Mercor $750 Million Series D $20 Billion Valuation Explained: How the AI Training Startup Doubled in a Year

on

Mercor 750 million Series D 20 billion valuation explained — San Francisco, where Mercor is headquartered

Illustrative image: the San Francisco waterfront. Mercor is headquartered in San Francisco. Image: Wikimedia Commons / Dllu, CC BY-SA 4.0

TL;DR: Mercor raised $750 million in a Series D at a $20 billion valuation, announced on 10 November 2026 — roughly double the $10 billion it was worth after its $350 million Series C in October 2025. The Mercor $750 million Series D $20 billion valuation explained in one line: investors are paying up for the human expertise behind AI training, as Mercor shifts from serving frontier AI labs to selling evaluations and post-training to every enterprise.

Key Highlights

  • Mercor announced a $750 million Series D at a $20 billion valuation on 10 November 2026, with an employee tender included in the round.
  • Investors include Benchmark, Felicis, General Catalyst, NVIDIA, Salesforce, Zoom and funds managed by affiliates of Apollo — no lead investor was named.
  • The money funds evaluation ("evals") and post-training infrastructure, so enterprises can test and train AI agents on their own work.
  • Mercor's network counts more than five million experts; reported gross revenue hit $614 million in the first half of 2026, with a $2 billion annualised gross run rate by June.
  • The risks are real: about 91% of that reported revenue came from foundation-model makers, and Mercor confirmed a March 2026 supply-chain breach affecting some contractors.

Mercor $750 Million Series D $20 Billion Valuation: the Deal in Numbers

Mercor, the San Francisco AI data company, announced the round on Tuesday, 10 November 2026, in a company blog post and a Business Wire release. It was the largest funding round announced that day, and one of the biggest AI raises of the month. At roughly ₹6,750 crore (about $750 million at ~₹90 to the dollar), it is a large cheque even by 2026's inflated AI standards — and the $20 billion valuation, about ₹1.8 lakh crore, is what makes it a story rather than a routine raise.

The valuation has roughly doubled in about thirteen months. Mercor raised $350 million (about ₹3,150 crore) in October 2025 at a valuation of about $10 billion, according to Reuters. This time the round drew in strategics as well as venture firms: NVIDIA and Salesforce joined alongside Benchmark, Felicis, Fundomo, funds managed by affiliates of Apollo, G Squared, General Catalyst, Glade Brook Capital, Sapphire Ventures, StepStone Group, XTX Ventures and Zoom. The round also included an employee tender, letting staff sell some shares — a common feature when a private company is growing fast but not yet public.

Detail Mercor Series D
Amount raised $750 million (about ₹6,750 crore)
Valuation $20 billion (about ₹1.8 lakh crore)
Round Series D, announced 10 November 2026
Previous round $350 million Series C, October 2025, at about $10 billion
Investors Benchmark, Felicis, Fundomo, Apollo-affiliate funds, G Squared, General Catalyst, Glade Brook Capital, NVIDIA, Salesforce, Sapphire Ventures, StepStone Group, XTX Ventures, Zoom
Employee tender Yes — included in the round
Use of funds Evaluation and post-training infrastructure for enterprises
Founded / HQ 2023, San Francisco; offices opened in New York and London in 2026
Expert network More than 5 million experts; about 30,000 active contractors reported

What Does Mercor Actually Do?

Mercor began in 2023 as an AI-powered hiring platform, founded by Brendan Foody, Adarsh Hiremath and Surya Midha. It then pivoted to a more valuable job: supplying the human expertise that trains and tests AI models. Frontier AI labs need doctors, lawyers, engineers, bankers and scientists to write hard questions, grade model answers and show models how professional work is actually done. Mercor recruits those experts — it says more than five million are in its network — vets them with AI-run video interviews, and matches them to labs and enterprises.

That is the training side. The Series D is about the next step: evaluation and post-training. A model can score well on a public benchmark and still fail inside a real company, because every business has its own systems, policies and edge cases. As co-founder and co-CEO Adarsh Hiremath put it in the announcement, "a model can score well on a public benchmark and still fail on the systems, policies, and edge cases that define how a specific business runs." Mercor builds custom benchmarks — "evals" — with each enterprise, measures models against that company's own standards, and then post-trains the model on the gaps.

To do that at scale, Mercor acquired Deeptune earlier in 2026, a company whose team spent two years recreating hundreds of enterprise applications as simulated environments where AI agents can practise professional workflows. It also runs APEX, a public benchmark series that measures how AI performs on economically valuable work. The pitch to enterprises, in co-founder and co-CEO Brendan Foody's words: "Every job in the economy will need its own AI agents, and every company will need a way to train and evaluate them on its own work."

Mercor 750 million Series D 20 billion valuation explained — data centre infrastructure behind AI training and evaluation

Illustrative image: the data-centre infrastructure behind AI training and evaluation. Image: Wikimedia Commons / Christopher Bowns, CC BY-SA 2.0

Why the Valuation Doubled in About a Year

Valuations double when revenue grows faster than expected, and Mercor's reported numbers are striking — with an important caveat about what they measure. The Information, citing internal documents, reported that Mercor booked about $614 million (roughly ₹5,500 crore) in gross revenue in the first half of 2026, and crossed a $2 billion annualised gross run rate in June 2026, up from $1 billion just four months earlier. Those figures are reported from internal documents, not audited financials — Mercor is private and publishes no audited accounts.

The word "gross" matters. Mercor bills customers for expert work and then pays most of that money to the experts doing it: contractors keep roughly 60–70% of the gross figure, according to the same reporting, which put Mercor's gross margin at about 33% in the second quarter of 2026. So the $2 billion run rate is payment volume flowing through the marketplace, not revenue Mercor keeps — its own net take is closer to $600–800 million on an annualised basis, by reported estimates. That is still a large, fast-growing business for a company founded in 2023, and it explains why NVIDIA, Salesforce and Zoom wanted in: they are customers and partners as much as investors.

The wider market helps too. According to Crunchbase, AI-driven startups raised $102 billion — 64% of all global venture capital — in the third quarter of 2026. When nearly two-thirds of the world's venture money chases one sector, the companies supplying its scarcest input — expert human judgement — get priced like infrastructure. The same day Mercor announced its round, TipRanks reported that TSMC's board had approved a $1 billion investment split between the Anthropic IPO and OpenAI share offerings — a sign of how much capital is now circling the AI stack, from chips to training data.

Why It Matters for India Readers

Mercor announced no India office, customer or plan with this round, and this article does not invent one. The relevance for India is the playbook, not the postcode. Indian enterprises adopting AI agents face exactly the problem Mercor sells the solution to: a model that aces a demo can still fail on a company's own workflows, documents and approval chains. Evaluation and post-training on a company's own work is becoming a standard step before deploying agents in banking, insurance, healthcare and customer support — all large Indian employers of exactly the kind of professional expertise Mercor recruits.

There is also a workforce angle. Mercor's expert network spans more than 45 countries, and its best-paid listings favour credentialed specialists — doctors, lawyers, engineers, finance professionals — paid by the hour in dollars. For Indian professionals, that is a real, if selective, way to earn from AI training work, though listings are often limited to residents of particular countries and the work is contractor-based, not employment. And for Indian startups, the funding climate matters: when a three-year-old AI data company can double to $20 billion in thirteen months, capital is clearly still flowing to AI infrastructure — the same wave behind Nominal's $240 million Series C at a $3 billion valuation earlier in the same week.

The Risks and Open Questions

A $20 billion valuation for a private company invites scrutiny, and there are three honest caveats. First, concentration: about 91% of Mercor's reported first-half revenue came from foundation-model makers, according to The Information, with OpenAI, Anthropic and Google DeepMind named as the top customers. If those labs build their own expert networks or cut data spending, Mercor's growth story weakens fast — which is precisely why the company is pushing into enterprise customers now.

Second, trust. In March 2026 Mercor confirmed it was affected by a supply-chain attack involving LiteLLM, a commonly used open-source tool. The company said only a very limited subset of its nearly five million experts had sensitive information affected, according to its 25 June 2026 update. Lawsuits followed, and were stayed for mediation in July 2026, with no settlement on the court docket as of 7 November. The claims are allegations, not rulings — but for a company whose product is other people's expertise, data handling is part of the product.

Third, the valuation itself. Twenty billion dollars is roughly ten times the reported $2 billion gross run rate, and far more than ten times the net revenue Mercor keeps. That multiple only makes sense if enterprise evals and post-training become as big as frontier-lab training — which is the bet this round is funding, not a proven outcome. Mercor has not announced an IPO, and this article does not speculate on one.

Should You Care?

✅ Verdict: worth knowing, not worth chasing. If you follow AI, this round is a useful signal: the money is moving from building models to measuring and improving them inside real companies. If you are an Indian professional, Mercor's expert marketplace is worth a look — but check the listing terms, the pay and the country eligibility before applying. If you are an investor, remember this is a private company: there is no public share price, and the $20 billion is a private-mark valuation, not a market verdict.

FAQs

What is Mercor?
Mercor is a San Francisco AI data company founded in 2023. It connects frontier AI labs, big tech companies and Fortune 500 enterprises with a network of more than five million experts who train and evaluate AI models on real professional work, from law and medicine to engineering and finance.

Why is Mercor worth $20 billion?
Investors priced the Series D at $20 billion largely on growth: Mercor reported $614 million in gross revenue in the first half of 2026 and a $2 billion annualised gross run rate by June, up from $1 billion in February. The valuation also reflects a bet that enterprise evaluation and post-training — its new focus — becomes a large market. The figures are reported from internal documents, not audited financials.

Who invested in Mercor's Series D?
Benchmark, Felicis, Fundomo, funds managed by affiliates of Apollo, G Squared, General Catalyst, Glade Brook Capital, NVIDIA, Salesforce, Sapphire Ventures, StepStone Group, XTX Ventures and Zoom. No lead investor was named, and the round included an employee tender.

What will Mercor do with the $750 million?
It will accelerate investment in evaluation and post-training infrastructure — custom benchmarks for enterprises, research, and realistic simulated environments where AI agents can learn complex professional workflows, building on its Deeptune acquisition from earlier in 2026.

Takeaway

Mercor turned the people who train AI into a $20 billion business — now it has to prove the same model works for every enterprise, not just the frontier labs.

For the funding wave in context, read our explainer on Applied Compute's $350 million raise at a $3.25 billion valuation, where Mercor is a named customer.

Sources

Comments

Post a Comment