Worldmodeldata, a Cambridge-based artificial intelligence startup, has raised £7 million in seed funding to develop and scale its platform for generating training data from video games. The round was led by Iona Star Capital.
The company addresses a growing challenge in AI development: the scarcity of high-quality datasets needed to train world models and robotics systems. Worldmodeldata’s approach leverages video game environments as sources for synthetic training data, capitalizing on the rich, controlled settings that games provide for teaching AI systems to understand and reason about physical environments.
Building AI Training Data at Scale
The newly raised capital will be deployed across three primary areas: accelerating product development, expanding the team, and establishing new data licensing agreements with partners. The company has set an ambitious target to develop a library containing one million hours of training data by 2025, positioning itself as a significant resource for organizations developing next-generation AI systems.
Rhea Loucas, associated with the company, explained the rationale behind the approach: “World models represent a significant shift in AI, but they require large-scale datasets that enable systems to understand and reason about physical environments. Video games provide rich, controlled environments that can generate the data needed to train these models, and our goal is to make that data available at scale.”
The use of synthetic data from gaming environments offers several advantages over traditional data collection methods. Video games can generate consistent, labeled data at scale while providing controlled conditions that would be difficult or costly to replicate in real-world scenarios. This approach has gained traction across the AI industry as companies seek to train increasingly sophisticated models without the limitations and expenses associated with manual data annotation.
Strategic Focus on Robotics and World Models
World models—AI systems designed to predict and understand how physical environments will evolve—represent a frontier in artificial intelligence research. These systems require extensive training on visual and temporal data to function effectively, making access to large, diverse datasets critical for development. The robotics sector similarly depends on high-quality training data to enable machines to navigate and interact with their surroundings.
Worldmodeldata’s focus on establishing data licensing agreements suggests a business model centered on providing data infrastructure to other organizations rather than building end-user applications directly. This positions the company within the growing ecosystem of AI infrastructure providers that support the broader development community.
European AI Infrastructure Development
The funding reflects continued investment in European AI infrastructure companies, particularly those addressing fundamental technical challenges in the field. Cambridge has established itself as a significant hub for AI research and development, building on the region’s academic strengths and growing concentration of AI-focused companies. The success of funding rounds like Worldmodeldata’s demonstrates investor confidence in European teams tackling critical bottlenecks in AI development, from data availability to model training approaches.