Mercor vs Surge AI: marketplace hours or finished datasets
Mercor and Surge AI sell expert human judgment to the same handful of frontier labs, and both name OpenAI, Anthropic, Meta, and Google among their customers. Underneath that overlap they are built differently in almost every respect: how the work is delivered, how the companies are financed, what each reports as revenue, and what each publishes about its own governance. A buyer who treats them as interchangeable will be surprised by at least one of those. This comparison lays out the differences that change a purchasing decision.
Key takeaways
- Mercor bills expert hours at a markup and you generally manage the work. Surge delivers finished post-training data and manages it itself.
- Their revenue figures are not comparable. Mercor's widely cited 2 billion dollars is gross marketplace volume before contractor payouts; Surge's 1.2 billion is company revenue.
- Financing is the opposite in each case. Mercor has raised roughly 484 million dollars and carries a 10 billion dollar priced valuation. Surge has raised nothing and has no priced valuation.
- On published governance, Mercor is modestly ahead: it operates a trust center with a SOC 2 Type 2 report. Neither posture is strong enough to skip your own assessment.
- Neither publishes customer-facing IP indemnification, and both sets of public terms are unfavorable to the customer in different ways.
- Mercor does some real physical collection through partner businesses. Surge publishes none.
About Mercor
Founded in January 2023 by Brendan Foody, Adarsh Hiremath, and Surya Midha, based in San Francisco. It raised a 100 million dollar Series B at 2 billion in February 2025 and a 350 million dollar Series C led by Felicis at 10 billion in October 2025. A round at roughly 20 billion has been reported in talks since July 2026, with Nvidia named as a potential investor in August 2026, and as of this writing no report says it has closed.
The core business is an expert marketplace. Candidates complete an AI-run interview of about twenty minutes with up to three retakes, and those who qualify are matched to client projects with Mercor billing a marked-up hourly rate. Mercor states an average contracted rate of 107 dollars per hour, and published listings range from around 60 dollars for language tutors to 450 for oncologists. It also sells data monetization, where enterprises license their own operational data to AI labs through Mercor's extraction and de-identification pipeline, and off-the-shelf data.
Mercor publishes its own APEX benchmark family alongside partner benchmarks built with Harvey, Ramp, Cognition, and Sierra. It acquired Sepal AI in February 2026 and Deeptune in July 2026.
Unlike Surge, Mercor does operate a genuine physical collection line: it supplies capture kits to partner businesses, employees opt in and wear cameras during shifts, and Mercor licenses the footage to AI and robotics companies, paying 12 to 18 dollars per approved hour.
About Surge AI
Legally Surge Labs, Inc., founded in 2020 by Edwin Chen, based in San Francisco. Surge has taken no outside capital and has been profitable since the start. Bloomberg reported 1.2 billion dollars in 2024 revenue, and Forbes reports Chen holding roughly 75 percent of the company. Surge sought outside funding in 2025 at valuations reported between 15 and 25 billion dollars, but no round closed, so there is no priced valuation.
Surge sells nine post-training product lines: RL environments and agents, rubrics and verifiers, RLHF, supervised fine-tuning, human evaluation, expert professional domains, internationalization, multimodal work across image, audio, and video, and off-the-shelf data. It claims a workforce including more than 200,000 PhDs across over 500 disciplines, though it names no countries or regions.
Its strongest public credential is research. Surge shipped 13 benchmarks across 2026 with 8 accompanying arXiv papers, and the labs cite them: OpenAI referenced GDP.pdf in its GPT-5.6 release and Anthropic referenced GDP.pdf and Riemann-bench in a system card. Its blog documents dated engagements with Anthropic and Microsoft.
Surge publishes nothing about physical or real-world collection. Its multimodal work is digital image, audio, and video performed by remote contractors.
What buyers report
Neither has a review base a buyer can use, and in both cases the available ratings measure the wrong population.
Mercor holds 4.4 out of 5 on G2 across 114 reviews, but its G2 profile sits under applicant tracking, AI recruiting, and job search categories, so those reviews come from candidates and contractors rather than enterprises buying data. Its Trustpilot reviews are contractors and its Glassdoor reviews are employees. It has no Gartner Peer Insights listing.
Surge has a single G2 review, no Gartner Peer Insights listing, and no Capterra presence found as of October 2026. G2's own profile text notes there are not enough reviews to offer buying insight.
Both sell top-down to a small set of large customers, which explains the absence without excusing it. Plan on a paid pilot for evidence.
The comparison
| Mercor | Surge AI | |
| Delivery | Sources and matches experts, bills hours at a markup | Delivers finished post-training data |
| Who manages quality | Typically you | Surge |
| Founded | January 2023 | 2020 |
| Outside capital | Roughly 484 million dollars raised | None |
| Valuation | 10 billion, priced October 2025; around 20 billion reported in talks | No priced round exists |
| Headline financial figure | 2 billion dollars gross annualized revenue, June 2026 | 1.2 billion dollars revenue, 2024 |
| Compliance | SOC 2 Type 2, via a published trust center | Substantially less published |
| Public terms | No Mercor-side indemnity published | No reciprocal indemnity published |
| Network disclosure | 30,000-plus active experts, 107 payout countries published | 200,000-plus PhDs claimed; no geographies published |
| Physical collection | Yes, wearable capture via partner businesses | None published |
| Open research | APEX benchmark family and partner benchmarks | 13 benchmarks, 8 arXiv papers, cited by OpenAI and Anthropic |
The revenue figures are not the same kind of number
This is the comparison most often made badly, including in coverage that puts the two side by side.
Mercor's 2 billion dollar figure was reported by The Information in June 2026 and labeled gross annualized revenue. On a marketplace, that means total customer spend before contractor payouts. Contractor share is estimated at 60 to 70 percent, so the company's net revenue is a fraction of it. Mercor does not publish net revenue, and outside estimates span roughly 180 million to 800 million dollars, a range too wide to be useful. Mercor's own chief executive gave growth timeframes in a single Fortune interview that conflict with each other, which the publication noted without resolving.
Surge's 1.2 billion is company revenue for 2024, reported by Bloomberg and corroborated by Forbes, from a business with no outside investors. Surge has published no 2025 or 2026 figure.
Set side by side without those definitions, Mercor looks like the larger company. On a like-for-like basis the comparison is unresolved, and possibly reversed. If vendor scale matters to your procurement, use the funding record and the customer evidence, both of which are documented, rather than the revenue headlines.
Which to choose
Choose Mercor when what you lack is access to credentialed people rather than the capacity to define and manage the work, when speed to staff a project matters, or when you need the data monetization or wearable-capture lines, which Surge does not offer. Ask what certifications exist beyond SOC 2 Type 2 and what changed after the 2026 incident.
Choose Surge when you want finished data rather than staffed hours, when evaluation and rubric design is the actual deliverable, and when independence from outside investors matters to you. Expect to negotiate security and IP terms from scratch and to run your own security assessment, because there is nothing published to start from.
Why HumanSignal is the better choice for designed capture
Mercor and Surge both sell expert human judgment, and between them they cover most of what a frontier lab needs from people who already know a domain. The limit they share is that judgment applies to something. Mercor's robotics line comes closest to creating new material, and it is opportunistic capture at partner businesses: cameras on whoever opted in, doing whatever that business happens to do that day. That produces footage. It does not produce a dataset designed to close a specific gap in a specific model.
HumanSignal Services starts from the gap. We write the collection protocol, recruit for the characteristics it calls for, run the capture in our studios, on site, or in the field with synchronized sensor, video, and audio streams, and verify delivery against the spec. The difference from opportunistic capture is control over the distribution: which tasks, which conditions, which failure modes, how many repetitions, which edge cases deliberately included rather than hoped for.
The supporting evidence is public. We publish contributor pay ranges, which makes our sourcing claims checkable rather than rhetorical. Our customer agreement gives the customer exclusive ownership of customer data and of insights produced specifically for them. We hold SOC 2 Type 2 and HIPAA compliance, documented on our security page. And quality workflows run in Label Studio Enterprise, so inter-annotator agreement and reviewer decisions are reportable to you during the work.
The capture operation came in through the acquisition of Erud AI in November 2025, and it runs on the quality infrastructure Label Studio has been built on since 2019. If the constraint is designed capture rather than expert hours, physical AI data services is the place to start.
Which company is bigger
It depends on the measure and the honest answer is that the public figures do not settle it. Mercor reports 2 billion dollars in gross annualized revenue, which is marketplace volume before contractor payouts. Surge reported 1.2 billion dollars in company revenue for 2024 with no outside investors. Mercor has more capital and a higher priced valuation; Surge has a longer operating history and reports a cleaner number.
Do they compete for the same work
Substantially yes, in expert-generated post-training data, evaluation, and RL environments for frontier labs, and both name OpenAI, Anthropic, Meta, and Google. They diverge at the edges: Mercor sells data monetization and wearable capture, Surge sells benchmark-grade evaluation design.
Can either collect physical or robotics data
Mercor can, through its robotics line, which licenses wearable-camera footage captured at partner businesses. Surge publishes no physical collection capability. For designed capture such as teleoperation or instrumented on-site recording, neither is a collection specialist.