Arm’s AI boom just hit a supply wall

The good news: demand is not the problem

Arm came in hot on fiscal fourth-quarter earnings, and the market briefly liked what it heard. Then the company did the classic “great news, but…” routine: demand for its AI data center CPUs is still screaming, but supply is tight enough that Arm says it can only fulfill roughly half of current demand.

That’s the kind of problem most companies would love to have, but Wall Street doesn’t hand out participation trophies. If you can’t ship enough product, the revenue hockey stick starts looking more like a garden rake.

The AI pitch is getting bigger

CEO Rene Haas spent plenty of time talking up the next wave of AI infrastructure, especially agentic AI — the kind where software isn’t just answering questions, but actually doing stuff on your behalf. That matters because all those little AI errands need CPUs to coordinate tasks, manage memory, and keep the whole machine from turning into a digital junk drawer.

Arm says that opportunity could top $100 billion by 2030, and it’s already seeing more than $2 billion in customer demand for its AGI CPU across fiscal 2027 and 2028. Haas also said future designs could scale from 136 cores today to 256 or even 512 cores. Translation: the chip arms race is getting even more ridiculous.

Why investors are watching the plumbing

The company isn’t just bragging about demand — it’s trying to build enough industrial plumbing to meet it. Arm says it’s working on more wafer, packaging, memory, and testing capacity, while still expecting initial production revenue in the fourth quarter of the fiscal year.

That’s the real tension here:

• AI demand is strong
• the ecosystem is expanding
• but supply constraints could delay the payoff

So yes, the long-term Arm story still looks spicy. But in the short term, the stock is basically telling you that even the hottest AI names can get tripped up by boring things like manufacturing capacity.

Big picture

Arm’s platform is becoming more central to cloud and AI infrastructure, with hyperscalers like Amazon, Google, Microsoft, Meta, and others deepening their use of Arm-based chips. The question now isn’t whether Arm has a place in the AI boom — it’s whether it can actually build fast enough to cash in.