The Product Lead

Figures converted from CHF at historical FX rates — see data/company.json.fx_rates. Ratios, margins, and multiples are unitless and unchanged.

The Product Lead

Sonova's recent share gains trace to one product cycle: the Phonak Sphere Infinio platform and its dedicated on-device DEEPSONIC AI chip, launched in August 2024. The technology lead is real and measurable — a 10.2-decibel signal-to-noise gain against 6.4 before — but it was won while research spend fell and every large rival shipped its own AI. Sonova's growth engine is an accelerating, high-margin Wholesale franchise (+9.5% local, segment margin up to 23.7%) but its ~7pp lead over Demant (+2% organic) was delivered on falling R&D intensity (6.8% -> 6.0% of sales), so the Sphere lead is a timing edge to be re-won every two to three years rather than a structural, spend-driven moat.[1][2][3] The 23.7% figure is the Wholesale segment margin, detailed in Segment Economics.

That distinction matters for the two-year return. Per What the Price Implies, the exit multiple, not earnings, drives that return — the scenarios move the multiple by roughly 50% against a normalized-EPS change of about 19%. If Wholesale local-currency growth converges toward the 4–6% market rate as rivals ship their next platforms, the premium compresses toward the bear 16x case — about a 15% two-year total loss — while a sustained premium is what supports a re-rating. The counter-fact sits in the same frame: rivals already ship on-device AI (Demant's Oticon Intent shipped in early 2024, before Sphere), and the falling R&D intensity may be operating leverage on a front-loaded, six-year DEEPSONIC build that Sonova has already extended with Infinio Ultra in October 2025 — not a hollowing-out of the pipeline.

What the lead actually is

The advantage has a specific engineering shape. Sonova's August 2024 launch was a pair of platforms: Phonak Infinio, built on a new ERA connectivity chip, and Phonak Audéo Sphere Infinio, which adds a second, purpose-built processor — the DEEPSONIC deep-neural-network chip — running alongside ERA in a dual-chip system [4]. DEEPSONIC was six years in development, trained on 22 million sound samples, and carries 4.5 million neural connections performing 7.7 billion operations a second — enough to separate speech from background babble in real time, on the device, rather than in the cloud [5].

The reason that matters is that speech-in-noise is the problem hearing-aid users most want solved, and the gain here is quantified rather than asserted. Sonova's prior best noise-cancelling delivered about a 6.4-decibel improvement in signal-to-noise ratio; Sphere Infinio reaches up to 10.2 decibels, which the company translates into users being two to three times more likely to understand every word against leading competitors [6]. At first fit, 93% of users preferred Infinio over a key competitor [7]. A dedicated real-time AI/DNN chip, exclusive to the Sphere line, is the piece rivals cannot simply copy in software [8].

That lead has shown up where it counts — in orders. The Wholesale business, which sells the devices, grew 9.5% in local currencies in FY2025/26 and won share globally, explicitly on the reception of the Infinio and Sphere platforms; management also flagged that the new products pulled in major-market customers who had not ordered the previous generation [9][10]. The quality-of-growth split is covered in Segment Economics; the point here is narrower — that outperformance traces to a single, datable product event.

AI is now table stakes, not a moat

The complication is that Sonova is not the only maker with a neural network in its hearing aids. Demant's Oticon brand launched Oticon Intent at the start of 2024 — before Sphere — with what Demant calls "powerful artificial intelligence" for speech clarity and noise cancellation, and Demant states it has been "investing significantly in AI-based signal processing for many years" [11]. Demant went on to launch first- and second-generation deep-neural-network devices across all its brands in early 2025, and in 2025 added Oticon Zeal, an AI-driven in-the-ear model [12][13]. The other large makers moved in the same window: Starkey shipped its Genesis AI and Edge AI platforms across 2023–24, and WS Audiology and GN round out a Big Five in which on-device AI is now the norm rather than the exception. What separates the makers is therefore not the presence of AI — everyone has it — but whether a particular implementation converts into durable share.

On the evidence so far, it does, and by a wide margin. Sonova's Wholesale business grew 9.5% in local currencies in its year to March 2026; Demant's Hearing Aids business grew 2% organically in calendar 2025 [14][15]. The periods differ by a quarter and the definitions differ slightly, but a seven-point gap against the maker most committed to AI is too large to be a measurement artifact.

Sonova Wholesale +9.5% local per Annual Report 2025/26 [16]; Demant Hearing Aids +2% organic per FY2025 Annual Report [17]. Fiscal periods and growth definitions differ; Sonova's year ends 31 March, Demant reports calendar years.

Demant is candid about what comes next: it expects "an intense competitive environment where everyone will fight even harder to compensate for the soft market" [18]. That is the honest frame: Sonova is ahead on execution this cycle, but the field is armed with the same category of technology and motivated to close the gap.

A lead delivered on a shrinking R&D budget

Here the numbers cut against a simple "innovation machine" story. Sonova produced its best product cycle in years while its research intensity fell. R&D costs ran at 6.8% of sales in FY2021/22 and stepped down to 6.0% by FY2025/26 [19]. Demant, the closest pure-play rival, spends a near-identical share of revenue on R&D, so raw budget is not what separates the two.

Source: Annual Report 2025/26, Five-Year Key Figures [20]. Dollar figures are converted from francs at fiscal-year-end rates, so their absolute path is distorted by the franc's strength against the dollar; the percentage-of-sales trend is currency-neutral and is the reliable signal.

The decline reads two ways, and both are legitimate. The bullish reading is that DEEPSONIC was a multi-year investment — six years to develop — that is now generating returns without needing more spend, so falling R&D intensity against rising share is operating leverage working exactly as intended [21]. The skeptical reading is that in an industry Demant describes as "very product-driven markets where significant R&D initiatives underpin market positions," trimming intensity to 6.0% of sales while three or four rivals sharpen their own AI is choosing near-term margin over the next platform. Which reading is right depends on whether the next launch cycle arrives on time and lands as well as this one — something no current filing can settle.

Cadence, and what keeps a lead sticky

Two things soften the "one cycle" risk without removing it. First, Sonova is already extending the platform rather than resting on it. It introduced Infinio Ultra in October 2025 — the company's phrase is "expanding innovation leadership" — advancing the AI a further step and widening the range into new form factors such as Virto R Infinio [22][23]. Management also notes the DEEPSONIC chip keeps being trained on new listening situations for future releases, so the same silicon is meant to carry more than one product generation [24].

Second, a prescription hearing aid is fitted, tuned, and serviced by a professional over years, which is the patient switching cost described in Moat and OTC Threat. A rival that ships a better chip next year does not instantly reclaim a patient who has been fitted on a Phonak device inside a clinic relationship. That installed-base friction is why product leadership tends to translate into a multi-year order tail rather than a single strong quarter — but it is friction, not a wall, and it works in rivals' favour too when they are the ones taking the fit.

What would change this read

The measured conclusion is that the product lead is genuine and is being extended, but it is a timing advantage rather than a structural one: it must be re-won every two to three years against rivals with the same category of technology and a similar budget. On the evidence to date — a seven-point Wholesale growth gap over the most AI-committed peer, a quantified sound-quality edge, and an already-launched successor platform — the advantage looks likely to persist through the near-term forecast horizon, which places the recent share gains on the durable rather than the fluke side of the ledger.

The strongest fact against that read is the R&D trajectory: Sonova is defending a technology position with declining research intensity (6.8% to 6.0% of sales) in a market its own competitor calls R&D-driven, and the entire recovery narrative leans on one launch that any of four rivals could answer. The read would weaken if Wholesale local-currency growth fell back toward the 4–6% market rate as competitors ship their next platforms; if the next Phonak flagship slipped or landed without a measurable performance step; or if R&D intensity kept sliding while share stalled. Each is visible in the semi-annual releases and the annual R&D line within the next two reporting cycles.