After the explosion of interest in text- and image-generators, using AI to generate videos is considered the next frontier and start-ups and Big Tech companies alike are investing in the space. Video tools from OpenAI and Google aren’t publicly available yet, so start-ups like Pika are moving fast to expand before the bigger companies put out their own commercial and consumer-focused tools. But creating a video with AI is much more technically difficult than making a still image, and requires a huge amount of computer processing power, making it an expensive and slow process.
Pika’s new funding round, which includes investments from Spark Capital, Lightspeed Venture Partners and Greycroft, brings its valuation to $470 million. Pika has built its own “foundation model,” meaning it doesn’t rely on tech from other AI companies. But that’s expensive and the company needs money to continue training its AI while also improving the features in its video-editing tool.
Pika’s tool works by letting people generate short videos based on text prompts. Users can change the aspect ratio, make the videos longer, trim out specific sections and change certain elements of a video, such as the kind of clothes a character is wearing.
Pika’s team of 13 people includes former Google, Meta and Uber AI researchers. Founders Demi Guo and Chenlin Meng have PhDs in AI from Stanford. In an interview, Guo said she believes the company has the AI expertise to compete with better-funded organizations.
GET CAUGHT UP
Summarized stories to quickly stay informed
“Our technical team is coming from all the top AI places,” Guo, who is also Pika’s CEO, said. “We are very confident in building the best video foundation model.”
Though AI video tools are still relatively rudimentary, static AI images are already common on social media platforms. Last week, an AI-generated image of a refugee camp and the words “all eyes on Rafah” went viral around the world as social media users signaled their support for Palestinians.
But the most popular and lucrative social media platforms in recent years have been video-based. Dozens of apps exist to help regular people edit videos to post on Instagram or TikTok and the big social media platforms have begun experimenting with putting AI features into their apps to let people create their own images.
The proliferation of AI image generators has already led to the tools being used for propaganda purposes. In India, deepfake images and audio were used by many candidates during the country’s election. Deepfake detection tools have been developed, but they often struggle to correctly identify an AI image or audio clip.
The funding round brings Pika’s total funding to $135 million. Bigger AI companies such as OpenAI have access to billions of dollars and hundreds of staff. OpenAI has begun building a consumer business, and video-editing tools could be part of that, creating significant competition for smaller companies like Pika.
Guo compared her company to other AI start-ups, like French company Mistral AI, that have less funding but still develop important and sought-after technology. The way Pika designs its AI means it doesn’t need the huge budgets that bigger companies chew through, she said.