Lululemon Athletica

Lululemon sells expensive stretchy clothes to people who exercise, and to a larger number of people who would like to look as if they do. That is not a sneer; it is the whole business, and for twenty years it was a magnificent one. A pair of Align leggings lists at up to $128, the fabric is proprietary, the stores are the company's own, there is almost no wholesale, and the gross margin — 56% on the trailing twelve months — is the kind of number a luxury house would recognise. Revenue went from $2.3 billion to $11.1 billion in the decade to FY2025, earnings per share from $2.21 to $13.26, and the share count fell 13% while it happened. Readers who want the mechanics of a brand moat could do worse than study those ten years.

Then the second quarter of fiscal 2026 arrived on 3 September, and the honest description of a share of LULU today is this: a debt-free, cash-rich, vertically integrated retailer of a single brand whose core product just stopped selling. Comparable sales fell 9%. In the Americas, two-thirds of the business, they fell 12%. Women's leggings — the franchise, the reason the company exists — fell roughly 20% in the quarter (approx. — earnings call). China, the growth engine that grew 30% in the first quarter, grew 4% in the second and went negative in constant currency. Guidance was cut for the third time in six months. The stock fell 17% the next day, to its lowest since 2018, and it has kept falling: $96.13 as we write, roughly 81% below its December 2023 peak (approx.).

The question this report exists to answer is the one Buffett got wrong on Dexter Shoe and got right on See's Candies: is this a durable consumer franchise having a bad year, or a fashion cycle that wore the costume of a moat? We asked the same question about Nike in July and answered "watch the turn". Lululemon is the purer test — one brand, one category, no wholesale to hide behind — and, as we will show, the harder one.

Two lessons for an owner. First, the growth was real and the plan was not: the company that promised $12.5 billion for this year will report about $10.4 billion, and the gap is not a recession — it is the difference between a brand that was winning and one that is now defending. Second, founders who leave rarely leave. Wilson has spent thirteen years outside the building and still picked two of eleven directors this summer; the executive chair, an Advent-era director who left after 21 years, a Levi's CEO, an ESPN marketer, the former co-CEO of a competitor, a $1 billion activist with its own CEO pick, and a muzzled founder are all now supervising a chief executive whom none of them chose with enthusiasm. That is a board of many fathers, and it is Part VIII's problem.

This is the exact trap Buffett described after Dexter Shoe: "what I had assessed as durable competitive advantage vanished within a few years." He paid with Berkshire stock for a shoe company whose moat turned out to be a cash-flow forecast, and he later called it the most gruesome deal he ever made. The lesson is not "never buy apparel" — Berkshire owned Nike for years and owns Brooks Running today. The lesson is that a brand's moat has to be proven every season, and the burden of proof rises the moment the volume response to price turns negative. It just did.

FY2025 revenue (year to 1 February 2026), by product and by channel, from our statements feed and the 10-K:

Geography, and a feed fault to name. The Americas were 71% of FY2025 revenue ($7.8B: United States $6.3B, Canada $1.4B, Mexico $0.1B), China Mainland 16% ($1.75B) and the rest of the world 13% ($1.5B) (10-K). Our geographic feed prints China at $3.7 billion — more than double the filed figure, almost certainly a double-counted row — so we use the 10-K and flag the feed, as we did for Chubb's segment data. The mix matters because the two largest markets are the two in trouble: Americas comps −12% and China comps −8% in constant currency in Q2, while the 13% "rest of world" grew a healthy mid-single digits.

Why the model cuts both ways. Yesterday we published McDonald's, where U.S. visits fell 4.5% and corporate operating income rose 3%, because the landlord collects a percentage of the tenants' sales and owns no kitchens. Lululemon is the opposite architecture: it owns every kitchen. Vertical retail gives it a 56% gross margin and full-price control when the product is winning; when the product is losing, every store lease, every sales associate and every unsold legging is the company's own problem. That is why a 4% revenue decline became a 13% operating-income decline in Q2 and why Q3 is guided to an operating margin near 6.5%, against 17% a year ago (approx. — call). Jefferies put it in one sentence: the fixed cost base is too big for a shrinking top line. Management's own words were "we are taking a deeper look to rightsize the cost base."

The tariff seam, which flattered the quarter. Lululemon owns no factories; 40% of product is made in Vietnam. Tariffs cost $275M gross in FY2025 and were planned at $380M for FY2026 — until the Supreme Court struck down the IEEPA duties in February and refunds began. $134.5M of refunds landed in Q2, adding 560 basis points to gross margin and $0.86 to EPS. Strip it out and the reported $2.92 becomes about $2.06 — still above guidance, but a 34% decline rather than a 6% one, and the underlying gross margin fell ~360 basis points rather than rising 200. The replacement Section 301 duties (12.5% on Vietnam, uncapped, no sunset) are permanent. Nike's Q4 had the same refund-inflated margin; both "beats" were the government giving money back.

Buffett's practical test for a franchise is the ten-percent question: raise prices ten percent and hold; if customers largely stay, you have a franchise; if they flee, you have a commodity. Lululemon raised prices in 2025 and lost a fifth of its legging volume in 2026. That is not a commodity result — the gross margin is still 56%, and nobody sells a $128 commodity — but it is no longer an unambiguous franchise result either. Coca-Cola raised prices through 2023 and kept its volume; Nike raised prices and lost its running share to On and Hoka; Lululemon has just joined the second group. Score: 5 — a real brand, with real fabric and real stores, whose customers have started to re-decide every season. The moat is exactly as wide as the next product cycle.

Every fallen angel on our board carries one question the price turns on. For Nike it was whether a self-inflicted strategy wound could self-heal. For Lululemon it is starker, because nothing here was a strategy: did the customer leave because the product was wrong for two seasons, or because the brand's moment passed?

The case that it is a cycle. Management attributes much of the damage to things that end: a proxy fight and a PFAS investigation that generated "negative commentary in the media and social channels", a mis-handled event in China, and — the one that matters — a product assortment that missed a shift toward looser, "away-from-body" silhouettes. The fix is under way and measurable: new styles rising from 23% to 35% of the assortment, the product cycle cut from 18–24 months toward 12–14, 20% more chase volume, and the styles that do match the trend (Groove wide-leg, Daydrift trousers) selling. Add a fortress balance sheet that can fund a two-year repair without borrowing, and a CEO who built Nike's women's business into a multi-billion division. On this reading, $96 is what Michael Burry called it on the morning of the crash: a "fat pitch".

The case that it is a peak. Americas comps have been flat or negative for more than ten straight quarters — that is not two bad seasons, it is a trend older than the tariffs, the proxy fight and the Great Wall combined. The share is going to named competitors with momentum: Alo, valued near $10 billion, and Vuori, at $5.5 billion, are exactly the "cooler, newer" brands Lululemon itself once was (approx. — press reports). The founder's diagnosis in the Wall Street Journal — that the company lost its creative edge to "technocratic MBAs" — is self-interested and also the most-quoted sentence about the company this year. And the shift the company is chasing (looser silhouettes) cannibalises the highest-margin product it owns. On this reading, the P/E of 8 is what a fashion cycle looks like on the way down, and the historical analogue is not Nike in 2024 but Under Armour in 2017.

Our read: the honest answer is that nobody knows yet, including the company, and the price is not low enough to pay us for that ignorance. What we can say is which reading each data point supports. The third guidance cut, the ten-quarter Americas trend and the share data support "peak". The balance sheet, the cash generation and the specificity of the product fix support "cycle". The tiebreaker is the one number we do not have: the first quarter under O'Neill, reported in December. A stabilising Americas comp — anything better than the guided mid-teens decline — would settle the argument for "cycle" at a price that will probably still be cheap. Another leg down would settle it the other way at a price that would look expensive in hindsight. We are content to wait for that print.

The pattern is the uncomfortable one for a brand investor: the share is not going to the big, dumb, cheap competitor — it is going to two small, expensive, cool ones. If Uniqlo were taking the volume, this would be a value story and pricing power would be the diagnosis. Alo and Vuori taking it at similar prices means the diagnosis is desirability, and desirability is the one thing a balance sheet cannot buy. The consolation is that both challengers are private, roughly a tenth of Lululemon's size, and have not yet proved they can run 800 stores at a 56% gross margin without the same fashion risk. Lululemon proved that for fifteen years. It now has to prove it again.

Integrity: no veto. The company has been candid about its problems, sometimes to a fault, and there is no Easterbrook here. Capital allocation is where the record is mixed and Buffett's rule bites hardest: what is smart at one price is stupid at another. Lululemon has repurchased more than $5.5 billion of stock since FY2015 — $1.6 billion in FY2024 and $1.2 billion in FY2025, the latter above that year's free cash flow of $922 million — at prices between $300 and $500. The share count fell 13% in a decade despite that outlay, because so much of it went to offsetting stock compensation and to buying at the top. Mirror ($500M in, $443M written off) is the one large acquisition, and it was a mistake. Now, at $96, with $712 million of authorisation left and guidance to match last year's ~$1.2 billion, the buyback is finally cheap: executed at these prices it would retire roughly 11% of the company in a year (approx.). That would be the first genuinely accretive repurchase in the company's history — if the cash is still there after the cost cuts and the tariffs. Dial: 5. A new CEO we cannot score, a board assembled by three factions, a buyback record that was expensive when it was easy and is now cheap when it is hard.

Read the balance sheet first, because it is the whole bull case. Cash of $1.4 billion, no bank debt, no bonds, an undrawn $600 million revolver, a current ratio above 2, an Altman-Z near 5. The "net debt" our feed prints is $1.5 billion of store leases, which are real obligations but not the kind that force a company to do anything in a hurry. Whatever this turnaround costs — markdowns, store closures, a cost-base reset, a bad Christmas — Lululemon can pay for it out of what it has. That is the difference between a cheap stock and a value trap with a maturity wall, and it is why Financial Strength scores 9 while Growth scores 2.

Three feed faults, named. First, the DCF: our model prints an intrinsic value of $303, +216% — the most broken figure on the board since NNN's +245%. It capitalises FY2025's $13.26 of earnings as permanent while management has guided ~$9.60 and the street sees $8.67 by FY2028; a DCF built on the earnings a company is leaving is not a valuation, it is an obituary read as a forecast. Second, the estimates feed still carries FY2026 consensus EPS of $13.05 — the pre-crash number from 18 analysts who have since cut; the current guide is $9.48–9.73, and we use it. Third, the geographic segments print China at $3.7 billion against a filed $1.75 billion (Part IV). None of these is a reason to distrust the platform; each is a reason to read the notes.

No dividend to prove, so we test the buyback instead. The six questions we ask of every dividend payer have no subject here — Lululemon has never paid one — but the capital does go somewhere, and the test is Buffett's: is the repurchase price below intrinsic value? For a decade it was not, and the company bought anyway; in Q2 FY26 it bought 2.7 million shares at an average near $120 (approx. — 10-Q), its first purchases near a decade low. If the FY2026 programme runs at the guided ~$1.2 billion at prices near today's, it retires ~11% of the count — accretive for the first time. The condition is that free cash flow holds: FY2025's $922 million was already below the buyback; FY2026's, on a guided ~14.5% operating margin and $680–700 million of capex, is unlikely to cover $1.2 billion without dipping into the cash pile. That is affordable once. It is not a policy.

Here is the whole valuation in one sentence: you are paying eight times last year's earnings for a company that has told you next year's will be a quarter lower, and the analysts think the year after will be lower still. The trailing multiple is a photograph of a business that no longer exists; the forward multiple, ten to eleven times, is fair for a no-growth brand and cheap for a recovering one. Which it is depends on Part VI, and we do not know.

So we price both outcomes. If the brand stabilises — Americas comps flat by late 2027, operating margin back to the high teens, earnings of $11–12 a share — a durable premium brand with a fortress balance sheet deserves 15–18× and the stock is worth $165–215, roughly double. If it does not — a slow fade at a mid-teens margin, earnings of $8–9, share count shrinking — it is worth 10–12× plus the cash: $90–120, which is where it trades. The market, in other words, is pricing the fade and giving you the recovery free — and Buffett's margin-of-safety tiers say that for a business whose intrinsic value we cannot reliably estimate, the required discount is 40–50% or "do not invest". Forty percent off the fade value is ~$70–75. That is the price at which we would own it without needing the turnaround to arrive. Above it, we need the print.