HeyGen: AI avatar videos for marketers, from single-digit millions to $100M recurring revenue
HeyGen turns a photo or short clip of a person into sales and marketing videos with an AI avatar. Its camera-shy founder built it to skip recording altogether; by November 2025 the company said it had $100 million in recurring revenue.
- Product
- HeyGen
- Founder
- Joshua Xu (co-founder and CEO)
- Launched
- n/a
- Model
- Subscription
Published · 2 min read · 2 sources
Key takeaways
- HeyGen said in November 2025 that it had reached $100M in recurring revenue and almost 200,000 paying customers, up from single-digit millions of revenue the year before.
- Co-founder Joshua Xu started it after his own fear of recording himself on camera.
- Competition is heavy: Synthesia raised $200M at a $4B valuation around the same time.
What it is
HeyGen customers upload a photo or video clip of themselves, which is turned into a sales or marketing video, such as a personalized pitch or a TikTok product tutorial (Forbes). The company has also released a "video agent" that handles voiceovers, visuals and rough scripts (Forbes).
Timeline
- April 2023: HeyGen reaches $1 million in ARR, according to an October 2025 newsletter summarizing the company's announcement (The Signal).
- 2024: HeyGen starts the year "in the single-digit millions of revenue" (Forbes).
- End of June 2025: HeyGen passes $80 million ARR and releases Video Agent (The Signal).
- October 2025: HeyGen announces ARR above $100 million, 29 months after $1 million (The Signal).
- November 2025: HeyGen says it has grown to $100 million in recurring revenue and almost 200,000 paying customers (Forbes).
Why it worked
The founder had the problem. Joshua Xu, "a camera-shy engineer", started HeyGen after realizing AI would let him skip recording entirely (Forbes).
One thing done well. "We've been laser focused on one thing, which is building the highest quality, most realistic avatar that people can use for videos," Xu said (Forbes).
A broad, underserved need. Investor and board member Sarah Guo said "a small minority of people have access to professional production and are comfortable in front of a camera" (Forbes).
Speed as a strategy. Xu's internal playbook describes two-month roadmaps, daily releases and experiments that go from hypothesis to launch within days: "in the AI era, the team that learns fastest wins" (The Signal).
What to learn
- Build for a fear you share. Camera shyness was Xu's own problem, and it describes many of his customers (Forbes).
- Pick a buyer with a budget. HeyGen sells to businesses making marketing videos, not to casual users (Forbes).
- Expand from your wedge. After nailing avatars, HeyGen is going after "the entire video production process" (Forbes).
Risks and what didn't work
- Self-reported numbers. The $100 million figure is what HeyGen "said", not audited revenue (Forbes).
- Crowded market. Synthesia raised $200 million at a $4 billion valuation, and OpenAI's Sora app brought AI video to the mainstream (Forbes).
- Taste is hard to automate. Forbes calls it "an open question" whether AI can replace the human creative taste needed to edit video (Forbes).
FAQ
Short answers drawn from the sections above.
What is HeyGen?
Why did HeyGen work?
- The founder had the problem. Joshua Xu, "a camera-shy engineer", started HeyGen after realizing AI would let him skip recording entirely (Forbes).
- One thing done well. "We've been laser focused on one thing, which is building the highest quality, most realistic avatar that people can use for videos," Xu said (Forbes).
- A broad, underserved need. Investor and board member Sarah Guo said "a small minority of people have access to professional production and are comfortable in front of a camera" (Forbes).
- Speed as a strategy. Xu's internal playbook describes two-month roadmaps, daily releases and experiments that go from hypothesis to launch within days: "in the AI era, the team that learns fastest wins" (The Signal).
What can indie builders learn from HeyGen?
- Build for a fear you share. Camera shyness was Xu's own problem, and it describes many of his customers (Forbes).
- Pick a buyer with a budget. HeyGen sells to businesses making marketing videos, not to casual users (Forbes).
- Expand from your wedge. After nailing avatars, HeyGen is going after "the entire video production process" (Forbes).
What didn't work for HeyGen, and what are the risks?
- Self-reported numbers. The $100 million figure is what HeyGen "said", not audited revenue (Forbes).
- Crowded market. Synthesia raised $200 million at a $4 billion valuation, and OpenAI's Sora app brought AI video to the mainstream (Forbes).
- Taste is hard to automate. Forbes calls it "an open question" whether AI can replace the human creative taste needed to edit video (Forbes).
Sources
Every link cited above, in order of first use.
- 1.Forbes (forbes.com)
- 2.The Signal (readthesignal.co)
Related case studies
More in AI tools →Perplexity: AI answer engine at a reported $20B valuation, with ARR approaching $200M
Perplexity answers search queries with conversational, AI-written responses instead of a list of links. Three years after founding it had raised about $1.5 billion, was reportedly valued at $20 billion and even offered to buy Chrome.
Suno: AI songs from a prompt, $200M in annual revenue, and a lawsuit from all three major labels
Suno lets anyone generate a song from a text prompt. Two years in, it said nearly 100 million people had made music on it, reported $200 million in annual revenue and raised at a $2.45 billion valuation, while fighting copyright suits from Sony, Universal and Warner.
Synthesia: $100 million ARR, then a $4 billion Series E
Victor Riparbelli, Steffen Tjerrild, Matthias Niessner, and Lourdes Agapito founded Synthesia in 2017 to turn scripts into AI-avatar videos. The company said it passed $100 million in annual recurring revenue in April 2025 and, in January 2026, raised a $200 million Series E at a $4 billion valuation.