Key Takeaways

  • LiveKit has raised $100 million in a Series B funding round led by Altimeter Capital.
  • The company’s infrastructure is currently powering OpenAI’s Advanced Voice Mode, validating its capability in low-latency environments.
  • The capital will fuel the expansion of its programmable network, specifically designed to support multimodal AI agents rather than just traditional video conferencing.

LiveKit, a developer of infrastructure software for real-time AI voice and video applications, has announced the raise of $100 million in funding. The Series B round was led by Altimeter Capital, with participation from existing investors like Redpoint and Benchmark. While nine-figure funding rounds are becoming slightly more common again in the AI sector, this one signals a specific shift in where investors believe the bottlenecks lie. It is no longer just about the models; it is about the plumbing required to make those models communicate instantly.

The state of voice interaction on the internet has historically faced significant hurdles regarding latency. When a user speaks, data travels to a server, gets processed, and returns. In a Zoom call between humans, a 500-millisecond delay is annoying but manageable; speakers naturally pause, talk over each other, and adjust. However, when talking to an AI, that half-second delay breaks the illusion of intelligence completely.

This is where LiveKit has positioned itself. The company is not building Large Language Models (LLMs); it is building the programmable network that carries audio and video data to and from the model with ultra-low latency. If the model is the brain, LiveKit acts as the nervous system.

Many users may have already utilized the technology without realizing it. OpenAI uses LiveKit to power the Advanced Voice Mode in ChatGPT. That seamless, interruptible quality—where a user can cut the AI off mid-sentence and it reacts instantly—is largely a function of how the data is being transported. Previously, developers building voice AI were forced to combine legacy telephony stacks or wrestle with raw WebRTC (Web Real-Time Communication).

WebRTC is a standard that has been available since 2011. While open and free, implementing it at scale is notoriously complex. It handles the handshake between devices but does not solve for routing, scaling, or the specific requirements of AI audio, which demands higher clarity than an average VoIP call. LiveKit effectively wraps that complexity in a developer-friendly API.

The raise brings the company's total funding to roughly $137 million. The substantial capital injection is driven by the industry shift toward "multimodal" agents. The market is moving away from simple text-box chatbots toward enterprise applications involving AI agents that can see and hear. Use cases include customer support agents that can view a video feed of a product to guide repairs in real-time, or language tutors that analyze mouth movements to correct pronunciation.

Executing these functions requires massive bandwidth and edge computing capabilities. LiveKit plans to use the funds to expand its SIP (Session Initiation Protocol) capabilities—bridging the gap between modern AI and traditional telephone lines—and to bolster its global network infrastructure.

There is also a philosophical angle regarding open source. LiveKit started as an open-source project, allowing developers to inspect the code and run it on their own servers. However, the commercial cloud offering—which removes the burden of server management—drives revenue. It is a classic "open core" business model that is resonating with the enterprise market.

Tech visionaries have predicted voice as the future interface since the days of the original Siri, but the experience was often clunky. Users often felt they were issuing commands rather than having a conversation. With latency minimized and model intelligence increasing, the industry may be at a tipping point where speaking to software feels natural.

The market is competitive, with traditional communication APIs like Twilio or Agora monitoring the space closely. However, LiveKit’s specific focus on the unique needs of AI—handling interruptions, function calling via voice, and maintaining context—provides a distinct advantage.

Altimeter’s involvement serves as a market signal. Brad Gerstner, the firm’s founder, has been vocal about the "supercycle" of AI. Investing in the infrastructure layer suggests a belief that the application layer is about to expand rapidly, requiring reliable methods to transmit voice and video data. For developers, the $100 million funding ensures the longevity of their toolset, while for end-users, it suggests that computer interactions are about to become significantly more conversational.