Companies / Data vendor
Scale AI
Scale AI provides training data (annotations and RLHF), model evaluations and red-teaming, and full-stack AI systems to AI labs, enterprises, and governments. Recorded here: 3 clauses of its contract in its own words, what it sells, in its own words, 3 funding events, 2 revenue figures, 14 public prices and 1 market record.
San Francisco, California, USFounded 2016scale.comLatest record 16 Oct 2026
How Scale AI buys data
Scale AI, data partnership page, 15 Oct 2026- What it wants
- Structured, de-identified records from data warehouses, chat, email, documents, work tools and business platforms.
- What it publishes about price
- "$10K–$1M+ illustrative value per data partnership", scaling with how often data is delivered.
- Who it takes
- "40+ employees is our preferred starting point", assessed case by case. Established systems and real operational workflows.
- How the holder is paid
- Not published. A scoping call first; a partnership can begin with a very limited scope.
- Who removes personal data
- Scale anonymises the data inside the company's own environment before extraction, and each package needs the company's approval.
- What it pays for an introduction
- Up to $50,000, paid only if Scale signs a partnership agreement.
We are not Scale AI's agent, partner or representative. If you hold data it might buy, we can put it in front of Scale AI and every other buyer that fits at the same time.
What Scale AI sells
In its own words, read on its own site 15 Oct 2026. On the vendor register
- What it sells
RL environments: “Scale AI provides RL environments for training agents on desktop VMs, MCP servers, and workflows.” scale.com
Datasets: “High-quality training data, annotations, and RLHF to power the world's most advanced AI models.” scale.com
Expert and human data: “We source contributors with precision (25% have advanced degrees) and deliver at the bar frontier AI demands.” scale.com
Evaluations and benchmarks: “Rigorous model evaluations and red-teaming to measure, benchmark, and improve AI performance.” scale.com
Annotation: “Access customized data annotations, model evaluations, and RLHF data via API, SDK, or web frontend.” api-reference.scale.com
Simulation: “Train AI agents in realistic simulated worlds.” scale.com
- Where its data comes from
Made by experts and contractors: “Domain experts curate artifacts that capture the complexity, ambiguity, and edge cases of real professional work.” scale.com
Replicas of real software: “They mirror real computer and API-based systems, supporting rich logs and application state that can be used for programmatic and rubric-based evaluation.” scale.com
- Buys data from owners
- Its programme is set out above.
- Areas its pages name
- Coding, Computer use, Business work, Robotics, Medical
What Scale AI's contract says
The clauses Scale AI publishes for the people it buys data from, in its own words, the costliest first. “Our reading” is ours, not Scale AI's. These terms change: read the page itself before you sign.
When you are paid
Designed as a continuing partnership with the chance of recurring revenue; payment timing is not published.
“Not a one-time file transfer. ... Yes, Scale designs its partnerships so there's opportunity for recurring revenue.”
Our reading: A recurring promise is worth only what the contract guarantees.
scale.com, read 11 Oct 2026Who removes personal data
Scale anonymises inside your own systems before anything is extracted.
“Scale anonymizes everything inside your environment, before any data is extracted. PII is stripped and records are processed so they can't be traced back to individuals or accounts.”
Our reading: Better than a raw transfer, since nothing leaves before it is scrubbed; ask who certifies it.
scale.com, read 11 Oct 2026Other
Scale's own job post says it is still defining how these deals are priced.
“You'll work with Scale’s legal team to define the first version of how these transactions get priced.”
Our reading: There is no standard price yet, so a competing offer sets it.
job-boards.greenhouse.io, read 12 Oct 2026
Not published: Licence term, Exclusivity, Sublicensing, Payment timing. Ask for each in writing.
Public prices involving Scale AI
- Authoring one hard terminal-based task with an automated verifier for Terminal-Bench (coding, machine learning, security, systems, hardware, operations and sciences), Outlier with Scale AI, terminal-bench expert page
Paid only for a task that stumps frontier models; a ceiling, not a typical payout. The same programme's other page (outlier.ai/experts/tbench, held as outlier-tbench-task) still states up to $1,350 per task on the same day, so the two live pages differ by $450; at 6 to 7 hours the ceiling is about $257 to $300 an hour. Undated page, dated by the read.
The payer's own page, Outlier AI (with Scale AI), Terminal-Bench expert page
Up to $1,800 USD/taskper task, typically 6 to 7 hours
- Annotation task bought from Scale AI, typical per-task price across complexity
Second column of Vendr's Appen vs. Scale AI table: Scale's typical per-task price runs about two to five times Appen's. Benchmark for commodity annotation tasks, not expert or coding tasks.
Press report, Vendr marketplace, Appen pricing (Appen vs. Scale AI comparison)
$0.05–$1.00+ depending on complexityper task
- Structured, de-identified company records (Scale AI)
An illustrative range that scales with how often data is delivered. Its preferred starting point is a company of 40 or more employees.
The payer's own page, Scale AI
$10K to $1M+per partnership
- Terminal-style coding task that stumps frontier models (Outlier, run with Scale AI)
Contributor payout ceiling, not the lab's price; the page says rates vary by expertise and lower rates apply to non-core work.
The payer's own page, Outlier (Scale AI)
Up to $1,350 USD/taskper task (paid to the contributor)
- Cumulative payouts to Outlier (Scale AI) contributors
A payout total, not a rate; Outlier publishes no hourly schedule on its homepage.
The payer's own page, Outlier homepage
$500M+ Paid to expertscumulative total, all contributors
- Research Advisor, Human Frontier Collective (Scale AI): senior finance, legal and medical experts
Part-time contract for partners, senior counsel, board-certified MDs and bulge-bracket finance seniors; the role also represents Scale with clients. The HFC Fellow roles on the same board state no rate.
The payer's own page, Scale AI careers (Greenhouse), Research Advisor, Human Frontier Collective (US)
$300/hrper hour (paid to the worker)
- Expressive voice recording for AI training, Bengali speakers in India (Outlier, via Braintrust)
The same Outlier voice project pays by the speaker's language and country: $7.50 for Bengali in India, $10 Thai, $11 Vietnamese, $20 Simplified Chinese, $25 French. Rate as the platform states it, carried by AITraining.jobs, which has the same owner as AIData.Broker.
A catalogue or marketplace listing, AITraining.jobs, Bengali Expressive Voice AI Trainer at Outlier via Braintrust
up to $7.50 USD per hourper hour, ceiling for core project work (paid to the speaker)
- Expressive voice recording for AI training, Simplified Chinese speakers (Outlier, via Braintrust)
One of a set of identical Outlier voice projects priced by language; open in Australia, Canada, the UK, Malaysia, New Zealand and Singapore. Rate as the platform states it, carried by AITraining.jobs, which has the same owner as AIData.Broker.
A catalogue or marketplace listing, AITraining.jobs, Expressive Voice AI Trainer - Simplified Chinese at Outlier via Braintrust
up to $20 USD per hourper hour, ceiling for core project work (paid to the speaker)
- Expressive voice recording for AI training, French speakers (Outlier, via Braintrust)
The top rate in a set of identical Outlier voice projects priced by language. Rate as the platform states it, carried by AITraining.jobs, which has the same owner as AIData.Broker.
A catalogue or marketplace listing, AITraining.jobs, French Expressive Voice AI Trainer at Outlier via Braintrust
up to $25 USD per hourper hour, ceiling for core project work (paid to the speaker)
- Expressive voice recording for AI training, Thai speakers (Outlier, via Braintrust)
One of a set of identical Outlier voice projects priced by language. Rate as the platform states it, carried by AITraining.jobs, which has the same owner as AIData.Broker.
A catalogue or marketplace listing, AITraining.jobs, Thai Expressive Voice AI Trainer at Outlier via Braintrust
up to $10 USD per hourper hour, ceiling for core project work (paid to the speaker)
- Expressive voice recording for AI training, Vietnamese speakers in Vietnam (Outlier, via Braintrust)
One of a set of identical Outlier voice projects priced by language. Rate as the platform states it, carried by AITraining.jobs, which has the same owner as AIData.Broker.
A catalogue or marketplace listing, AITraining.jobs, Vietnamese Expressive Voice AI Trainer at Outlier via Braintrust
up to $11 USD per hourper hour, ceiling for core project work (paid to the speaker)
- Effective pay for a generalist tasker on some Outlier (Scale AI) projects
One tasker's account after Meta's Scale AI deal; shows generalist rates falling while expert rates held.
Press report, Business Insider (on AOL)
around $20 an hour, down from $50per hour (paid to the worker)
- Google's planned 2025 spend with Scale AI for human-labeled training data
Scale's largest customer at the time; Google planned to cut ties after Meta took 49% of Scale.
Press report, TechCrunch (citing Reuters)
$200 million this yearper year (2025, planned)
- Expert-written benchmark question for Humanity's Last Exam (eval set)
Prize money from a $500,000 pool paid by the eval's builders (CAIS and Scale AI) for questions frontier models could not answer.
The payer's own page, Center for AI Safety
$5,000 each for the top 50 questions, $500 each for the next 500per accepted question
What Scale AI earns
Revenue, run rates and ARR as each was published. Run rates are often a best month times twelve, and gross figures include what passes through to experts.
- FY2025MeasureguidanceFigure$2B
- FY2024MeasurerevenueFigure$870M
Funding and acquisitions
- 12 Jun 2025EventStrategicAmount$14.3BValuation$29B
- 25 Apr 2025EventTender offerAmount$150MValuation$25B
- 21 May 2024EventSeries FAmount$1BValuation$13.8B
In the brokerage market
Oct 2026, referral terms
up to $50,000
Paid only if Scale signs a partnership agreement.
Cited in these 8 guides
- Which AI Data Buyers Take Holders Outside the US?
- AI Training Data Companies: Who Does What in 2026
- How Big Is the AI Training Data Market? What the Records Show
- micro1, Handshake, Turing, Mercor: Data Programs Compared
- How Do AI Companies Get Their Training Data?
- How Much Is Your Company's Data Worth to AI Companies?
- What Are RL Environments? Who Builds, Buys and Prices Them
- How to Sell Your Codebase to AI Companies (and What It's Worth)
Cite this
AIData.Broker, “Scale AI in the AI training data market”, reviewed 16 Oct 2026. https://aidata.markets/companies/scale
Free to reuse under CC BY 4.0 if you credit AIData.Broker and link to this page. Each row names the publication that stated it, and words quoted from a source keep that source's own terms.