Nanoveu (ASX:NVU) hits sub-1mW voice sensing and finally has silicon to sell

Bone-conduction keyword spotting on ECS-DoT cracks the noise, power and latency trade-off in a US$62 billion market.

Nanoveu (ASX:NVU) has put out a technical milestone today that is more commercially interesting than it first appears. Its EMASS subsidiary has demonstrated always-on, real-time keyword spotting on the ECS-DoT edge AI chip using bone-conduction vibration sensing, running at under 1 milliwatt of computational power.

In plain English, the chip can recognise a spoken command by sensing the vibration in your jawbone rather than by recording the air around you. No microphone sits in the detection path. Ambient noise never enters the signal, so wind, factory floor and crowded train carriage get filtered out at the source rather than by burning power on noise cancellation.

This advances the bone-conduction hearables reference design EMASS announced in March 2026. The earlier version used motion to wake an audio subsystem. The new version skips that step and performs keyword spotting directly on the vibration signal.

For investors who have followed the ECS-DoT story since the TSMC 16nm tape-out, today’s announcement shifts the chip from a promising specification sheet to a demonstrated solution for a named problem in a US$62 billion market.

Why solving the three-way trade-off is worth real money

Wearable voice control has been stuck on a trilemma for years. Designers can pick any two of noise immunity, low power and real-time response, but not all three. Always-on microphones are responsive but burn battery and capture everything around the user.

EMASS has effectively side-stepped the trilemma rather than balancing within it. Because the vibration signal does not contain ambient noise in the first place, there is no noise to cancel and no audio to record. The sub-milliwatt power envelope then lets the chip stay on continuously without draining the battery.

The hearables market is forecast to grow from US$62.22 billion in 2026 to US$107.1 billion by 2031, with voice becoming the primary control interface. If voice is the interface, the trade-off EMASS just resolved is exactly the constraint holding back the next generation of AR glasses, earbuds and industrial comms gear.

ECS-DoT is starting to look like a platform, not a product

What stood out to us in the CEO commentary is the deliberate framing of ECS-DoT as a platform for extracting signal from mechanical vibration, not just a hearables chip. The same architecture is being positioned for AR glasses, VR headsets, professional comms, and defence and critical-infrastructure systems.

That matters because Nanoveu’s broader story now spans semiconductors, the Spinoff Robotics drone-autonomy stack, and the NTU GPS-free swarm IP under exclusive evaluation. The common thread is ECS-DoT performing AI inference at power budgets that make compact, battery-constrained devices viable.

We think the read-through to the drone work is also worth noting. A chip that can pull meaningful signal from vibration at under a milliwatt is doing exactly the kind of always-on inference defence and inspection drones need on tight thermal budgets.

What we would want to see before getting too excited

Our concern is the standard one for early-stage semiconductor stories. Demonstrations on a benchtop and design-ins at a tier-1 hearables or AR customer are different things, and the gap between them has historically been where small-cap chip companies burn the most cash.

The announcement points to active engagement with device manufacturers and module makers. That language is encouraging but it is not yet a named customer or a binding agreement. At the March quarter, Nanoveu held A$6.67 million against a burn of roughly A$2.16 million per quarter.

The Investors Takeaway for Nanoveu

Today’s demonstration meaningfully strengthens the case that ECS-DoT is a genuine ultra-low-power edge AI platform rather than a single-application chip. Solving the noise-power-latency trilemma in wearable voice is a concrete commercial hook, and the privacy angle of never capturing ambient audio is the kind of feature regulated and enterprise buyers will pay extra for.

The next milestone investors should watch is a named customer engagement or design-in announcement in either hearables or AR glasses, ideally before the NTU evaluation period concludes in September 2026. Until then, this remains a story stock where technical milestones are running ahead of revenue. For our previous coverage on the drone-autonomy pivot, readers can see our earlier piece at stocksdownunder.