It was one of those Wall Street evenings where the traders told a different story than the numbers. Arm Holdings rode a surge of optimism into its fourth-quarter results, closing Tuesday up almost fourteen percent. After the call started and the earnings declined, the stock started to move in the opposite direction in a matter of minutes. In after-hours trading, shares had dropped more than 6% by the time most East Coast residents had finished their dinner. Despite exceeding expectations and setting a record for the quarter, disappointment persisted. That’s the kind of paradox that makes Arm such an odd company to follow at the moment.
The headline figures were really impressive. Earnings per share came in at sixty cents, slightly higher than what analysts had predicted, and revenue came in at $1.49 billion, up twenty percent year over year. During the call, CEO Rene Haas, who has spent the better part of two years transforming Arm from a quiet British IP licensor into something more akin to a front-line AI player, boldly declared that chip demand had nearly doubled to roughly $2 billion. That should have delighted investors. A few did. However, Haas also acknowledged—almost simultaneously—that Arm was unable to truly meet all of it. Memory, wafers, packaging, and testing are all in short supply. They might be able to fulfill half. At that point, the atmosphere changed.
Observing this unfold gives the impression that Arm has fallen prey to its own story. Since January, its stock has increased by more than 130%, significantly surpassing Nvidia, AMD, and Broadcom. Anything less than flawless begins to appear problematic when a business is operating at such high levels. Jay Goldberg of Seaport did a good job of capturing it: Arm did a good job, but not well enough. The company had just not kept up with expectations.
The pivot, however, is the larger story that lies beneath the commotion. Arm worked in the background for decades, earning meager royalties on almost every smartphone that was sold. The architecture was sophisticated, low-power, and nearly imperceptible to customers. With Meta as the primary client, Haas now intends to take the business straight into chip manufacturing and sell its own AGI CPUs. By 2031, Arm anticipates chip sales of about $15 billion annually. The licensing business that founded the company would be dwarfed if that were to land.
It’s a huge wager that carries some risk. In essence, Arm is trying to convince investors that it can be both a direct rival to its own clients and a neutral licensor akin to Switzerland. The tension hasn’t entirely subsided. Long-term investors were attempting to determine whether this was a buying dip or the beginning of something more complex in some Reddit threads following earnings. The forward P/E ratio is still staggering, exceeding 300 by some metrics, which is the kind of figure that only makes sense if you completely agree with the AI infrastructure thesis.
And perhaps that’s how Arm should be viewed right now. No matter how frequently the outdated framing continues to appear in coverage, the company is no longer a smartphone story. Its chips are at the forefront of a shift in the industry toward data centers that must maximize every watt, energy-efficient compute, and AI agents that rely more on CPUs than GPUs. According to UBS analyst Tim Arcuri, Arm’s instruction set may have the biggest market share in the future. Maybe. Whether the supply issues Haas mentioned will be a one-quarter hiccup or something more persistent is still unknown.
The speed at which Arm has transitioned from the periphery to the center stage is difficult to overlook. The majority of retail investors could not have told you what the company actually did a few years ago. The chip industry as a whole is now affected by its earnings. The answer is unlikely to be available this quarter, whether the stock can increase to that valuation or if Tuesday’s after-hours selloff was a warning sign. With stories this large, it hardly ever does.
