ElevenLabs raises $500M Series D at $11B valuation, hits $330M ARR with enterprise voice agents

Summary

ElevenLabs raised $500 million in Series D funding at an $11 billion valuation, led by Sequoia Capital with Andreessen Horowitz and Iconic tripling and quadrupling their positions respectively. Andrew Reed from Sequoia joined the board. The company reached $330 million ARR by the end of 2025, adding roughly $100 million in new annual revenue every few months.

Enterprise voice agents

The growth came from deploying conversational AI agents at scale. ElevenLabs started with text-to-speech and speech-to-text models in 2022, then expanded into music generation and dubbing. The inflection point arrived in 2025 when voice agents began operating in production environments. Deutsche Telekom uses them for customer support. Square and Deliveroo use them to call restaurants and train new riders. Revolut built voice agents serving 4 million European users for customer care. These are real-time, multilingual voice interactions that were operationally impossible a year ago.

ARR trajectory

The company took 20 months to reach $100 million ARR, 10 more months to $200 million, then 5 months to $330 million. Co-founder Mati Staniszewski attributes the acceleration to enterprise focus on latency and reliability, the opposite problem that killed earlier chatbot waves. Between 2015 and 2019, voice agents were vendor bloat imposed on consumers. Now customers request voice interfaces first. The change reflects model quality. When Claude or GPT can reason coherently, callers stop asking to speak to a human.

Product strategy

ElevenLabs is staying anchored to audio and voice while expanding into adjacent areas such as knowledge bases, telephony systems, testing, and monitoring that sharpen the voice experience. ElevenLabs Reader, a consumer tool for AI-narrated audiobooks, spans research, enterprise agents, and creative applications. Spotify allows ElevenLabs users to publish AI audiobooks. Audible blocks them unless Amazon created the AI content. That friction may not persist if consumer demand for audio-first content continues.

User education

Reed sees the next phase as teaching people how to use these systems. Most users interact with voice agents as if using a phone tree, unaware these systems can orchestrate across an entire company's capabilities in natural language. Once users see what leading implementations do, adoption patterns could shift rapidly. Ukraine's government deployed a voice agent for citizen services, letting residents call and discover programs they didn't know existed. That use case works as access, not friction.

Market position

In public software, companies are discarding proven models. Private AI-native companies with real revenue and unit economics are outpacing public SaaS expectations. ElevenLabs is the counter-example cited when portfolio companies claim to be AI native. It has the numbers.