$122 Billion! OpenAI Sets Record for Largest Single Funding Round

By Yu Yang and Meng Yao

I read the filing this morning while watching a warehouse robot struggle to pick up a single box without crushing it. The numbers on paper are staggering, but out here in the field, we measure success by uptime and unit economics, not valuation spikes. OpenAI just closed a $122 billion round, pushing its post-money valuation to $852 billion. That is 841.154 billion RMB for the cash and roughly 5.9 trillion RMB on paper.

For the first time, they opened this round to individual investors through banking channels, pulling in over $3 billion from retail backers. The market is already pricing in an IPO, but I’m more interested in whether these funds will actually build reliable systems or just buy more chips for demos that don’t work outside the lab.

OpenAI Sets Its Largest Financing Record in History

The capital came from a coalition of giants: Amazon, Nvidia, and SoftBank led the round. Nvidia and SoftBank each put up $30 billion, while Amazon contributed $50 billion. Microsoft, our long-time partner in this space, is also on the list. While they didn’t disclose the specific amount for this latest tranche, we know Microsoft had already invested over $13 billion cumulatively by last year’s end.

I think valuations don’t fix broken actuators or reduce latency in real-world deployments.

There was a structural shift here too: that extra $3 billion from individual investors marks the first time OpenAI tapped retail capital directly. This moves them away from being solely dependent on a few tech giants toward a more dispersed funding model. It’s a safer bet for them, but it also means more pressure to show returns to people who don’t understand robotics engineering.

Under this $852 billion valuation, the internet is calling it a huge gamble on scaling laws having no upper limit. I’ve seen enough scaling law papers to know that physical constraints always win eventually.

Sam Altman now has to justify this massive price tag. Recently, OpenAI cut features like Sora to consolidate compute and control costs. That’s a sign of strain, not strength. They claim confidence based on user growth: fastest to 10 million users, then 100 million, and soon 1 billion weekly active users.

They state that within one year of launching ChatGPT, revenue hit $1 billion. By the end of 2024, quarterly revenue was also $1 billion. Currently, monthly revenue stands at $2 billion. They claim their growth rate is four times that of Alphabet or Meta during their respective eras.

In the field, user numbers are vanity metrics if the underlying infrastructure can’t handle real-world edge cases.

However, last year’s total revenue was $13.1 billion, and they still aren’t profitable. In a separate announcement, they revealed an advertising pilot exceeded an annualized run rate (ARR) of $100 million in just six weeks. ToB revenue now makes up over 40% of their income and is expected to match ToC by 2026.

Looking ahead, OpenAI wants to build “AI Super Apps”—systems that understand intent, execute actions, and operate across workflows. That sounds good on a slide deck, but until I see a robot navigate a cluttered kitchen without calling for help, I’ll remain skeptical.