A new frontier in artificial intelligence, focused on building ‘world models’ to understand and interact with the physical environment, faces a critical challenge: a scarcity of suitable training data. Unlike large language models with vast text corpora, these next-generation AI systems require comprehensive datasets illustrating physical movement and object interaction. Addressing this need, Origin Lab has emerged with an innovative solution, securing an $8 million seed funding round led by Lightspeed Ventures, with participation from SV Angel, Eniac, Seven Stars, and FPV, alongside angel investors Kevin Lin (Twitch co-founder) and Kyle Vogt (Cruise founder).

Origin Lab’s premise centers on the video game industry as an untapped reservoir of highly valuable data. Co-CEO and co-founder Anne-Margot Rodde, along with co-founders Antoine Gargot and Colin Carrier, assert that the intricate digital worlds within video games contain precisely the kind of information AI systems need to comprehend physical mechanics. The startup functions as a crucial marketplace, enabling world-model-focused AI labs, such as Yann LeCun’s AMI Labs or Fei-Fei Li’s World Labs, to acquire high-quality, licensed data. On the supply side, video game companies can generate additional revenue by monetizing their existing digital assets, which Origin Lab then converts into suitable training data, ranging from simple rendering runs to automated analyses of extensive gameplay footage.

Historically, AI researchers have shown interest in video game footage, but challenges related to licensing and data quality have often impeded its widespread use. Origin Lab aims to resolve these long-standing issues by providing the necessary infrastructure to connect these two industries. Faraz Fatemi, a partner at Lightspeed who spearheaded the investment, highlighted the growing market for data vendors serving major AI labs. He noted the substantial revenue scaling potential for companies that can supply high-quality data, emphasizing that data remains a primary bottleneck for many well-capitalized AI development efforts.