Nvidia Weighs Backing Its Own Data Supplier: Mercor at a $20 Billion Valuation

Nvidia is discussing an investment in Mercor, the AI data-labeling and expert-data startup, as part of a funding round that would value the company at $20 billion — double the $10 billion valuation it commanded just ten months ago, according to a report by The Information’s Phoebe Liu and Julia Hornstein published August 19 and confirmed by multiple outlets including Reuters and Yahoo Finance.

The reported details are straightforward but consequential: existing investor General Catalyst has been in talks to lead the new round, and Nvidia — which relies on Mercor as a supplier of expert training data for its own open-source model development — is weighing participation. If completed, the deal would crystallize one of the fastest valuation doublings in the current AI cycle, from $10 billion in October 2025 to $20 billion in mid-2026.

Why Mercor keeps compounding

Mercor was founded in 2023 by Brendan Foody, Adarsh Goel, and Surya Midha — Thiel Fellows who started the company while still in college and who became, at the time of the Series C, among the youngest self-made tech billionaires ever. The company began life as an AI-powered recruiting platform, but it has since pivoted into what may be the single most strategically important corner of the AI stack: supplying frontier labs with expert human data.

The business model is deceptively simple. Frontier models have run through the public internet; what remains scarce is professional-grade data — the judgment of doctors, lawyers, investment bankers, software engineers, and research scientists, captured and formatted for model training. Mercor recruits these domain experts, vets them through AI-driven interviews, and pays them — often hundreds of dollars per hour — to produce reasoning traces, evaluations, and RL environments for the major labs. Its client list is reported to include OpenAI, Anthropic, Meta, and Nvidia itself.

The revenue numbers explain the valuation fervor. According to Forbes, Mercor crossed $2 billion in annualized revenue in June 2026 — doubling from $1 billion in February, which itself was up from roughly $100 million in ARR a year earlier. A company that quadrupled sales in four months, serving the most capital-intensive buyers in the world, is exactly the kind of asset that investors believe deserves a premium multiple.

The Nvidia angle: investing upstream

What makes this round more than another eye-popping AI valuation is who is reportedly kicking the tires. Nvidia doesn’t just want a financial stake — it is one of Mercor’s customers. Seeking Alpha’s coverage notes that Mercor “helps it develop open-source models,” meaning the chipmaker depends on Mercor’s expert-data pipeline for its own model work, most visibly around its Nemotron family of open models.

For Nvidia, the strategic logic mirrors moves it has made across the AI stack: invest in the suppliers whose success drives demand for GPUs. The company has already deployed capital into AI infrastructure plays and signed financing agreements with some of the largest pools of capital in the world to fund data-center buildouts. Backing a data supplier at the application layer of the stack extends the same flywheel thesis — every dollar spent generating expert training data ultimately pulls through compute demand.

There is also a defensive dimension. High-quality human data has become a bottleneck resource for frontier-scale training, and the labs that control it gain a durable edge. By taking an equity position in a key supplier, Nvidia buys itself both influence over, and visibility into, a pipeline it cannot easily replicate in-house.

The $20B question

Skeptics will note the pattern: a supplier whose revenue is concentrated among a handful of mega-labs, valued at 10x annualized sales, in a round led by an existing investor. Mercor’s $2 billion run-rate depends on the continued willingness of OpenAI, Anthropic, Meta, and others to spend aggressively on human data — and on Mercor’s continued ability to win that spend against rivals like Scale AI (majority-acquired by Meta in a $14.3 billion deal) and Surge AI.

But the bull case has data behind it. The market for expert human data is not shrinking — it is shifting from bulk annotation toward high-value reasoning data, exactly Mercor’s strength. And unlike the model labs themselves, Mercor is supplier-agnostic: whoever wins the frontier race, they will need expert data, and Mercor sells to all of them.

The round is not final, and terms could shift before close. But the signal is clear: the strategic value of the human-data layer is still being repriced upward, and the biggest compute vendor in the world wants a piece of it.

Sources

• The Information — Nvidia Discusses Funding Its AI Data Supplier Mercor at a $20 Billion Valuation
• Yahoo Finance — Nvidia Reportedly Explores Stake in AI Data Provider Mercor At $20B
• Seeking Alpha — Nvidia considers funding AI startup Mercor, which helps it develop open-source models
• Forbes — AI Data Labeler Mercor In Talks To Raise $500 Million
• CNBC — AI hiring startup Mercor valued at $10 billion with new funding round