Emulate, a stealth-mode British artificial intelligence startup founded by three former Google DeepMind researchers, is in advanced talks to raise up to $700 million in seed funding at a post-money valuation of roughly $3.7 billion, according to reports first surfaced by Bloomberg on September 17, 2026. Index Ventures and Lightspeed Venture Partners are jointly leading the round, which values a company that was only incorporated last month and has no shipped product, published website or disclosed headcount.
A Genie-Era Team Reborn As A World-Model Company
Emulate was formed in August 2026 by Jack Parker-Holder, Matthew McGill and Philip Ball, all of whom worked on DeepMind's Genie 2 and Genie 3 world models. Genie showed that a single foundation model could generate playable, physically consistent 3D environments from an image or a text prompt, and the trio's new venture aims to turn that research direction into a commercial platform for robotics, autonomy and reinforcement-learning teams that need to test agents in synthetic worlds before deployment.
Who's Chasing The World-Model Category
The Emulate raise lands in a category that has already attracted enormous capital: Fei-Fei Li's World Labs has raised more than $1.2 billion at a roughly $5 billion valuation, Yann LeCun has launched AMI Labs to pursue joint embedding-based world models, and Generalist has hit a $3 billion valuation on the strength of its own DeepMind alumni. The bet across all of them is that Physical AI depends on simulators that respect physics well enough to close the sim-to-real gap for humanoids, drones and self-driving vehicles.
Why Investors Are Paying Up Pre-Product
Terms of the round mirror a growing pattern in which frontier-AI teams command billion-dollar valuations on the strength of pedigree alone. Lightspeed and Index have both been aggressive in the space this year, backing several Physical AI and HardTech rounds, and Lightspeed also co-led the recent Series C for Chinese chip and robotics company D-Robotics. Investors see world-model infrastructure as a wedge into the training stack for humanoid robotics, embodied agents and next-generation robotaxi platforms, similar to how Vantora is positioning its venture studio around the same thesis.
What Emulate Has To Prove
Emulate has yet to publish a technical paper, name a chief scientist beyond its three founders, or ship a benchmark result under its own brand. To justify a $3.7 billion post-money valuation, the team will have to convert Genie's academic breakthroughs into a training-and-simulation product that robotics customers can plug into their existing pipelines. If it can, the round positions the UK as one of the most capital-rich hubs for world-model research, a rare bright spot in a European scale-up landscape that has struggled to hold onto its AI winners.
Reporting based on coverage from Bloomberg, Finimize, Dealroom and TechFundingNews.