Origin Lab raises $8M to help game companies monetize data for building AI worlds.

As artificial intelligence expands its reach into the physical world, a new generation of laboratories is actively developing "world models" designed to operate physical robotics or accurately simulate objects within physical environments. Crucially, unlike large language models, these emerging models lack an easily accessible data source, compelling many research institutions to laboriously compile the necessary training datasets.

Now, a promising startup is emerging with an innovative solution, tapping into an unexpected yet rich data reservoir: the video game industry.

This pioneering premise underpins Origin Lab, which recently announced the successful closure of an $8 million seed funding round. Lightspeed Ventures led the investment, with additional participation from SV Angel, Eniac, Seven Stars, and FPV. Angel funding also came from notable figures such as Twitch co-founder Kevin Lin and Cruise founder Kyle Vogt.

Anne-Margot Rodde, co-CEO and co-founder, articulated the fundamental challenge to TechCrunch: “The AI systems that are being built now need to understand how the physical world works and how things move.” She then identified the key insight: “That data essentially lives in video games.”

In essence, Origin Lab will function as a specialized marketplace. It will enable labs focused on world models, including prominent entities like Yann LeCun’s AMI Labs or Fei-Fei Li’s World Labs, to acquire high-quality, licensed data. Conversely, video game companies will gain a new avenue to generate additional revenue from the digital assets they have already created. Origin Lab's crucial role in the middle involves converting these video game assets into a format suitable for AI training data, a process that could range from straightforward rendering operations to the complex automation of extensive walkthrough footage.

Rodde further elaborated on the market opportunity: “It became clear that the video game industry was sitting on some incredibly valuable data, but there was no real way or infrastructure to basically connect AI labs and the video game industry.” She concluded, “So essentially, we built that bridge.”

While research labs have long recognized the potential of video game footage as a data source, widespread utilization has historically been hindered by licensing complexities and data quality concerns. A notable incident occurred in December 2024 when the initial version of OpenAI's Sora video-generation model seemingly regurgitated footage from popular video games and streamers, presumably due to training on Twitch streams. Amazon has also openly acknowledged its interest in leveraging Twitch footage for model training.

Origin Lab's success in fundraising signals the rapid expansion of a market not just for specialized training data, but also for startups capable of serving as essential suppliers to major AI laboratories. Faraz Fatemi, a partner at Lightspeed who spearheaded the Origin investment, observed that the proven success of companies like Scale.AI has made this emerging opportunity impossible to overlook.

Fatemi conveyed to TechCrunch the significant revenue scaling potential in this sector, stating, “We’ve seen how sharp the revenue scaling can be for data vendors that are serving the major labs.” He underscored the critical bottleneck these well-capitalized businesses face: “These are very well-capitalized businesses, and the bottleneck for all of them is data.”