Reports emerge that Thinking Machines Lab is set to close a $5 billion funding round at a $40 billion pre-money valuation, with Accel leading the round. The deal, which could close as soon as this week, would more than triple the startup’s valuation from its last public round, a $2 billion seed raised last year. Neither side has confirmed the terms: the lab declined to comment, and Accel did not respond.
## The terms, as reported
The round is led by Accel, an investor in the lab, according to multiple sources familiar with the matter. It values Thinking Machines at $40 billion before new money arrives, a climb from the more than $10 billion valuation attached to its seed round last year. The figures remain reported terms, not a closed deal, until one side confirms them.
The same terms surfaced two months ago, when September reports described a $5 billion to $6 billion raise at a valuation of at least $40 billion, with Accel in talks to lead. Those reports put Nvidia’s investment at about $2.5 billion. Whether Nvidia’s money sits inside the total or alongside it is unclear, and no investor list has been published.
## A repricing, not a first raise
Thinking Machines has chased a higher price before and come away empty-handed. Last year it sought a round at a $50 billion valuation, but those talks were abandoned. The new figure sits below that ask, making this round a repricing as much as a raise: the market’s second verdict on a lab still building a business on its models.
The lab was founded by Mira Murati, OpenAI’s former chief technology officer, and its first product is Tinker, a system for fine-tuning language models. A new version of Tinker arrived on Thursday, with the company saying it has seen gains in AI coding. Its second-generation model, [Inkling 2 Small](https://geroundtable.com/thinking-machines-inkling-2-small/), launched the same day, with open weights promised for next week.
## The talent problem
The money arrives while the lab is losing people. About a third of its founding team has left, according to a May report, lured away by pay packages from rival labs. Named departures include Barret Zoph, Luke Metz and Andrew Tulloch. The exits are the other side of the valuation: investors are paying more for a team that keeps shrinking.
## Our Read
A $40 billion valuation is a bet on the team and the models, not on a business. The open-weights strategy is the differentiator: Thinking Machines is the only American lab shipping frontier-class open models, and that scarcity commands a premium while rivals stay closed. [Sand.ai](https://geroundtable.com/sand-ai-200-million-unicorn/) raised $200 million this week on the same bet, and investors are buying it widely.
The departures cut against it. A lab that has lost a third of its founders is asking investors to trust who remains, and the pay war with rivals will only get more expensive. The round settles it: $5 billion buys time, talent and compute. Whether it buys a lead depends on next week’s weights and on Tinker earning its keep.
## Why It Matters
This is the biggest lab financing since the summer, and it sets the price for every other frontier lab raising. If it closes at $40 billion after losing a third of its founders, rivals can cite the number and demand their own. [Zipline](https://geroundtable.com/zipline-1-5-billion-drone-delivery-network/) raised $1.5 billion this week too: money has not dried up, it has concentrated on trusted names.
## Background
Thinking Machines emerged from stealth last year as a new lab founded by former OpenAI leaders. Its first model, Inkling, arrived in July, and Inkling 2 Small followed this week with weights promised next week. The funding round, if it closes, would be the lab’s first priced raise since the seed — and a test of its $40 billion price.
Source: [the full relay of Business Insider’s report](https://zot.news/article/thinking-machines-lab-set-close-5-billion-accel-led-round-40-ifg1f) and [Forkast’s September report](https://forkast.news/thinking-machines-lab-seeks-40b-valuation-as-nvidia-extends-its-capital-allocator-role-into-model-space/)
0 Comments