more than 700% from that floor to $27.79. A portion of that is indicative of actual operational advancement, such as the launch of facilities, the signing of contracts, and the acceleration of revenue. A portion of it is a reflection of the market’s desire for anything related to spending on AI infrastructure. It’s possible that the current price assumes a level of execution that the company still needs to demonstrate, and separating those two forces is more difficult than it seems. That question will be greatly addressed in the coming quarters as Delta Forge 1 moves closer to completion and the ChronoScale separation takes place. As of right now, the megawatts are online, the trucks are operating, and the losses are growing. Investors are placing bets on which of those three facts is most important.