Fortell’s $163 million hearing-aid bet is really a software bet

Hearing aids have sat in an awkward spot for years. The hardware is tiny, regulated, and expensive. The software has been cautious. Most devices still do the obvious thing: make everything louder. That works poorly in the places people care about most, like a noisy restaurant or a family dinner.

Fortell thinks that gap is the opportunity.

The startup, led by Matthew de Jonge, has raised $163 million from Founders Fund, Thrive Capital, and Antonio Gracias’ Valor Equity Partners. Its pitch is simple enough to explain and hard enough to build: make hearing aids that understand the sound environment in real time, pick out what matters, and suppress the rest. That means a model-driven audio pipeline, custom silicon, and a product plan built around the idea that the old hearing-aid industry left a lot on the table.

The industry is also tightly concentrated. Five companies make about 97% of hearing-aid products. That tends to produce polished distribution and slow change. It also gives incumbents room to move cautiously, since the buyer is often older, insurance coverage is patchy, and replacement cycles are long.

Why hearing aids lagged behind consumer audio

It’s easy to lump hearing aids in with headphones or earbuds. They’re all audio devices, so how different can they be?

Different enough to stall a category for years.

A hearing aid has to do low-latency signal processing on a tiny battery while dealing with changing acoustic scenes, feedback suppression, speech enhancement, and user-specific hearing loss profiles. It’s not enough to identify speech. The device has to decide which speech matters, how to handle competing speakers, and how hard to filter environmental noise without making everything sound clipped or artificial.

That’s the problem Fortell is going after. Traditional hearing aids amplify broadly, then lean on fitted profiles and static processing. In practice, people still struggle in complex environments. Real-world sound overlaps, bounces, and changes faster than a basic hearing test can predict.

If Fortell can do better, the upside is obvious. People don’t stop wearing hearing aids because they forget them. They stop because the experience is awkward, socially annoying, or not worth the hassle.

The real bet is edge AI

The most interesting detail is the architecture. Fortell says its system uses a custom chip designed with help from a former Google TPU designer. That tells you where the company is spending its energy.

A hearing aid can’t depend on a cloud round trip to classify sound. Latency would be too high, and connectivity would be unreliable. The device needs local inference, on a very tight power budget and with careful thermal constraints. A custom chip makes sense if the company is serious about running real-time audio models on-device instead of dressing up a few rules as AI.

That brings the usual trade-offs into a very small box:

• Model quality vs. power draw: Better recognition usually means larger models, but battery life matters more in a hearing aid than in a phone.
• Latency vs. accuracy: Real-time audio enhancement can’t wait for slow inference passes.
• Personalization vs. deployment complexity: Hearing loss is highly individualized, which pushes the system toward adaptive calibration and more complicated fitting workflows.
• On-device processing vs. cloud features: Privacy improves when audio stays local, but updates and model improvements are harder to ship.

That last point matters more than vendors usually admit. If the product depends on ongoing model tuning, the firmware, chip, and mobile app all become part of a tightly coupled stack. That starts to look a lot more like medical-device engineering than consumer audio.

Apple is a competitor and a proof point

De Jonge’s take on Apple’s AirPods makes sense. He’s treating them as both a hearing-aid alternative and an at-home hearing test, not as a direct existential threat.

That’s a fair read. Apple has already normalized the idea that consumer earbuds can do some hearing-related work. Hearing tests and amplification-style features train people to think about hearing support as software, not just as a clinical device. That lowers the barrier for everyone else.

AirPods also show the ceiling of a general-purpose product. They’re not built for persistent, highly personalized hearing correction. They’re a broad platform with some hearing use cases attached. Fortell’s pitch is narrower and harder. If it works, it should beat consumer earbuds at actual hearing assistance. If it doesn’t, people will stick with Apple because it’s already there.

That’s the awkward part of this market. Consumers compare medical-ish hearing products with devices they already own and understand. They’re not shopping in a vacuum.

Distribution may matter as much as the model

The technology is only half the fight. The other half is getting the product into people’s ears.

Hearing aids are still shaped by regulation, telehealth rules, and fitting workflows. Fortell is dealing with remote hearing-aid fittings, and state-level telemedicine laws affect how easily it can scale. That’s a classic healthcare software problem: the product may work in one state, then hit licensing or remote-care restrictions in another.

For a startup, that creates a real operational burden. You can’t just ship nationally the way a SaaS company can. You need workflows for compliance, support, clinical oversight, and probably a lot of handholding during onboarding. If the company wants to cut old friction, it can’t replace it with an equally annoying setup process.

Insurance is another drag. In many cases, it won’t cover the cost. That means Fortell has to sell directly, or at least convince buyers the out-of-pocket price is worth it. In consumer health hardware, that usually means one of two things: either the product feels better fast, or it gets ignored after the novelty wears off.

There’s not much room for “pretty good” here. If the sound quality is only fine, people won’t tolerate the price.

Why investors like this category now

Founders Fund, Thrive Capital, and Antonio Gracias aren’t backing hearing aids because the category sounds glamorous. They’re betting on technical defensibility and market neglect.

The defensibility comes from the stack. Custom silicon, real-time audio inference, fitting data, and medical-device workflow integration are all hard to copy quickly. A startup that can build all of that has a wider moat than a pure software company.

The market case is simple. Hearing loss is common, the incumbent market is concentrated, and the user experience has been underworked for decades. That’s the kind of setup investors like when a new platform shift makes an old product class feel stale.

Still, a good thesis doesn’t make a reliable company. Hardware timelines slip. Audio quality is subjective. Regulatory assumptions change. And hearing-aid users are not easy early adopters for a product that may require adaptation, support, and trust.

Then there’s data. Real-time audio systems improve with usage signals, but hearing data is sensitive. Any company in this space has to treat privacy seriously, especially if it stores listening patterns, tuning profiles, or diagnostic information. Get casual about that and people will walk.

Why builders should pay attention

This is one of the few consumer-health stories that actually rewards people who think in systems.

If Fortell works, it shows how far edge AI has moved past demos and into constrained devices with real stakes. It’s a test of whether custom inference hardware, on-device models, and product-specific tuning can beat generic compute in a category that has been boring for too long.

The pattern shows up anywhere software has to sit close to the human body and far from a reliable network: wearables, assistive devices, medical monitoring, even some industrial audio systems. The technical bar is high, but the payoff is real when the product feels useful instead of merely clever.

For now, the interesting thing about Fortell is that it isn’t trying to ship a louder hearing aid. That would be a weak idea. It’s trying to build something more selective, more adaptive, and harder to make. That’s a better bet, even if the road from funding round to a device people wear every day is packed with ugly engineering work.