Last Updated: November 14, 2026
Summary: Scale AI's main competitors in 2026 are Surge AI, Mercor, Snorkel AI, Labelbox, Appen, Turing, Invisible Technologies, Toloka and Prolific. Surge AI, Mercor and Snorkel AI compete hardest for frontier AI lab budgets with expert human data, while Labelbox, Toloka and Prolific suit teams that want to run labeling or evaluation work themselves. Most of these vendors quote custom prices, so the real comparison comes down to data type, expert depth, neutrality and contract terms.
The market changed in June 2025, when Meta took a 49% stake in Scale AI and hired its founder. Big customers moved work elsewhere, and rivals raised money at multibillion-dollar valuations. This guide covers what happened to Scale, profiles nine Scale AI alternatives for training data, data labeling and model evaluation, and compares them on funding, customers and pricing. Figures were checked in November 2026.
Free download: The AI Stats Sheet 2026. 36 key AI numbers for business on four printable pages, with the source under every figure. Join 2,000+ business leaders who read AI Business Weekly and get it in your inbox right away.
In this guide
What Does Scale AI Do?
Scale AI sells human-made training data, model evaluation and AI deployment services to AI labs, enterprises and governments. Founded in 2016, Scale AI became the best-known name in data labeling.
Scale AI built its business on human-made training data for AI labs, companies and governments. Photo: Getty Images / Unsplash+
Scale AI now has two sides: a data business for frontier labs and an applications business for enterprises and government. In May 2026, Scale described products including the Scale Data Engine for computer vision, the Scale GenAI Platform for fine-tuning and evaluating models, and Scale Donovan, a decision-support tool for defense users.
What Happened to Scale AI After Meta's Investment?
Meta invested $14.3 billion in Scale AI in June 2025 for a 49% stake without voting power, and Scale founder Alexandr Wang left to join Meta. Within weeks, OpenAI said it was winding down its work with Scale, Google was reported to be cutting ties, and Scale laid off about 14% of its staff.
The timeline, as reported:
June 12, 2025: Scale announced a "significant new investment" from Meta that valued the company at more than $29 billion, according to Scale's press release on Business Wire. Chief Strategy Officer Jason Droege became interim CEO; Wang stayed on the board.
June 18, 2025: CNBC reported the deal terms ($14.3 billion for 49%, no voting power) and Droege's memo calling Scale "unequivocally, an independent company." OpenAI said it had been wrapping up its Scale work over six to 12 months, not because of Meta. Reuters reported Google was cutting ties.
July 2025: Scale cut about 200 full-time employees, roughly 14% of staff, plus about 500 contractors, TechRepublic reported.
September 2025: Droege told Axios that Scale had "grown every single month since the deal happened" and that its applications business brought in $200 million to $300 million.
May 6, 2026: Scale said the U.S. Department of War's Chief Digital and AI Office raised the ceiling on its agreement from $100 million to $500 million, according to Scale's announcement. Axios described Droege as CEO and said Scale has more than 1,300 employees.
The practical issue for customers is neutrality. When one AI lab owns nearly half of a data supplier, rival labs worry about what their data requests reveal. That worry explains much of the growth of the Scale AI alternatives below. Meta's own AI push is covered in our guide to what Meta AI is.
Who Are Scale AI's Biggest Competitors?
Scale AI's biggest competitors for frontier lab spending are Surge AI, Mercor and Snorkel AI, all expert human data suppliers with multibillion-dollar valuations or $1 billion-plus revenue. Turing, Invisible Technologies, Appen and Toloka compete at smaller sizes, and Labelbox and Prolific offer platforms teams run themselves.
Scale AI rivals split into expert data services, platforms with a workforce, and labeling software. Photo: Getty Images / Unsplash+
Sort them by what you are buying:
Expert data as a service: the vendor recruits experts and delivers finished datasets or evaluations. Surge AI, Mercor, Snorkel AI, Turing, Invisible and Appen fit here.
Platform plus workforce: you design tasks on the vendor's software and tap its contributor pool. Toloka and Prolific fit here.
Labeling software: you bring your own labelers or buy services as an add-on. Labelbox started here.
Surge AI: The Largest Independent Rival
Surge AI is a data labeling and RLHF company that, according to Reuters, generated more than $1 billion in revenue in 2024, more than Scale AI's $870 million. Surge AI was bootstrapped and profitable when Reuters reported in July 2025 that it was seeking its first outside capital.
What it does: custom human data for training and evaluating large language models, including RLHF, with labeling in 30+ languages, per Surge AI's pricing page.
Customers: Google, OpenAI and Anthropic, according to Reuters (July 1, 2025).
Funding: Reuters reported Surge sought up to $1 billion at a valuation above $15 billion, and Bloomberg reported on July 30, 2025 that talks valued it at $25 billion or more. A closed round was not confirmed in the sources we checked.
Pricing: enterprise projects are priced per task after an intro call. Individuals can sign up free to launch a labeling project. No rates are published.
Surge AI pitches itself as independent. After the Meta deal, Surge wrote in a company blog post that clients "need independent partners."
Mercor: The Fastest-Growing Expert Network
Mercor connects AI labs and enterprises with subject-matter experts who train, test and evaluate AI models, and on November 10, 2026 it raised $750 million at a $20 billion valuation. Mercor's valuation doubled from about $10 billion after a $350 million round in October 2025, Reuters reported.
What it does: human data, evaluations, benchmarks and RL environments built from real professional work, plus enterprise deployment with forward-deployed engineers, per Mercor's website.
Scale: more than five million experts in its network, according to Mercor's Series D announcement, which lists Benchmark, Felicis, General Catalyst, Nvidia and Salesforce among participants.
Pricing: no published buyer prices. The site routes buyers to "Book demo" and "Contact sales."
Mercor fits best when you need credentialed professionals, such as doctors or lawyers, rather than general annotators.
Snorkel AI: From Software to Expert Data
Snorkel AI is a Stanford-born company that shifted from selling data software to supplying finished expert datasets and RL environments, and in September 2026 it raised $350 million at a $3.5 billion valuation. Snorkel's annualized revenue run rate passed $350 million, up from about $20 million a year earlier, Reuters reported on September 22, 2026.
What it does: data as a service to label, refine and evaluate domain-specific datasets, custom AI agents, and public leaderboards for model evaluation, per Snorkel's website.
Customers: frontier AI labs, cloud providers, enterprises and the U.S. federal government, with coding data among its biggest areas of demand (Reuters).
Funding: led by Insight Partners and S32, nearly triple its $1.3 billion valuation of May 2025.
Pricing: not published. Buyers talk to a strategist or data researcher.
Labelbox: Labeling Software With Services on Top
Labelbox is a data labeling platform that meters usage in Labelbox Units (LBUs) and sells labeling services as an add-on. Free Labelbox accounts get 500 LBUs a month, according to Labelbox's billing documentation.
How the meter works: annotating a standard data row uses 1 LBU, a live multimodal or LLM data row uses 20 LBUs, and Catalog storage uses about 1 LBU per 60 data rows a month.
What 500 free LBUs buys: roughly 500 standard annotated rows, or just 25 LLM data rows (500 divided by 20). Teams doing LLM work will outgrow the free tier fast.
Plans: Labelbox's pricing page lists Free, Starter and Enterprise tiers, with custom pricing and volume discounts for Enterprise, which includes fully managed human data services.
Labelbox suits teams with in-house labelers who want control of the workflow.
Appen: The Public-Market Veteran
Appen is an Australian-listed AI training data company with more than 30 years in the business, and it is the only publicly listed company on this list. Appen reported first-half 2026 revenue of A$119.9 million, up 17.5%, according to Staffing Industry Analysts (August 28, 2026).
The split: Appen China revenue rose 80.4% to A$76.2 million while Appen Global fell 26.9% to A$43.7 million. Appen reaffirmed full-year 2026 revenue guidance of A$270 million to A$300 million.
What it offers: frontier alignment and evaluation, coding and agentic AI data, cybersecurity, speech and audio, data collection and physical AI. Appen says it has more than one million contributors in 170+ countries covering 235+ languages.
Pricing: not published. Buyers "Talk to an AI Data Strategist."
Appen fits multilingual and speech projects.
Turing, Invisible, Toloka and Prolific
These four are more specialized Scale AI alternatives: coding data, enterprise AI operations, flexible per-project data, and research-grade human feedback.
Smaller rivals compete on niches such as coding data, AI operations and research-grade human feedback. Photo: Alex Shuper / Unsplash+
Turing
Turing supplies software engineers who write and review code for training large language models. Turing raised $111 million at a $2.2 billion valuation in March 2025, led by Malaysia's Khazanah Nasional, and had about $300 million in annualized revenue run rate at the time, TechCrunch reported. OpenAI is a customer, and Turing says it works with about four million coders. Its site sends enterprise buyers to "Talk to an expert."
Invisible Technologies
Invisible Technologies provides expert-made AI training data, RLHF and evaluation alongside a modular enterprise AI platform (Neuron, Atomic, Meridial, Synapse and Axon). Invisible raised $100 million in September 2025 led by Vanara Capital and reported $134 million in 2024 revenue, according to Invisible's announcement. SiliconANGLE, citing Bloomberg, put the valuation above $2 billion. It says it has trained models for Cohere, Microsoft and AWS. Its site sends buyers to "Book a demo."
Toloka
Toloka helps train and evaluate AI models with networks of human experts and testers. Toloka raised $72 million in May 2025 in a round led by Jeff Bezos's Bezos Expeditions, with parent Nebius keeping a majority economic stake but giving up voting control, PYMNTS reported, citing Reuters. Customers include Amazon, Microsoft and Anthropic. Toloka's pricing page says each project gets an estimate based on task complexity, expertise tier and volume, with no minimums or long-term contracts.
Prolific
Prolific is a human data platform used for AI evaluation, RLHF data collection and academic research, with more than 200,000 active participants, according to Prolific's AI page. Prolific is the most transparent on price: corporate customers pay a platform fee of 42.857% on top of participant rewards, and academic and non-profit users pay 33.3%, per Prolific's help center. The recommended minimum pay is $12 an hour.
Worked example: a 15-minute evaluation study with 1,000 participants paid $12 an hour costs $3,000 in rewards (1,000 x $3). The 42.857% corporate fee adds $1,285.71, for a total of about $4,286.
Scale AI Alternatives Compared
The table compares Scale AI competitors on focus, latest reported funding and pricing model.
Company | Best for | Latest reported funding or revenue | Pricing model |
|---|---|---|---|
Scale AI | Large data programs, government and defense | Valued above $29B after Meta deal (June 2025) | Custom |
Surge AI | Frontier LLM training data and RLHF | Over $1B revenue in 2024 (Reuters, July 2025) | Per task, quote-based |
Mercor | Credentialed domain experts and evals | $750M at $20B (Nov 10, 2026) | Contact sales |
Snorkel AI | Expert datasets, RL environments, coding data | $350M at $3.5B (Sept 22, 2026) | Contact sales |
Labelbox | Teams running their own labeling | Not reviewed for this guide | Free, Starter, Enterprise tiers; 500 free LBUs a month |
Appen | Multilingual, speech and audio data | A$119.9M H1 2026 revenue (Aug 2026) | Project quotes |
Turing | Coding data for LLMs | $111M at $2.2B (March 2025) | Talk to an expert |
Invisible | Enterprise AI operations plus training data | $100M, valued above $2B (Sept 2025) | Book a demo |
Toloka | Flexible projects without long contracts | $72M led by Bezos Expeditions (May 2025) | Per-project estimate, no minimums |
Prolific | Human evaluation and research studies | Not reviewed for this guide | Rewards plus 42.857% fee (corporate) |
Two patterns stand out. The money has moved to expert data, as Mercor's and Snorkel's 2026 rounds show. And only Labelbox, Toloka and Prolific explain pricing before a sales call. For context on AI budgets, see our AI spending statistics.
Free download: The AI Stats Sheet 2026. 36 key AI numbers for business on four printable pages, with the source under every figure. Join 2,000+ business leaders who read AI Business Weekly and get it in your inbox right away.
How to Choose a Scale AI Alternative
Match the vendor to your data type first, then check expert depth, neutrality and how much work you want to manage yourself.
Name the data you need. RLHF, coding, expert answers, images and speech each have different best vendors. Our explainer on AI fine-tuning covers where this data fits in model building.
Decide managed or self-serve. If you have no labeling team, pick a managed vendor like Surge AI, Mercor or Snorkel AI. If you have ML engineers who want control, Labelbox, Toloka or Prolific cost less to start.
Ask about neutrality. If you compete with Meta, ask any data vendor who owns it, who else it serves, and how it walls off customer data.
Test expert quality. Run a paid pilot of a few hundred tasks with two vendors and score the output blind.
Get pricing in writing. With most vendors quoting per task or per project, ask for unit prices, rework terms and quality guarantees before you sign.
Many of these vendors also appear on our list of top AI startups of 2026, alongside the companies on our biggest AI companies list that buy their data.
Frequently Asked Questions
Who are Scale AI's main competitors?
Scale AI's main competitors are Surge AI, Mercor, Snorkel AI, Labelbox, Appen, Turing, Invisible Technologies, Toloka and Prolific. Surge AI, Mercor and Snorkel AI compete most directly for frontier AI lab budgets with expert human data.
Does Meta own Scale AI?
No. Meta invested $14.3 billion in June 2025 for a 49% stake without voting power, and Scale AI says it remains an independent company. Scale's founder, Alexandr Wang, left to work at Meta and stayed on Scale's board.
Is Surge AI bigger than Scale AI?
By 2024 revenue, yes. Reuters reported in July 2025 that Surge AI generated more than $1 billion in 2024 revenue, compared with $870 million for Scale AI.
What is the best Scale AI alternative for data labeling?
For managed expert data, Surge AI, Mercor and Snorkel AI are the leading Scale AI alternatives. For teams that want to run labeling themselves, Labelbox offers software with a free tier of 500 LBUs a month, and Toloka offers per-project pricing with no minimums.
How much do Scale AI competitors charge?
Most Scale AI competitors, including Surge AI, Mercor, Snorkel AI and Appen, quote custom prices per task or project. Prolific is an exception, publishing a 42.857% platform fee on top of participant pay for corporate customers.
The Bottom Line
Scale AI still runs large data and government programs, with a Pentagon agreement now worth up to $500 million, but Meta's 49% stake pushed many AI labs to spread their data spending across rivals. Surge AI, Mercor and Snorkel AI have been the biggest winners, while Labelbox, Toloka and Prolific give smaller teams a lower-commitment, hands-on route.
The right pick depends on fit: the data you need, the experts behind it, and whether you trust the vendor with your roadmap. Pilot two vendors before committing.
Related Articles
Top AI Startups 2026: the fastest-growing AI companies to watch.
Biggest AI Companies: the largest players in AI by value and revenue.
AI Fine-Tuning Explained: how training data turns a base model into a specialist.
What Is Meta AI?: Meta's assistant and AI strategy explained.
AI Spending Statistics: how much companies are investing in AI.
⭐ Get more AI Business Weekly in Google: add us as a preferred source on Google.
Written by Sameer Khan with AI assistance, fact-checked against the sources linked above.