Thinking Machines Lab $40B Valuation: Murati's Raise

The private-market bidding war for frontier AI talent has a new headline number. Thinking Machines Lab — the research company founded by former OpenAI chief technology officer Mira Murati — is in talks to raise about $1 billion at a valuation of at least $40 billion, according to reporting published September 3, 2026. Existing backer Accel is in discussions to lead the round. Neither the company nor the firm has confirmed the talks, which remain in progress and could still change in size or price.

If it closes near those terms, the round would more than triple the valuation Thinking Machines carried just over a year ago and place a company with a comparatively small revenue base among the most valuable private startups on earth — a vivid measure of how much investors are willing to pay for a seat at the frontier-model table.

The numbers behind the round

Thinking Machines is reportedly seeking roughly $1 billion in fresh capital at a post-money valuation of $40 billion or more. The company’s annualized revenue run rate is reported at more than $100 million — a figure that implies a valuation multiple in the hundreds, the kind of pricing reserved for businesses investors expect to compound for years.

The new figure is a step up from where the company started but a step down from where it once aimed. In July 2025, Thinking Machines closed what was then described as the largest seed round in AI history: about $2 billion at a $12 billion post-money valuation. Late last year, the company reportedly sought a valuation closer to $50 billion in subsequent talks. A $40 billion mark would land below that ambition while still representing more than a 3x markup on the seed price in a little over a year.

Accel’s reported role as lead is notable. The firm is an existing investor, and follow-on leadership from an insider is often read as a signal of conviction — though it can also reflect how concentrated the pool of investors willing to underwrite frontier-lab economics has become.

Who Thinking Machines is

Thinking Machines Lab was founded in 2024 and is headquartered in San Francisco. Its founder, Mira Murati, served as OpenAI’s CTO and briefly as its interim CEO during the company’s 2023 leadership crisis; she left in 2024 to start her own lab. The founding team was stacked with senior OpenAI alumni, including co-founder John Schulman as chief scientist and Barret Zoph as CTO, alongside researchers such as Lilian Weng, Andrew Tulloch, and Luke Metz.

That roster is also a reminder that the talent market cuts both ways. In January 2026, Zoph and Metz departed Thinking Machines and rejoined OpenAI along with other researchers — a churn that underscores how fiercely the largest labs compete for a small number of senior scientists, and how much of a startup’s valuation rests on retaining them.

What the company actually sells

Unlike labs racing to ship a single flagship chatbot, Thinking Machines has bet on customization and openness. Its first product, Tinker, launched in October 2025 as a managed fine-tuning platform: it lets developers adapt open-weight models to their own data without having to manage distributed training infrastructure themselves. The company has since lifted the waitlist and expanded Tinker to support larger reasoning models, vision-language systems that combine images and text, and inference that is compatible with the OpenAI API.

In July 2026, the company released Inkling, its first open model — a move that positioned Thinking Machines as a U.S. alternative in open-weight AI at a moment when many of the most capable openly released models have come from Chinese labs. The strategic thesis is that enterprises want models they can inspect, fine-tune, and run on their own terms, rather than a single one-size-fits-all API. That approach differentiates the company from the capex-heavy scale race being run by the largest labs, but it also means monetization looks more like a developer-platform business than a consumer-subscription juggernaut.

Why raise now, and why so much

For a frontier research lab, capital is the raw input that everything else depends on. Training and serving competitive models requires enormous quantities of scarce accelerators and the power and data-center capacity to run them. A company that wants to keep pace with the frontier — while also standing up a platform business — needs a balance sheet deep enough to fund both simultaneously.

Raising at a $40 billion valuation also does something strategic beyond the cash: it sets a price. A high, insider-led mark helps a company recruit against deep-pocketed rivals, retain the researchers who are its core asset, and signal staying power to enterprise customers deciding whose models to build on. In a market where the largest AI labs are marching toward public offerings, a fresh private round buys time and optionality without the scrutiny of the public markets.

The timing fits a broader pattern. The AI funding environment in 2026 has rewarded a handful of frontier labs with valuations that dwarf their current revenue, on the bet that today’s run rates are the earliest innings of much larger businesses. Thinking Machines is being priced on that same logic.

The risks

The bull case is that Thinking Machines becomes a durable, independent platform for customized AI — the layer enterprises reach for when a generic API is not enough. The risks are equally clear.

• Revenue-to-valuation gap. A reported $100 million+ run rate against a $40 billion price implies a multiple that leaves little room for execution missteps. The trajectory has to justify the mark.
• Talent concentration. The company’s value is bound up in a small group of senior researchers. The January 2026 departures showed how quickly that base can shift when rivals come calling.
• Monetizing openness. Betting on open-weight models and fine-tuning tools is differentiated, but open releases are, by design, harder to monetize directly than a closed flagship. Turning developer adoption into durable revenue is unproven at this scale.
• Compute economics. Like every frontier lab, Thinking Machines is exposed to the cost and availability of accelerators and the capital intensity of staying competitive.

What it means

A $40 billion valuation on nine-figure revenue tells you less about Thinking Machines’ current business than about the market it operates in. Investors are underwriting a small number of frontier labs at prices that only make sense if AI capabilities — and the revenue attached to them — keep compounding steeply. Accel’s reported willingness to lead an insider round at more than triple the seed price is a vote that Thinking Machines belongs in that group.

The winners, if the round closes, are Murati and her team, who gain the capital to compete and a price that helps retain talent, and early backers sitting on a rapid markup. The losers are harder to name today, but the strategy carries a specific tension: Thinking Machines is pursuing an open, customization-first path in a market whose richest valuations have so far gone to closed, scale-first labs racing toward the public markets. Whether the platform thesis can support frontier-lab economics is the open question.

What to watch next: whether the round closes at the reported $40 billion or moves on final terms; how much of the $1 billion goes to compute versus hiring; the reception and adoption of Inkling and Tinker among enterprises; and whether the company can stabilize its senior research bench after a year of high-profile comings and goings. Each will say more about the durability of this valuation than the headline number does.