Former Snap software engineer Joshua Xu believes AI-generated video is about to have a moment like Snapchat or Instagram had in the early days of the mobile photography revolution.
As the first proof of this, he points to his own company HeyGen. After launching its AI-powered video creation app last September, HeyGen reached $1 million in annual recurring revenue in March, and then $10 million in August. Today that amount has risen to $18 million, says Xu, co-founder and CEO Forbes.
“Snapchat is a camera company where everyone creates content through the mobile camera,” said Xu. “We think AI can create the content. AI could become the new camera.”
On Wednesday, HeyGen announced $5.6 million in new venture capital funding led by Sarah Guo’s Conviction Partners. The round values the Los Angeles-based company at $75 million; As part of the deal, Guo will take a board seat in place of HongShan (formerly Sequoia China), while HeyGen takes steps to distance itself from its Chinese heritage.
HeyGen is also launching a new product that will make it easier for people to create custom AI avatars that appear in the videos. Previously, HeyGen’s personalized photorealistic avatars required professional photography to create and a process that could take days, although it also offers more than 100 ready-made avatars. Xu said the new product can generate AI avatars using smartphone images in just five minutes – an upgrade he attributes to a breakthrough in the architecture of HeyGen’s AI model.
Xu and Wayne Liang, the Chief Product Officer, studied together at Tongji University in Shanghai, and then at Carnegie Mellon University as master’s students. They both then ended up on the West Coast, Xu at Snap and Liang as a product designer for karaoke app startup Smule and TikTok parent ByteDance. Xu became stuck in China in 2020 after a visit home coincided with the government’s Covid-19 travel restrictions. When he left Snap to found HeyGen later that year, he raised capital from leading companies in his native country, including Sequoia China and ZhenFund.
Since then, rising tensions between the US and China have changed the technology landscape, as evidenced by Sequoia’s legendary geographic split. Xu said it was always his intention to move back to Los Angeles for the company. Since its product launch last year, HeyGen has focused on Western markets (the product is banned in China for reasons Xu says he doesn’t know, though he suspects it’s similar to Beijing’s censorship of ChatGPT).
“The geopolitical situation has changed dramatically in the past year and a half,” says new investor Guo. “[Xu] was extremely firm in saying that we will be very clear about our investor base, our user base, our data centers and that we will have no government influence.”
HeyGen, which now employs 25 people, quickly implemented advances in “diffusion” generative AI models – the technology underlying popular image generators like OpenAI’s Midjourney or Dall-E. Xu said the company has built its own AI models for video, while integrating major language models from OpenAI and Anthropic for text purposes, and offerings from Eleven Labs for audio.
“We have 0.1% penetration among people who even understand this is possible, let alone adopt it.”
Other AI-for-video startups are popping up, like Runway and Pika, which let users generate and edit videos by typing in a text prompt. But those companies focus on creatives and consumers. HeyGen has focused on the business market and the endless need for marketing, training and instructional videos. And that’s going well, but Xu hopes that the new product can give HeyGen additional traction among content creators on YouTube and TikTok.
Still, stabilizing sales flow is a top priority for HeyGen. Last year, Xu described the company to TechCrunch as Jasper for video production. The AI marketing copywriting startup he mentioned has since become a canonical Silicon Valley example of a popular company whose revenue growth declined after its novelty wore off. While HeyGen brings in business customers, many of them are people who work within companies, and not at the companies themselves. HeyGen appointed its first salespeople in November and plans to double its workforce over the next year, with a focus on pursuing enterprise contracts.
In doing so, HeyGen will inevitably face off against rivals like Synthesia, a London-based AI avatar software maker that has raised more than $150 million in VC funding. Xu hopes to differentiate HeyGen with an emphasis on features such as avatar personalization. Guo isn’t worried. She thinks competition won’t be an issue for a while due to the sheer number of untapped customers for different use cases. For example, if Synthesia controls the education market, HeyGen has plenty of open space for marketing videos.
“We have a 0.1% penetration rate among people who even understand this is possible, let alone adopt it,” she said.