Investors Go Hard for Voice AI With a String of Big January Deals
We have fresh scoops on previously unreported funding rounds for voice AI startups and customer service agents.
At our Cerebral Valley London summit last June, GV managing partner Tom Hulme predicted that voice would be the next frontier for an AI interface — and six months later, a string of hot voice funding rounds shows that he was likely right.
Last week, ElevenLabs officially announced a $500 million Series D round at an $11 billion valuation. That followed voice customer support startup Decagon, led by CEO Jesse Zhang (pictured), closing a $250 million Series D from Coatue and Index, and voice model startup Deepgram raising a $130 million Series C three weeks ago. Gradium, based in Paris, raised a $70 million seed in December.
While voice AI is a horizontal tool, customer service voice agents are one of the most popular use cases for voice AI, where companies can have AI handle the bulk of their inbound calls. Beyond Decagon, the most notable entries include Bret Taylor’s unicorn startup Sierra and MavenAGI.
One startup, Phonic, is building both its own voice models and the orchestration software to help its customers successfully run voice agents for all kinds of use cases. The largest so far is for customer support, but its models have been used for trucking logistics management and healthcare back-office work.
Lux Capital backed their $4 million seed round in 2024, and we heard from two sources that they’ve now raised a Series A round from First Round Capital and Felicis.
Existing customer support startups are also embracing voice. Netomi, for example, launched voice support as part of its product offerings, and has successfully landed large enterprise customers like Paramount+, United Airlines, and DraftKings. The startup has been backed by WndrCo, Index Ventures, Salesforce Ventures, Y Combinator, and Fin Capital, and we’ve heard that they’re close to closing a new 9-figure Series C round.
OpenAI has built its own voice models, but efforts stalled a bit last year and researchers complained internally that their performance was lagging behind its text-based models. Its latest voice efforts are expected to be released sometime in Q1 this year, per The Information.
Not all of the big labs are as bullish on voice. Anthropic chief product officer Mike Krieger expressed some skepticism around voice as the next big breakout medium at our Cerebral Valley AI summit in November last year, instead noting the staying power and usefulness of the text box. Claude has been given some voice capabilities as a way to enable hands-free use.
Early-stage VCs are keen to back smaller, vertical voice support startups where a founder can have a unique advantage when navigating a regulated market.