Sofwave Medical: How Recurring Is the Usage-Fee Engine, Really?

Real Recurrence, but Not the Kind the Headline Implies

The main article was right to mark Sofwave's jump to 43% of revenue from usage fees as one of the most important parts of the 2025 read. This follow-up isolates only that issue: what exactly is recurring here, how clean it is, and how much of the line reflects true installed-base monetization rather than convenient accounting framing.

The short conclusion is fairly sharp. Sofwave's usage fees are more real than the headline suggests, but also less "subscription-like" than the language of recurrence can imply. This is not revenue that is locked in by a fixed contract. It is revenue that appears only if the physician keeps treating patients, keeps consuming pulses, and keeps paying for those pulses.

In 2025 the company reported $87.6 million of revenue, of which $37.7 million came from usage fees, or 43% of total revenue. In 2024 usage fees were $23.3 million, so the annual increase was much steeper than the growth in system sales themselves. The fourth quarter pushed the point further: usage fees reached $13.2 million, about 45.7% of quarterly revenue, while capital equipment revenue was $15.7 million.

Those numbers do support a real installed-base engine. The presentation points to roughly 2,700 systems sold since commercialization and roughly 770,000 treatments completed. The annual report adds that revenue is spread across the United States, Europe and the Middle East, East Asia and other markets, and that in 2023 through 2025 no single customer represented more than 10% of sales. At year-end 2025 no single customer represented more than 10% of receivables either. That matters: the usage engine is not being inflated by one clinic chain or one distributor.

But this is also where precision matters. Breadth is not the same thing as a contract. A broad customer base means the usage pattern is diversified. It does not mean it is locked in. The right way to read the 43% is as installed-base monetization through actual procedures, not as ARR in disguise.

The Accounting Is More Conservative Than the Headline

The revenue-recognition note explains why the usage-fee line looks cleaner than the broad phrase "recurring revenue" suggests. For each system sale, the contract usually contains only one performance obligation, the sale of the system. The transaction price includes two layers: a fixed amount when control over the system is transferred, and variable amounts tied to the purchase of additional pulses. Those additional pulses are not treated as a separate performance obligation. They are treated as usage-based variable consideration.

That sounds technical, but it is the core point. At system delivery, the company does estimate how many additional pulses the customer may buy in the future. Even so, for the periods presented it concluded that there was not a high enough probability to avoid a significant reversal, so it constrained that variable consideration and recognized revenue from additional pulse packages only upon actual payment.

That last figure matters a great deal. In 2025 the company recognized only about $26,993 from systems sold in prior periods, versus about $16,949 in 2024 and about $6,457 in 2023. So the "change in transaction price" mechanism does exist, but it barely moves the story. The 43% headline is not being built on aggressive pull-forward of future pulse revenue. If anything, the accounting read is conservative because most of the recognition is deferred until the customer actually pays.

That is a critical distinction. Anyone reading the recurring-revenue narrative as a story of generous accounting estimates is reading the filings the wrong way. If anything, the filings lean in the opposite direction.

What the Pulse Economics Really Mean

This is where the operating economics become clearer. The company describes two main revenue pipes: sale of the system itself, and ongoing revenue from pulse sales. Each system comes with a prepaid bank of uses, but every treatment consumes pulses, and a single treatment generally requires 150 to 200 pulses. Once that pulse balance runs out, the customer must buy additional packages to keep using the system.

The more important point is channel control. Pulses are sold only by the group, through its online store, and cannot be purchased from third parties. That turns the installed base into a real monetization asset. It is not accidental that both the presentation and the annual report connect broader system usage to the growth in usage fees.

But precision still matters here. Usage fees are not a subscription. They are a price tag on activity. If treatment volume rises, the revenue recurs. If treatment volume weakens, there is no hard contractual minimum that forces the customer to keep buying pulses just because the system is already in place. So the increase in usage fees in 2025 is a very positive signal on installed-base depth, but not proof of a fixed long-term contracted stream.

That distinction is especially visible in the fourth quarter. Usage fees rose 83% year over year, while capital equipment revenue rose 42%. That is strong evidence that the existing base generated more activity, not only that Sofwave shipped more boxes. But it is still evidence of usage, not of contractual lock-in.

The company also tries to extend this logic to Pure Impact, where the model is based on treatment hours purchased before treatment. The strategic message is clear: management wants a platform where revenue does not end with the system sale. But as of year-end 2025, the direct numeric disclosure still sits mainly in the Sofwave pulse engine, not in a second usage layer disclosed at similar scale.

This Is Still a Heavy Commercial Engine, Not Revenue That Arrives by Itself

This is the layer that a short headline about recurrence can obscure. Sofwave ended 2025 with $66.0 million of gross profit and a 75.3% gross margin, but the cost structure does not look like a light software model at all. Selling and marketing expense reached $38.7 million, versus $12.8 million of R&D and $7.6 million of G&A.

The detail makes the point sharper. Selling and marketing included $19.3 million of payroll, $6.6 million of sales agents and consultants, $8.4 million of exhibitions, advertising and promotion, plus logistics and shipping expense. On the cost-of-sales side, the company still carried $12.3 million of system cost, $3.5 million of shipping and freight, and $3.2 million of warranty and service expense.

The implication is straightforward. Recurring revenue, yes. Autonomous revenue, no. Sofwave's usage-fee engine sits on top of a medical-device platform that still needs sales coverage, training, marketing, freight and service. So the right label is not "the company became a subscription model." The right label is "the company built a recurring monetization layer on top of a device platform." That is strong, but it is not the same thing.

That is also why the 43% usage-fee line should not be translated into the same language investors use for enterprise software. The key question here is not only what recurs, but how much commercial and operating effort is still required to keep that recurrence going.

Collection Quality Looks Better in the Headline Than in the Tail

The last layer is cash quality. On the surface, the picture looks cleaner. Net receivables fell to $7.615 million at the end of 2025 from $8.668 million at the end of 2024, while revenue rose to $87.637 million from $59.651 million. On a rough calculation, gross receivable days fell to about 34 days from about 54 days. That is a real improvement.

But the detail is less comfortable. Receivables older than 120 days rose to $1.571 million from $1.129 million a year earlier. At the same time, the allowance for doubtful accounts rose to $475 thousand from $231 thousand, nearly doubling, and the allowance rate rose to about 5.9% of gross receivables from about 2.6% the year before.

So the picture is mixed. On one side, the usage-fee line itself looks relatively clean because in the periods presented it is recognized only when payment is actually made. On the other side, the balance sheet still says that not every layer of the company's collection profile is equally clean. The right line is not "collection quality deteriorated" and not "collection quality improved." The right line is that the balance improved at the headline level, while the problematic tail became heavier.

That matters in a follow-up about recurrence. If the usage-fee line is going to be treated as a genuinely high-quality engine, investors need to see not only that it grows, but also that this older tail stops widening.

Conclusion

The precise way to read Sofwave's 43% usage-fee mix is this: it is a real recurring engine, but one tied to actual use rather than to fixed commitment.

The supportive side of the thesis is clear. The company has a broad installed base, sells the pulses itself, recognizes most of the additional revenue only upon actual payment, and in 2025 did not rely on meaningful accounting catch-up from old systems. In other words, growth in this line looks more economic than accounting-driven.

The limiting side is just as clear. Customers buy pulses only if they keep treating patients, so this is not locked revenue. Beyond that, the whole model still sits on a heavy sales and service layer, and the balance sheet shows a better overall receivables picture alongside a weaker overdue tail and a higher allowance.

So the 2026 test should not be whether usage fees remain large. It should be whether they keep growing faster than system sales without forcing the company to pay for that growth through a heavier collection tail and without leaving too much of the economics behind in the commercial layer. If that happens, the usage-fee engine will look not only recurring, but higher quality. If not, the headline will remain stronger than the economics underneath it.