A startup incorporated last month, operating in stealth with no public product and no website, is reportedly about to close one of the largest seed rounds in AI history. Emulate, a London company founded by three former Google DeepMind researchers, is in advanced talks to raise up to $700 million at a valuation of about $3.7 billion including the new capital, the Financial Times reported on September 17.

The round is expected to be led by Index Ventures and Lightspeed Venture Partners. If the terms hold, investors will have priced a company with no shipped technology near the going rate for a mature frontier lab. The headline number is striking, but the more interesting story is what it says about where AI money believes the next frontier is.

The deal, as reported#

According to the FT's reporting, Emulate is negotiating a raise of up to $700 million. Two valuation figures have circulated: roughly $3.7 billion including the new capital, and a $3 billion target before the new money — a reminder that the round is in advanced talks, not signed. Briefs reports that Creandum is expected to participate alongside the two lead firms. None of the named parties has confirmed anything publicly: Emulate, Index, and Lightspeed did not respond to press inquiries, and Creandum declined to comment.

What is known is the shape of it: a seed round in the hundreds of millions for a company incorporated in August that remains in stealth, with no product page, no website, and no disclosed headcount. Dealroom notes the deal would rank among the largest seed rounds on record, in any sector.

The founders and the Genie lineage#

Emulate's founders — Jack Parker-Holder, Matthew McGill, and Philip Ball — worked on DeepMind's Genie family of world models. Genie generates realistic videos and interactive 3D environments from a short prompt. When it was unveiled in January, the demonstration rattled public markets: the FT noted billions of dollars in market value evaporating from video game companies including Take-Two, Roblox, and Unity, as investors weighed what generative 3D worlds would do to their business.

Emulate is the third major spinout from DeepMind's London offices this year to command mega-financing. Ineffable Intelligence, launched by DeepMind veteran David Silver, raised $1.1 billion in April to build what it calls a "superlearner." Recursive Superintelligence reportedly secured $600 million at a $4 billion valuation. The pattern is unmistakable: venture capital is pricing DeepMind London alumni like a franchise system for frontier talent.

Why world models are the bet#

A world model is an AI system that learns to simulate how the physical environment behaves: how a glass breaks when dropped, how a robot arm should move to avoid a collision, how a scene evolves over time. Where language models predict the next token, world models predict the next state of a world. A robotics team could rehearse thousands of scenarios in simulation before running a single policy on physical hardware.

The field is drawing attention precisely because it is less crowded than large language models. The Emulate bet is that understanding physical space — not just language — is the next frontier, with implications across several industries:

  • Robotics: training control policies in simulation before touching real hardware.
  • Games and film: generating interactive 3D environments from prompts instead of hand-building them.
  • Simulation and digital twins: forecasting physical processes for industry and research.

What investors are actually buying#

None of Emulate's founders has run a company, and the technology has not been publicly demonstrated — Dealroom's sober caveat. So the bet is on pedigree and timing: the team behind Genie, working in a lane with real but unclaimed upside, at a moment when frontier talent is priced like infrastructure. China blocked Meta's $2 billion Manus acquisition this month; Google has committed tens of billions to Anthropic; SpaceX holds a reported option to buy Cursor for $60 billion. Against that backdrop, $3.7 billion for a world-class world-model team looks like the going rate for optionality.

The closest comparison is World Labs, Fei-Fei Li's spatial-intelligence company, which has raised $1.23 billion and carries a near-$5 billion valuation. The two companies are now the best-funded pure-play bets that world models, not chatbots, define the next AI chapter.

What to watch#

  • Whether the round closes at the reported terms — and which other investors join the syndicate.
  • Whether Emulate can turn the Genie research lineage into shipped products in games, film, or robotics.
  • Whether world-model valuations survive contact with public demos and real customers.
  • Who leaves DeepMind London next. The talent pipeline itself is now the story: three mega-spinouts in nine months is a pattern, not a coincidence.