Arm is the invisible toll booth on every computing device on Earth â 30 billion+ chips shipped per year, royalties on every one. The Q4 FY26 print on May 6 is the first chance to hear FY27 guidance, CSS royalty rate progression, and whether the data center royalty displacement of smartphone is ahead of schedule. That's the entire call in three sentences.
Arm Holdings (NASDAQ: ARM) designs the processor architectures that run virtually every smartphone, tablet, and an accelerating share of AI data center infrastructure on Earth. The company does not manufacture chips â it licenses instruction set architectures (ISAs), CPU/GPU/NPU cores, and increasingly complete Compute Subsystems to semiconductor companies, who pay an upfront license fee and then a royalty on every chip they ship. That asset-light model produces a ~97% gross margin and positions Arm at the center of every major computing trend without taking manufacturing risk.
Founded in 1990 as a joint venture between Acorn Computers, Apple, and VLSI Technology in Cambridge, UK, Arm was acquired by SoftBank in 2016 for $32 billion. It relisted on NASDAQ on September 14, 2023 at $51/share â the largest IPO of that year â raising ~$4.87 billion, with SoftBank retaining approximately 90% ownership. CEO Rene Haas, who took over in 2022 from Simon Segars, has repositioned Arm's narrative from "smartphone royalty engine" to "compute platform of choice from milliwatts to megawatts."
Arm's revenue model has two components: licensing revenue (upfront fees from chip designers for access to IP), which is lumpy quarter-to-quarter but signals future royalty streams; and royalty revenue (per-unit fees on every chip shipped using Arm IP), which is recurring and scales with chip volumes and architecture mix. The key valuation driver is the architecture transition from Armv8 to Armv9 â v9 pays approximately 2x the royalty rate of v8, and as the v9 mix in shipping chips increases, royalty revenue grows without requiring more chip volumes.
Arm's strategic agenda is concentrated on two structural transitions: the CSS royalty rate step-change and the Armv9 mix inflection. Both are happening simultaneously and both are systematically under-modeled by sell-side consensus.
Arm's revenue trajectory reflects two dynamics: licensing revenue is episodic and lumpy (deal timing drives quarter-to-quarter swings), while royalty revenue is structurally growing as v9 mix increases and data center ramps. The last four quarters:
| Quarter | Revenue | Royalty | License | Non-GAAP EPS | YoY Growth |
|---|---|---|---|---|---|
| Q1 FY26 (Jun 2025) | $1.05B | ~$590M | ~$472M | $0.35 | +12% |
| Q2 FY26 (Sep 2025) | $1.14B | $620M | $515M | $0.39 | +34% |
| Q3 FY26 (Dec 2025) | $1.24B | $737M | $505M | $0.43 | +26% |
| Q4 FY26 Guide (Mar 2026) | $1.47B ±$50M | up low-teens YoY | up high-teens YoY | $0.58 ±$0.04 | ~+18% |
Royalty Revenue as the quality signal: Royalty revenue is the recurring engine â $737M in Q3, up 27% YoY with data center growing >100%. The Q4 guide calls for royalties up "low teens" YoY â approximately $650â670M. Sequential decline Q3âQ4 is normal seasonality. Watch whether Q4 royalty beats or misses the implied ~$655M consensus.
License Revenue lumpiness: Q3 license came in at $505M vs. $519M consensus â the stock dropped 7%+ despite beating on total revenue. ACV (Annualized Contract Value) grew 28% YoY in Q3. Q4 license guidance implies ~$815-830M (high teens YoY on the Q4 FY25 base of ~$700M). Any miss on license will likely trigger a similar stock reaction even with a headline revenue beat.
FY26 full year estimate: Q1+Q2+Q3 actual ($3.43B) + Q4 guided midpoint $1.47B = ~$4.90B FY26 revenue (vs. $4.01B FY25 = +22% YoY). Non-GAAP OpEx guided to ~$745M in Q4 â $2.9B+ annualized. TTM Free Cash Flow: $893M as of Q3 FY26.
Arm competes in the invisible layer: it doesn't build chips, it provides the blueprints. This makes its competitive position simultaneously more durable (no fab risk, no design execution risk) and more vulnerable (a superior ISA alternative would threaten the entire royalty stream, not just one product line).
Get the post-print update. We'll email you within 24h of the print with actuals, thesis verdict, and an updated bull/bear range.
Arm's risk profile is unusual: not execution risk (it doesn't build anything) and not market risk (compute demand only grows). It's structural risk â the risk that the computing industry migrates to ISAs that don't pay Arm royalties. Near-term, the risks are valuation and guidance disappointment.
The sell-side consensus on ARM focuses on unit volumes and total revenue. It systematically misses the rate story â how CSS and v9 mix change the per-chip royalty Arm earns even when volumes are flat. These four observations are where most bull cases are underestimated.
| Metric | Company Guidance | Street Consensus | Actual (Post-Print) |
|---|---|---|---|
| Total Revenue | $1.47B ±$50M | ~$1.47B | TBD |
| Non-GAAP EPS | $0.58 ±$0.04 | ~$0.57â0.59 | TBD |
| Royalty Revenue | up low-teens YoY | ~$650â670M | TBD |
| License Revenue | up high-teens YoY | ~$815â830M | TBD |
| FY26 Full Year Revenue | â (no guide given) | ~$4.87â4.95B | TBD |
| FY27 Revenue Guide | â | ~$5.8â6.0B est. | TBD |
The single most important number on this call: the combination of total royalty guidance for Q1 FY27 and any quantification of FY27 framework revenue. If Haas says "we expect FY27 revenue of $6B+" with specific drivers, the stock re-rates immediately. If he hedges, expect disappointment regardless of Q4 numbers.
Get the post-print update. We'll email you within 24h of the print with actuals, thesis verdict, and an updated bull/bear range.
Vektor delivers the full 7-section brief â bull/bear ranges, consensus tables, call watch items â for any ticker in 40 seconds. $29/month.
No spam. Unsubscribe anytime.