ElevenLabs, the voice AI company founded in 2022 by two Polish friends and now headquartered in London and New York, has doubled its valuation to $22 billion in eight months. The company said on 30 September that it had completed a $300 million employee tender offer, co-led by Wellington Management and T. Rowe Price, at twice the $11 billion valuation it reached in a $500 million Series D in February.
The money lets staff sell vested shares rather than funding the business, a route fast-growing AI companies increasingly use to keep people without going public. New backers in the deal include EQT, Goldman Sachs, Singapore's GIC, Ontario Teachers' Pension Plan, Sapphire Ventures and BDT & MSD, alongside existing investors Andreessen Horowitz, Lightspeed and ICONIQ.
What changed in those eight months is the mix of the business. Enterprise customers now account for more than 55% of revenue, and the company's conversational agents handle over 15 million conversations a week, three times February's level, with revenue from the agents platform also up more than threefold.
From Voices To Agents
ElevenLabs built its name on synthetic speech convincing enough for audiobooks and film dubbing, and that remains the foundation. Its newer business sells complete voice agents that answer calls, hold a conversation and resolve a request, which puts it in direct competition with contact centre software rather than with media tools.
The company says voice agents resolve issues 31% faster than chat agents on average, a claim that reflects how people use the two: a caller explains a problem in one breath, where a chat exchange takes several rounds of typing. It has also built versions aimed at specific industries, including financial services, healthcare, government, retail and telecoms, along with Scribe v2 Medical for clinical transcription.
Who Is Buying
The customer list explains the valuation better than the technology does. ElevenLabs says it works with five of the ten largest technology companies, five of the ten largest insurers and four of the ten largest telecoms, with named customers including Stripe, Deutsche Telekom, the insurer Admiral, Customers Bank, the design software firm Cadence and the governments of Ukraine and Greece.
Those are the organisations that run the highest call volumes, and they buy slowly but at scale. Revenue has followed: ElevenLabs passed $600 million in annual recurring revenue by June, according to research firm Sacra, up from $330 million at the end of 2025.
Voice Became The Interface Businesses Will Pay For
The rise of voice agents marks a shift in where enterprises see value in AI. Chat assistants have been widely deployed but are often treated as a cheaper website search, while a voice agent replaces a measurable cost, namely the time human agents spend on routine calls.
Voice is also the interface that reaches people who will not use an app. That matters for utilities, insurers, banks and public services, and it extends to machines and vehicles where hands and eyes are busy, the direction we traced in how the cellular module is becoming an AI computer.
The money has followed. Voice AI startups raised about $7 billion in the first quarter of 2026, against under $1 billion a year before that, and one market estimate puts the voice generation market on a path from $4.16 billion in 2025 to $20.71 billion by 2031.
The Competition Is Thick
ElevenLabs is not alone, and its rivals come from two directions. Specialists including Deepgram, Cartesia and Rime compete on speech models and latency, while the large AI labs offer realtime voice capabilities inside their own platforms, where voice is one feature among many rather than the product.
ElevenLabs argues that treating audio as the main business produces better results than treating it as an add-on, and its enterprise share suggests buyers currently agree. That advantage would narrow if a frontier lab's bundled voice became good enough for routine calls, since many companies already pay those labs for other work.
What Buyers Should Check Before Committing
The enthusiasm comes with a warning from the analysts. Gartner expects task-specific AI agents in 40% of enterprise applications by the end of this year, up from under 5% in 2025, and also expects about 40% of agentic AI projects to be cancelled by 2027 over runaway costs, unclear returns and governance problems.
Voice agents carry specific risks worth testing before a rollout. Costs scale with call minutes rather than seats, so a successful deployment raises the bill. Regulated industries need records of what the agent said and clear escalation to a human. Voice cloning raises consent and fraud questions, particularly where a synthetic voice could be mistaken for a named employee. And a wrong answer delivered confidently in speech is harder for a customer to check than the same answer in text.
The practical questions for a buyer are what share of calls the agent resolves without a human, what happens when it cannot, how conversations are logged and retained, and what the cost looks like at full volume rather than in a pilot.
A European AI Company At Scale
ElevenLabs has become one of Europe's most valuable private technology companies, with more than 800 employees and offices in London and New York, which makes it a rare case of a European AI firm competing directly with US rivals in an enterprise market rather than being acquired early.
An employee tender also keeps that independence intact for now. It gives staff liquidity without the disclosure and scrutiny of a listing, and lets the company keep raising at higher marks from institutional investors who want exposure before any IPO.
The Call Centre Is The Test Bed For Agents
ElevenLabs' valuation reflects a bet that voice will be where businesses first trust AI agents with real work, because the task is well defined, the volume is large and the saving is easy to count. Fifteen million conversations a week suggests that bet is landing with the kinds of companies that do not move quickly.
The harder test comes next, as those deployments grow from pilots into the main route customers take to reach a company. If resolution rates hold and costs stay predictable at that scale, voice agents will become standard infrastructure. If they do not, ElevenLabs will find itself inside Gartner's cancellation statistic alongside everyone else selling agents.
