Snips raises $13M Series A for its smart assistant platform which works without the cloud
Snips announced today that it has raised $13 million to boost its launch of a new voice platform designed to give hardware makers an alternative to Google's Home and Amazon's Alexa.
Context & Ripple Effects
Snips' $13M Series A lands in the middle of a 2016–2017 funding wave for the component parts of voice: Vesper raised $15M for always-on, low-power microphones and SoundHound pulled in $75M from NVIDIA and Samsung Catalyst to push voice AI beyond phones. What distinguishes Snips is its wedge — a smart-assistant platform that runs entirely without the cloud, pitched directly at hardware makers who don't want to build on Google Home or Amazon's Alexa.
The bet was ultimately validated by exit rather than scale: Sonos paid $37M for Snips in 2019, folding the on-device stack into a speaker maker that had sold 6.1M units that year. That outcome reframes this raise as early infrastructure for device-owned voice, not a challenger platform in its own right.
First-order effects
- Hardware makers gain a licensable, privacy-forward assistant they can embed without routing user speech through Amazon or Google — the first credible off-ramp from the Alexa/Home duopoly for smaller device vendors.
- The raise funds Snips' commercial launch, moving it from research project to competing directly with Alexa and Google Home for design wins in third-party hardware.
Second-order effects
- Edge-voice component suppliers line up behind the same thesis: Vesper's always-on mics only reach their efficiency promise when recognition runs locally, so on-device platforms like Snips expand the market for such silicon.
- Amazon and Google's control of the assistant layer is contested at the licensing level — if device makers can buy voice capability outright, the incumbents' leverage shifts from distribution to raw model quality.
Third-order effects
- Voice assistants are consolidating into device-maker acquisitions rather than standalone platforms: Sonos bought Snips, and Snap later paid $70M cash plus $50M stock for voice-assistant startup Voca.ai — a pattern where the assistant layer gets absorbed vertically instead of remaining an independent product category.
- If local processing keeps improving, the default architecture for ambient voice splits between cloud-first giants and embedded stacks, with privacy positioning as the durable differentiator for hardware brands.
The trend: Voice AI is splitting into cloud platforms owned by a few giants and licensable on-device stacks that hardware makers increasingly acquire outright.