There’s a difference between building a product and building the rails the future runs on. LiveKit isn’t just piping data from A to B, they’re rebuilding how we speak to machines, how machines speak to each other, and how voice becomes the new command line for everything from ChatGPT to 911 dispatch. And now, they’ve got $45 million more reasons to scale the hell out of it.
Congrats to co-founders Russ d’Sa (ex-Twitter, code whisperer turned comms architect) and David Zhao (ex-Motorola, who’s forgotten more about infrastructure than most engineers ever learn) for locking in a Series B that plants their flag at a $345 million valuation. Altimeter Capital is back in the lead seat, joined by Redpoint Ventures and Hanabi Capital, with Mike Volpi stepping in with his first bet from the new fund. They’re not tossing coins here. This is conviction capital for a category that’s not just heating up; it’s setting fire to the old way of doing things.
LiveKit started like most real revolutions do: open source, no fluff, just function. During the pandemic, they dropped a tool that made WebRTC sing. Then Spotify and Oracle knocked on the doo,r asking for a cloud version. Fast-forward: LiveKit Cloud is now processing 3 billion calls a year, powering 25% of all U.S. 911 calls, and serving as the real-time backbone for OpenAI’s Voice Mode, Meta, Microsoft, Adobe, Skydio, and whoever else is serious about making latency vanish.
They’ve got HIPAA compliance, SOC2 creds, and a platform so lean it shaves time off every syllable with sub-25 millisecond latency for AI-driven voice. Under the hood? A cross-platform SDK spread, an SFU architecture optimized for scale, and a SIP stack holding up national emergency infrastructure. WebRTC, Kubernetes, and that rarest of things in AI: actual production usage.
But let’s talk game theory. LiveKit isn’t chasing shiny demos, they’re building Agents 1.0 with multilingual support, semantic turndetection, and multi-agent orchestration built for real enterprise workflows. That’s loanservicing, customer support, healthcareops, all stitched together with voice interfaces that don’t just understand you, they respond like they’ve been briefed.
This raise? It’s jet fuel. They’re expanding from 12 to 15 global data centers, hiring AI / ML and infraspecialists like the next world war depends on it, and rolling out elastic agent compute so your voiceAI doesn’t flinch when it scales from 1 user to 100,000 in real time.
The voice-AI market is pushing $160 billion. LiveKit’s not trying to ride the wave; they’re pouring the concrete under it. Think Stripe for communications, but open-source and built for the GPT4o world. Their trajectory isn’t hype. It’s infrastructure-level inevitability.
Discussion about this post
No posts