Thinking Machines Lab, the AI company founded by former OpenAI CTO Mira Murati, is in advanced talks to raise $1 billion at a pre-money valuation of at least $40 billion. Accel is expected to lead the round, with Nvidia eyeing participation.

To put that number in perspective: Thinking Machines was valued at $12 billion when it closed a $2 billion seed round less than 18 months ago. It has roughly tripled its valuation since then, on the back of a product that actually generates revenue.

Why This Matters More Than Another Funding Headline

AI funding rounds stopped being surprising a long time ago. What makes this one worth paying attention to is the revenue story underneath it.

Thinking Machines Lab is reporting over $100 million in annualized revenue, generated primarily through its Tinker API platform and the Inkling open-weight model it released in July 2026. That’s real product revenue, not compute credits or grant funding.

This marks a shift in how the most serious AI labs are being built and valued. The playbook is no longer “raise enormous amounts, train frontier models, worry about revenue later.” Thinking Machines is making money while raising money, which is a meaningfully different kind of company.

The Tinker and Inkling Model

For those who haven’t followed Thinking Machines closely: Inkling is an open-weight multimodal model that enterprises can fine-tune on their own proprietary data. The Tinker platform is where that fine-tuning actually happens, with usage-based compute fees making up most of the company’s revenue.

This model gives businesses something the closed frontier labs don’t offer: genuine ownership. When you fine-tune Inkling through Tinker, you end up with a version of the model shaped by your data and workflows, running in environments you control. The revenue Thinking Machines collects is for the compute, not a perpetual licence or ongoing subscription to capabilities you don’t own.

The $100M ARR figure suggests that enough enterprises have decided this tradeoff is worth paying for.

The Valuation Jump in Context

A jump from $12 billion to $40 billion in roughly a year is significant. But it’s worth separating the signal from the market noise.

AI valuations have been running hot since late 2024. OpenAI’s most recent round valued it above $300 billion. Anthropic has crossed $900 billion in reported valuation discussions. In that environment, $40 billion for a revenue-generating AI lab with a credible product and a former OpenAI CTO at the helm is arguably conservative.

The more telling signal is that the valuation is being justified by real commercial traction, not projected future capabilities. That’s the direction the AI market is moving: away from pure research bets and toward companies that can show a line of revenue alongside their benchmark scores.

What This Means for Business

Three things stand out for business leaders thinking about their AI strategy:

Open-weight models are becoming a serious enterprise option. The commercial success of Inkling suggests that enterprises are genuinely interested in owning their AI capabilities, not just renting them. If you’re evaluating AI vendors today, the build-vs-buy question now includes a third option: fine-tune a capable open-weight model on your own data.

Revenue justifies runway. Thinking Machines’ ability to raise at this valuation while generating meaningful revenue is a proof point that AI can be a real business, not just a technology. For businesses still treating AI as a cost centre or a pilot programme, that framing is becoming harder to defend.

Nvidia’s continued involvement is strategic. Nvidia is reportedly eyeing participation in this round, on top of a multiyear partnership announced in March that involves Thinking Machines deploying at least one gigawatt of Nvidia chips for training and inference. When the world’s dominant chip maker keeps co-investing with an AI lab, it signals confidence in the lab’s long-term compute demand. That kind of partnership shapes which models get access to the best hardware.

The Broader Picture

Thinking Machines occupies an interesting position in the AI landscape. It’s not trying to be OpenAI, and it’s not positioning itself as a narrow enterprise software vendor either. The combination of open-weight models, a revenue-generating compute platform, and a high-profile founder puts it somewhere between a frontier lab and a commercial AI platform.

Whether it gets to $40 billion on the strength of that positioning will depend on whether the Tinker platform continues to attract enterprise fine-tuning workloads at scale. But the direction of travel is clear: the AI labs that survive the next phase of consolidation will be the ones that found a way to turn capability into commercial value.

Thinking Machines is one of the few AI startups that appears to be doing both at once.

---

For businesses evaluating AI infrastructure decisions, Enterprise DNA’s Omni Advisory service helps cut through the noise on AI vendor selection, open-weight versus closed model strategies, and where to build versus buy.