Mercor eyes $20B valuation in AI data labeling boom

Key Takeaways

• Mercor is negotiating a $500 million raise at a $20 billion valuation, roughly 10x its previous round
• The company's three co-founders, all around 23 years old, are now among the world's youngest self-made billionaires
• The funding reflects explosive growth in demand for human-labeled training data as AI models scale

Mercor, the AI data labeling company founded by three twenty-somethings now counted among the world's youngest self-made billionaires, is negotiating a $500 million funding round at a $20 billion valuation, according to Forbes Middle East. If the deal closes, it would mark roughly a 10x jump from the company's previous valuation and cement Mercor as one of the most valuable AI startups globally.

The raise comes as demand for high-quality human-labeled training data continues to surge. Every major AI lab, from OpenAI to Anthropic to Google DeepMind, relies on human annotators to rate model outputs, correct errors, and provide the feedback that powers reinforcement learning from human feedback (RLHF). Mercor has positioned itself as the matchmaker, connecting AI companies with a reported 300,000+ workers who perform these tasks.

Who founded Mercor and how did it grow so fast?

Brendan Wang, Amar Bhatia, and Charlie Duan started Mercor in 2023. Wang dropped out of Harvard to build the company. All three founders are approximately 23 years old, and the $20 billion valuation, if finalized, would make them billionaires on paper.

The speed of Mercor's ascent is striking even by AI startup standards. Founded less than two years ago, the company has gone from zero to a valuation that rivals public software companies with decades of revenue history. The trajectory reflects both genuine demand and the frothy capital environment around anything labeled AI infrastructure.

Why data labeling commands such high valuations

AI models learn from examples. The better the examples, the better the model. This simple fact has turned data labeling from a cost center into a strategic asset. Companies like Scale AI, which went public via SPAC discussions, and Labelbox have built multi-billion-dollar businesses on this premise.

Mercor's approach leans on a global distributed workforce. Workers on the platform annotate images, rate AI-generated text, and perform the granular tasks that shape how models behave. For AI teams building products, the quality and speed of this feedback loop directly affects how quickly they can ship improvements.

The valuation math here is speculative, but the underlying bet is clear: as models get larger and more capable, the volume of human feedback required doesn't shrink. It grows. Synthetic data can fill some gaps, but for tasks requiring human judgment, actual humans remain essential.

What this means for AI builders

For product teams training or fine-tuning models, the Mercor news is a reminder that data quality remains a competitive moat. The companies willing to invest in rigorous human feedback loops tend to ship models that behave more predictably and fail less embarrassingly.

It also signals rising costs. If data labeling companies can command $20 billion valuations, the underlying service isn't cheap. Teams working on smaller budgets may need to get creative, whether through active learning techniques that reduce labeling volume or tools like Airtable and Notion to manage annotation workflows in-house before scaling to a dedicated vendor.

The risk behind the hype

A $20 billion valuation for a two-year-old company requires enormous future cash flows to justify. Mercor will need to demonstrate that its workforce model scales profitably, that clients stick around as AI shifts, and that competitors don't undercut pricing. Scale AI, the incumbent, has years of enterprise relationships and deep integrations with defense and government clients.

There's also the synthetic data question. If AI models can generate their own training data, or if techniques like constitutional AI reduce reliance on human raters, the total addressable market for labeling shrinks. Mercor is betting that day remains distant.

Frequently Asked Questions

What does Mercor do?

Mercor connects AI companies with a global network of workers who label data, annotate images, and rate AI outputs. This human feedback is used to train and improve AI models.

How old are Mercor's founders?

Co-founders Brendan Wang, Amar Bhatia, and Charlie Duan are all approximately 23 years old. If the $20 billion valuation closes, they would rank among the youngest self-made billionaires.

How does Mercor compare to Scale AI?

Both companies provide data labeling services for AI training. Scale AI is the established incumbent with deep enterprise and government relationships, while Mercor has grown rapidly with a reported 300,000+ worker network.

Why are AI data labeling companies valued so highly?

AI models require massive amounts of human-labeled data to train effectively. As models scale, demand for quality annotations grows, making labeling infrastructure a critical and valuable layer of the AI stack.

Source: Forbes Middle East / Forbes Middle East