Nvidia is in talks to invest in Mercor, the AI data-labeling startup, at a valuation of roughly $20 billion, according to The Information. The report puts Nvidia in the position of funding a company that already sits inside its own supply chain, and that detail is what makes this deal worth paying attention to.
Mercor sells the raw material that makes frontier models smarter: high-quality, expert-generated training data. Think human annotators, domain specialists, and structured feedback that labs use to fine-tune models and push past the limits of scraped web text. A $20 billion valuation for that kind of business would have sounded absurd two years ago. It doesn’t now.
What Mercor actually does
Modern AI progress runs on two fuels: compute and data. Nvidia owns the compute layer. Mercor plays in the data layer, connecting AI labs with the human experts who label, grade, and generate the specialized examples models learn from.
As the open web gets exhausted as a training source, labs are paying up for curated, expert data instead. Doctors, lawyers, coders, and PhDs are now grading model outputs and writing the reference answers that push reasoning quality higher. Mercor’s pitch is that it can source and manage that expert labor at scale, and the market clearly believes the demand curve is going straight up.
Why Nvidia funding its own supplier matters
Here’s what stands out. Nvidia already invests aggressively across the AI stack, from model labs to cloud providers to robotics. Backing a data supplier extends that strategy one layer deeper, into the pipeline that feeds the models its chips train.
This is significant for a few reasons:
- Vertical reach. Nvidia isn’t just selling shovels anymore. It’s taking stakes in the mines, the miners, and now the ore. Owning a piece of the data layer gives it visibility into where model training is heading.
- Signal to the market. When the most powerful company in AI puts money into expert data, it tells every lab and investor that data quality, not just GPU count, is becoming the bottleneck.
- Circular financing questions. Nvidia funding companies that ultimately buy or enable more Nvidia compute is a pattern regulators and analysts are watching closely. This deal fits that pattern.
The bigger shift
For most of the current AI boom, the story was compute. Whoever had the most GPUs won. That story is changing. The frontier is moving toward data quality, and specifically toward expert human data that can’t be scraped for free.
Mercor rode that shift fast. The company was tiny not long ago. A $20 billion valuation would place it among the most valuable AI infrastructure startups in the world, built almost entirely on the premise that great data is now scarce and expensive.
That repricing tells you something about where the money thinks the next round of model gains will come from. Not bigger clusters alone. Better data.
What to watch next
The deal isn’t closed, and The Information frames it as ongoing discussions, so terms could move. A few things to keep an eye on:
- Whether it closes at $20 billion. Valuations in AI data have been climbing fast, and a confirmed number would reset the market’s sense of what this layer is worth.
- Other backers. Nvidia rarely moves alone. Watch for labs or major funds joining the round, which would deepen the ties between compute, data, and models.
- Competitive response. Rival data providers and labs building their own data operations will feel the pressure once a deal this size lands.
If you build with AI, the takeaway is simple. The value in this industry keeps migrating toward whoever controls scarce inputs, and expert data just moved up the list. Nvidia clearly wants a seat at that table before anyone else locks it up. Full details are available at the original report from The Information.