Arm Holdings (ARM)

By both methods, the value today (DCF $26 · Multiples $31) is below the market price ($252).

Today — expensive, no margin of safety: at $252 trades ~718.1% above its value discounted to today (~$31); the expected return does not even reach the risk-free rate (4.5%).

At 5 years — Sobrevalorado: the expected total return is negative — the price already discounts a demanding scenario that, if not met, results in a loss.

The bridge: the return at 5 years falls below the risk-free rate (4.5%) — which is why there is not even a discount to today's value. To require a 15% annual return, it would need to be bought at ~$19.

Educational / informational. Does not constitute investment advice.

The inverse approach: instead of projecting growth to obtain the value, the market price ($252) is taken as given and it solves for what annual owner-earnings growth would need to hold for 5 years for the present value —at the method's rate (4.5%, no-growth terminal)— to equal that price. It is the disconfirmation test: the expectations the price already pays for, contrasted against the method's projection.

The price discounts growth (89.1%/year) above our base case (15.0%/year) → it is priced for a demanding scenario and leaves little cushion against a slowdown.

That growth implies ~$13.7 bn of owner earnings in year 5 (vs ~$1.1 bn of our base case). It recalculates if the DCF assumptions are edited.

Year-by-year projection of the selected scenario. From each year, two versions of the flow are derived: growth FCF (operating flow − total capex, the cash surplus) and maintenance FCF (the owner earnings: what the business yields if it only sustains its capacity). The flow is returned almost in full (dividend + buyback) or redeployed into the operation, so that EV stays roughly flat and multiples compress because the metric grows, not because of cash accumulation. The valuation is done on EV/NOPAT. In edit mode, revenue, margins, capex, and exit multiples can be adjusted.

Arm is a capital-light semiconductor intellectual property licensing business with two revenue engines: licenses (payment for design access, recognized upfront or over the contract term depending on the agreement type) and royalties (a percentage or flat amount per chip shipped, which accumulates for years over the installed base). Year 0 is the TTM closed on June 30, 2026 (rolling grid, §5 R1 bis), because Arm does tag its interim results via 6-K with XBRL: revenue US$5.156bn, operating income US$0.877bn (margin 17.0%), net income US$1.044bn. The valuation metric is EV/NOPAT: GAAP operating income already expenses stock-based compensation (22.4% of revenue, 76.6% of reported free cash flow — directive of the initial classification), so no further adjustment is needed to avoid giving away that cost to the shareholder. Balance sheet invested capital (equity US$8.630bn + debt 0 − cash and short-term equivalents US$3.888bn = US$4.742bn) gives a ROIC of 14.4% on TTM NOPAT (US$0.685bn) — above the 10% bar, in the 'good' band (10-15%), without reaching 'excellent'. The operating margin has been under sustained compression (FY23 25.0% → FY25 20.7% → FY26 18.3% → TTM 17.0%, with the most recent quarter at only 7.1%), not from a one-off but from deliberate R&D investment (+29% year-over-year in the quarter) for the entry into production silicon (Arm AGI CPU, announced March 2026) and the integration of DreamBig Semiconductor (acquired for US$265M, closed July 1, 2026). The base case models a gradual margin recovery (17.0%→22.5% by year 5) supported by the Armv9 royalty mix (higher rate per chip) and by revenue growth beginning to outpace operating expense growth. Revenue growth starts at ~20% in year 1 (below the TTM's 25.1%, which is inflated by the one-time doubling of Arm China's licenses — +141% in related-party licenses, attributed by the company's own note to the timing and size of multiple high-value license agreements, a one-time, non-recurring recognition pattern) and decelerates smoothly to 10% by year 5, within the 'quality compounder with runway' band (8-15%, §5 R2). The 6-K dated July 29, 2026 (quarter closed June 30, 2026) carries no numerical revenue or earnings-per-share guidance for the full FY2027 fiscal year — the term for earnings guidance was explicitly searched for and does not appear, consistent with Arm's historical practice of not giving formal guidance —, so the path anchors on the three-year trajectory rather than on guidance. Remaining performance obligations (RPO, license portion only — royalty is excluded by the company's own policy) total US$2.1226bn as of June 30, 2026, with 24% expected to be recognized within the next 12 months and 23% in the following 13-24 months: it is a single-point-in-time figure (with no historical series in the factsPack), so it is disclosed as qualitative support for license growth and not used as a quantitative anchor for the path. With no debt (Arm carries no financial debt) and no share buybacks (there is no program; the diluted count has been growing slightly from the net issuance of RSUs — 1,063M FY25 → 1,068M FY26 → 1,078M TTM), excess cash does not accumulate: it is redeployed into R&D, M&A (DreamBig), and equity investments (stakes in Ampere Computing, Raspberry Pi Holdings), so the model uses reinvestsExcess:true (cash stays ~stable, analogous to the method's growth reinvestor) and does not carry a sharesPath (dilution from stock-based compensation is already expensed in the metric — modeling it in the share count as well would be double-counting, §6). The base exit multiple (21x) falls within the semiconductor archetype band [18,25]x, in a mid-to-high position that reflects the wide moat (switching costs, scale of the licensing model, patents) tempered by the declared eroding direction (RISC-V competition, even among Arm's own licensees, and customers designing their own silicon) and by a mid-range terminal growth rate (10%) within the compounder band.

Today's elevated multiple is the price of growth: if the business grows, the entry point cheapens on its own going forward (the metric grows while EV stays roughly flat). The exit multiple at 3 years is higher than the terminal at 5 years —at 3 years there is more growth still ahead—, so the value curve shows whether value creation is concentrated in the early or the later years. The required return is applied to the base scenario.

They are valued separately, with their own rationale, and are not incorporated into the base or the verdict (they are excess return). When assigning them value — in Editmode —, the total with optionalities updates live, without moving the base.

The verdict, the base CAGR, and the margin of safety are always calculated on the base; optionalities do not alter them (with optionalities at $0 they do not move).

Each card fixes a required annual return and answers: if the business is worth $38 in 5 years, what is the maximum that can be paid today to obtain that return? Since it now trades at $252, the margin of safety is how much cheaper the market is than that maximum. The three thresholds: 4% covers inflation (the floor), 10% is the long-term average return, and 15% is the level of a great investment.

With a target price of $38 in 5 years and a required return of 4.5% annually, the maximum to pay today is $31. Against the current market price ($252), the margin of safety is -718.1% (trades above the maximum → a premium is paid) and the total return at that price would be -31.4% annually.

• Entry multiple today. ~53x TTM revenue and ~400x TTM NOPAT — extremely rich even for an IP licensor with a wide moat.
• Exit multiple vs archetype band. The base case uses 21x EV/NOPAT, within the semiconductor band [18,25]x.
• CAGR at market price (base case). -31% — well below the 10% hurdle.
• CAGR in the favorable scenario. -31% — even with generous growth and margins, the 5-year value remains well below the current price.
• Margin of safety (Graham, today). -1220% — deeply negative under the discount at the floored risk-free rate.
• Reverse approach (reverse DCF). The growth implied by the current price far exceeds what the disciplined favorable scenario can sustain, evidence that the price is paying for exceptional execution expectations.
• Decline from the 52-week high. A 43% decline is not enough to close the gap between the price and a disciplined 5-year projection.

Each margin over sales, year by year: historical (solid line) → projection (dotted).

• ✓ ROIC exceeds the cost of capital (~10%)
• ✓ CFROIC backs up the ROIC (83%, cash vs. accruals)
• ✓ Healthy balance sheet (low corporate debt)
• ✓ Durable competitive moat (multiple advantages)

• ROIC vs the 10% bar. 14.4% on the balance sheet's real invested capital (equity + debt − cash) — in the 'good' band (10-15%).
• Invested capital. US$4.742bn (equity US$8.630bn, zero debt, cash and short-term equivalents US$3.888bn); a capital-light business with no external funding.
• Cash generation. TTM operating cash flow of US$2.094bn, with owner earnings (NOPAT + D&A − maintenance capex) of ~US$0.567bn after charging growth capex.
• Earnings quality. Net income includes non-operating gains/losses from Arm's equity investments (Ampere, Raspberry Pi Holdings), a volatile component unrelated to the licensing business.
• Operating margin. Under sustained compression for three and a half years (25%→17%), from deliberate R&D investment, not accruals or accounting distortion.
• Reinvestment runway. The AGI CPU and DreamBig open new reinvestment runways, but without evidence yet of a return on the capital invested in them.
• Working-capital superpower. The business does not depend on negative working capital or float; its capital advantage is the scale of the shared-cost model, not third-party funding.

Annual levels from the official filing (10-K); the % over each bar is the year-over-year (YoY) growth vs the prior year.

Revenue breaks down into license (payment for IP access) and royalty (recurring charge per chip shipped), with royalty as the more organic 53% and license as the 47% more subject to the lumpiness of one-time high-value agreements. The backlog of remaining performance obligations (RPO, license portion only — royalty is excluded by the company's own policy) totals US$2.1226bn as of June 30, 2026, with 24% expected to be recognized within the next 12 months and 23% in the following 13-24 months — a single-point-in-time figure in the factsPack, with no historical series, so it is disclosed as qualitative support for contracted license growth rather than a quantitative anchor for the path.

The % are the annual (year-over-year) growth: each year —historical and projected— vs the prior one; the TTM (trailing 12m) vs the TTM of a year ago, to avoid overlapping windows. The historicals are exact figures from the official filings; the projected years come from the year-by-year model (the intermediate years without their own series are interpolated between the anchors). The projected columns (+1y…+5y) are 12-month windows counted from the TTM close (30-jun-2026): the projection starts from the most recently reported data, not the fiscal year. The projected base is realistic and unbiased — the risk discount is applied at the end, via the required return. The rationale for each metric is in the (i).

• FY2026 revenue growth. Revenue +23% year-over-year, with both royalty (+21%) and license (+25%) revenue growing robustly.
• Growth composition. The 82% increase in related-party revenue is concentrated in one-time licenses with Arm China, not recurring volume.
• RPO / contracted backlog. US$2.1226bn in remaining performance obligations (license only), with 47% expected to be recognized within the next 24 months.
• TAM expansion (physical AI, networking). The Arm AGI CPU and the DreamBig acquisition extend the platform into physical computing and datacenter networking, without material revenue yet.
• Formal guidance. The company does not publish numerical revenue or earnings-per-share guidance for the current fiscal year.
• Geographic dependence on China. ~18% of revenue comes from the People's Republic of China (including Arm China), exposed to growing export controls.
• Internationalization. 70% of the quarter's revenue comes from outside the U.S., with diversified exposure across Japan, Taiwan, South Korea, and other markets.

• Revenue recurrence. The royalty is charged for years on chips already shipped, providing long-term visibility independent of new license sales.
• Differentiated product vs commodity. Arm's IP is not interchangeable: the design and its associated software ecosystem are architecture-specific.
• Pricing power. The Armv9 royalty mix (with a higher rate per chip) raises revenue per unit without needing higher volume, evidence of real pricing power.
• Operating leverage. Operating leverage is currently reversed: R&D spending is growing faster than revenue due to investment in production silicon.
• Customer concentration. The top five customers (including Arm China and SoftBank) are ~57% of revenue; Arm China alone ~16%.
• Recession behavior. The semiconductor cycle is cyclical by nature, but Arm's royalty cushions part of the impact by being charged on chip volume across multiple verticals (mobile, IoT, automotive, cloud).
• Capital cycle / competitive entry. RISC-V, an open, free architecture, is being actively driven by several of Arm's own customers, the most direct way capital is entering to compete against the existing moat.

• ✓Leverage (net debt / EBITDA)Net cash $3.9 bn
• –Interest coverage (EBIT / interest)no data
• –Liquidity (current ratio)no data
• ✓Cash quality (CFROIC vs ROIC)CFROIC backs 83% of ROIC
• !Value creation (ROIC − 10% bar)+4pp
• !Malinvestment test (capex vs incremental ROIC)Capex/D&A 2.2x
• –Float / working capitalNeutral WC
• ✕Dilution (SBC % of revenue + shares)SBC 22.4% of revenue

A traffic-light interpreted by the method (not generic): float (negative WC) adds up, capex is judged by incremental ROIC (malinvestment test), and a lender is not subjected to corporate solvency. The (i) shows the derivation of each number.

• Debt. Zero financial debt; Arm is an unlevered business.
• Cash and liquidity. US$3.888bn in cash and short-term equivalents as of June 30, 2026, sufficient for more than 12 months of operations according to the company itself.
• Working capital. Accounts receivable and contract assets moved favorably in the most recent quarter, releasing cash.
• Legal contingencies. Corporate governance litigation by Arm China's former CEO against its own management could lead to changes Arm does not control.
• Export regulatory risk. U.S./U.K. export controls on advanced compute processors to the People's Republic of China already require licenses for certain designs.
• Fixed-charge coverage. TTM operating income (US$0.877bn) comfortably covers financial expense, which is practically nil given the absence of debt.

Who owns the shares — the alignment and whether there is a controlling shareholder.

• Controlling shareholder alignment. SoftBank Group controls ~86.4% of the ordinary shares, long-term alignment with governance risk disclosed in the company's own risk factors.
• Capital allocation. 100% of free cash flow is redeployed into R&D, growth capex, and complementary M&A (DreamBig); no dividend or buyback.
• Candor about setbacks. The company explicitly discloses the margin compression and its cause (R&D investment for production silicon) in its own results discussion.
• Compensation and dilution. Stock-based compensation is 22.4% of revenue and 76.6% of reported free cash flow, a high level that requires expensing it in the valuation metric.
• Conflict of interest from the CEO's dual role. Rene Haas took on the role of CEO of SoftBank Group International in April 2026, in addition to CEO of Arm, which raises a governance question about dedication and potential conflict of interest.

• The 43% decline from the 52-week high coincides with SoftBank's announcement of a partial sale of its majority stake, which introduced uncertainty about the controlling shareholder's commitment and created technical supply pressure on the stock.
• Licensing guidance for FY2027 decelerated versus the pace of FY2026, which the market read as a sign of exhaustion in the cycle of high-value licenses with Arm China.
• The compressed operating margin (17.0% TTM, 7.1% in the most recent quarter) creates the perception that investment in production silicon is destroying profitability without a proven return yet.

None of these three reasons alone sustains a classic 'missing buyers' or 'motivated sellers' thesis: they are real business risks, not short-term noise disconnected from value. Even with the 43% decline, the current price still demands execution far superior to what a disciplined 5-year projection can sustain —even the favorable scenario leaves a value well below today's price— so the honest conclusion is that Arm is not cheap: it trades expensive even after the correction.

The annual return (CAGR at 5 years) in each scenario, with the total period return below — the margin of safety made visual: upside range wide, downside range narrow.

Even in the bear scenario, the return holds at -43%/year (-94% total): the margin of safety protects the downside. The bull (-19%/year, -66% total) exceeds it comfortably — a favorable asymmetry, with a narrow downside range and a wide upside range.

• RISC-V adoption among Arm customers. Several Arm customers actively drive RISC-V, including a 2023 automotive joint venture; they could migrate volume away from the Arm architecture.
• Customers designing their own processors. Key customers could opt for pure architecture licenses and design their own core, drastically reducing the fees they pay Arm.
• Concentration in Arm China. ~16% of revenue via an exclusive IPLA with an entity Arm does not control and from which it has historically had difficulty obtaining timely financial information.
• U.S./U.K.-China export controls. Restrictions on advanced compute processors to the People's Republic of China could tighten and materially affect a business that has not yet suffered that effect.
• Execution in production silicon. The entry into finished-chip business (AGI CPU, DreamBig) brings new manufacturing, supply-chain, and working-capital risks foreign to the historical pure-licensing model.
• Controlling shareholder conflict of interest. SoftBank Group, with ~86.4% of ordinary shares and indirect participation in Arm China, has interests that the filing itself acknowledges may not align with those of minority holders.

Each thinker's analytical framework applied to our data.

• Buffett / Graham Quality + margin of safety

  Fails the quality gate: wide moat, eroding.

• Peter Lynch Growth at a reasonable price (GARP)

  A fast grower growing 15% at a multiple/growth of 26.1 → expensive for its growth.

• Joel Greenblatt Cheap and high-return (Magic Formula)

  Earnings yield 0% (EBIT/EV) + ROIC 14% → falls outside the Magic Formula.

• Howard Marks Perception vs reality + cycle

  The price discounts 89% vs our 15%: priced for perfection, no favorable mispricing.

• Seth Klarman Capital protection (bear scenario)

  Bear-scenario floor -43%/yr over 5y (material loss) → risk of permanent capital loss.

• Pat Dorsey Moat strength (Five Rules)

  A wide moat, eroding; sources: switching costs, efficient scale, network effects, intangibles, cost advantage → partially passes the Five Rules.

• Aswath Damodaran Expectations implied by the price

  The price demands 89% growth, far above our 15% — heroic assumptions.