The archive · Developer & Business Tools · Product decision · 2021–2026
LiveKit's voice-AI infra bet: open-source WebRTC to a $1B unicorn
LiveKit bets real-time voice/video becomes the default AI interface; its open-source stack powers ChatGPT Voice Mode and a $100M Series C at $1B.
LiveKit
What the business is
An open-source, real-time audio/video infrastructure company (San Jose, CA): self-hostable WebRTC servers and SDKs, plus LiveKit Cloud and an agents framework for building voice and video AI applications.
Starting capital：$45M Series B led by Altimeter (April 2025), then a $100M Series C led by Index Ventures at a $1B valuation (January 2026).
How it started
In 2021, early Twitter engineer Russ d'Sa and former Motorola engineering director David Zhao created LiveKit as an open-source software package for building apps that transmit real-time audio and video, born in the Zoom-heavy pandemic era. It grew as free developer infrastructure until large companies experimenting with it - Spotify, Oracle, Reddit - asked for a cloud-hosted version; that request turned the project into a startup.
What happened
LiveKit Cloud launched, and the voice-AI wave arrived: OpenAI built ChatGPT Voice Mode's realtime stack on LiveKit, and in October 2024 a Show HN post ('Open source framework OpenAI uses for Advanced Voice') drew 266 points. A Series A followed in 2024; in April 2025 a $45M Series B led by Altimeter came with 500+ paying customers, a claimed $10M+ run rate and 100k+ developers, with the platform powering roughly 25% of US 911 emergency calls. In January 2026, a $100M Series C led by Index Ventures valued the company at $1B, with customers including OpenAI, xAI, Salesforce, Tesla, Spotify and Meta.
How it ended up
Still live and scaling: LiveKit reached unicorn status with the January 2026 Series C, and GitHub growth continues (20.6k stars, trending at #2 Go in Dec 2025 and #3 in Feb 2026).
Background
LiveKit is an open-source, real-time audio/video infrastructure company founded in 2021 by Russ d'Sa, an early Twitter engineer, and David Zhao, a former Motorola engineering director, based in San Jose, California. Its bet: real-time voice and video would become the default interface for AI agents, and the plumbing for that world would be developer-first, open-source infrastructure rather than closed SaaS.
The wedge was deliberately bottom-up. LiveKit shipped as a free, self-hostable WebRTC server with SDKs, and developers adopted it for streaming audio and video without any sales team. The company formed when large enterprises experimenting with the open-source project - Spotify, Oracle, Reddit - asked for a managed cloud version, which d'Sa describes as 'Cloudflare, but for media streaming.'
The voice-AI boom turned that infrastructure bet into a business. OpenAI built ChatGPT Voice Mode's realtime stack on LiveKit, and in October 2024 a Show HN post about the open-source framework behind it drew 266 points. A Series A landed in 2024, a $45M Series B led by Altimeter followed in April 2025 (500+ paying customers, a claimed $10M+ run rate, 100k+ developers, and roughly 25% of US 911 emergency calls running on the platform), and in January 2026 a $100M Series C led by Index Ventures valued the company at $1B.
LiveKit now positions itself as the 'AI-native cloud' for real-time communications - 'what Stripe did for payments, LiveKit is doing for communications' - with customers from OpenAI, xAI, Salesforce and Tesla to Spotify, Meta, Microsoft and Character AI. On GitHub the main repo holds 20.6k stars as of September 2026, with continued Go-trending runs in late 2025 and early 2026.
What has to be true
- Open source removed the adoption barrier: developers self-hosted and evaluated the WebRTC stack for free, building a 100k+ developer base before any sales motion.
- Enterprise pull, not push: Spotify, Oracle and Reddit asking for a managed version proved the market before LiveKit spent on distribution.
- It rode the AI wave instead of fighting it: the same realtime stack built for conference calls became the transport layer for ChatGPT Voice Mode and thousands of voice agents.
- It kept expanding the wedge - from self-hosted server to LiveKit Cloud to the agents framework - so each round funded a wider slice of the realtime AI stack.
What can be applied
Give away the infrastructure, sell the managed version: the free open-source stack built the developer base, and enterprise demand for hosted WebRTC made LiveKit a $1B company when voice AI arrived.
Aftermath
As of Sept 2, 2026, LiveKit is a scaling unicorn: a $100M Series C at a $1B valuation (Jan 2026), 200k+ developers on the platform per PYMNTS, 20.6k GitHub stars with continued Go-trending runs, and customers spanning OpenAI, xAI, Salesforce, Tesla, Spotify and Meta. It powers ChatGPT Voice Mode and a large share of US 911 emergency calls, and is expanding from voice agents into video and physical-AI (robotics) infrastructure. The open question is whether realtime voice becomes as ubiquitous as d'Sa predicts - LiveKit's valuation is a bet on that answer.
Sources
- LiveKit's tools power real-time communications, including OpenAI's Voice Mode
- Voice AI engine and OpenAI partner LiveKit hits $1B valuation
- LiveKit Raises $100 Million for Voice AI Developer Platform
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card