Mati Staniszewski: How He Grew ElevenLabs to $330M ARR in 46 Months

ElevenLabs went from a stealth London flat to $330M ARR and an $11B valuation in 46 months — surviving two deepfake scandals along the way. The five-move playbook, run four times in a row.

TL;DR

• ElevenLabs went from $0 to $330M ARR in 46 months, reaching an $11B valuation by February 2026 — from stealth incorporation to one million users in five months, to a $1.1B unicorn 12 months later.
• The company survived two deepfake scandals in 12 months — a 4chan celebrity-voice abuse incident in January 2023 and a fake Biden robocall in January 2024 — and converted each into a public-trust audit it refused to lose.
• Every funding round was a product bundle. Six rounds in a row each shipped alongside a model release or new product — pre-seed through the $500M Series D — turning one announcement budget into multi-category press coverage every time.

The Numbers

ARR for early 2024 is a Sacra/Sifted estimate; the $90M, $200M and $330M figures are company-disclosed. The valuation curve compresses each leg — $1.1B to $11B in roughly 25 months — and by year-end 2025, enterprise was approaching 50% of revenue, with Deutsche Telekom and Revolut publicly cited as customers.

What they did differently

Move 1: Lock the thesis to a multilingual, emotion-preserving frame from day one

Mati Staniszewski (ex-Palantir) and Piotr Dabkowski (ex-Google) grew up in Poland watching badly dubbed American films. The original product idea, sketched in 2020, was fixing that — voice that crosses languages with the speaker's emotion intact.

That single insight pre-committed ElevenLabs to two architectural choices most competitors did not make. First, multilingual from day one: the January 2023 beta shipped with English and Polish, Eleven Multilingual v2 covered nearly 30 languages by August 2023, and by Eleven v3 in June 2025 the count was 70+. Second, emotion as a first-class output — not "TTS that sounds OK" but voice that conveys feeling across language barriers. The audio-tag syntax in v3 — [excited], [whispers], [laughing] — is the natural endpoint of that thesis, six years from the original sketch.

Competitors with a generic text-to-speech framing optimized for narration quality. ElevenLabs optimized for emotional cross-lingual transfer, and the framing constrained the roadmap so that every release — v2, then v3, then audio tags — felt like progress on the same promise rather than a scattershot feature list. The company incorporated in April 2022 and did not ship until January 23, 2023: nine months of model training before any user touched the product.

Move 2: Treat misuse as a forced-trust audit, not a PR crisis

The 4chan incident is the first thing that should have killed the company. Within a week of the public beta, 4chan users cloned Emma Watson, Joe Rogan and Ben Shapiro to spew abusive content.

ElevenLabs responded the next business day. It shipped paid-only voice cloning behind ID verification, an AI-detection classifier, and account-level traceability — all within days. The first crisis became the first public proof of operating discipline. The company did the exact same thing 12 months later: when Pindrop traced a fake Biden New Hampshire-primary robocall to ElevenLabs, the account was suspended within 72 hours and the company gave a clear public statement. The Biden episode became a case-study citation in every "responsible AI" panel for the rest of the year.

Most companies hide misuse. ElevenLabs treated each incident as a chance to publicly demonstrate that the platform had auditable controls. The trust narrative ended up doing real GTM work — large enterprise contracts in 2025 cited operational discipline as a reason to commit.

— From the ElevenLabs growth teardown

Enterprise procurement reads the response, not the incident. That posture requires unusual founder calm — most CEOs would have hidden both stories.

Move 3: Bundle every funding round with a product launch, then move up the stack

Look at the cadence: a $2M pre-seed with the public beta; a $19M Series A with new voice products; an $80M Series B bundled with Voice Marketplace, Dubbing Studio and a Mobile SDK; a $180M Series C ten weeks after Conversational AI v1 shipped; and a $500M Series D bundled with the $330M ARR disclosure and the first public IPO talk. Six rounds, six bundled milestones — every announcement window doubled as a product window.

The platform pivot inside that cadence is what mattered most. Conversational AI v1 (November 2024) integrated text-to-speech, speech-to-text and LLM orchestration into one agent stack. In November 2024 ElevenLabs was selling against other TTS APIs; by mid-2025 it was selling against contact-center incumbents — a much larger market with much larger contracts. The Series C strategic-investor list tells the same story: new checks from Deutsche Telekom, NTT DOCOMO Ventures, RingCentral Ventures, HubSpot Ventures and LG Technology Ventures. Telco, CRM and consumer electronics — not creator tools. The 11-week jump from Conversational AI v1 to a Series C with telco strategics on board is the textbook version of moving up the stack before competitors do.

What you can copy

1. Lock your thesis to a constraining frame on day one. A founding insight that constrains the roadmap — multilingual, emotion-preserving — makes every release feel like progress on one promise instead of a feature grab-bag.
2. Front-run misuse instead of hiding it. If your product gets abused, ship the controls publicly and fast. Enterprise procurement evaluates your response, and a visible audit trail becomes a sales asset.
3. Never announce funding alone. Bundle every round with at least one product launch. Same press budget, three to four times the coverage surface — and ElevenLabs ran this six rounds without breaking it.
4. Move up the stack before competitors do. An API is a smaller market than a platform. ElevenLabs jumped from TTS API to conversational-agent platform in weeks, and the larger contracts followed.
5. Make the product output the share unit. Voice clips autoplay on X, Instagram and TikTok. A generous free tier turns creators into a brand-extension layer you do not pay for — the experiments become Reels and shorts on their own.

What probably won't work for you

The playbook is real, but four preconditions kept it from being available to most teams.

Your category needs an output that is itself a share unit. Audio clips autoplay on every social platform — a Darth Vader clip travels native on TikTok. Most categories — code, data pipelines, infrastructure dashboards — do not have this property. If your output is not a share unit, the creator flywheel will not ignite, and the cheap brand work disappears.

The multilingual founding insight was structural, not bolted on. The dubbing-pain-point origin gave ElevenLabs a pre-committed reason to ship 70+ languages. A US-only TTS startup would not have built that surface, and cannot retrofit it as a marketing claim.

Front-running scandals requires unusual founder temperament. Most CEOs would have hidden the 4chan and Biden-deepfake stories. ElevenLabs absorbed both publicly — a posture that demands calm most founders do not have under live press pressure.

Telco and enterprise demand had to be timed to the platform pivot. The Conversational AI launch hit exactly when Deutsche Telekom-class buyers were procuring voice infrastructure. A year earlier the same product would have had no enterprise buyers. And the headline ARR hides the unknowns — ElevenLabs has never disclosed free-to-paid conversion, and voice AI is one of the most crowded, margin-pressured AI infrastructure categories. Copy the distribution moves; do not assume the unit economics are as clean as the press cycle.

Sources & references

• ElevenLabs raises $2M pre-seed and announces AI speech platform — ElevenLabs blog
• AI voice firm 4chan cloned celebrity voices — Vice
• Voice cloning startup ElevenLabs lands $80M, achieves unicorn status — TechCrunch
• ElevenLabs bans account blamed for Biden audio deepfake — Bloomberg
• ElevenLabs raises $180M Series C at $3.3B valuation — TechCrunch
• ElevenLabs CEO says voice AI startup crossed $330M ARR — TechCrunch
• ElevenLabs raises $500M from Sequoia at an $11B valuation — TechCrunch
• Full timeline: The ElevenLabs Growth Story — every round, model release, and scandal, charted

This case study is part of GrowthHunt's growth teardown series. For another AI rocket built on bundled rounds, read the Lovable teardown; for the lean-and-profitable opposite, the Gamma teardown. Track the fastest-growing AI founders live on GrowthHunt Velocity.

More from the growth playbook.

See it run in the wild.