Thinking Machines at $40B: A 3.3x Step-Up That Is Also a Haircut

Accel is reportedly in talks to lead a $1 billion round for Mira Murati's Thinking Machines Lab at a valuation of at least $40 billion. Read one way, that is a 3.3x markup on the largest seed round in history, fourteen months later. Read another, it is a ~20% climb-down from the ~$50 billion the company reportedly sought late last year. Both readings are true, and neither is a price yet.

The Setup

The Information reported Thursday — and TechCrunch corroborated with its own source — that Thinking Machines is in discussions to raise $1 billion at a valuation of at least $40 billion, with existing backer Accel in talks to lead. The company's annual revenue run rate stands at over $100 million, per a source with knowledge of its financials. Accel and Thinking Machines did not respond to requests for comment.

The context that matters: the AI lab, founded in February 2025 by former OpenAI CTO Mira Murati, closed a record $2 billion seed at a $12 billion valuation in July 2025, led by Andreessen Horowitz and joined by Nvidia, GV, Lightspeed and Conviction Partners. In July 2026 it introduced Inkling, an open-weight model monetized through usage-based compute fees for adapting models on its Tinker platform. It has also absorbed several high-profile departures, with some co-founders — including Lilian Weng and Luke Metz — returning to OpenAI.

The Concept

A round "in talks" is the least binding of all private-market numbers. Until it closes, it is a negotiating position that leaked. The discipline that keeps you honest is the same one we applied to Anthropic's three numbers: a closed round price is something somebody paid; a reported target is something somebody hopes; and the gap between successive targets — here, $50B sought versus $40B discussed — is often the most informative number on the page, because it tells you how the negotiation is actually going.

The Read

$12B — the only price on the tape: last July's seed. That number is fourteen months stale, which in this cycle is geological time.
$40B+ — the reported ask. A 3.3x step-up from the seed, and roughly 400x the reported $100M+ run rate. For calibration, that multiple sits above even the frothiest closed AI rounds on our tape; it is a bet priced entirely on the lab's trajectory and Murati's bench, not on current revenue.
~$50B — what the company reportedly sought late last year and did not get. The $40B discussion is a fifth off that ambition — the rare data point where a hot AI lab's price found resistance.
$1B on $40B — about 2.4% dilution. This is a top-up at a mark, not a war chest raise; compare OpenAI and Anthropic raising tens of billions. Either the burn is genuinely modest, or a larger structure comes later.

Two receipts from our own ledgers are worth holding beside the headline. On the QL Founder Ledger, Murati's one graded public claim — shipping a first product "in the next couple of months," pledged July 2025 — resolved kept (Tinker landed October 1, about two and a half months later, with the gap noted). Delivery so far has roughly matched the words. Against that: the co-founder departures are the kind of signal a $40B price has to argue past, because the seed was explicitly underwritten on the team.

The Action

Treat $40B as a bid range, not a mark. Our Thinking Machines page keeps the company at its last priced $12B with the reported talks annotated — because that is the honest state of the world. If the round closes, the mark moves; not before.

Watch what the multiple implies for the sector, not just the name. A closed round near 400x run-rate revenue re-anchors every private AI lab's comparables. Holders of exposure to the peer set — directly or through funds — get repriced by this print whether or not they own Thinking Machines.

If you track the name, track it through the roster. The Pre-IPO Bridge shows the graded score, the public-sleeve read-through and the receipt trail as evidence lands.

What to Watch Next

Whether the round closes, and at what number. $40B closing confirms the step-up; a quiet shelving would echo the $50B episode.
Who leads. Accel leading a $1B AI-lab round would be a statement position for a firm that was a smaller participant in the seed.
Inkling's revenue trajectory. Usage-based fees are the first durable revenue engine here; the next reported run-rate figure tells you whether 400x is compressing or inflating.
Further bench moves. The seed was priced on people. Departures — or senior arrivals — move this thesis more than most companies'.

The Counter

The bull reading of the same facts: revenue went from zero to a reported $100M+ run rate within roughly a year of first product — that pace, not the level, is what a 3.3x step-up prices. The $50B-to-$40B walk-down can be read as discipline rather than weakness: the company waiting until it had revenue to support a bigger ask, and investors paying for evidence instead of pedigree this time. And a $1B raise at modest dilution is exactly what a lab that doesn't need the money would do. If Inkling's fee model compounds, today's 400x becomes tomorrow's rounding error — that is the trade Accel is reportedly weighing.

Key Terms

Step-up — the multiple between a company's successive round valuations. Here: $12B → $40B ≈ 3.3x in ~14 months, if it closes.
Run rate (ARR) — the most recent revenue period annualized. A growth-adjusted snapshot, not audited trailing revenue.
Open-weight model — a model whose weights are downloadable and adaptable; Thinking Machines monetizes the adaptation compute rather than the weights.
Reported / in talks — sourced reporting on a live negotiation. Terms, and the round itself, can change or vanish before any close.

Primary Sources

QuantLogix is not a registered investment advisor, broker-dealer, or financial planner. This article is for informational and educational purposes only and does not constitute financial advice or a recommendation regarding any security or offering. The round described here is sourced reporting on a live negotiation — neither Accel nor Thinking Machines has confirmed the terms, the figures are unaudited, and the discussions may change or end without a transaction. Past performance does not guarantee future results.