Arm Holdings (ARM)

1. Business Overview

Arm Holdings is one of the most consequential technology companies in the world, and yet it remains one of the least understood by the general investing public. Unlike Intel, which makes the processors in your laptop, or Nvidia, which sells the GPUs powering AI data centres, Arm sells neither a finished product nor a service in the traditional sense. It sells the architectural blueprints — the intellectual property — upon which the vast majority of the world’s chips are built.

Founded in Cambridge in 1990, Arm designs Reduced Instruction Set Computing (RISC) processor architectures and licenses them to semiconductor companies, which then manufacture their own chips based on Arm’s designs. Apple’s A-series and M-series chips, Qualcomm’s Snapdragon, Samsung’s Exynos, and the chips inside virtually every smartphone on earth all trace their lineage to Arm’s instruction set architecture (ISA). By some estimates, over 99% of the world’s smartphones run on Arm-based chips — a concentration of market presence that most companies can only dream of.

Arm’s revenue model operates through two primary channels. The first is licensing revenue, whereby companies pay upfront fees to gain access to Arm’s processor designs, tools, and related software. These agreements can range from access to a basic processor core to full “architectural licences” that allow a customer — think Apple or Qualcomm — to design custom processors that remain compatible with the Arm ISA. The second and more valuable stream is royalty revenue: a per-chip fee that Arm earns on every processor sold that incorporates its designs. Arm’s royalty revenue is directly tied to the success of its clients’ chips and the volumes shipped, and the long-term nature of these agreements means that Arm can still collect significant revenues for IP designs more than ten years old.

For its complete 2025 fiscal year, Arm booked $4.0 billion in revenue, marking the first time the company crossed the $4 billion threshold. In its most recent reported quarter, total revenue reached $1.053 billion, up from $939 million in the same period a year prior.

Then, just yesterday — March 24, 2026 — Arm made an announcement that fundamentally changes the nature of this analysis. For the first time in its 35-year history, Arm has announced it will make physical silicon of its own, moving from exclusively licensing chip architecture to producing finished processors. The new product, called the AGI CPU, is an Arm-designed data centre processor built for agentic AI workloads. Meta Platforms is the lead partner, having co-developed the chip alongside Arm, and customers include OpenAI, Cloudflare, SAP, and SK Telecom. CEO Rene Haas has projected the new chip will generate roughly $15 billion in annual revenue within about five years, with overall company revenue reaching $25 billion by 2031. This is, to put it plainly, a historic strategic pivot — and one we will examine in depth throughout this report.

2. Industry Context

Arm sits at the intersection of two of the most consequential structural shifts in modern technology: the proliferation of connected, power-sensitive computing devices, and the rise of artificial intelligence as a core computing workload. Both of these dynamics play directly to Arm’s architectural strengths.

The global semiconductor industry is broadly segmented between firms that design chips (fabless companies), those that manufacture them (foundries like TSMC and Samsung), and the rare companies that do both (Intel, Samsung). Arm is unusual in that it occupies an even further upstream position: it designs the architecture that others then design chips around. This asset-light model means Arm captures value from nearly the entire chip ecosystem without bearing the capital intensity of fabrication.

Within this ecosystem, Arm’s principal historical domain has been mobile computing, where its RISC architecture provides efficiency advantages over the x86 architecture dominating desktop and server computing. Performance per watt — the ability to do more computation while consuming less power — is the defining competitive dimension for mobile, IoT, and increasingly, AI workloads. Arm’s architecture excels here.

Arm has successfully infiltrated the lucrative cloud server market with its Neoverse platform, breaking the x86 duopoly through strategic alliances with hyperscalers and a symbiotic relationship with Nvidia. Amazon Web Services’ Graviton processors, powered by Arm’s Neoverse architecture, have demonstrated that Arm-based server CPUs can meaningfully undercut x86 alternatives on cost-per-compute while matching or exceeding performance for cloud workloads.

The competitive landscape on the architectural level, however, is shifting. RISC-V has emerged as the primary challenger to Arm’s dominance, particularly as industries seek to insulate themselves from rising licensing costs and geopolitical volatility, with an estimated 20 billion cores in operation by the end of 2025. The widespread adoption of the RVA23 standardization profile has effectively addressed the historical criticism of RISC-V fragmentation, allowing major operating systems and AI frameworks to run natively across diverse RISC-V hardware. Meanwhile, x86 remains dominant in high-performance desktop and traditional server computing, though Arm’s inroads are undeniable.

Arm’s new chip announcement also positions it in direct competition with Intel and AMD in the data centre CPU market — a segment experiencing explosive growth driven by agentic AI workloads.

3. Economic Moat

Arm possesses one of the most durable and complex moats in the technology sector, built not from a single competitive advantage but from the reinforcing interplay of several.

Intellectual Property. Arm’s core asset is its instruction set architecture — the standardized language by which software communicates with hardware. Decades of investment have resulted in a rich portfolio of processor designs, tools, software development kits, and ecosystems. Crucially, the ISA itself acts as an industry standard. Software written for Arm runs on all Arm-based chips; the architecture is the universal language of mobile computing. This is not easily replicated.

Switching Costs. Leaving Arm’s ecosystem is extraordinarily costly. Chip designers who have built decades of engineering expertise around Arm’s toolchains, who have accumulated vast libraries of Arm-compatible IP, and whose customers have written enormous software stacks optimized for Arm, face a restructuring cost that is measured not in millions of dollars but in years of engineering time. A virtuous cycle exists where Arm’s IP licensing model fosters a vast partner and developer ecosystem, which in turn drives widespread adoption and creates a formidable strategic moat.

Network Effects. The Arm ecosystem exhibits classic network effects. The more developers write code optimized for Arm, the more valuable Arm hardware becomes; the more hardware shipped on Arm, the more developers optimize for it. This dynamic is self-reinforcing and has taken decades to build.

Performance-per-Watt Leadership. The inherent performance-per-watt efficiency of Arm’s RISC architecture has become the critical currency for power-hungry AI workloads. As AI data centres increasingly constrain their architectures around power envelopes rather than raw silicon, this efficiency advantage grows more — not less — important. Arm’s AGI CPU claims double the performance per watt of equivalent x86 rack configurations.

Scale and Ubiquity as a Moat. When your architecture is embedded in hundreds of billions of deployed devices, the prospect of migrating away becomes practically inconceivable for most purposes. The installed base creates a gravitational pull that draws new designs toward Arm compatibility.

That said, Arm’s moat is not impregnable. The RISC-V threat is real: Arm is now forced to innovate more aggressively on its licensing terms and technical performance to justify its premium pricing. The moat is wide but not infinitely deep, and the open-source nature of RISC-V removes the royalty overhead that is Arm’s fundamental business model. To be a shareholder in Arm Holdings today, you would need to believe that the company’s technology is indispensable for future AI and data centre growth, and that it can defend premium royalty rates even as competition intensifies.

4. Financial Quality

Arm’s financials are a study in exceptional gross-level economics constrained by heavy investment requirements.

Revenue Growth. The company reported a 23.9% revenue increase for its fiscal year ending March 2025, with recent quarters showing continued momentum. Annual gross profit for fiscal 2025 was $3.886 billion, a 26.21% increase from 2024. Growth is real, it is accelerating, and it is driven by higher royalty rates as newer, more complex chip designs attract better terms.

Profitability. Here the picture is genuinely exceptional at the gross margin level and more complicated below it. Gross margins run at approximately 97%, indicating effective cost management and high core profitability. This is among the highest gross margins in all of publicly traded technology — a direct reflection of the asset-light, IP-licensing model. However, operating margins have been volatile, swinging significantly quarter to quarter, primarily driven by enormous R&D expenses which consumed over 60% of revenue in some periods. As the company pours capital into next-generation architectures and now into physical chip development, the gap between gross profit and net income remains wide.

Operating Margin Dynamics. The operating margin on a reported basis stood at approximately 20.7% in recent filings, though non-GAAP figures run higher. R&D intensity is the dominant variable: in the quarter ending June 2025, R&D spending was $650 million against revenue of $1.053 billion — a staggeringly high ratio that reflects the competitive reality of remaining at the frontier of processor design.

Valuation. The market prices this business for perfection. With a P/E ratio of approximately 143x and an EV/EBITDA of approximately 108x, Arm trades at valuations that price in years of compounding growth without error. Its Enterprise Value to Sales ratio of 43x is extreme even by the standards of high-growth technology, and represents a significant premium to even its most valued semiconductor peers.

Balance Sheet. Arm carries a strong balance sheet with significant cash and low financial leverage, consistent with its capital-light heritage. The new chip initiative will increase capital requirements substantially, and execution risk on that dimension will be worth monitoring closely as the company transitions from a pure IP licensor to a hybrid model.

5. Management & Capital Allocation

Rene Haas has served as CEO of Arm since February 2022 and is based in California, reflecting the company’s increasingly American-oriented growth strategy despite its Cambridge headquarters. In 2025, Haas was named among the 100 Most Influential People in AI by Time Magazine. He is regarded by peers as a commercially-minded leader with deep technical credibility — a rarer combination than it sounds in the semiconductor world.

The strategic decision to move into physical silicon deserves particular scrutiny here. Arm spent $71 million and roughly 18 months building three new lab rooms at its campus in Austin, Texas, where a team has grown to over 1,000 engineers. The initiative began in 2023 and has culminated in a functioning, customer-committed chip. This is disciplined capital allocation: a targeted bet on a new revenue category, not a sprawling acquisition spree.

The relationship with SoftBank introduces a structural complexity. Following its IPO in September 2023, Arm qualifies as a controlled company under Nasdaq rules due to SoftBank Group’s majority shareholding. Masayoshi Son sits on Arm’s board. This means minority shareholders have limited governance rights — decisions of strategic importance can be made without their explicit consent. SoftBank’s own financial pressures and strategic priorities can, in theory, exert influence on Arm’s direction in ways that may not always align perfectly with minority shareholders. This is a real, if often underappreciated, governance risk.

Capital allocation priorities have historically leaned toward reinvestment — heavily so. Buybacks and dividends are minimal; R&D and strategic investment dominate. Given Arm’s growth opportunities, this appears rational. The pivot into silicon, however, represents a significant change in the capital model, introducing hardware development cycles, inventory risk, and supply-chain dependencies that were absent from a pure IP licensing business.

6. Risks & Red Flags

RISC-V and Open-Source Architecture. This is, in the long run, the most existential challenge. RISC-V has emerged as the primary challenger to Arm’s decades-long hegemony, particularly as industries seek to insulate themselves from rising licensing costs and geopolitical volatility. The RVA23 standardization profile has silenced earlier fragmentation concerns, enabling RISC-V to move beyond microcontrollers into complex, high-performance computing. Crucially, China and India have funneled billions into RISC-V development to avoid potential sanctions that could cut off access to Western proprietary architectures. Arm cannot meaningfully participate in China’s semiconductor ambitions over the long term — RISC-V can.

Geopolitical Exposure. China has historically represented a significant portion of Arm’s licensing revenue. As tensions between Western governments and China intensify, Arm faces a double bind: it risks losing Chinese customers to RISC-V alternatives while simultaneously being constrained by export-control regimes. This is not a near-term catastrophe, but it is a meaningful long-term erosion risk.

Customer Concentration. Arm’s royalty revenue depends on the health and market share of a relatively small number of anchor customers — Apple, Qualcomm, Samsung, and increasingly the major hyperscalers. Disruption at any of these companies, or a strategic shift toward in-house architectures, could meaningfully impact Arm’s royalty stream.

The New Silicon Business Model. Moving from exclusively licensing chip architecture to competing with companies like Apple, Nvidia, Amazon, and Google in physical silicon is a profound strategic shift — and not without peril. Arm risks cannibalizing the trust of its most important customers. If Apple believes Arm is now competing with it, not just serving it, the relationship dynamic changes. Nvidia, Amazon, and Google all have deep architectural relationships with Arm and may reconsider how much proprietary chip design they commit to an architecture owned by a nascent competitor.

Valuation Risk. Even granting Arm every benefit of the doubt on execution, the current multiple leaves almost no room for disappointment. A single missed earnings quarter, a slowdown in AI infrastructure spending, or a larger-than-expected shift to RISC-V in a key market segment could compress the multiple sharply.

Free Cash Flow Volatility. Revenue growth of 23.9% has been robust, while free cash flow growth has sharply declined, raising concerns about cash generation efficiency. Heavy R&D and now hardware development spending are squeezing the cash conversion that the high gross margins would otherwise imply.

7. SWOT Analysis

Strengths. Arm’s dominant position in mobile computing is without peer — its architecture underpins virtually all smartphones globally, a position built over decades and defended by profound switching costs and network effects. The gross margin profile, at approximately 97%, reflects the pricing power that only comes from owning an architectural standard. The efficiency characteristics of its RISC architecture are increasingly aligned with the dominant computing paradigm: AI inference, edge computing, and power-constrained data centres all reward what Arm does best. The new AGI CPU launch brings technical credibility — up to 64 AGI CPUs, containing some 8,700 cores, can fit in a single air-cooled rack, delivering double the performance per watt of an x86 equivalent, a specification that should resonate with data centre operators facing energy constraints.

Weaknesses. Arm’s financial quality below the gross profit line remains inconsistent. R&D spending is enormous in absolute terms and ratio terms, and the translation of impressive architecture design into sustained earnings growth has been uneven. The company’s governance structure, as a SoftBank-controlled entity, reduces the influence of public market shareholders meaningfully. The new chip business introduces operational complexity that Arm has never managed before — supply chains, manufacturing yields, customer support for hardware, inventory management — all areas in which Arm has no track record. The stock’s valuation is simultaneously a reflection of quality and a source of fragility.

Opportunities. The AI inflection is genuinely transformative for Arm’s addressable market. A unified architecture that allows developers to create AI applications that can be deployed seamlessly across the entire computing spectrum, from cloud to device, represents a structural advantage that compounds over time. Haas expects the IP business itself to double over the next five years, independently of the new silicon business — implying a compound opportunity of potentially historic proportion. Automotive represents a particularly promising frontier: the rise of software-defined vehicles is creating entirely new categories of automotive processors, most of which are being designed on Arm architecture. IoT and industrial computing offer similar long-cycle growth.

Threats. The RISC-V transition is not hypothetical. The royalty-free RISC-V architecture is being actively supported by major technology firms including Google and Meta — the same Meta that is Arm’s launch customer for the AGI CPU. The strategic irony is uncomfortable: Arm’s most important hardware partner is simultaneously funding its deepest architectural competitor. Chinese domestic chip ambitions represent a secular headwind. And the entry into physical silicon, while potentially rewarding, risks triggering defensive architectural decisions from the hyperscalers who currently represent Arm’s fastest-growing royalty base.

8. Investment Thesis

The Bull Case. Arm is the closest thing the semiconductor world has to a toll road. Every chip shipped on its architecture generates revenue without Arm bearing manufacturing costs, and as chips grow more complex and more capable — as they inevitably do — the royalty rates rise. The AI era, rather than threatening this model, supercharges it: the demand for power-efficient, scalable compute aligns perfectly with Arm’s architectural strengths. The new silicon business, if successfully executed, could expand Arm’s addressable market from a few billion dollars to tens of billions, while the IP business continues to compound quietly in the background. The CEO’s projection of $25 billion in revenue and $9 in earnings per share by 2031 implies a business six times its current size, built on both organic licensing growth and the new chip revenue. At that scale, the current valuation, while extreme by today’s numbers, might look more reasonable in hindsight.

The Bear Case. The current valuation embeds an enormous amount of optimism that is already well understood by the market. RISC-V is not going away; it is accelerating. The new chip business introduces risks that have tripped up companies with far more hardware experience. SoftBank’s ownership creates governance uncertainty. And if the AI infrastructure spending cycle turns or consolidates around fewer, more proprietary architectures, Arm could face a simultaneous hit to both its royalty business and its new silicon ambitions. At a P/E of over 140x, the margin of safety is essentially zero.

What Type of Investor Does This Suit? Arm is not a value investment by any conventional definition. It suits investors with a genuinely long-term horizon — five to ten years — who are comfortable with elevated valuations in exchange for structural, high-quality growth. It is best suited to investors who understand semiconductor ecosystems and are prepared to monitor the RISC-V competitive dynamic closely and over time. Growth-oriented institutional investors and technology-focused managers are its natural long-term holders. For the valuation-sensitive or those with shorter time horizons, the asymmetric downside at current prices is difficult to justify, however impressive the underlying business.

Arm Holdings is, in summary, a genuinely exceptional business trading at a genuinely demanding price, at a moment of genuine strategic transformation. The combination of those three “genuines” makes it one of the most intellectually interesting equities in the market — and one of the harder ones to size with conviction.

This analysis is for informational and research purposes only and does not constitute investment advice. All figures are based on publicly available filings and data.