What Oura's filing teaches hardware startups about attaching a subscription

Oura filed publicly for a US listing on 3 September and disclosed numbers that are unusual for consumer hardware. Slightly more than $1.2 billion of revenue for the nine months to 30 June 2026, up 74 percent on the same period a year earlier. Net income over the comparable periods rising from roughly $1.5 million to close to $60.8 million. Five million paying subscribers. Hardware companies do not usually produce that profit curve, and the reason is worth working through, because the structure is more portable than the product. The default economics of consumer hardware are punishing and well understood. Revenue arrives in lumps tied to product launches. Gross margin is exposed to component costs, freight and tariffs. Growth requires persuading someone to buy a physical object again, and the replacement cycle is measured in years. A company in that position is valued as a manufacturer, which is to say cautiously. Attaching a recurring software fee changes several of those properties at once. A portion of revenue continues whether or not the customer upgrades. The margin on that portion is software margin. Customer lifetime value stops being a single transaction. And the revenue becomes predictable, which is the property public markets pay for above almost any other. Oura's profit swing is what that looks like when it works. Growing revenue 74 percent while moving net income from about $1.5 million to nearly $60.8 million is operating leverage: subscription revenue scaling without a proportional increase in cost of goods sold. The trap is that this reads as a formula and it is not one. A great many hardware companies have attached a subscription and found that nobody paid for it, and the difference is not marketing. Three conditions have to hold together. The device has to generate data continuously rather than once, because a subscription to interpret a measurement taken at purchase is a manual, not a service. The interpretation has to improve or accumulate with time, so that a customer twelve months in has something a new customer does not. And cancelling has to cost the customer something they can feel, usually their own history. A sleep and recovery ring satisfies all three almost by accident. It measures every night, the analysis is longitudinal by nature, and a member who cancels loses their baseline. A device that measures once, or whose output is the same on day 400 as on day 1, satisfies none of them, and no amount of packaging fixes that. The second lesson in the filing is less comfortable and gets less attention. Oura reports distribution across 56 markets through roughly 8,400 physical stores. That is a substantial retail footprint, and building it is expensive, slow and operationally heavy: channel margin, inventory, returns, merchandising, and the working capital that sits under all of it. The subscription did not replace that work. It was built on top of it. A recurring revenue stream requires a large installed base of devices, and getting devices onto people requires distribution. Founders drawn to the software economics of this model frequently underestimate that the hardware business has to be won first, at hardware's cost structure, before the software economics arrive. The third lesson is about what the filing does not settle, which is the question any investor will focus on. Five million paying subscribers is a stock. It says nothing about what proportion of device buyers subscribe, what monthly churn looks like, how many subscribers survive a device replacement cycle, or how much of the growth is new device sales rather than retention of existing members. Those figures determine whether the subscription is durable or is a lagging indicator of a strong hardware year. The competitive pressure sits in the same place. Samsung has entered the smart ring category, and the broader wearables market is led by companies for which health tracking is a feature of a device people already own. Oura's defence is that a ring is worn overnight in a way a watch often is not, and that the analysis rather than the sensor is the product. Whether that holds as larger companies close the gap is the strategic question underneath the financial one. Some coverage has cited an annual revenue figure around $1.4 billion alongside the $1.2 billion nine-month figure. Those are different periods rather than a contradiction, and the disclosed nine-month number is the one anchored to the filing. For a founder, the usable version of all this is short. Ask whether the device produces continuing data, whether interpretation compounds, and whether cancelling costs the customer their history. If all three are true, the subscription is a real business and the hard part is distribution. If any is false, the subscription is a price increase with extra steps.