ELEVENLABS IS NOW REPORTEDLY WORTH $22 BILLION FROM MAKING AI VOICES
ElevenLabs started with software that could make a computer sound human. Four years later, the company is reportedly being valued at around $22 billion — and its biggest business is no longer simply generating voices.
IT STARTED WITH A VOICE
ElevenLabs was founded in 2022 by Mati Staniszewski and Piotr Dabkowski.
The original idea was deceptively simple:
Type text
↓
AI speaks it
But the company focused obsessively on making the result sound natural.
Not robotic.
Not obviously synthesized.
Not like the old text-to-speech systems people had been hearing for decades.
That made ElevenLabs useful for creators, games, audiobooks, dubbing and other forms of audio production.
Then something much larger happened.
Businesses started using the same technology to talk to customers.
$22 BILLION WITHOUT A NEW $22 BILLION INVESTMENT
The headline valuation needs some context.
In July 2026, Reuters reported that ElevenLabs was exploring a secondary share sale that could value the company at around $22 billion. The proposed transaction would allow employees to sell some of their existing shares to investors rather than giving the company a large new injection of capital.
That is different from saying:
“ElevenLabs raised $22 billion.”
It didn't.
The company's previous major financing was a $500 million Series D in February 2026, which valued ElevenLabs at $11 billion.
So the reported progression is:
2025 employee tender
→ $6.6B valuation
February 2026 Series D
→ $11B valuation
reported 2026 secondary sale
→ ~$22B potential valuation
If completed around that valuation, the company's paper value would have roughly doubled in months.
THE REAL SURPRISE IS ITS REVENUE
A voice generator becoming valuable isn't particularly surprising anymore.
The more interesting number is the business ElevenLabs has built around it.
The company said it ended 2025 with more than $350 million in annual recurring revenue, and that it had already surpassed $500 million ARR in the first four months of 2026.
In a September 2026 interview, CEO Mati Staniszewski told TechCrunch that ElevenLabs had reached approximately $600 million ARR, with more than 55% of the business coming from traditional enterprise customers.
That tells you something important.
ElevenLabs isn't primarily becoming valuable because millions of people want to make funny AI voices.
It is becoming valuable because companies are paying to use voice AI inside their operations.
THE VOICE BECAME THE INTERFACE
Imagine calling customer service.
Normally:
You → phone system → employee
ElevenLabs wants another version:
You → phone system → AI agent
The AI can listen.
↓
Understand what you said.
↓
Respond naturally.
↓
Access information.
↓
Perform actions.
↓
Continue the conversation.
That is a much larger market than voice generation for content creators.
The voice becomes the interface through which software communicates with customers.
ELEVENAGENTS IS THE BIGGER BUSINESS
ElevenLabs has expanded from voice generation into ElevenAgents, its platform for businesses building conversational AI agents.
The company says organizations including Deutsche Telekom, Square, Revolut and the Ukrainian government use its technology for customer support, conversational commerce, citizen engagement, internal training and inbound sales.
This changes the product from:
“Make me a voice.”
to:
“Give my company an AI employee that can talk.”
That is an enormous difference in potential spending.
A creator might pay for audio generation.
A large enterprise might pay for thousands or millions of automated conversations.
THE PHONE CALL IS BECOMING SOFTWARE
For decades, businesses had to build phone infrastructure around humans.
A customer calls.
↓
The system routes the call.
↓
An employee answers.
↓
The employee searches for information.
↓
The employee responds.
AI changes the economics.
Customer calls
↓
AI answers
↓
AI retrieves information
↓
AI responds
↓
AI completes the task
The company doesn't necessarily need a human involved in every conversation.
That doesn't mean humans disappear.
It means businesses can decide which conversations require humans and which don't.
ELEVENLABS IS SELLING MORE THAN A VOICE
This is why the company's expansion beyond text-to-speech matters.
ElevenLabs now describes its technology across:
→ text-to-speech
→ speech-to-text
→ dubbing
→ music
→ sound effects
→ conversational AI
→ voice agents
The company says its ElevenAPI is used by companies including Meta, Epic Games, Salesforce, MasterClass and Harvey, with those integrations reaching more than one billion users.
That turns ElevenLabs into infrastructure.
And infrastructure businesses can become much larger than individual creative tools.
META IS ACTUALLY A CUSTOMER
This is one of the more interesting details.
Meta uses ElevenLabs technology through its API.
That means one of the world's largest technology companies can use ElevenLabs as part of its own products rather than building every voice capability internally.
The business model becomes:
ElevenLabs builds difficult voice technology
↓
other companies integrate it
↓
their users experience the technology
↓
ElevenLabs earns from the infrastructure
The user may never even know ElevenLabs was involved.
That is a powerful position for an AI company.
ENTERPRISE IS CHANGING THE ECONOMICS
Consumer AI products can have huge user numbers but relatively uncertain monetization.
Enterprise contracts are different.
A business might pay because the AI:
→ answers customer calls
→ books appointments
→ handles support
→ qualifies leads
→ follows up with customers
→ translates content
→ trains employees
The company can compare the cost of the AI against the cost of human labor or lost sales.
That makes the value proposition much easier to calculate.
Cost of AI conversation
versus
Cost of human conversation
If the AI can provide an acceptable experience at a fraction of the cost, businesses have a financial reason to adopt it.
ELEVENLABS IS TRYING TO MAKE AI SOUND EMOTIONAL
The next problem is not simply accuracy.
It's emotion.
A human doesn't speak every sentence at the same speed.
People pause.
They hesitate.
They become frustrated.
They become excited.
They change their tone.
Staniszewski told TechCrunch that ElevenLabs is working toward conversational AI that understands emotional cues and can change how it speaks accordingly.
The ambition is essentially:
Don't just understand the words.
Understand the person saying them.
That is much harder.
THE COMPANY HAS AN UNUSUAL DATA PROBLEM
ElevenLabs says it has access to millions of hours of customer-service calls through its enterprise deployments.
But having recordings isn't automatically enough to create better AI.
The company told TechCrunch that a significant part of its work involves annotating conversations — including what was said, how it was said, when people spoke and what emotions were involved. It also uses voice coaches to help identify accents accurately.
That creates an unusual labor layer behind AI.
The machine-learning model may be automated.
But teaching the model what makes a voice sound natural can still require humans.
Raw audio
↓
annotation
↓
specialized training
↓
better models
↓
better conversations
IT IS ALSO BECOMING A CREATIVE COMPANY
Enterprise isn't the only expansion.
ElevenLabs' ElevenCreative platform allows creators and brands to generate and localize audio in more than 70 languages.
The company says brands including Duolingo, NVIDIA and TIME use its technology for content creation and localization.
That creates another economic opportunity.
A company doesn't need to record an entirely new advertisement for every country.
Instead:
one piece of content
↓
AI localization
↓
multiple languages
↓
multiple markets
The cost of globalizing audio can fall dramatically.
DUBBING IS ANOTHER HUGE MARKET
Imagine a video originally produced in English.
Traditionally, localization could require:
→ translators
→ voice actors
→ recording studios
→ editing
→ synchronization
AI can compress much of that workflow.
ElevenLabs can preserve characteristics of a speaker while producing speech in another language.
For streaming, education, advertising and online creators, that can turn one piece of content into a much larger international asset.
THE COMPANY IS MOVING UP THE STACK
There is a subtle but important business strategy here.
At first:
ElevenLabs → voice model
Then:
ElevenLabs → voice API
Then:
ElevenLabs → creative platform
Then:
ElevenLabs → enterprise voice agents
The company keeps moving closer to the actual business problem.
Instead of merely selling the underlying technology, it sells the system that uses the technology.
That usually creates a stronger relationship with customers.
AND IT IS NOT ONLY ABOUT ELEVENLABS' OWN MODEL ANYMORE
Staniszewski told TechCrunch that customers can choose different reasoning models for voice agents.
That means ElevenLabs is increasingly sitting between:
AI reasoning
and
human communication.
A customer could potentially use one model to determine what should be said while ElevenLabs handles the voice, conversation orchestration and related infrastructure.
That makes ElevenLabs less dependent on being the only company with the best general-purpose AI model.
Its specialized advantage becomes:
making AI sound and behave naturally in conversation.
THAT MAKES THE $22 BILLION NUMBER EASIER TO UNDERSTAND
The valuation isn't really a bet on people wanting funny celebrity voices.
It is a bet on a future in which computers increasingly communicate with humans through speech.
If that happens, the market isn't just:
AI voice generation
It becomes:
AI communication infrastructure.
Customer service.
Sales.
Healthcare.
Government.
Education.
Games.
Entertainment.
Advertising.
Personal assistants.
Every one of those categories involves humans talking.
EVEN GOVERNMENTS ARE USING IT
ElevenLabs says the Ukrainian government is among organizations using ElevenAgents.
In an example discussed by Staniszewski with TechCrunch, ElevenLabs described a deployment in Poland where AI agents call patients to remind them about healthcare appointments. The company said the system was designed around local data-residency requirements.
That is a revealing example.
The product isn't being used because people want an entertaining AI voice.
It is being used because a phone call has an operational purpose.
An appointment reminder can reduce missed appointments.
That is a measurable business or institutional outcome.
THE STRANGE PART IS THAT VOICE MAY BECOME INVISIBLE
The ultimate success of ElevenLabs might actually mean people stop noticing ElevenLabs.
You call a company.
An AI answers.
You don't think:
“This is ElevenLabs technology.”
You think:
“The company answered my call.”
That is exactly how infrastructure businesses often work.
The technology disappears into the product.
BUT ELEVENLABS ALSO HAS TO DEAL WITH TRUST
The more convincing an AI voice becomes, the more important disclosure becomes.
Staniszewski told TechCrunch that businesses should currently tell customers when they are speaking to an AI agent. He argued that people are not yet accustomed to these interactions and may feel deceived if they discover it later.
That creates a strange paradox.
The better AI voices become:
the more human they sound
↓
the harder it becomes to distinguish them
↓
the more important disclosure becomes
The company's technical achievement therefore creates a social problem at the same time.
FROM $1.1 BILLION TO A REPORTED $22 BILLION
ElevenLabs became a unicorn in 2024 after an $80 million funding round reportedly valued it at around $1.1 billion.
By September 2025, the company had already completed a $100 million employee tender at a $6.6 billion valuation. ElevenLabs itself confirmed that transaction.
Then came:
$11B
in February 2026.
And now:
~$22B reported
in connection with a potential secondary transaction.
The speed of the valuation increase reflects how quickly investors' view of the market has expanded from AI-generated voices to AI agents communicating with the world.
MAACAT PERSPECTIVE
ElevenLabs' biggest business may no longer be making computers sound human.
It is making businesses capable of talking to humans without putting a human on every call.
The progression is fascinating:
voice generator
↓
voice infrastructure
↓
AI voice agents
↓
enterprise automation
↓
communication layer for AI
That is why a company that began with synthetic speech can now be discussed at a reported $22 billion valuation.
The voice was only the entry point.
The bigger business is the conversation.
