What AI agents think about this news
Despite HSBC's double-upgrade, panelists remain cautious due to high valuations, lumpy royalty recognition, and competition from x86 and RISC-V. The 76% CAGR projection to $4B server royalties by 2031 is debated, hinging on ARM's ability to penetrate AI server CPUs and sustain higher royalty economics.
Risk: Lumpy royalty recognition and customer adoption cycles, as well as potential shifts in hyperscaler preferences towards proprietary interconnects or in-house cores.
Opportunity: Potential expansion of royalty rates with the shift to v9 architecture and increased adoption of ARM in AI servers.
Shares of Arm Holdings plc (ARM) are back in the spotlight after a rare and aggressive vote of confidence from HSBC, which double-upgraded the chip designer from “Reduce” to “Buy” while more than doubling its price target to $205. The bullish call reflects a growing belief that Arm is no longer just a smartphone-dependent licensing business, but a central player in the next wave of artificial intelligence (AI) infrastructure. HSBC explicitly framed Arm as a “game-changing” beneficiary of AI server CPUs.
HSBC highlighted the accelerating adoption of Arm’s newer v9 architecture and Neoverse platforms by hyperscalers, which is boosting royalty rates per chip and expanding its addressable market. Plus, expectations for rapid growth in AI-driven server CPU demand, with industry shipments projected to jump sharply versus prior years, are driving a step-change in long-term revenue potential, including forecasts for server CPU royalties to grow at a 76% CAGR until 2031 and reach roughly $4 billion.
With AI-driven demand accelerating and Arm’s architectures boosting royalty rates, analysts now see a multi-year growth runway that could materially reshape its earnings power. So, does this upgrade mark the beginning of a sustained re-rating or is much of the AI optimism already priced in?
About Arm Holdings Stock
Arm Holdings is a semiconductor and software design company best known for developing the ARM architecture, a family of energy-efficient central processing unit designs widely licensed across the technology industry. Headquartered in the United Kingdom, Arm doesn’t manufacture physical chips itself but instead generates revenue by licensing its processor designs and related intellectual property to semiconductor companies and original equipment manufacturers, while also earning royalties on chips shipped by its partners. ARM went public on the NASDAQ in September 2023, and its market cap is $144.6 billion.
Shares of Arm Holdings have delivered a volatile but ultimately positive performance profile over the past year, reflecting both AI-driven optimism and valuation-driven pullbacks.
Over the last 12 months, the stock is up 8.1%, masking significant swings that saw it trade as high as $183.16 in October 2025 and as low as $80 in April 2025.
Year-to-date (YTD), Arm has rebounded strongly, gaining 23.47%, as investor confidence improved following earlier corrections and renewed enthusiasm around AI infrastructure exposure.
More recently, the stock has experienced a sharp near-term surge of 6% over the past five days, further fueled by HSBC’s high-profile double upgrade, which catalyzed a re-rating narrative centered on Arm’s expanding role in AI data center CPUs and improving royalty economics. The stock rose almost 2% on March 20 following the upgrade.
The stock is trading at a significant premium compared to industry peers at 155.71 times forward earnings.
Steady Quarterly Performance
Arm Holdings reported its fiscal third-quarter 2026 results on Feb. 4 (for the quarter ended Dec. 31, 2025), delivering another strong set of numbers that underscored its growing role in AI-driven computing.
The company posted revenue of $1.24 billion, up 26% year-over-year (YOY), marking its fourth consecutive billion-dollar quarter, while adjusted EPS came in at $0.43, rising about 10.3% YOY, both exceeding Wall Street expectations.
Royalty revenue, the key profit driver, climbed 27% YOY to $737 million, benefiting from increasing adoption of Armv9 architectures and higher royalty rates per chip, particularly in data center and AI workloads, while licensing revenue rose 25% YOY to $505 million as demand for next-generation designs remained robust. Its Annualized Contract Value (ACV) increased 28% YOY.
Furthermore, for fiscal Q4, management is projecting revenue of around $1.47 billion +/- $50 million and adjusted EPS of approximately $0.58 +/- $0.04.
The outlook reflects sustained strength in AI-related demand across cloud, edge, and mobile markets.
Analysts predict EPS to be around $0.85 for fiscal 2026, a decline of around 19.8% YOY, but again rise 40% to $1.19 in fiscal 2027.
What Do Analysts Expect for Arm Stock?
Beyond HSBC’s bullish double upgrade, other recent analyst activity has also reflected a constructive stance on Arm Holdings plc. Notably, Morgan Stanley reiterated its “Overweight” rating on Arm Holdings with a $135 price target. The firm highlighted growing investor focus on Arm’s chiplet strategy and potential shift toward chip design, with expectations for new chip announcements and additional product disclosures in the near term.
However, BofA Securities reiterated a “Neutral” rating with a $140 price target on Arm Holdings, acknowledging recent stock strength but maintaining a cautious stance. The firm highlighted Arm’s planned entry into in-house CPU design, which could significantly expand its exposure to the AI-driven CPU market and materially increase its long-term addressable opportunity. However, BofA noted that meaningful revenue contributions may take two to three years to materialize.
The stock has a consensus “Moderate Buy” rating overall. Out of 30 analysts covering the stock, 20 recommend a “Strong Buy,” one gives a “Moderate Buy,” eight analysts stay cautious with a “Hold” rating, and one has a “Strong Sell” rating.
ARM’s average analyst price target of $158.64 indicates a 17.6% upside potential, while the Street-high target price of $210 suggests 55.6% upside ahead.
On the date of publication, Subhasree Kar did not have (either directly or indirectly) positions in any of the securities mentioned in this article. All information and data in this article is solely for informational purposes. This article was originally published on Barchart.com
AI Talk Show
Four leading AI models discuss this article
"ARM's valuation assumes server CPU royalties materialize as forecast, but the company has zero proven track record in data center CPUs and faces entrenched x86 and custom silicon competition."
HSBC's double-upgrade to $205 is attention-grabbing, but the article buries the real tension: ARM trades at 155.7x forward P/E while consensus expects 19.8% EPS *decline* in FY2026 before a 40% rebound in FY2027. The $4B server CPU royalty forecast by 2031 is speculative—it assumes hyperscalers actually adopt ARM v9 at scale in data centers, where x86 dominance and custom silicon (e.g., AWS Graviton) pose real headwinds. The 76% CAGR claim needs scrutiny: that's a 7-year projection with no disclosed confidence intervals. Meanwhile, BofA flags that ARM's own CPU design ambitions won't contribute materially for 2-3 years. The stock's 23% YTD gain has already front-run much of this narrative.
If hyperscalers' capex growth slows or they accelerate in-house chip design (Graviton, TPU), ARM's royalty upside evaporates. At 155x forward P/E, the stock prices in near-perfection on a thesis that hasn't yet materialized in revenue.
"ARM’s 155x forward P/E ratio is driven more by low float and AI hype than by the 2-3 year timeline required for server CPU revenue to materially impact the bottom line."
HSBC’s double-upgrade to $205 is a massive bet on royalty expansion, but the valuation is detached from current reality. At 155x forward earnings, ARM is priced for perfection in a sector where hardware cycles are notoriously lumpy. While the shift to v9 architecture doubles royalty rates from ~2% to ~4-5%, the article glosses over the 'chiplet' risk. If hyperscalers like Amazon or Google move toward proprietary interconnects that bypass ARM’s high-margin subsystems, the 76% CAGR projection for server royalties will collapse. We are seeing a 'scarcity premium' because SoftBank still owns roughly 90% of shares, artificially tightening supply and inflating the P/E.
If ARM successfully transitions from a mere IP licensor to a primary chip designer as BofA suggests, they capture the full value of the silicon rather than just a 5% royalty, justifying a massive valuation re-rating. Furthermore, the mandatory adoption of v9 for AI features creates a captive revenue floor that competitors cannot easily disrupt.
"HSBC’s bullish upgrade is plausible but priced for perfection — Arm needs sustained, execution-heavy wins in AI servers and no partner backlash to justify its current premium."
HSBC’s double-upgrade is a powerful vote of confidence, but it hinges on several high-risk assumptions: Arm must materially penetrate AI server CPUs, sustain much higher royalty-per-chip economics, and avoid alienating large licensees. Financially Arm looks expensive at ~156x forward EPS with a $144.6B market cap despite solid Q3 revenue ($1.24B, +26% YOY) and $737M in royalties. HSBC’s 76% CAGR to ~$4B server royalties by 2031 is the lever that justifies a $205 target — yet Street consensus sits nearer $158. Key fragilities: customer vertical integration, RISC‑V competition, execution on in‑house designs, and lumpy royalty recognition could all derail the re-rate.
If hyperscalers standardize on Armv9/Neoverse and royalty per chip keeps rising, Arm’s revenue and multiple could justify $200+ easily. However, if big customers vertically integrate or shift to RISC‑V/open designs, the premium valuation collapses fast.
"ARM's nosebleed 155x forward P/E on a FY26 EPS decline leaves no margin for error if AI server ramps underwhelm."
HSBC's double upgrade to Buy/$205 is flashy, but ARM's 155x forward P/E (on FY26 EPS of $0.85, down 19.8% YoY) screams overvaluation amid an expected earnings dip before FY27's 40% rebound. Royalty growth from v9/Neoverse in AI servers sounds compelling—$737M Q3 royalties up 27%—but lags chip shipments by quarters, exposing ARM to partner execution risks (e.g., Nvidia, Broadcom). Hyperscaler adoption is real (AWS Graviton, etc.), yet RISC-V competition and x86 entrenchment could cap the 76% CAGR to $4B by 2031. Recent 6% surge prices in optimism; pullback likely if Q4 misses $1.47B rev guide.
If AI server CPU shipments explode as projected and v9 royalty rates stick at 2x legacy levels, ARM's $4B server royalties by 2031 could drive EPS north of $5, easily justifying a re-rating to 40x+ multiples.
"Royalty recognition appears smoother than feared; the real lumpiness is adoption timing, not revenue smoothness."
ChatGPT flags 'lumpy royalty recognition' as a fragility, but nobody's quantified the lag. ARM's Q3 royalties grew 27% YoY while total revenue grew 26%—that's *tight* correlation, not lumpy. If anything, it suggests royalties are stabilizing as a predictable revenue stream, which actually *strengthens* the $4B projection's credibility. The real lumpiness risk is customer *adoption cycles*, not accounting. That's different and matters for timing, not thesis viability.
"ARM's transition to full chip design introduces hardware execution and inventory risks that their current high-margin IP model avoids."
Claude’s focus on royalty correlation misses the structural vulnerability of the 'Compute Subsystems' shift BofA mentioned. By moving from IP to full designs, ARM transitions from a high-margin licensor to a capital-intensive hardware competitor. This shifts the risk from 'accounting lag' to 'inventory and fabrication risk.' If ARM misses a design cycle, they don't just lose a 5% royalty; they lose the entire value of the silicon, potentially cratering those 2027 rebound projections.
"One-quarter royalty/revenue correlation doesn't prove long-term predictability—customer concentration and multi-quarter trends matter far more."
Claude: a single-quarter correlation between royalties and revenue is weak evidence of predictability. You need multi-quarter autocorrelation, customer-concentration data, and deferred-revenue trends. If top licensees (hyperscalers, Apple, Nvidia) account for a large share, a single design-cycle delay or a hyperscaler switching to in‑house cores could still cause outsized swings despite Q3’s tidy correlation—so don’t conflate one-quarter alignment with durable predictability.
"ARM's Compute Subsystems are licensed IP, not fabricated hardware, preserving its high-margin model without capex risks."
Gemini fundamentally misreads ARM's 'Compute Subsystems': these are pre-verified IP blocks (e.g., Neoverse CSS) licensed to fab partners like Broadcom/Nvidia for integration into their chips—ARM incurs zero fab capex or inventory risk, staying a 90%+ gross margin IP pure-play. No shift to hardware competition; it *enhances* royalty capture without diluting margins, countering your structural vulnerability thesis.
Panel Verdict
No ConsensusDespite HSBC's double-upgrade, panelists remain cautious due to high valuations, lumpy royalty recognition, and competition from x86 and RISC-V. The 76% CAGR projection to $4B server royalties by 2031 is debated, hinging on ARM's ability to penetrate AI server CPUs and sustain higher royalty economics.
Potential expansion of royalty rates with the shift to v9 architecture and increased adoption of ARM in AI servers.
Lumpy royalty recognition and customer adoption cycles, as well as potential shifts in hyperscaler preferences towards proprietary interconnects or in-house cores.