As artificial intelligence begins to engage with the physical world, innovative labs are emerging to create world models that facilitate the operation of robotics and the modeling of physical objects. Unlike traditional large language models, acquiring the necessary data for these world models has proven challenging, leaving many researchers in search of viable training datasets.
Enter Origin Lab, a startup poised to leverage an unexpected data source: the video game industry. Recently, Origin Lab announced it has successfully raised $8 million in seed funding, spearheaded by Lightspeed Ventures, with participation from SV Angel, Eniac, Seven Stars, and FPV. Notable angel investors include Twitch co-founder Kevin Lin and Cruise founder Kyle Vogt.
"AI systems being developed today must comprehend the dynamics of the physical world and the movement of objects," explained Anne-Margot Rodde, co-CEO and co-founder of Origin Lab. "This critical data is predominantly found within video games."
Origin Lab aims to function as a marketplace, allowing labs focused on world models--such as Yann LeCun's AMI Labs and Fei-Fei Li's World Labs--to purchase high-quality licensed data. For video game companies, this initiative presents an opportunity to generate additional revenue from existing digital assets. Origin Lab will facilitate the transformation of video game assets into formats suitable for training data, which could range from simple rendering tasks to automating extensive gameplay footage.
Rodde noted, "It became evident that the video game sector harbored valuable data, yet there was no established infrastructure to link AI labs with the gaming industry. We essentially built that bridge."
Historically, labs have sought video game footage as a data resource; however, licensing and quality concerns have often hindered progress. In December 2024, OpenAI faced scrutiny when its Sora video-generation model appeared to replicate footage from popular video games and streamers, likely due to training on Twitch streams. Amazon has also expressed interest in utilizing Twitch content for model training.
The successful fundraising round for Origin Lab signals a burgeoning market for training data and highlights the need for startups that can provide essential resources to leading AI labs. Faraz Fatemi, a partner at Lightspeed who oversaw the investment in Origin, remarked on the growing recognition of this opportunity, stating, "The revenue potential for data vendors serving major labs is undeniable."
"These businesses are well-capitalized, and the primary challenge they face is acquiring data," Fatemi added.
As AI continues to evolve and integrate into various sectors, the collaboration between gaming and AI data could pave the way for groundbreaking advancements in understanding and interacting with the physical world.