Cambridge-based World Model Data raises £7m led by Iona Star Capital
This article covers World Model Data, a Cambridge Science Park AI startup, raising £7m in a seed funding round as it emerges from stealth. The funding will support its effort to assemble large-scale, game-derived datasets to train world models for researchers and engineering teams building physical AI systems, including robotics and autonomous vehicles.
World Model Data, a Cambridge Science Park AI startup, has raised £7 million in a seed funding round as it emerges from stealth. The company says the cash will accelerate its effort to assemble large-scale, game-derived datasets for training “world models” — datasets its founders argue are essential for AI systems that must understand physical dynamics and operate safely in the real world.
Why it matters
World models — internal simulators that let AI predict how environments change over time — are increasingly cited as the next step beyond large language models for tasks that require planning and physical reasoning. But training those models needs richly annotated, action-conditioned data that is scarce outside controlled environments. That scarcity is a practical constraint for industries such as robotics and autonomous vehicles, where trial and error in the real world is costly or unsafe.
By harvesting and structuring gameplay data, World Model Data targets that gap. If it can scale the dataset as planned, researchers and engineering teams building physical AI systems could access a broader foundation for training models to predict outcomes and plan actions.
Inside the product
World Model Data says it collects gameplay data from titles built on engines such as Unreal and Unity and delivers it as curated datasets. The company emphasises formal licensing agreements with game developers and players as the mechanism for sourcing data rather than automated web scraping, and it says the approach enables the gaming community and asset creators to monetise their work.
The startup aims to assemble one million hours of training data by the end of 2026; it contrasts that target with what it calls the current largest comparable database of about 40,000 hours. World Model Data positions these datasets for use by labs developing world models, companies building physical AI systems, and robotics teams that need realistic, human-driven behaviour in dynamic simulations. An example use case given by the company is enabling a world model to act as an internal simulator for self-driving cars to predict pedestrian movement and navigate traffic.
Founder perspective
In the announcement, Rhea Loucas, Founder and CEO, World Model Data, said:
World models represent a fundamental paradigm shift in AI, but progress like this needs fuel: internet-scale data to help AI systems make predictions and reason in physical environments.
Such a comprehensive dataset does not yet exist; however, video games, as safe, controlled environments, are the perfect setting for generating the action-conditioned data needed to train the next generation of AI at the required scale. Worldmodeldata is built to bridge that gap.
Who are the investors?
The £7 million seed round was led by Iona Star Capital, a London-based venture capital fund that focuses on early-stage companies at the intersection of AI, data and technology. The company has not disclosed additional participating investors.
The funding will be used to develop the product, expand the team, and secure further data-sourcing agreements. The announcement also names Lord Richard Allan — a UK technology policy specialist and former Meta vice president of public policy — as a supporter who joins the board as Chairman.
In the announcement, Richard Allan, Chairman, said:
We are proud to anchor ourselves in the UK’s AI ecosystem, a strategic choice driven by the urgent push for sovereign AI capabilities and the robust infrastructure that powers them. The deep expertise of this team is hugely exciting and positions them perfectly to lead this charge.
In tackling critical gaps that LLMs cannot bridge, this isn’t about improving AI model training, but building an essential foundation for deploying AI in sectors where the demand is vast but the solutions remain limited.
If you're researching potential backers in this space:
- Explore our list of ai venture capital firms in the UK
- Browse our directory of ai angel investors in the UK
The bigger picture
The deal lands amid broader interest in physical AI and training-data plays. Seedcamp has recently set up a fund aimed at physical AI, signalling investor appetite for companies that provide the datasets and tooling needed for embodied intelligence. For the UK, projects that stitch together domestic data capacity with licensing-based sourcing speak to both commercial and policy priorities around sovereign AI capability.
If World Model Data can reach its stated scale and secure robust licensing relationships, it would offer a distinct alternative to synthetic or scraped datasets. That would be relevant to UK and European labs and companies looking to avoid overreliance on third-party web data and to train models for real-world safety-critical applications.
The outcome of this initiative will be worth watching for anyone tracking how the UK builds infrastructure — both commercial and regulatory — for the next wave of AI that must operate in physical environments.
They've invested in World Model Data
Get to know these ai investors
| Investors | Investment Focus | Startup Investments | Round Size | Connect |
|---|---|---|---|---|
 Iona Star Capital( ) | ||||
| All investors | All investor sectors | All funded startups | All funding rounds |
Click here for a full list of 7,589+ startup investors in the UK