# Genesis AI is in talks to raise $500 million at a $3 billion valuation just a year after emerging from stealth

> Source: <https://startupfortune.com/genesis-ai-is-in-talks-to-raise-500-million-at-a-3-billion-valuation-just-a-year-after-emerging-from-stealth/>
> Published: 2026-07-26 16:37:24+00:00

*The Franco-American robotics startup has gone from a $105 million seed to a reported $3 billion valuation in twelve months, a 28x markup that signals AI-for-robotics is producing the kind of nosebleed numbers once reserved for pure LLM plays.*

Bloomberg reported on July 23 that Genesis AI is negotiating a roughly $500 million funding round, with Premji Invest expected to lead and HSG and Northzone also in talks. The pre-money valuation being discussed sits at $3 billion. That's not a typo. Twelve months ago the company didn't exist publicly. It emerged from stealth in July 2025 with a $105 million seed backed by Eclipse Ventures, Khosla Ventures, Bpifrance, Eric Schmidt, and Xavier Niel. The round isn't closed, and terms in late-stage negotiations shift. But even the fact that this number is being said out loud, by people serious enough for Bloomberg to print it, tells you something about where the money is moving.

Genesis AI's pitch is full-stack physical AI. Not just a model that tells a robot what to do, but the entire chain: proprietary tactile-sensing gloves that map human hand movements one-to-one onto a robotic hand, a simulation engine to accelerate training, and GENE-26.5, the foundation model that ties it together. In May, TechCrunch covered a live demo of the system. The robotic hand cooked a 20-step meal, solved a Rubik's Cube mid-air, and played piano. Genesis claims its data-collection setup costs a hundred times less than typical alternatives while capturing five times more usable training data. You can argue about benchmark claims, but the demo was real hardware doing things most robotic grippers can't.

## Why the valuation makes sense - if you squint

The 28x markup in a year sounds insane until you put it next to the LLM funding wave of 2023 and 2024. Anthropic went from a $4.1 billion valuation to over $60 billion in roughly the same timeframe. OpenAI's trajectory was steeper. The pattern is repeating in robotics. Marquee technical backers signal credibility to the next tier of investors. Rapid model-shipping velocity proves the team can execute. And the category narrative is large enough that any VC who misses it faces an awkward meeting with their LP base in three years. Genesis has Schmidt and Niel on the cap table, shipped GENE-26.5 within months of its seed close, and is targeting automotive, electronics, pharmaceuticals, and logistics - industries where conventional robotic grippers have been failing at fine manipulation for decades.

That combination is now the template. It doesn't mean the company is worth $3 billion on any discounted cash flow basis. It means enough capital is chasing physical AI that a startup with the right pedigree and a working demo can command a valuation that would have seemed absurd for a company at this stage even two years ago. Frankly, investors aren't pricing what Genesis is worth today. They're pricing what they think the category does in five years and back-calculating from there.

## A crowded field, and one genuine differentiator

The competitive picture makes this less surprising than it looks. Physical robotics is jammed. Figure AI raised at a $2.6 billion valuation in early 2024 and has been rumoured at higher since. Physical Intelligence, the Google-backed startup founded by former OpenAI and Google Brain researchers, raised $400 million at a $2.4 billion valuation last year. 1X Technologies, Apptronik, Agility Robotics - the list of well-funded humanoid and manipulation-focused startups is long enough that it's starting to resemble the early LLM market, where every serious team had a nine-figure check and a plausible claim to be building the foundational layer.

Genesis's differentiator, if it holds, is that full-stack bet. Most competitors are either model companies that rely on third-party hardware or hardware companies that license or build their own AI layer separately. Genesis is trying to own both simultaneously, which is expensive and operationally complicated but creates the kind of vertical integration that's hard to replicate quickly. Whether that moat is real or theoretical is something the next 18 months will answer.

The Northzone connection is worth noting too: the firm was already in the seed round. If the Series A reporting is accurate, they're doubling down rather than taking dilution. That's a different signal than a new name leading a round. It suggests the people closest to the company's internals like what they've seen since July 2025.

None of this is a guarantee. Late-stage negotiations collapse, terms shift, and the robotics category has a long history of promising demos that don't survive contact with real factory floors. Genesis hasn't shipped product at commercial scale yet. The gap between a convincing manipulation demo and a system that works reliably in an automotive plant is still wide. But the same was true of every LLM startup before the market decided it didn't care about that gap and started writing checks anyway. If the $500 million round closes near the reported terms, Genesis AI will enter the second half of 2026 as one of the best-capitalised robotics companies in the world, with a mandate to find out whether physical AI actually works at scale, or whether it's the next category that raised a decade's worth of capital in two years and then had to figure out the hard part.

**Also read:** [Claude shared chats have been indexed by Google and anyone with a search bar can find them](https://startupfortune.com/claude-shared-chats-have-been-indexed-by-google-and-anyone-with-a-search-bar-can-find-them/) • [Anduril is chasing a $100 billion valuation before its last round has even settled](https://startupfortune.com/anduril-is-chasing-a-100-billion-valuation-before-its-last-round-has-even-settled/) • [Singapore's Ropedia raises $30M to teach robots how to do chores by watching humans](https://startupfortune.com/singapores-ropedia-raises-30m-to-teach-robots-how-to-do-chores-by-watching-humans/)
