5-1 Koishikawa
Bunkyo, Tokyo
Japan
In 2011, Patagonia ran an advertisement in the New York Times under the headline "Don't Buy This Jacket." Sales rose. The company had asked its customers for less and received more — more trust, and in consequence more business. I keep returning to this episode because it compresses a pattern I see everywhere in the hardware companies worth studying: the durable ones are defined less by what they build than by what they decline. The refusals are not a cost of the strategy. They are the strategy.
Consider the refusals at Teenage Engineering, a Stockholm company of somewhere between fifty and eighty people with revenue estimated around $25–30 million. No discounts, no sales, ever — a $59 Pocket Operator and a $1,999 OP-1 field, each priced without apology. Five years passed between the company's founding and the OP-1, and only after the product had built its authority did the collaborations begin. Even the products refuse things: the OP-1's deliberate constraints — the tiny screen, the four tracks, the limited storage — forced its users to become inventive, and the inventive users became the community that no marketing budget could have bought. The constraint is the moat.
Arc'teryx shows the same refusals holding at a scale where they should have collapsed. The company runs a no-deadline culture — products ship when they reach the standard, not when the calendar arrives, and its avalanche airbag pack took thirteen years. Its CEO's line is the purest statement of the genre: "The world does not need Arc'teryx to make athleisure products." When Off-White sent hacked-up Arc'teryx jackets down a runway, the company publicly distanced itself from the tribute. When gorpcore made it a trend, it declined to participate in its own moment — and when the trend peaked and faded, revenue kept climbing, to roughly $2.1 billion in 2024, growing 36 percent in a year, precisely because the brand had never attached itself to the wave that was receding. Its largest refusal was structural: in about four years it inverted from 80 percent wholesale to roughly three-quarters direct-to-consumer, which is to say it refused other people's shelves, and other people's ideas about what belongs on them. In Japan the discipline is priced openly — the Beta AR jacket sells for ¥110,000, above its US price, across twenty-four stores, to customers who scrutinize stitching and provenance and pay the premium because the rigor is visible in the object.
Balmuda is the control case, and it is instructive because the first act was flawless. The Toaster of 2015 — ¥25,850, five times the price of an ordinary toaster — launched with no advertising and earned three-month waiting lists on word of mouth, carried by a genuine technology: calibrated steam that produced measurably better toast. Then the refusing stopped. A kettle followed, then a speaker, a lantern, a microwave, a rice cooker; a 2020 IPO arrived at peak enthusiasm and brought the growth expectations that public markets bill for; and in 2021 came a smartphone, built with Kyocera, launched into territory held by Apple, Samsung, and Sony. The reviews were poor, the product was halted within months, and the stock crashed. Each step was individually defensible as growth. Jointly they converted "the company that reinvented toast" into a nice appliance maker — the scarcest asset spent to make the ordinary kind of revenue.
Patagonia's endgame shows how far refusal can be taken: in 2022 the Chouinard family gave the company away — 98 percent to an environmental nonprofit, the 2 percent of voting stock to a purpose trust that exists solely to keep the mission from ever being sold. Yvon Chouinard's verdict on the conventional exit was that an IPO "would have been a disaster." Even the aftermarket is a refusal: Patagonia's Worn Wear repairs gear rather than replacing it, and Arc'teryx's equivalent program repaired 28,000 garments in 2024, with technicians working behind glass where shoppers can watch — the anti-obsolescence position performed in public, and rewarded with loyalty no campaign could rent.
The mechanism under all of this is not mysterious. What a design-led hardware company actually owns is trust — the buyer's belief that everything carrying the name received the maker's full attention. Every standard growth instrument converts that trust into near-term revenue: the discount converts it into a transaction, the line extension converts it into shelf space, the trend converts it into reach, the borrowed channel converts it into volume. Each conversion looks free and compounds negatively, and the balance sheet records the revenue while the loss lands somewhere accounting cannot see. Refusal reads as leaving money on the table. It is closer to declining to sell the only asset that cannot be repurchased.
What I take from these companies is not their scale but the opposite observation: none of the refusals require scale. Declining to discount, declining the adjacent category, declining the deadline, declining the borrowed shelf — these are available to a studio of one on its first day, and unlike nearly everything else in business they get cheaper the earlier they are adopted, because nothing has yet been promised to anyone. The hard part is not affording the refusals. It is meaning them when the money is actually on the table.