AI Startups

General Intuition Hits $6B, Lambda Eyes $3B IPO: AI Startups to Watch for August 25, 2026

General Intuition triples to $6B, Lambda nears a $3B pre-IPO, and robotics labs keep raising—today’s AI startups to watch.

Daniyal

· 3 min read
A robot hand rising from a grid of video-game data frames in gold and blue

Money is chasing physical AI, and the multiples are getting aggressive. A world-model startup nearly tripled its valuation in eight weeks, a Nvidia-backed neocloud is eyeing a public debut, and robotics labs keep stacking nine-figure rounds. These are the AI startups to watch this week.

General Intuition: nearly tripled to $6 billion in eight weeks

New York-based General Intuition, which trains foundation models on hundreds of millions of hours of video-game footage to teach agents how to move through space and time, is raising at a $6 billion pre-money valuation from Valor Equity Partners, Point72 Ventures, and Seven Seven Six, with Khosla Ventures and General Catalyst re-upping, per TechCrunch. That is nearly triple the $2.3 billion mark set just eight weeks ago in a $320 million round, and the deal is oversubscribed.

Why it's disruptive: the startup, spun out of Medal in October by CEO Pim de Witte, is putting its model into robotic embodiments using CoreWeave compute. If game-clip world models generalize to real robots the way ChatGPT generalized to text, it attacks the entire robotics control stack at once. The WSJ's Christopher Mims profiles this wave as the missing runtime layer between perception and physical control.

Lambda: a $3 billion pre-IPO at a $12 billion valuation

Nvidia-backed neocloud Lambda is in talks to raise up to $3 billion at a valuation above $12 billion, roughly doubling its Series E mark from November, per Bloomberg. Sources say 2026 revenue is on track to top $1.5 billion, positioning the AI cloud provider for a public debut later this year or in early 2027.

Why it's disruptive: Lambda sits between the hyperscalers and the model labs, renting GPU capacity to developers. A public Lambda would give investors one of the purest plays on AI compute demand and put direct pressure on the megacap clouds' pricing.

Generalist: another $200 million, eleven weeks after its last round

Robotics foundation-model startup Generalist raised around $200 million in a round led by 8VC, Axios reports, barely eleven weeks after its $400 million Series B closed at a $2 billion valuation in June. The San Mateo lab, founded by ex-DeepMind researcher Pete Florence, released its GEN-1 physical-task model in April.

Why it's disruptive: Generalist, General Intuition, and Physical Intelligence are now in a three-way sprint for the foundation model that lets a single robot do many tasks instead of one. Whichever lab gets the generalist model first controls the upgrade path for every hardware maker.

Quintessent: $40 million for lasers that cut AI-cluster power

Optical-interconnect startup Quintessent closed a $40 million oversubscribed Series A led by Cycle Capital, with Goldman Sachs, Susquehanna, and Osage University Partners, following an $11.4 million seed. The company builds a gallium-arsenide comb laser whose quantum dots emit eight wavelengths in parallel, claiming 40 percent less power and better high-temperature reliability than silicon-based rivals.

Why it's disruptive: inside a data center the links between AI accelerators burn a huge share of the power budget. A laser that does the same job at 40 percent lower power attacks one of the least glamorous but most expensive bottlenecks in AI infrastructure.

Inherent: DeepMind alumni exit stealth with a $50 million seed

London-based Inherent emerged from stealth with a $50 million seed round, founded by Google DeepMind alumni including chief scientist Edward Hughes. The lab says its Faraday agent, running on Qwen 3.6 and OpenAI's GPT-5.5 Codex for coding, outperforms Claude Opus 4.8 and GPT-5.5 at independently reproducing findings from published scientific papers.

Why it's disruptive: this is the research-automation bet that quietly threatens both academic workflows and proprietary-lab moats. An agent that can reproduce a paper's results is a step toward agents that generate them, a lever every frontier lab wants to own.

What the pattern says

Across these five, two themes hold. Money is pouring into physical AI, the world models and robotics labs that want to give machines a general understanding of space and action. And it is also flowing to infrastructure no one glamorizes, from GPU clouds to data-center lasers. That is a healthier mix than last year's pure model-lab funding, and founders should read it as a signal: the outsized rounds are now going to whoever removes a physical constraint, not whoever ships another GPT clone.

For the industry-wide picture behind these deals, see today's General AI News digest, and compare with who was raising on August 23 in AI Startups to Watch. Browse the full AI Startups tag for the archive.

If this story matters to you, these earlier editions cover the same ground in more detail:

More on AI Startups

Browse every AI Startups story →