No Product, No Website, $3.7 Billion: The Genie Creators' Month-Old Startup Emulate Redefines the AI Seed Round

A British AI startup founded just last month by three former Google DeepMind researchers is nearing a $4 billion valuation — despite having no product, no website, and no publicly demonstrated technology. According to a Financial Times report published Thursday, September 17, 2026, London-based Emulate is in advanced talks to raise as much as $700 million in a seed funding round co-led by Index Ventures and Lightspeed Venture Partners, in a deal that would value the company at roughly $3.7 billion including the new capital. Sources caution that the terms have not yet been finalized and could still change.

If the round closes anywhere near those numbers, Emulate will instantly rank among the largest seed rounds in European history — and among the fastest paper valuations ever assigned to a company that has not yet unveiled a single artifact.

Who is behind Emulate

Emulate was founded by Jack Parker-Holder, Matthew McGill, and Philip Ball, all veterans of DeepMind’s world-models effort and key contributors to the Genie project. Parker-Holder led the development of Genie 2, while all three worked on Genie 3, the system DeepMind has described as a general-purpose world model capable of generating interactive, explorable 3D environments from short text prompts. Genie 3’s debut in January 2026 rattled the video-game industry, wiping billions of dollars in market value from companies like Take-Two, Roblox, and Unity in a single day, as investors grasped what on-demand generation of playable worlds could mean for content businesses.

The trio’s new company is working on so-called world models — AI systems designed to simulate and predict how physical environments behave. Where large language models such as GPT and Gemini learn to predict and generate text, world models aim to learn how actions change an environment: physics, object permanence, spatial relationships, cause and effect. It is a research direction that DeepMind itself has pointed toward robotics and autonomous systems as the natural application space.

The appeal is concrete. A robotics company could test thousands of movement strategies in a simulated warehouse — with photorealistic dynamics and none of the hardware cost — before ever deploying a physical robot. Autonomous vehicle developers, game studios, film production houses, and industrial simulation firms all face the same underlying bottleneck: generating realistic, interactive environments at scale is slow and expensive, and world models promise to make it cheap and instant.

Betting on “future capability,” not revenue

The most striking aspect of the deal is what it does not rest on. Emulate has no product, no website, no customers, and — as City AM noted — none of its three founders has previously run a company. Charlie Dai, vice president and principal analyst at Forrester, told City AM that the round reflects investors betting on “future capability creation, not present revenue or commercial traction.”

“Investors are increasingly underwriting elite AI research teams before product-market fit, especially in frontier domains where a breakthrough could create entirely new markets,” Dai said. “The funding reflects confidence in the team’s ability to create category-defining technology rather than any specific product currently in market.”

That framing has become the dominant logic of frontier-AI venture capital in 2026. When the people who built a state-of-the-art system at a major lab walk out the door, investors no longer wait to see what they build — they bid on the team itself, on the theory that the option value of a breakthrough exceeds any discounted cash-flow analysis. Emulate is the third AI lab to spin out of DeepMind’s London offices this year with hundreds of millions of dollars in financing, following fundraises by other ex-DeepMind teams. The most prominent comparison is Ineffable Intelligence, the startup launched by DeepMind veteran David Silver, which raised $1.1 billion in April to develop its “superlearner” — a system designed to generate knowledge from its own experience rather than from human-generated training data.

Dai argues the trend is less about DeepMind losing its edge than about the market repricing elite researchers. “As investors are willing to fund proven research leaders with hundreds of millions of dollars before product launch, some top talent will inevitably pursue independent ventures,” he said. “This reflects intense market demand for elite AI expertise rather than a decline in DeepMind’s technical leadership.”

The world-model land grab

Emulate’s raise also marks world models’ arrival as a distinct investment category. The field is far less crowded than the LLM arena dominated by OpenAI, Anthropic, and Google, and capital is now flowing in to change that. Fei-Fei Li’s World Labs has raised more than $1 billion to develop spatial-intelligence technology, and a growing cluster of companies — from Decart and Odyssey to the robotics-adjacent simulators — are targeting simulation and embodied AI rather than competing head-on with chatbot-scale language models.

The strategic logic is straightforward. Language models are approaching commodity economics at the frontier, with brutal price competition and overlapping capabilities. World models, by contrast, sit upstream of several industries that have proven resistant to LLM disruption: robotics, industrial automation, game development, and physical simulation. Whoever owns the best generative environment model arguably owns the training ground for every embodied AI system built on top of it — the same way whoever owned the best foundation model owned the application layer of the last cycle.

That positioning explains how a month-old company with three founders and no output can command a $3.7 billion price tag. Investors are not valuing what Emulate has made. They are valuing the Genie lineage — the demonstrated ability of this specific team to build world models that stunned an industry — and pre-paying for the next one.

What to watch

The round has not closed, and until it does, the numbers can move. But the signals to track afterward are clear. First, whether Emulate’s first technical reveal matches the Genie 3 bar its founders set at DeepMind — anything less will invite an immediate markdown in sentiment. Second, whether the company follows Ineffable toward self-improving learning systems or stays squarely on environment generation, where its founders’ proven edge lies. Third, how DeepMind and Google respond, given that their own world-model roadmap now has a well-funded competitor staffed by the people who built it.

One month from incorporation to a $3.7 billion valuation is not a normal seed round. It is a statement about where the market thinks the next frontier lies — and about how little evidence venture capitalists now require when the team is right.