HeyGen

The video that started everything was supposed to be a secret. In late 2021, Joshua Xu and Wayne Liang sat around a single laptop in a sparse office in Shenzhen. They typed a script in English, selected a Chinese-speaking avatar, and hit generate. What appeared on screen seconds later was not just a video — it was a vision of what global communication could become. Joshua, who doesn't speak Mandarin, played it for his mother. She understood every word. "How much did this cost?" she asked. When he said it was AI, her disbelief was all the validation they needed.

They'd built the proof of concept. Now they had to build the company. That meant navigating scrutiny from US lawmakers about their Chinese investors, dissolving their Shenzhen entity entirely in 2023, rebranding twice — from Surreal, to Movio, to finally HeyGen — and rejecting the AI UGC ad-farm positioning that would have grown faster but meant less. They chose to build something that mattered for business communication. The bet paid off: G2's #1 Fastest Growing Product of 2025, $95M ARR, $500M valuation, and 100,000+ businesses globally.

Why Avatar IV Changed Everything

HeyGen has shipped many model upgrades, but Avatar IV — released April 2025 — was different in kind, not just degree. Previous HeyGen avatars were good. Avatar IV is uncanny-valley-defeating. Upload a single photo. Write a script. What comes back has natural micro-expressions, fluid hand gestures, accurate lip sync across 175 languages, and the kind of eye contact that used to require a professional camera operator with thirty years of experience telling someone exactly where to look. The upgrade drove a measurable spike in enterprise pilots converting to annual contracts within 30 days of launch.

The Full Platform in 2026

• Video Translation — upload any video, get a dubbed, lip-synced version in 175+ languages in minutes; used by McDonald's, the United Nations Development Programme, and the translators who lip-synced Argentina's President Milei at Davos
• LiveAvatar — real-time streaming avatar for live calls, interactive customer service, and always-on virtual representatives; priced separately from the main subscription
• Talking Photo — turn any still image into a full speaking video with natural voice, lip sync, and expressive face dynamics from a single uploaded photo
• AI Podcast Generator — two AI avatar hosts, one topic, a complete podcast episode with B-roll and captions generated automatically
• Video Agents — interactive AI avatars that can answer viewer questions in real time and adapt content dynamically based on viewer responses; in active development
• MCP Integration — generate HeyGen videos directly from Claude, Manus, and OpenAI agents via Model Context Protocol, no API keys required
• HeyGen Mobile — full iOS and Android app with Creator plan access; includes a $10/week plan for users who want to test before committing monthly

Who's Actually Using It

The customer list reads like a deliberate attempt to prove that AI video isn't just for tech companies. McDonald's used HeyGen for its Grandma McFlurry campaign — a video showing families overcoming language barriers. The UN Development Programme used it for its Weather Kids climate initiative. Komatsu trains 65,000 employees in 14 languages. Coursera translates instructor-led courses with preserved teacher voice and emotion. Würth Group reduced video translation costs by 80% across 80+ countries. These aren't edge cases. They're the core use case: big organizations that need consistent, multilingual video at a scale that was economically impossible three years ago.

The Numbers

$65.6M raised total. $60M Series A led by Benchmark at $500M valuation in June 2024 — joining Conviction (Sarah Guo, who also took a board seat), Bond Capital, and Thrive Capital. $95M ARR as of early 2026. Profitable since Q2 2023, which is why the company raised on its own terms rather than out of necessity. 157 employees. Revenue per employee: north of $600K. Among the highest in the AI video category.