Arm Holdings (ARM): A Record Quarter, a 50% Crash — And Still 970x Owner Cash Flow

Arm Holdings plc (ARM) Q1 FY2027 — Arm Holdings (ARM), the UK-based company whose CPU architecture sits inside almost every smartphone on earth and a fast-growing share of AI data centers, reported fiscal Q1 2027 (quarter ended June 30, 2026) after the close on July 29: revenue rose 22% YoY to a record $1,289M, beating the $1.26B guide; royalty revenue +22% to $715M with data-center royalties more than doubling; license and other revenue +23% to $574M. Non-GAAP EPS of $0.45 (+29% YoY) beat both the $0.40 guide and the $0.40 Street estimate, and non-GAAP operating margin rose to 41.2%. Management raised Q2 guidance to $1.38B +/- $50M (Street ~$1.34B) and non-GAAP EPS to $0.47 +/- $0.04 (Street ~$0.43). But the GAAP books tell a different story: GAAP operating income was just $91M and the GAAP operating margin FELL to 7.1% from 10.8%, because Arm expensed $343M of share-based compensation plus $90M of employer taxes — 33.6% of revenue — in a single quarter. Of $270M GAAP net income, $128M was a non-cash equity-investment mark-up and $17M a tax benefit. Arm touts $1,397M of TTM non-GAAP free cash flow (+134%), but that treats $1,154M of TTM stock comp as free; subtract it and true owner free cash flow is roughly $243M, and on that basis Arm's FCF was negative in FY2024, FY2025 and FY2026. Annualized contract value rose only 13% versus reported licensing +23%, and Arm quietly stopped disclosing remaining performance obligations, Access licence counts and chip unit volumes. 30% of revenue ($388M) is related-party, principally Arm China. The stock closed at $224.89 on July 29 (down 8.1% on the day, pre-print) after a 50% collapse from its $452.70 June high, and traded near $225.84 after hours — essentially flat on a beat and raise. Our owner-earnings DCF with SBC honestly expensed lands at $66 (bear $28 / bull $105); the reverse DCF says today's price demands 37%/yr FCF growth for a decade on the company's own numbers, or 56%/yr with SBC expensed. Our call: SELL, 4/5 — a genuinely great business at a price we cannot defend. Wall Street is a Buy with a $318 average target (+41%), so we DIFFER, dramatically. Arm Holdings (ARM) just did something unusual: it printed a record quarter, beat on both lines, raised guidance — and the stock barely moved, because it had already fallen 50% in six weeks. Arm is the UK-domiciled company that designs the CPU architecture inside almost every phone on earth and, increasingly, inside every AI data center; it licenses that IP up front and then collects a royalty on every chip that ships, forever. Fiscal Q1 2027 (ended June 30, 2026): revenue +22% YoY to a record $1,289M, above the $1.26B guide; royalty revenue +22% to $715M with data-center royalties more than doubling and Neoverse passing 1.5 billion cumulative cores (the first billion took six years, the most recent 500 million took nine months); license and other revenue +23% to $574M. Non-GAAP EPS $0.45 beat the $0.40 guide and the Street's $0.40, non-GAAP operating margin improved to 41.2%, and Q2 guidance was RAISED to $1.38B +/- $50M and $0.47 +/- $0.04 — both above consensus. Demand for the brand-new Arm AGI CPU (Arm's own production silicon, launched in March) now exceeds $2B across FY27-FY28 versus the $1B management guided in May, with foundry capacity, not orders, as the binding constraint. So why are we cautious? Because of the second set of books. GAAP operating income was only $91M and the GAAP operating margin actually FELL, from 10.8% to 7.1% — Arm expensed $343M of share-based compensation plus $90M of employer taxes, 33.6% of revenue, and non-GAAP adds every dollar back. Of $270M GAAP net income, $128M was a non-cash mark-up on equity investments and $17M a tax benefit. Arm reports $1,397M of TTM non-GAAP free cash flow, up 134% — but that treats $1,154M of stock handed to employees over the same twelve months as costless; diluted shares rose to 1,078M. Subtract it and true owner free cash flow is about $243M, and by fiscal year, after expensing stock comp, Arm's free cash flow was negative in FY2024, FY2025 AND FY2026. Three more things almost nobody mentions: annualized contract value grew only 13% while reported licensing grew 23%, so this quarter's licensing line borrowed from the future; Arm stopped reporting remaining performance obligations, Access licence counts and chip unit volumes, so nobody outside can verify the 'higher royalty rate per chip' claim; and $388M — 30% of all revenue — is related-party revenue, principally from Arm China, an entity Arm does not control. Add the strategic tension of Arm now selling silicon in competition with the licensees who pay it royalties (Qualcomm, a former litigation opponent, just announced its own Arm-based data-center CPU), a 98% gross margin that cannot survive a hardware mix shift, capex already running $197M a quarter (+28%), RISC-V at the low end, and SoftBank's ~88% stake leaving a ~12% float on a $240B company. Our owner-earnings DCF, with SBC expensed as the real cost it is (22%-to-13% revenue CAGR, owner cash margin scaling from 4.7% to 26%, 9.5% discount, 3.5% terminal growth), lands at $66 versus $225.84 after hours — 71% below. Even a deliberately generous bull case reaches only ~$105, which is BELOW the market price. Inverted, the reverse DCF says today's price requires ~37% annual free-cash-flow growth for ten straight years on the company's own definition, or ~56% with stock comp expensed; revenue grew 22%. Our call: SELL, 4/5 — the franchise is an A, the price is indefensible, and a 50% crash took Arm from absurd to merely very expensive. Wall Street is a Buy (19 buy / 6 hold / 2 sell) with a fresh $318 average target implying +41%, so we DIFFER dramatically. Watch owner free cash flow, not the non-GAAP headline. Not financial advice. THE CALL: SELL (4/5, A GENUINELY GREAT BUSINESS AT A PRICE WE CANNOT DEFEND — EVEN AFTER A 50% CRASH) — base-case value ~$66.00 vs ~$225.84 today. What to watch: share-based compensation falling from 27% of revenue toward the low teens while revenue keeps compounding above 20% would close the gap between reported and owner free cash flow faster than anything else in the model, and would move our fair value more than any other single assumption — that, plus a price somewhere between $100 and $130, would flip us constructive; the risks to respect run the other way — the blended gross margin compressing as Arm AGI CPU silicon mixes into a business that has never paid a foundry, capex and inventory building, annualized contract value continuing to lag reported licensing, further reductions in disclosure, or any SoftBank sell-down into a float of only about 12% Also on YouTube: @ChargedAlpha DISCLAIMER: For informational and educational purposes only. Not financial advice. Do your own research before any investment decision.