ARM — FY2026 (Year Ended 31 March 2026) Financial Report Analysis

Revenue up 23%, operating profit up 8% — and 61% of the growth came from Arm's own parent

Arm Holdings plc reported revenue of $4,920 million for fiscal 2026, the year ended 31 March 2026 (Arm's own label; its financial year runs April to March), up 22.8% from $4,007 million. That is the headline. The composition underneath it is the story.

Of the $913 million of additional revenue, $559 million — 61% — came from a single licensing-and-consulting arrangement with an affiliate of SoftBank Group, Arm's controlling shareholder. Revenue from that arrangement was $704.4 million in fiscal 2026 against $145.5 million a year earlier (Note 20, Related Party Transactions). Over the same period, licensing revenue from ordinary third-party customers fell 9%, from $1,421 million to $1,298 million.

Meanwhile operating income — profit from running the business, before investment gains, interest and tax — rose only 8.3%, to $900 million, because research and development spending grew 34%.

Source: Consolidated Income Statements and Notes 20 and 21, Arm Holdings plc Form 20-F for fiscal year ended 31 March 2026. Figures in millions of US dollars.

Two revenue lines, two very different businesses

Arm licenses chip designs. It earns money twice: an upfront licence fee when a customer buys the right to use a design, and then a royalty — a small per-chip fee — on every chip shipped that contains that design. Royalties are the recurring, annuity-like half; licence fees are lumpy and depend on when large contracts happen to be signed.

Royalties are the clean growth story. Royalty revenue rose 20.5% to $2,613 million, and royalties from external customers alone rose 20.4% to $2,123 million. Management attributes this to "an improved mix of products with higher royalty rates per chip, such as Armv9 technology" — that is, Arm is being paid more per chip, not merely benefiting from more chips shipping. This is genuine pricing power with arm's-length customers, and it is the most durable thing in the results. Cumulative Arm-based chips shipped passed 350 billion as of 31 March 2026. Mobile application processors — the main chip in a smartphone — still generated roughly 43% of royalty revenue, so the royalty line remains tied to the smartphone cycle.

Licensing is where the concentration risk sits. Total licence revenue grew 25.4%, which reads well. But external licence revenue declined 9%. The entire increase, and more, came from related parties: licence and other revenue from related parties rose 141%, from $418 million to $1,009 million. Arm discloses two related-party sources:

• Arm China (formally, revenue under the IP licence agreement with Arm Technology China): $790.6 million, up from $670.4 million — a respectable 17.9% increase, and a long-standing arrangement.
• An affiliate of SoftBank Group, under a consulting agreement: $704.4 million, up from $145.5 million.

That second line went from immaterial to 14% of total revenue in one year. Arm also disclosed it is "engaged to advise on or design chips for certain existing customers and other third parties, including affiliates of SoftBank Group."

Takeaway: Strip out the SoftBank-affiliate arrangement and Arm's revenue grew roughly 9%, not 23% — and its third-party licensing business shrank. The recurring royalty engine is genuinely strong at +20% on better Armv9 pricing, but the headline growth rate is being set by a contract with Arm's own controlling shareholder, which is not the same thing as demand from the open market and should not be capitalised into a growth multiple as though it were.

Costs: the AI build-out is real money

Research and development rose $705 million, or 34%, to $2,776 million — now 56% of revenue. Arm names the cause directly: "investments in next generation products, such as the Arm AGI CPU," alongside headcount growth, share-based compensation, cloud and IT costs, and engineering expense. Selling, general and administrative costs rose 13% to $1,115 million. Cost of sales was flat at $121 million, which is what a licensing model looks like — Arm does not manufacture, so gross margin is near 98%.

The result: revenue grew 22.8% while total operating expenses grew 27.6%, and operating margin fell 2.4 percentage points to 18.3%. Operating margin is the share of revenue left after paying to run the business, before interest and tax. Arm is deliberately spending ahead of a product transition, which is defensible — but it means the current year's profit growth is not a guide to the underlying operating leverage in either direction.

Why net income growth understates the year — and why EPS flatters it

Net income rose 14.1% to $904 million and diluted EPS rose 13.3% to $0.85. Both figures sit awkwardly between the 22.8% revenue growth and the 8.3% operating income growth, and neither is a clean read, because three items below the operating line moved sharply:

• Tax swung by $325 million. Fiscal 2025 carried a tax benefit of $72 million (an effective rate of −10.0%); fiscal 2026 carried a tax expense of $253 million (21.9%), driven by changes in withholding tax, unrecognised tax benefits and share-based compensation tax effects. Pre-tax income actually rose 60.7%, from $720 million to $1,157 million. The 14% net income growth is almost entirely a tax-comparison artefact.
• A one-off disposal gain. Other non-operating income was $132 million versus $10 million, including a $131.0 million pre-tax gain on the sale of the Artisan foundation IP business to Cadence Design Systems, completed 26 August 2025. This will not repeat.
• A prior-year investment writedown reversed out of the comparison. Income from equity investments was +$14 million versus −$237 million, the prior-year figure having been driven by a $246 million fair-value loss on Ampere Computing. SoftBank Group acquired Ampere outright in November 2025.

Adjusting for the Artisan gain and the equity-investment swing, pre-tax income from continuing operations grew far less than 61%. The honest summary is that fiscal 2026 operating profit grew single digits and the reported bottom line was flattered by disposals and depressed by a normalising tax rate at the same time.

Stock-based compensation exceeds operating profit

Arm's share-based compensation cost was $1,052 million in fiscal 2026, up 28.3%, plus a further $160 million of employer payroll taxes on it. That is 21.4% of revenue, and more than the $900 million of GAAP operating income the company reported. Put plainly: Arm paid its employees more in shares during the year than the accounting profit the business generated.

This matters for anyone comparing Arm's GAAP results to an "adjusted" or non-GAAP profit figure. Adding back $1.05 billion of share compensation roughly triples apparent operating profit, but the cost is real — it is paid in newly issued shares, which dilutes existing holders. Diluted share count rose from 1,065 million to 1,078 million between the June 2025 and June 2026 quarters, so dilution is running near 1% a year; the effect is muted relative to the expense, but it is not zero.

Cash generation was strong regardless: net cash from operating activities was $1,524 million, nearly four times the prior year's $397 million, helped by the non-cash nature of that share compensation. Arm closed the year with $2,751 million of cash and equivalents plus $850 million of short-term investments, against no meaningful debt.

Foreign exchange was not a factor. This is worth stating explicitly for a UK-incorporated company: less than 2% of Arm's revenue is denominated in currencies other than the US dollar, and Arm describes the FX impact on fiscal 2026 revenue as immaterial. None of the growth above is a currency effect. Revenue from customers outside the US was 64%, up from 57%.

The strategic change: Arm is becoming a chip company

In March 2026 Arm announced it is expanding "into production silicon products with the Arm AGI CPU," with production expected by the end of calendar 2026. It had no material revenue effect in fiscal 2026 but it is the largest change to Arm's business model since its IPO.

Until now Arm sold designs to companies that build chips. Selling finished chips puts Arm into competition with some of those same customers — a risk Arm states itself: "some of our customers may face direct competition from us in silicon production products, such as with the Arm AGI CPU." It also changes the financial shape of the business, since silicon carries manufacturing cost and inventory risk against Arm's current near-98% gross margin. The $265 million agreed acquisition of DreamBig Semiconductor (networking technology, announced October 2025, expected to close by the end of Arm's fiscal second quarter of FY2027) points the same direction.

The 34% R&D increase is the cost of this shift showing up before any of the revenue does.

Forward read

Arm does not publish formal guidance in its 20-F. But the first quarter of fiscal 2027 has already been reported, and it sharpens the picture (Form 6-K furnished 29 July 2026, for the quarter ended 30 June 2026):

The pattern from fiscal 2026 has intensified. Revenue growth held above 22% and royalties accelerated slightly — the good news, and evidence the Armv9 pricing effect is continuing. But operating income fell 20% year over year and operating margin dropped to 7.1%, as R&D climbed another 29%. The doubling of net income to $270 million is not operational: it came from $128 million of gains on equity investments and a $17 million tax benefit, against an operating result that went backwards.

Remaining performance obligations — contracted revenue not yet recognised, excluding future royalties — were $2,122.6 million at 30 June 2026, up modestly from $2,071.4 million at 31 March 2026. Arm expects to recognise about 24% of that within twelve months, down from 28% on the March balance, meaning the contracted backlog is converting to revenue slightly more slowly.

My read: the royalty business is performing, and a 20%+ increase driven by rate per chip rather than unit volume is the highest-quality revenue Arm produces. Two things should be watched rather than assumed. First, the SoftBank-affiliate arrangement that supplied 61% of fiscal 2026's revenue growth is a contract, not a market — if it does not repeat at $704 million, fiscal 2027 growth reverts toward the underlying high-single-digit to low-teens rate that third-party revenue is actually running at, and the declining external licence line offers no cushion. Second, operating margin has now compressed for two consecutive reported periods and is at 7.1% in the most recent quarter; the AGI CPU spending is front-loaded by design, but Arm has not yet shown revenue on the other side of it. The company is asking to be judged on a product that reaches production at the end of calendar 2026. Until it does, the reported growth rate and the reported profitability are pointing in opposite directions, and the gap between them is being filled by the parent company.