FEATURED

ARM Is Up 11% Today. July 29 Is What Counts.

Here’s what made today interesting.

Arm Holdings (ARM) surged roughly 11% on July 9, making it one of the standout large-cap winners in a market that was otherwise dealing with geopolitical noise, oil moving sharply, and the usual reasons to stay cautious. When the broader environment is choppy and a single stock still surges 11%, it’s usually saying something. The question is whether you’re hearing the same thing the market is.

What it’s saying here is complicated.

What Actually Drove the Move

The clean version: investors are rotating back into premium AI names after a mid-year correction in tech and semiconductor equities, and Arm’s structural positioning inside AI infrastructure is getting harder to ignore. The messier version: ARM had sold off hard from its 52-week high of $452.70, and investors smelled a re-entry point ahead of the July 29 earnings report.

The anticipatory buying is real. So is the momentum underneath it. Arm posted a 29% year-over-year increase in licensing revenues for fiscal Q4 2026, driving overall revenue growth of 20% for the quarter, with royalty revenues rising 11% to $671 million, supported by broader adoption of its Armv9 architecture in AI, data center, and custom silicon markets. A new institutional investor position disclosed in regulatory filings added fuel to the move today.

The Company Most Investors Still Underread

Arm is not a chip company. It does not make chips. It licenses the architecture that chips are built on, a royalty business that has quietly become the default blueprint for everything from your iPhone to the AI servers running agentic workloads inside Amazon, Google, Microsoft, and Meta data centers.

The shift happening right now is the one that makes this interesting. On the Q4 FY2026 earnings call, CEO Rene Haas disclosed that committed customer demand for the AGI CPU exceeded $2 billion across fiscal 2027 and 2028, more than double the $1 billion figure stated at the product’s launch just six weeks earlier. The logic behind the acceleration: as AI transitions from prompt-and-response to continuous agentic workloads, the number of CPU cores required per data center gigawatt grows roughly 4x. Arm estimates a $100 billion CPU total addressable market by 2030. Citigroup projects the server CPU market could reach $132 billion by that same year, with agentic workloads driving much of the expansion.

Slight tangent, but it matters: the reason Arm is so well positioned for agentic AI specifically is that always-on, continuous workloads need energy efficiency above almost everything else. That is exactly where Arm’s architecture has always had a structural advantage over x86. This is not a new thesis. It just became the most important one in the data center almost overnight.

AWS, Google, Nvidia, and Microsoft are all building and accelerating on Arm-based silicon. That list is not incremental validation. That is the core of the AI industrial complex, all building on the same architecture.

The Numbers

• Q4 FY2026 revenue: $1.49 billion, up 20.1% year-over-year, with non-GAAP EPS of $0.60 beating the Street estimate of $0.58
• Full-year FY2026 revenue: $4.92 billion, up 23% year-over-year, marking the third consecutive year of 20%-plus growth
• Full-year royalty revenue: $2.61 billion, up 21%; full-year licensing revenue: $2.31 billion, up 25%
• Data center royalty revenue: more than doubled year-over-year in Q4
• Q1 FY2027 guidance (July 29 report): Street consensus calls for EPS of $0.36 on revenue of roughly $1.27 billion
• 52-week range: $100.02 to $452.70; beta: 3.76

The Valuation Tension Nobody Is Resolving

This is where it gets genuinely uncomfortable. According to S&P Global data covering roughly 40 analysts, the consensus rating on ARM is Buy, but the average 12-month price target sits around $296, which is below where the stock is trading today after today’s surge. That is an unusual configuration: a Buy-majority consensus with a mean target that implies downside from current levels.

The most bullish analysts are not timid. TD Cowen raised its target to $475, UBS to $470, and Mizuho to $500, targeting $15 billion in agentic AI CPU revenue by fiscal 2031. Those are bold numbers. They are also three to four years away from being provable.

The skeptics have a point too. The aggressive transition into designing custom silicon has forced non-GAAP operating margins down from 52.8% to 49.1%, driven by a 43% year-over-year surge in R&D expenses. That is real margin compression happening in real time. The business is investing hard into a bet that has not yet delivered the revenue scale to justify current price levels on any near-term fundamental screen. ARM currently trades at a forward price-to-sales ratio far above the semiconductor industry average, reflecting expectations that are priced well into the future.

The float situation adds a layer most investors are not watching closely enough. SoftBank maintains roughly a 90% ownership stake in ARM, leaving a highly constrained public float. SoftBank has also used ARM shares as collateral for margin loans to fund its AI investment activity, including commitments to OpenAI and Stargate. Any severe systemic drawdown could create forced selling pressure in a stock with very limited daily liquidity relative to the size of those pledged positions.

Bull / Base / Bear

Bull: Arm is pushing deeper into AI infrastructure with its AGI CPU, securing more than $2 billion in customer commitments for fiscal 2027 and 2028 and targeting up to $15 billion in annual AI CPU revenue by 2031. If those commitments convert and royalty rates expand alongside Armv9 adoption, the current valuation becomes retrospectively cheap.

Base: Street consensus projects roughly 20% revenue growth for Q1 FY2027, with growth expected to accelerate modestly into late fiscal 2027. Steady 20%-plus growth gets priced in, and the stock trades sideways until the silicon revenue materializes at scale.

Bear: Risks include the Qualcomm/Nuvia trial scheduled for late 2026, potential U.S. semiconductor tariff impacts, and an ongoing FTC antitrust investigation into Arm’s licensing practices. Any one of those could disrupt the licensing relationships that underwrite the entire royalty model. Continued friction with key partners also risks pushing major tech clients toward open-source RISC-V alternatives.

Technical Overlay

Arm is trading below its 20-day moving average (around $355), above the 50-day at roughly $304, and well above the 200-day at approximately $179, confirming a medium- and long-term bullish structure. Today’s 11% surge after a prolonged correction is the kind of move that either confirms a bottom or sets up a head-fake. The honest answer is you don’t know which until July 29.

A golden cross pattern that emerged in April, when the 50-day moving average crossed above the 200-day, continues to hold. That’s constructive. But the stock’s beta of 3.76 means any macro shock between now and earnings will amplify in both directions.

What Investors Should Watch

The July 29 earnings report is the only number that matters right now. Specifically: does management reaffirm the $2 billion-plus AGI CPU demand figure, or does that number grow? Fulfillment of AGI CPU demand is currently limited by wafer, memory, and packaging capacity, so the key question on the call will be whether supply constraints are easing or widening. If supply is catching up to demand, the royalty inflection accelerates. If not, guidance gets complicated fast.

Also worth tracking: the Qualcomm/Nuvia trial timeline. ARM faces a high-stakes legal battle with Qualcomm over Nuvia architecture licensing set for trial in late 2026. That outcome could meaningfully reshape the licensing business in ways the current consensus has not fully modeled.

Bottom Line

Arm’s 11% move today is easy to explain as a dip-buy in a recovering semiconductor sector. What’s harder to explain is whether the stock at current levels is pricing in a business that will exist in 2031 or one that exists right now. The honest answer is both. The investment case on Arm Holdings is not primarily about what the company earns today. It is about whether a royalty-and-licensing business that has compounded at 20%-plus for three consecutive years can layer a $15 billion silicon revenue stream on top of an IP business projected to double, without disrupting the relationships that underwrite the entire royalty model.

That tension does not resolve on July 9. It resolves on July 29.