Origin Lab has closed an $8 million funding round to build a marketplace connecting video game studios with AI companies seeking training data. The platform lets game developers monetize their existing assets while providing AI labs access to high-quality, licensed synthetic data.
The startup addresses a growing bottleneck in AI development. World models, which aim to simulate physical environments and predict how systems evolve over time, require massive amounts of diverse, labeled data to train effectively. Video games offer a natural source: they contain complex 3D environments, physics simulations, and rich contextual information already optimized for rendering.
Game studios sit on terabytes of unused assets, game engines, and environmental data accumulated across decades of development. Origin Lab's platform lets them monetize these resources without releasing proprietary game code or assets. Instead, game companies can license their data or render synthetic training datasets directly through the marketplace.
The timing reflects broader industry trends. OpenAI's Sora, Runway, and other generative video models have demonstrated that synthetic video data accelerates model training. Frontier AI labs increasingly need diverse, controllable data sources to build better world models. Simultaneously, game studios face economic pressure and layoffs, making new revenue streams attractive.
Origin Lab operates as a broker. It handles licensing negotiations, ensures legal compliance, and manages the technical pipeline for data delivery. Game developers set their own pricing. AI labs browse available datasets and purchase what they need.
The $8 million round supports platform development, recruiting, and initial data licensing partnerships. The startup competes indirectly with data companies like Scale AI and Hugging Face, but focuses specifically on the game-to-AI bridge.
The model assumes alignment between video game realism and AI training needs. Games excel at rendering photorealistic environments and simulating physics, but may not always match the distribution of real-world data required for specific AI applications. Origin Lab's value