Arm Stock Slides Despite Smashing Earnings, Here’s the Hidden Reason That’s Spooking Wall Street

Imagine you’ve just sold $2 billion worth of tickets to a concert you can’t yet stage. The demand is real. People want in. But the venue only holds half that many seats, and the contractor hasn’t finished the stage, locked in the lighting, or even confirmed the speakers will arrive on time. Your would-be audience is ecstatic, your phone won’t stop ringing… and your shareholders are suddenly really, really nervous.

That’s the near-perfect metaphor for what just happened to Arm Holdings.

Late Wednesday, Arm reported blockbuster numbers: fourth-quarter revenue of $1.49 billion, smashing the $1.47 billion Wall Street consensus, powered by an AI-driven surge in data center chip demand and licensing income up an eye-watering 29% year over year. The stock initially soared 12% in after-hours trading, a euphoric bounce that lasted about as long as it takes for a conference-call microphone to warm up.

Then management began its call with analysts. Within an hour, Arm shares erased the entire gain, fell into the red, and kept sliding. By morning trading, the stock had shed 5.6% to roughly $224, translating into more than $12 billion in market value evaporated, not because the quarter was weak, but because reality poked a hole in the ceiling of hope.

What spooked the market? Two words that have become the shadow antagonist of the entire AI boom: supply constraints.

And right now, that bottleneck is hungry enough to swallow a $2 billion order book whole.

1. The After-Hours Rollercoaster, What Actually Happened Here?

Let’s rewind the tape, because the whiplash is a story in itself.

On Wednesday, May 6, Arm closed its regular trading session at $237.30, up 13.63% on the day, driven by pre-earnings euphoria and a fresh wave of analyst upgrades, at least 14 brokerages raised their price targets after the report dropped.

Then came the shareholder letter. And the data was genuinely impressive:

• Q4 revenue: $1.49 billion versus $1.47 billion expected (beat by ~$20 million)
• Adjusted EPS: $0.60 versus $0.58 expected
• Full fiscal year 2026 revenue: $4.92 billion, up 23% year over year, that’s three consecutive fiscal years of 20%+ growth since Arm’s IPO
• Licensing and other revenue: $819 million (blowing past the $774 million consensus)
• Q1 2027 revenue guidance: approximately $1.26 billion, modestly above the $1.25 billion Street estimate
• Q1 EPS guidance: $0.40, versus the $0.36 that analysts had modeled, a roughly 11% beat

So far, so good. The stock spiked 12% after hours before the conference call even started.

Then, during the investor Q&A, CEO Rene Haas and CFO Jason Child dropped the revelation that would become the dominant headline less than 24 hours later: For the AGI CPU, Arm has only locked in enough supply to meet the first $1 billion of demand. The company has not yet secured the wafers, memory, packaging, and testing capacity required to fulfill the second $1 billion.

CFO Jason Child put it plainly: Arm is maintaining its official $1 billion revenue guidance for the AGI CPU while it “pursues supply chain capacity,” with first production revenue not expected until the fourth quarter of fiscal 2027 (calendar early 2028).

The market’s algorithmic reaction was swift and brutal: from $237 on the close, shares tumbled overnight into the low $220s — restoring a significant discount to the optimism that had built in over the preceding weeks.

Portfolio manager Jay Goldberg at Seaport Research Partners summed it up with the sharp honesty this moment demanded: “It was a very tough setup for them, the expectations were just so high. They were good numbers, but not good enough.”

That phrase, “not good enough”, is maybe the fairest post-mortem you’ll read anywhere. Arm didn’t fail; it simply couldn’t jump over a bar the market set at an Olympic height.

2. Why Demand Alone Isn’t Enough, The AGI CPU Supply Chain Logjam

Here’s the thing about chips: wanting to make them is not the same as being able to make them.

Arm’s AGI CPU, a purpose-built data center processor designed for so-called “agentic AI” workloads where autonomous software agents orchestrate tasks with minimal human oversight, is being manufactured by Taiwan Semiconductor Manufacturing Co. (TSMC), the world’s most advanced contract chipmaker, using state-of-the-art 3-nanometer process technology. This is not a commodity product you can simply shuffle to another factory. TSMC currently has a near-monopoly on fabrication at this performance tier.

And TSMC’s 3nm lines are, to put it mildly, completely slammed.

TrendForce reports that AI demand has been surging since 2023, pushing 3nm-to-2nm wafer capacity and 2.5D/3D advanced packaging into structural shortages. CoWoS packaging bottlenecks in particular have never let up, every hyperscaler on the planet is fighting for allocation. SemiAnalysis warns that AI-related workloads could consume nearly 60% of TSMC’s N3 output in 2026 alone, climbing toward 86% by 2027.

And here’s the critical detail that makes Arm’s position especially precarious: while Arm was busy designing chips, its own ecosystem partners,  Nvidia, AMD, Broadcom, even Apple, have been lining up at TSMC’s door for years. Nvidia’s Blackwell and upcoming Vera platforms, Google’s TPU systems, Amazon’s Trainium chips, they all rely on TSMC’s advanced nodes, and they all compete for the exact same capacity allocation that Arm now needs to manufacture its brand-new AGI CPU.

Put another way: Arm entered the queue at a restaurant where the waitlist already wraps around the block, and some of the guests are valued regulars ordering half the menu.

This isn’t just a “bottleneck” in the abstract sense. It’s a bottle with a very narrow neck and a dozen greedy straws crammed into it.

Haas acknowledged this candidly on CNBC Thursday morning, telling Jim Cramer: “These chips take a while to design. They take a while to build. They take a while to get through the entire pipeline. So we are working with our partners in TSMC for months. I’m confident we’ll get that supply, and I’m also very confident demand will continue.”

The good news? Demand isn’t “perishable,” as Haas put it. This is firm, structural demand driven by AI agents that are fundamentally dependent on CPU orchestration, not disappearing hype.

The uncomfortable news? The market hates waiting. And right now, Arm is asking investors to price in certainty on a timeline no one fully controls.

3. Smartphone Headwinds & The Royalty Miss, A Tale of Two Markets

While the supply chain drama between Arm and TSMC stole the spotlight, there’s a second undercurrent worth flagging, one that introduces a slower-burning drag on the stock.

Arm’s designs power virtually every smartphone in the world. That has historically been the company’s crown jewel. But right now, that jewel is losing a little of its luster. Management predicted “slightly negative” unit growth in smartphones during the call, citing a persistent shortage of memory chips that has been driving up consumer electronics prices and suppressing device sales, a headwind that chipmaker Qualcomm flagged just a week earlier in its own cautious outlook.

The numbers corroborated the concern: Q4 royalty revenue came in at $671 million, missing the $697.1 million consensus estimate by about 4%. Morgan Stanley analysts called this out specifically, noting that the royalty miss and “weaker-than-expected” operating expenditure growth trend were the two most notable blemishes on an otherwise strong report.

Here’s how to think about this: Licensing revenue is the forward-looking signal (companies pay fees upfront to design products with Arm blueprints), and that signal is glowing green. Royalty revenue is the rearview mirror (what’s already been sold and shipped), and that rearview mirror shows a few potholes.

As smartphones wobble and data centers roar, Arm is becoming a dual-speed narrative: rearview is wobbly, windshield is open highway. Managing both narratives simultaneously is ultimately what makes the stock so volatile to trade right now.

4. From Licensing House to Chip Maker, Why This Pivot Changes the Game

For 35 years, Arm operated a gloriously asset-light business model: design the blueprint (IP), license it to partners, and collect a per-chip royalty on every device shipped using that blueprint. It’s the architectural equivalent of collecting a songwriting royalty, do the creative work once, then get paid every single time someone plays the song. That model powered gross margins in the high 90s, because someone else (TSMC, Samsung, etc.) paid for the factories.

The AGI CPU changes the equation fundamentally.

For the first time in its history, Arm is selling its own silicon, stepping onto the factory floor as a manufacturer, not just a design studio. Rather than simply licensing its Neoverse architecture to partners like AWS (Graviton), Google (Axion), and Microsoft (Cobalt), Arm is putting its own complete chip into the market under its own brand, on its own balance sheet.

And that shift, from a pure intellectual-property business to a hybrid IP-plus-chip company, introduces an entirely new category of risk. Physical supply chains. Capital intensity. Inventory management. Margin dilution during ramp periods. Customer conflict risk. When you sell chips to the same hyperscalers who also license your designs, the relationship becomes… complicated. No longer are you the neutral arms dealer selling blueprints to everyone; now you also compete in select categories with some of your own customers.

Goldman Sachs’ sell thesis leans heavily on this dynamic. Their analysts argue that “operating expenses are rising,” that the transition carries meaningful execution risk, and that the current P/E multiple of approximately 100x severely overstates near-term growth potential. Goldman maintains a $125 price target, implying downside of over 40% from the pre-report closing price, and believes the stock will remain “range-bound” in the near term.

That’s the bear case. But the bull case is arguably equally coherent.

5. The Analyst Civil War: Is This a $125 Stock or a $300 Stock?

If you thought the earnings report was disorienting, wait until you survey the analyst landscape. It’s a genuine schism:

That’s a spread of $175, more than 100% of the lowest target — and it reflects exactly how much uncertainty is embedded in the pivot from licensing house to chip company.

Goldman Sachs is pricing the company on what it currently is (a licensing-and-royalty business with a stretched multiple). KeyBanc and Jefferies are pricing it on what it could become (a CPU giant riding a $100B+ AI infrastructure wave). The truth, as it usually does, probably lies somewhere in the messy middle. But the burden of proof has shifted from “do you believe in AI demand?” to “can the company execute on supply?”

BofA’s note is particularly instructive: they model only 5-7% market share for Arm in AGI chiplets by 2031, versus management’s stated 15% target, pointing out that “key customers including Meta and OpenAI already have multi-sourcing CPU agreements.” This is the polite way of saying: even if demand is real, market share is not guaranteed.

6. What Comes Next, The $15 Billion Vision for 2031

Here’s the long-term picture that Arm painted for investors, because even amid the selloff, the thesis remains ambitious:

• AGI CPU annual silicon revenue target: $15 billion by fiscal 2031
• IP business expansion: Management expects IP revenue to roughly double to $10 billion annually
• Total company revenue target: $25 billion by fiscal 2031, translating to over $9 in earnings per share
• Server CPU TAM expansion: UBS forecasts the total addressable market will grow from ~$30 billion in 2025 to ~$170 billion by 2030, roughly a 5x expansion, and expects Arm to capture 40-45% total unit market share, up from ~15% today

That trajectory makes investor eyes light up, and at least partially explains why the stock was able to surge 75%+ between the AGI CPU launch on March 24 and the earnings print on May 6, ripping well ahead of any revenue reality.

But the supply chain constraint is a giant asterisk that sits squarely above every growth model. You cannot capture market share if you cannot manufacture and ship the product. And you cannot ship the product if TSMC’s N3 lines are booked by Nvidia’s Blackwell-Rubin families, Google’s TPU arrays, and Amazon’s Trainium chips, all of which operate at volumes that dwarf a brand-new entrant.

This is the paradox investors now have to live with: the demand story is one of the most compelling in semiconductors. The supply execution story is, at best, a “check back in 12-18 months.”

The Paradox Investor’s Checklist

Arm’s stock slide isn’t a verdict. It’s a volatility event driven by a dislocation between what the market hoped to hear and what the supply chain can realistically deliver today. The underlying fundamentals are strong: accelerating licensing, record revenue, a legitimate AI narrative, and a strategic pivot that could dramatically expand Arm’s TAM and margin profile over the next five years.

But the price of admission is sky-high, a trailing P/E hovering around 317, a forward multiple north of 120, and that rich valuation leaves absolutely no room for execution slippage. When the command from the market is “be perfect or be punished,” and a 4% royalty miss and a supply-chain caveat arrive in the same call, the sell button gets pressed without asking follow-up questions.

If you’re already holding ARM: The signal in the noise is the demand trajectory. Customer commitments doubled in 6 weeks, that doesn’t happen for a product nobody wants. The question you need to answer honestly: can I stomach 20-30% drawdowns on the path to a possible 2031 re-rating?

If you’re considering an entry: Watch for three catalysts, and three risks.

🟢 Catalysts (Reasons to Get Constructive)

• Supply chain announcements — any confirmation that wafers, memory, packaging, or test capacity have been secured beyond the initial $1B tranche
• Two consecutive quarters of in-line or beat royalty revenue — smoothing the smartphone headwind narrative
• Non-hyperscaler customer wins that demonstrate demand breadth beyond the early-adopter mega-caps

🔴 Risks (Reasons to Stay Patient)

• Valuation compression — if the growth narrative stalls, a forward P/E of 120x could drift toward the 50-70x range that even premium semiconductor stocks struggle to command
• Gross margin pressure — selling physical chips is a structurally lower-margin activity than licensing IP, and the blended margin story is unproven at scale
• Customer conflict escalation — the more Arm competes with its own licensees, the more complex those relationships become, and the more incentive hyperscalers have to diversify architecture commitments

Arm’s promise isn’t dead, but its perfect narrative is. And in a market that’s been paying perfection-level premiums, minor imperfections are all it takes to introduce some needed gravity.