ARR Leaderboard

Retell AI ARR: $60M, Up 650% Year over Year.

Retell AI sells the infrastructure layer for voice agents — a conversation API with turn-taking, interruption handling and telephony that developers use to build AI phone products. Sacra estimates its run rate at $60M, growing 650% year over year, about twenty-seven months after launch per Leonis Capital. The sourced numbers are below.

~$60M

Run rate, per Sacra

+650%

Year-over-year growth

27 months

Launch to $50M, per Leonis

Developers

The customer, not enterprises

The Retell AI Revenue Curve.

Annualized run rate in USD millions, 2024 to 2026. Retell has not disclosed exact revenue; the curve follows Sacra's estimates, including its 650% year-over-year growth figure.

Sources: Sacra, Leonis Capital.

Launch → $50M ARR

~27 months

Leonis Capital

YoY growth

+650%

Sacra, 2026

Run rate

~$60M/year

Sacra, 2026

Founder

Shangding Gu (Stanford)

Company

Every Reported ARR Milestone.

2023
$0M ARR
Founded by Shangding Gu, a Stanford PhD, to fix the hardest part of voice AI: not speaking, but listening — turn-taking and interruption in real conversation. — Company
2024
$4M ARR
The conversation API launches; developer teams building AI phone products adopt it instead of stitching speech-to-text, LLM and telephony stacks together. — Company / Estimated
2025
$15M ARR
Voice-agent products multiply across reminders, receptionists and outbound sales — each one metering Retell minutes. — Estimated
2026
$60M ARR
Sacra estimates $60M/year, up 650% YoY; Leonis Capital charts roughly twenty-seven months from launch to $50M. — Sacra / Leonis

Retell AI by the Numbers.

Latest estimated ARR

~$60M

Sacra, 2026

Growth rate

+650% YoY

Sacra, 2026

Time to $50M

~27 months from launch

Leonis Capital

Founded

2023

Company

Founder

Shangding Gu

Company

Pricing model

Usage-based, per minute

Company

Disclosed revenue

None; Sacra estimates

—

Metered Minutes: The AWS Pattern for Phone Agents.

Retell's insight is that voice AI products don't fail on the model's ability to speak — they fail on conversation mechanics: when to talk, how to handle interruption, how to stay natural across latency. Retell productized those mechanics into a conversation API with telephony attached, so a developer team can ship an AI phone line without assembling a speech stack.

Pricing is usage-based per minute, which makes revenue compound with every product built on top: each customer's success raises their spend automatically. That's the infrastructure pattern — metered consumption, no enterprise sales cycle — and it's why the 650% growth figure is structurally plausible rather than promotional.

Revenue Mix

Per-minute API usageMetered conversation minutes across voice-agent products.
Telephony & numbersCalling infrastructure billed alongside the API.
Volume tiersDiscounted rates as customer call volume scales.

FAQ

Retell AI Revenue, Answered.

What Is Retell AI's ARR in 2026?

Around $60 million per Sacra, growing about 650% year over year. Retell has not disclosed exact revenue; treat the figure as a research estimate. Leonis Capital independently charts roughly twenty-seven months from launch to $50M.

What Does Retell AI Sell?

A conversation API for building voice agents: turn-taking, interruption handling and natural pacing, plus telephony. Developers use it to ship AI phone products without assembling their own speech-to-text, LLM and calling stack.

Why Is It Growing 650% a Year?

Usage-based pricing on an infrastructure layer: every voice-agent product built on Retell meters minutes, and successful customers raise spend automatically as their call volume grows — no new sales motion required.

Who Founded Retell AI?

Shangding Gu, a Stanford PhD, founded the company in 2023. The founding focus was conversation mechanics — the listening side of voice AI — rather than model quality.

How Is Retell Different from ElevenLabs?

ElevenLabs leads with voice synthesis — making speech sound human. Retell leads with conversation orchestration — deciding when and how the agent talks within a live phone call. Developers often use both layers in the same product.

Usage-Based AI Infrastructure Needs Metered Billing.

Clink meters per-minute API usage, handles prepaid credits and thresholds, and consolidates metered invoices for infrastructure products that bill on consumption.

Meter API Revenue→