ElevenLabs
Usage-based + subscription API/SaaS: creator and prosumer tiers (per-character/credit) plus enterprise voice-agent and API contracts; land-and-expand from creative tools into customer-facing voice agents.
Primary rounds (Series A-D) are company-disclosed and press-verified; the Jul 2026 ~$22B point is an indicative secondary/liquidity mark from early tender talks, not a closed or audited primary raise. Series A valuation (~$100M) is an approximation from press reporting.
Earnings, margins, COGS & capex
One of the fastest-scaling software companies on record by ARR: ~$100M reached during 2024, ~$330M at YE2025, and $500M+ by spring 2026, with ~$150M net-new ARR added in the first four months of 2026. Growth is increasingly led by the voice-agent (ElevenAgents) and enterprise API business rather than the original creator TTS product. ~41% of the Fortune 500 reportedly use the platform, and ElevenAPI reaches 1B+ end-users through customer apps (company-stated). Profitability and unit economics are undisclosed; the business is compute- and research-intensive, so the case rests on a durable quality lead and expansion, not a disclosed earnings floor.
Revenue trend
Margins
pressured by inference/compute cost and price competition on commodity TTS
reinvesting for growth; likely negative to breakeven
very high vs SaaS norm; small headcount, high leverage
COGS structure
Primarily GPU inference/compute for model serving (largely NVIDIA hardware via cloud), plus cloud hosting and voice-talent licensing/royalties for the voice marketplace. Not itemized publicly.
Capex
No disclosed physical capex; compute is consumed as cloud opex. Heavy ongoing R&D/model-training spend funded by Series C/D proceeds.
Latest earnings
N/A (private)
No formal guidance. Management (CEO Mati Staniszewski) has signaled building toward an IPO but has named no timeline or venue; earlier reports of a specific IPO date or a Warsaw Stock Exchange listing are unconfirmed.
- ARR (spring 2026)
- $500M+
- Fortune 500 penetration
- ~41% (company-stated, early 2026)
- Employees
- ~530 across 50+ countries (company-stated; some third-party sources cite ~400)
- Category entry share
- ~95% of first-time voice-AI buyers start with ElevenLabs (YipitData mid-market panel, 3mo to Jan 2026, up from 81% prior year)
- Mid-market category spend
- ~98% of observed spend across ~900-company panel (Jan 2026, YipitData)
Growth drivers
- Shift from creative TTS/dubbing into higher-value enterprise voice agents (ElevenAgents) for support, sales, and hiring
- ElevenAPI embedding into partner apps reaching 1B+ end-users (company-stated; named customers include Duolingo, Epic Games, HarperCollins)
- Enterprise land-and-expand: ~41% of Fortune 500 reportedly on-platform; net-new revenue concentrated in agents
- Planned multimodal expansion (image/video generation integrated with audio)
- Strategic-investor distribution — NVIDIA, Salesforce, Deutsche Telekom, KPN, Santander as both backers and channels
Bull & bear
ElevenLabs is becoming the default voice layer of the internet, converting a dominant creative-tools funnel into a high-value enterprise voice-agent business that is compounding ARR faster than almost any software company ever; secondary talks near $22B (double the Feb 2026 primary mark) show investor conviction ahead of an IPO.
- $500M+ ARR with ~$150M net-new added in the first four months of 2026 -- reached in ~3 years from launch
- Category ownership: ~95% first-buyer entry share and ~98% mid-market spend give ElevenLabs a distribution moat competitors must overcome (YipitData)
- Voice agents (not commodity TTS) drive net-new revenue -- a stickier, contract-based, higher-ACV motion with ~41% of Fortune 500 already on-platform
- Strategic backers (NVIDIA, Salesforce, Deutsche Telekom, Santander, KPN) double as enterprise distribution channels
- Valuation momentum: $3.3B (Jan 2025) to $11B primary (Feb 2026) to ~$22B secondary talks (Jul 2026) -- repeated up-marks from top-tier crossover names
The core TTS product is commoditizing under hyperscaler and open-source pressure, margins are undisclosed and likely thin, and an $11B primary / ~$22B secondary mark (~22x-44x ARR) bakes in flawless execution on the unproven, competitive voice-agent transition -- leaving real downside if growth decelerates or a price war compresses the multiple into an IPO.
- Google Gemini TTS and OpenAI match quality at a fraction of the price; base TTS is a race toward zero and ElevenLabs' premium is eroding
- Margins/profitability never disclosed -- compute-heavy COGS plus GPU-supply and inference-price risk could mean weak or negative unit economics
- No infrastructure moat: fully dependent on rented NVIDIA GPUs and hyperscaler clouds that are also competitors
- Voice-cloning is a regulatory and litigation magnet (right-of-publicity, deepfakes, EU AI Act; the 2024 Biden-robocall episode) that can constrain the highest-value features
- A ~$22B secondary tender is a preliminary, unclosed, unaudited mark at peak AI enthusiasm (~44x ARR); a growth stumble or down-round would reprice sharply into an IPO
What it is worth
ARR multiple vs private AI-infra/app comps, triangulated to the last primary round ($11B post-money on ~$500M ARR = ~22x ARR, Feb 2026) and cross-checked against secondary-market marks (~$22B tender talks, Jul 2026, ~44x ARR -- preliminary, unclosed, unaudited).
~$6-9B
price war compresses margins, growth decelerates below 25%, multiple resets to ~8-12x ARR -- a down-round/repricing on commoditization and undisclosed-unit-economics concerns.
~$11-15B
growth moderates to 30-40%, ARR toward $700-800M, multiple compresses toward ~15-20x as TTS commoditizes but agents scale -- roughly flat-to-up from the last primary mark.
~$22-28B
ARR reaches ~$1B in 12-18mo, agent mix lifts gross margin and retention, ~22-28x ARR holds -- the secondary-talk level sustained into a strong IPO.
No public price. $11B is the Feb 2026 Series D primary mark on ~$500M ARR (~22x). A ~$22B secondary tender is in early talks (Bloomberg, Jul 2026) but had not closed as of July 2026 and is a liquidity/secondary mark, not a fresh primary raise. Either multiple is defensible only if 40%+ growth persists and the agent business proves durable margins; both are unproven. Repricing risk is asymmetric into an IPO.
SWOT
Strengths
- Best-in-class voice quality and breadth (TTS, dubbing, cloning, multilingual) with the strongest brand in voice AI
- Dominant top-of-funnel — ~95% of first-time voice-AI buyers start with ElevenLabs; ~90% of mid-market panel use it exclusively (YipitData)
- Extraordinary capital efficiency: ~$500M ARR on ~530 staff; among the fastest ARR ramps in software history
- Deep-pocketed, strategic cap table (Sequoia, a16z, ICONIQ, NVIDIA, Salesforce, Deutsche Telekom, BlackRock, Wellington) providing capital and distribution
Weaknesses
- Undisclosed margins/profitability; compute-heavy COGS exposed to inference-price and GPU-supply dynamics
- Core TTS is increasingly commoditized -- Google/OpenAI/Microsoft undercut on price at near-parity quality
- Heavy dependence on third-party cloud/GPU (mainly NVIDIA) with no owned silicon or datacenter moat
- Voice-cloning/deepfake exposure (e.g. the Jan 2024 Biden robocall tied to its tech) creates reputational, legal, and regulatory tail risk
Opportunities
- Voice agents as the durable, higher-margin, stickier layer (support/sales/hiring automation) -- the real TAM
- Enterprise + regulated-industry expansion (telecom, finance, government) via strategic investors as channels
- Multimodal expansion (image/video + audio) broadening from voice into full generative media
- IPO optionality provides a liquidity and currency path for M&A and talent; interim secondary tenders give employees liquidity
Threats
- Hyperscaler bundling — Google, Microsoft (Azure/OpenAI), and Amazon can give away 'good-enough' TTS inside existing clouds
- Open-source/near-free models (Meta, Kokoro, others) collapsing willingness-to-pay for base TTS
- Well-funded specialists (Cartesia, Inworld, Hume, Deepgram) attacking on latency, expressiveness, and price
- Regulation of synthetic voice / deepfakes (EU AI Act, US state right-of-publicity/deepfake laws) raising compliance cost and constraining cloning features
Moats, dependencies & bottlenecks
Moats
~95% first-buyer entry share; default choice, but stickiness is weaker for commodity TTS than for agents.
Real today but the gap is narrowing fast vs Google, OpenAI, Inworld, Cartesia.
Moderate-Strong Agent deployments embed into workflows/telephony; the durable moat if the agent business scales.
Marquee actors and a curated voice catalog are hard to replicate; also a legal-exposure surface.
NVIDIA, Salesforce, telecoms as investors-and-channels; helpful, not defensible alone.
Dependencies
Supplier / compute Inference and training run on NVIDIA hardware; NVIDIA is also an investor -- supply and pricing directly hit COGS.
Hosts serving/inference; the same vendors (AMZN, MSFT, GOOGL) are also direct competitors.
Marketplace and cloned voices require rights; disputes/regulation could restrict the catalog.
Revenue concentration Large-app embeds (1B+ end-users, company-stated) mean a few platform relationships carry meaningful volume.
EU AI Act + US state right-of-publicity/deepfake laws can constrain the highest-value cloning features.
Advantages
- Fastest ARR ramp in category with dominant brand and first-buyer share
- Land-and-expand from cheap creative funnel into high-ACV enterprise agents
- Elite, capital-rich cap table doubling as distribution
- Extreme capital efficiency (~$940K ARR per employee)
- Product breadth (TTS, dubbing, cloning, agents, API) across creator-to-enterprise
Weaknesses
- Compute-cost-exposed COGS with undisclosed and likely-pressured margins
- No owned silicon or datacenter -- fully dependent on competitor-controlled infrastructure
- Core TTS commoditizing under hyperscaler and open-source pressure
- Voice-cloning legal/regulatory tail risk
- Private $11B-$22B marks leave down-round/repricing risk if growth stumbles
Bottlenecks
- GPU/compute availability and cost as inference volume scales toward billions of end-users
- Commoditization of base TTS forcing a pivot to agents before the multiple resets
- Regulatory/compliance overhead for synthetic-voice consent and deepfake safeguards
- Talent competition with hyperscalers and other AI labs for voice/audio researchers
- Undisclosed unit economics -- proving durable gross margin under price competition
Top signals & trends
Top signals
Growth is accelerating, not decelerating -- the core bull data point.
Crossover/PE names entering ahead of an IPO signal institutional conviction.
Preliminary and unclosed; a secondary/liquidity mark, not a primary audited valuation.
Direct margin/pricing pressure on the core product.
The quality/latency lead is contestable, not owned.
Enterprise stickiness building where it matters.
Trends
The durable, high-ACV TAM ElevenLabs is pivoting toward.
Erodes creator-tier pricing power; forces the agent pivot.
Google/Microsoft/Amazon can undercut via distribution.
Expands ElevenLabs' surface beyond voice if it executes.
Compliance cost, but also a moat vs unsafe open-source clones for enterprises.
Supports the elevated marks and an eventual public exit.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
GPUs for training/inference; also a Series D strategic investor (via NVentures).
Cloud hosting/inference (also a competitor).
Cloud infrastructure (also a competitor).
Cloud infrastructure (also a competitor).
AI voice for language learning via API.
Customer and strategic investor; voice for CRM/agent products.
Named enterprise customer.
Enterprise customer and strategic investor (via T.Capital).
Epic Games / HarperCollins / MasterClass / TIME / Revolut / Harvey Private/enterprise customers across gaming, publishing, media, fintech, legal.
Gemini TTS competes on quality at a fraction of the price; bundled into GCP/Workspace/Android.
Azure AI Speech in Foundry plus preferred access to OpenAI models; enterprise/FedRAMP reach.
Private. TTS/Realtime voice APIs at competitive price/quality; huge developer distribution.
AWS Polly / Nova Sonic; commodity TTS bundled in AWS with distribution and price leverage.
Open-source voice/audio research collapses willingness-to-pay for base TTS.
Riva/NeMo speech stack; also ElevenLabs' key supplier and investor -- coopetition.
Private. Expressiveness/latency-focused TTS; ranks near the top of third-party voice-quality leaderboards.
Private. State-space Sonic models; sub-100ms time-to-first-audio, a latency leader.
Private. Speech-to-text + TTS for voice agents; developer/enterprise focus.
Private. Emotionally expressive/empathic voice; differentiates on affect.
Private. Next-largest category-entry driver but a small fraction of ElevenLabs' mid-market volume and declining (YipitData).