Mercor is a startup that connects large companies with experts in narrow domains for training and improving foundational AI models. Currently, the company is in talks with investors about a Series C round, with sources familiar with the negotiations confirming active discussions and the existence of marketing materials that were considered outside the company.
Felicis Ventures is weighing the possibility of increasing its contribution to the Series C, according to two sources. The firm declined to comment.
Mercor’s valuation now exceeds $10 billion, higher than the previous estimate of about $8 billion announced a few months ago. However, the final terms of the deal may still change during negotiations.
Sources say potential investors have already made offers valuing Mercor at more than $10 billion. It is also known that the company has brought in at least two new investors through special purpose vehicles (SPVs) to fund the possible deal.
The previous round was announced in February – a $100 million Series B at a valuation of around $2 billion, led by Felicis.
Mercor, founded in 2022, is approaching an ARR of about $450 million per year. In March, founder and CEO Brandon Foody said that ARR had surpassed $100 million, and previously in February reported ARR of $75 million.
The company also notes growing profitability: in the first half of the year, Mercor earned about $6 million, according to some business reports.
Mercor’s revenue is primarily generated by giving enterprises access to specialists – scientists, doctors, and lawyers – with an hourly rate for sourcing and approving their work. The company says it works with annotated data for five leading AI labs, including Amazon, Google, Meta, Microsoft, and OpenAI, as well as Tesla and Nvidia. According to sources, a substantial portion of the revenue may come from one of these brands, specifically OpenAI.
As part of diversifying its business model, Mercor is considering expanding reinforcement learning (RL) infrastructure – a method that allows testing and adapting model solutions through feedback. The company also plans to create an AI-powered recruitment platform in the future.
Among Mercor’s competitors are Surge AI, which has announced potential funding with an valuation of around $25 billion, as well as Turing Labs and Scale AI, which are also expanding RL services. Some analysts speculate that OpenAI may launch its own RL training service using human experts.
“We weren’t trying to raise funding,”
“We turn down proposals every month.”
ARR and Mercor’s financial dynamics
According to Foody, ARR indeed exceeds $450 million; he also noted that revenues include client fees for services before payments to contractors – a standard accounting practice recommended by auditing firms and used by competitors.
Team and founding history
Mercor was founded in 2023 by three members of the Thiel Fellows program and Harvard graduates – Brandon Foody (CEO), Adarsh Khirmath (CTO) and Surya Midha (COO). In 2024 the company appointed its first president – Sundip Jain, former Chief Product Officer at Uber.
Legal issues and market competition
There are reports of a Scale AI lawsuit against Mercor over alleged misappropriation of trade secrets: Scale claims that one of Mercor’s former employees allegedly stole more than 100 confidential documents related to Scale’s client strategies and other confidential information. The case is ongoing in the relevant jurisdictions.
Against the backdrop of rising competition, Mercor faces other data and service providers in the RL space that are also expanding their capabilities. The market continues to assess the potential for integrating the human factor into model training and creating new business models in the AI field.