Coursera's Chairman Started a Company With No Product. Coursera Paid $100 Million for a Third of It.

LearnVector's website is one page long. There is no product. There is no revenue. Nothing ships until early 2027.

Coursera paid $100 million for a third of it.

On July 28, 2026, Coursera announced a $100 million equity investment for a 33.33% fully diluted stake in LearnVector Inc., an AI-native learning company founded by Coursera chairman Andrew Ng. Class Central put it bluntly: Ng's LearnVector is a one-page website with no product, and Coursera, where he is chairman, just valued it at $300 million. The domain had been registered exactly one month earlier.

What you actually get for $300 million

LearnVector is based in Mountain View and is pre-product. Ng told Reuters it is still working out course offerings, pricing and user interface, with first products targeted for early 2027. It has no revenue. It isn't building its own AI models either — those come from other companies.

The plan: AI agents that function as personal tutors, adapting to each learner and practicing with them until they demonstrate they've mastered a concept. Ng told Reuters it will make individualized courses for white-collar workers that get harder as they improve, and said he expects to sell to corporations, governments, and higher education.

So not a consumer course business. An enterprise one.

The chairman sits on both sides

Because Ng chairs Coursera's board, the investment is a related-party transaction, and Coursera routed the decision through a committee of directors with no stake in LearnVector — the standard process for that kind of conflict. That independent committee unanimously approved the deal. Ng continues as chairman of the board.

It isn't his only overlapping seat. Workera, a skills-assessment startup Ng chairs and helped incubate, integrates with Coursera and last December announced a partnership with Udemy; Ng also still leads DeepLearning.AI, whose courses run on Coursera.

Class Central's reading of the announcement: Coursera is supplying the cash, the content, and the distribution, and getting a third of the company in return.

The sentence worth reading twice

Here is the part that should matter to anyone selling a recorded course.

Ng, in the announcement: "With LearnVector, we're putting AI to work for human development, transforming learning from the traditional one-to-many model to one-to-one."

Coursera's own blog said the same thing in its own voice. Education has largely been built around a one-to-many model — one instructor, one curriculum, one path delivered to many learners — which dramatically expanded access, but AI now gives the opportunity to reinvent learning from the ground up.

One-to-many is the format. It is the thing you record once and sell forever. It is the entire premise of the evergreen course business.

And the company publishing that line is the company that just bought the world's largest library of one-to-many courses. Coursera completed its acquisition of Udemy on May 11, creating a combined platform with more than 290 million learners, 18,000 enterprise customers and 95,000 content creators. The merger was pitched as creating a MOOC giant valued at $2.5 billion, with roughly $1.5 billion in annual revenue and $115 million in overlapping costs eliminated.

Ten weeks later, it wired $100 million to a startup whose pitch deck starts with the limits of that catalog.

The market didn't love it

The timing was brutal. The LearnVector news landed the day before Coursera's first earnings report since the Udemy close.

Shares tumbled 17% on Thursday, falling $1.04 to $5.14. Normalized revenue declined 1% year-over-year when including Udemy's pre-acquisition results, revealing the reported 60% surge was entirely acquisition-driven. Investing.com noted the related-party structure raised governance concerns among investors already wary of capital-allocation discipline post-merger, with no near-term revenue benefit given the early-2027 timeline. Some investors questioned the rationale outright, worried it may benefit insiders and that the funds could have gone to buybacks or dividends.

The money exists. Coursera ended the quarter with $982 million in unrestricted cash, cash equivalents and marketable securities, and zero debt. It had also repurchased roughly $140 million of its own stock in Q2 and the first four weeks of July. So $100 million is about a tenth of the pile — a bet it can afford, placed on a company that currently consists of a thesis.

The number course sellers should circle

Buried in the same quarter is the detail nobody put in a headline.

Consumer revenue of $158.6 million was up 29% as reported but down 5% on a normalized basis, reflecting pressure in transactional offerings. That decline came despite paid subscribers surging 44% to more than 1.65 million.

Transactional means the one-off course purchase. Pay once, own the recording. That is the line under pressure, while subscriptions grow.

Read the two decisions together and the logic stops looking strange. The buy-a-recording business softens. The subscription business holds. And the chairman raises nine figures on the argument that the future is a tutor that stays with you until you can prove you learned something — sold to employers, not to individuals scrolling a feed.

LearnVector's own framing pairs that one-on-one tutoring with what it calls trustworthy learning — which, in practice, appears to mean Coursera's content. The library isn't the product anymore. It's the raw material.

First LearnVector products are due in early 2027. Until then, 95,000 instructors keep selling recordings on a platform whose chairman has told everyone what he thinks comes next.