Thinking Machines Lab in talks for $1B Series A at $40B valuation

What's the deal? Thinking Machines Lab, the AI developer led by former OpenAI chief technology officer Mira Murati, is in talks to raise $1 billion in a Series A round at a $40 billion pre-investment valuation. Existing backer Accel is in talks to lead, with Nvidia considering participation.

How big is this? The round sits in the 99th percentile by amount among all-time Series A deals in its sector and country. That puts it among the largest early-stage rounds on record — remarkable for a company founded in early 2025.

What does it do? Thinking Machines releases open-weight models — its first, Inkling, launched in July — but makes money selling tools that let companies customise AI models with their own data on its Tinker platform. It has also released AI specialised for voice interactions.

The numbers: The startup generates at least a few hundred million dollars in annualised revenue, according to one person; another source put the run rate at over $100 million. Either figure makes the $40 billion valuation an extraordinarily high revenue multiple.

Why now? The round follows Inkling's release and comes six months after Nvidia said it was making a significant investment and planned a "long-term, gigawatt-scale strategic partnership" — likely tied to chip access. The proposed valuation is a leap from July 2025, when the company raised $2 billion at a $10 billion pre-investment valuation, led by Andreessen Horowitz.

What could go wrong? The target is below the $50 billion-plus valuation the company sought late last year. It has also seen turnover: co-founder Lilian Weng rejoined OpenAI, the fourth co-founder to leave in the past year. Murati holds a board vote weighted heavier than all other directors combined, a control provision that could deter some investors.

The signal: Thinking Machines is among the highest-profile of a cluster of AI startups known as neolabs, raising billions to train foundation models and recruit staff. Rather than competing head-on with closed-source developers like OpenAI and AnthropicDealroom has a profile for this one. Try Dealroom →, it targets cloud and inference providers helping businesses customise models — a bet that open-weight tooling can carve out its own market.

Read more: The Information, TechCrunch

Image credit: Thinking Machines Lab