On May 5, ElevenLabs added a third close to its Series D, locking in an $11 billion valuation and lifting the round from the $500 million announced in February to more than $550 million. The more revealing number, though, was not the financing total. It was revenue.
ARR reached $500 million in Q1 2026, up from $350 million before year-end. That is a 43% jump in three months.
This is not the kind of ARR figure an AI company waves around at a launch event. For a voice synthesis company to add $150 million of annual recurring revenue in one quarter, enterprise buyers have to be purchasing in volume.
Investors: institutions, strategics and celebrities
The Series D investor mix says almost as much as the size of the round.
Institutional capital: BlackRock, Wellington, D.E. Shaw and Schroders. Put together, those names suggest late-stage investors are treating ElevenLabs as a company nearing the public-market track.
Strategic capital: Nvidia's NVentures, Salesforce Ventures, Deutsche Telekom and Santander. This group sends a clearer operating signal. Nvidia backs voice AI for the same reason it backs AI data centers: compute demand. Salesforce needs a voice layer for Agentforce, and building it from scratch would take too long. Deutsche Telekom and Santander point to telecom and finance, two industries that care intensely about voice quality and compliance, already viewing ElevenLabs as a long-term vendor.
Celebrity capital: Jamie Foxx, Eva Longoria, Squid Game director Hwang Dong-hyuk and earlier investor Matthew McConaughey.
Celebrities have been investing in AI companies for years, but voice AI is a different case. These investors control voice IP; their voices are commercial assets. A stake in ElevenLabs is a wager on the copyright infrastructure behind future AI dubbing, not only a financial bet.
The growth engine is enterprise, not consumers
CEO Mati Staniszewski framed the core issue directly:
Enterprise adoption requires genuinely human-level AI voice models, not robotic-sounding systems.
ElevenLabs' customer list explains the point: Meta, Salesforce, Revolut and Klarna. Their use cases cluster around four business functions: customer service, sales, recruiting and marketing.
| Use case | Traditional setup | ElevenLabs setup |
|---|---|---|
| Customer service | Recordings plus IVR | Real-time conversational agents that can handle interruptions |
| Outbound sales | Human outsourcing | AI calls that do not sound obviously robotic |
| Recruiting screens | Online questionnaires and video | AI phone interviews with automated assessment |
| Marketing | Studios and voice actors | Voice cloning plus multilingual generation |
Those areas share three traits: high contract value, large volume and extreme sensitivity to whether speech sounds human. Klarna has previously said AI customer service helped save the equivalent of more than 700 full-time roles; the voice layer behind that kind of shift depends on suppliers at ElevenLabs' level.
The stickiness matters even more. Once ElevenLabs is wired into a support workflow, switching costs become high. Business process, prompts, voice style and compliance audit trails all get tied in. In SaaS terms, ElevenLabs' net retention is likely well above 130%.
Is $11B expensive on $500M ARR?
The valuation implies roughly 22 times ARR. In the current AI market, that is not extreme. OpenAI trades around 25 times revenue, Anthropic around 42 times and Cursor has discussed roughly 25 times. Voice AI also has a broad ceiling: call centers, audiobooks, game dubbing, film and TV localization, education, vehicles and smart homes together form a trillion-dollar addressable market.
Two risks still matter.
First, open-source voice models are catching up. Projects such as Coqui and StyleTTS2 have improved enough for everyday use. Mid-market customers could move to open-source stacks quickly. ElevenLabs has to keep widening the gap at the premium, almost-indistinguishable-from-human end.
Second, OpenAI and Anthropic are building voice. GPT-5.5's realtime voice API already competes directly with ElevenLabs Conversational, with lower pricing and tighter integration with text models. Claude voice features are only a matter of time. The unresolved question is whether the voice AI vertical can avoid being swallowed by general-purpose model companies.
By bringing in BlackRock and NVentures early, Staniszewski has added one more moat before a possible listing: with those names on the cap table, ElevenLabs can either file for an IPO or raise again at a higher valuation.
One detail stands out
The announcement also included a small but important line: ElevenLabs completed its second employee share sale in less than a year, with two rounds of $100 million each.
Two separate $100 million liquidity events for employees are rare in Silicon Valley. They imply three things:
- Cash flow is strong enough that the company does not need to use illiquid equity as its main retention tool.
- Employees own enough stock that the company needs to provide liquidity proactively.
- Secondary-market buyers are valuing ElevenLabs in the same range as the Series D, meaning the $11 billion figure has been tested by real secondary demand rather than set only by VCs.
That detail says more about ElevenLabs' position than the $550 million round itself.
Sources: ElevenLabs lists BlackRock, Jamie Foxx, and Eva Longoria as new investors (TechCrunch); CocoLoop, ElevenLabs adds BlackRock, Nvidia and Jamie Foxx to $550M+ Series D (Tech.eu); ElevenLabs Surpasses $500M ARR as BlackRock, Nvidia and Celebrity Investors Join $500M Series D (The AI Insider)