Vertical B2B · USA

Mercor

A marketplace selling expert labor to AI labs; the payment flow and the profit are real, but the headline "$2 billion ARR" is gross volume, not net revenue.

Verdict: Gaps remain

Updated: 2026-10-06

Profile

What it does: Expert labor marketplace: it matches doctors, lawyers, bankers and engineers to model training and evaluation work at AI labs; benchmark products such as APEX; reinforcement learning environments through the Deeptune acquisition [4].
Founded: January 2023. It began as a hiring marketplace matching Indian engineers to startups and shifted from 2023 toward post-training expert work for AI labs (secondary compilation) [7]
Founders: Brendan Foody (CEO), Adarsh Hiremath, Surya Midha [6]
Pricing model: Commission on gross volume; a 30% recruiting fee structure per Sacra (secondary data) [6]. 60–70% of top-line revenue is paid out to contractors [2].
Funding: February 2025: $100M Series B at a $2B valuation [5]
27 October 2025: $350M Series C at a $10B valuation; led by Felicis, with Benchmark, General Catalyst, Robinhood Ventures [5]

Growth timeline

2023-09
About $1M ARR in nine months, before funding (secondary compilation)
[7]
2025-02
$100M Series B at $2B
[5]
2025-10-27
$350M Series C at $10B; claim of more than $1.5M paid to experts per day
"more than $1.5 million per day"
[5]
2026-03
Security incident through the LiteLLM supply-chain attack
"A malicious actor published compromised versions of LiteLLM designed to exfiltrate credentials"
[3]
2026-06
$2B gross annualized revenue (The Information report)
"hit over $2 billion in gross annualized revenue in June 2026, double its pace at the start of the year"
[2]
2026-07-09
Deeptune acquisition; claim of more than five million domain experts
"more than five million domain experts"
[4]

Growth loop

1Lab demand
2High hourly pay
3Expert supply grows
4Matching speed and quality
Leak: 60–70% of gross revenue to contractors

A two-sided marketplace loop: demand raises pay, pay grows supply, supply speeds matching, speed wins new demand. The red arrow shows the gap between the headline figure and net revenue.

The drawing shows the shape of the loop; each node is explained in the list below. The red arrow marks value leaking out of the loop.

Mechanism

Acquisition channels: Founder-led direct sales (cold emails to OpenAI leadership), customer referrals and deep partnerships with a few labs (secondary compilation) [7]. A demand shock: OpenAI and Google DeepMind reportedly cut ties with Scale AI after Meta invested $14B in it and hired its CEO [5]. The channel is a gap in market structure plus founder sales (opinion).
Loop: Lab demand, pay above $85 an hour, more than 5 million registered experts, better and faster matching, larger contracts [5] [4]. The supply side is fed by more than $1.5M in daily payouts [5].
Activation: For a lab, delivery of the first expert cohort within days; for an expert, the first payout. No figures disclosed.
Retention evidence: No claim. Customer concentration (a handful of labs) is an open risk; no figure disclosed.
Role of pricing: The headline figure is gross, before contractor payouts. For net revenue, Sacra's estimate for the first half of 2026 is $180–250M (secondary data) [6]; dealroom's annualized calculation is $600–800M [2]. The word "ARR" is used here for gross volume; that is the main lesson of this case (opinion).
Distribution moves: APEX benchmark publications open lab doors through research reputation; Deeptune bundles experts with environments [4]. The official Series C post carries no figures; the figures went to the press [1] [5].
Team: About 400 full-time employees (secondary data) [6]. That puts gross volume per employee around $5M; with net revenue, $1.5–2M (our calculation, inputs are estimates) (opinion).

Real reading

The five questions of the Growth Claim Check, applied to this company's public claims.

01
What was measured, and by whom?
unclear
Figures come from the founder's X posts and The Information leaks; the official blog carries no figures.
02
Perception or outcome?
yes
The payment flow is real: daily contractor payouts. But the word "revenue" describes gross volume.
03
Is there a stability or quality metric?
no
Customer count, concentration and contract length are not disclosed.
04
Is there a counterfactual or cohort?
no
No cohort or counterfactual.
05
Nominal or real?
unclear
A 3–4x gap between gross and net; the $6M profit estimate for the first half of 2025 is a positive signal but an estimate.

Verdict

Gaps remain

There is a real business: profit and a payment flow. But the headline "$2B ARR" is gross volume, not net revenue.

Test your own claim with the same five questions: Growth Claim Check →

Shadow side

The gross-versus-net confusion: even dealroom writes that it is "a critique likely to intensify as the company touts multibillion-dollar milestones" [2].
Expert credentials leaked in the March 2026 LiteLLM supply-chain attack; the company offered TransUnion identity protection [3].
Demand depends on a few labs; labs building their own data teams or moving to synthetic data is a single-point risk (opinion).

What it teaches a founder in Türkiye

This section is opinion; a reading, not a source.

01
(opinion) If you are building a marketplace, do not call gross volume "revenue". Investors and the press will one day correct the multiple against net revenue; write two lines from the start: gross volume and net revenue.
02
(opinion) A break in market structure is the fastest channel. Be ready to receive customers the moment a large supplier is acquired or withdraws.
03
(opinion) Fast, high payouts to the supply side are the strongest lever in a two-sided marketplace; but without a cash cycle between collection and payout, growth burns cash.
04
(opinion) Supply-chain security is a growth topic. In a marketplace built on expert identity, a leak hits the supply side.

A question for you: remove this company's logo and drop the loop onto your own product. Which node stays empty?

Sources

Numbers are the references in the text. An 'opened' source was read by the researcher; 'secondary' ones are verified against their primary source before publication.

[1]
Unlocking human potential in the AI economy (Series C post)
Mercor (Brendan Foody) · 2025-10-27 · opened
“Knowledge is passed from one person to another.”
[2]
Mercor doubles to $2B gross revenue run rate as AI labs buy expert data
Dealroom (reporting The Information) · 2026-07-06 · opened
“Mercor pays out 60% to 70% of its topline revenue to contractors”
[3]
Update on Mercor security incident
Mercor (official blog) · 2026-06-25 · opened
“A malicious actor published compromised versions of LiteLLM designed to exfiltrate credentials”
[4]
Mercor to acquire Deeptune
Mercor (official blog) · 2026-07-09 · opened
“more than five million domain experts”
[5]
Mercor quintuples valuation to $10B with $350M Series C
TechCrunch · 2025-10-27 · opened
“more than $1.5 million per day”
[6]
Mercor company page
Sacra (analyst; estimates marked) · accessed 2026-10-05 · opened
“$6 million of profit in H1 2025”
[7]
Mercor's Growth Playbook: $1M to $1B ARR in 20 months
Startup Riders (secondary compilation) · accessed 2026-10-05 · opened
“pre-training stopped being the bottleneck”
[8]
Brendan Foody on X (September 2025 run-rate, July 2026 gross revenue)
X · accessed 2026-10-05 · secondary
Relayed wording: from US$647.1K to US$323.5M run rate in 17 months
Note: could not be opened; dealroom relay

Open questions

The exact text of Foody's X posts.
Customer concentration and contract structure.
Contractor-side complaints and payout delays.
Whether the $20B valuation round closed.