Scale AI
Unicorn · $29BYC S16YC Top CompanySan Francisco, US · Founded 2016 · 500 employees · Hiring · 61 known investors
Scale provides high-quality training data, model evaluations, and full-stack AI systems for enterprises and governments to build and deploy AI applications. The company supplies data annotation, RLHF, benchmarking, and red-teaming services for machine learning models.
Also known as Ava · Scale · Scale AI, Inc.
Founders & leadership· Y Combinator alumni (S16)
Scale AI was founded in 2016 by Alexandr Wang and Lucy Guo.
Investors · 61
Also in the syndicate · 21
Funding
SEC filings, press & company announcements$1B disclosed across 1 of 16 rounds · 2017–2026
- $1BSeries FMay 2024 · 3 sources
Accel (lead), Amazon, AMD, Cisco, Coatue, DFJ Growth, Founders Fund, Greenoaks, Index Ventures, Intel, Meta, Nvidia, ServiceNow, Spark Capital, Thrive Capital, Tiger Global Management, WCM, Wellington Management, Y Combinator
Source ↗
Source: company announcements and press reports — follow each round's link for the claim.
Valuation · disclosed
Disclosed eventsSource: SEC prospectus filings, and round valuations the company or its investors disclosed — follow each entry's link for the claim.
Company profile
researched Aug 2026Scale AI, Inc. is an American artificial intelligence infrastructure and software company headquartered in San Francisco, California. Originally focused on data annotation, it now offers reinforcement learning from human feedback (RLHF), large language model evaluation, red-teaming, and enterprise software suites used to build and deploy AI applications. Its product lines are grouped as Data at Scale (training data, annotations, RLHF), Evaluations (model benchmarking and red-teaming), and Applied AI (full-stack systems for enterprises and governments). Annotation products span image annotation, semantic segmentation, 3D point cloud annotation, and LiDAR and RADAR annotation, delivered through a proprietary "Data Engine" and the Scale GenerativeAI Platform.
The company's research arm, the Safety, Evaluation and Alignment Lab, works on evaluating and aligning LLMs and co-created the Humanity's Last Exam benchmark with the Center for AI Safety; Scale has also contributed to the EnigmaEval, MultiChallenge and MASK benchmarks. An in-house LLM red team performs human adversarial testing for vulnerabilities, biases, jailbreaks and agentic AI risks, and has worked with OpenAI, Google DeepMind and national AI Safety Institutes. Data labeling work is outsourced through two subsidiaries: Remotasks, focused on computer vision and autonomous vehicles, and Outlier, focused on LLM data annotation.
A substantial share of Scale's business is public sector. It supports computer vision and agentic generative AI programs across the U.S. Department of Defense, the Intelligence Community and federal civilian agencies, covering acquisition and contracting, mission planning and wargaming, ISR and threat analysis, cyber and security operations, and back-office functions, with access across CUI, SIPR and TS classification levels. Named government-related programs include Donovan (deployed on a classified network with the U.S. Army's XVIII Airborne Corps), Defense Llama, Thunderforge, Autonomous Perimeter Security, and a U.S. Army Robotic Combat Vehicle program.
Founding story
Scale was founded in 2016 by Alexandr Wang and Lucy Guo, who had worked together at Quora, and was built through the Y Combinator Summer 2016 batch. Early investors included Dragoneer Investment Group, Tiger Global Management and Index Ventures.
Business model
Scale sells data, evaluation and applied AI services and software to AI model developers, large enterprises and government agencies, typically via direct sales and contracts (including multi-year government contracts and marketplace availability such as AWS Marketplace for the U.S. National Security Community). Labeling and annotation work is performed by a global contributor workforce engaged through subsidiaries Remotasks and Outlier; the company states it has paid $1 billion to contributors globally.
Revenue comes from paid data annotation and RLHF data delivery, model evaluation and red-teaming engagements, and enterprise/government software and applied AI deployments, including large multi-year government contracts (e.g., a $250 million federal contract in January 2022 and a five-year Qatari government partnership).
Traction
Reported metrics include $870 million revenue in 2024 and an expectation to more than double sales to $2 billion in 2025; 15 billion human decisions used to train AI models; $1 billion paid to contributors globally; 1,000+ employees per the company site and 1,200 employees in 2025 per Wikipedia; a stated $29 billion valuation; and a claim that 90% of leading generative AI model builders use Scale. Startup Intros reports $16.1 billion raised across eight rounds.
Latest developments
In June 2025 Meta Platforms took a 49% non-voting stake for more than $14 billion; founder Alexandr Wang departed for Meta and Jason Droege became CEO. Scale has since announced the appointment of Francis deSouza, a Google Cloud executive, as its new CEO. Bloomberg reported in April 2025 that Scale expected to more than double sales to $2 billion in 2025. Recent product and customer news includes the launch of Scale Labs, a healthcare collaboration with Mayo Clinic, work with Morgan Stanley on evaluations, and continued public sector programs such as Thunderforge and a geospatial collaborative with the University of Missouri–St. Louis.
▸Full profile — market position, technology, go-to-market, geography, history, risks & controversies
Market position
Scale describes itself as powering 90% of the world's leading generative AI model builders and operates leaderboards and private benchmarks used to measure frontier models. Reported revenue was $870 million in 2024, with a valuation of $29 billion cited on its own site and roughly $14 billion at its May 2024 round; Meta's 49% stake was valued at more than $14 billion in June 2025.
Scale positions itself on the combination of frontier research, large-scale high-quality data production with vetted contributors (25% stated to hold advanced degrees), independent evaluation and benchmarking authority, and deployment experience spanning commercial and classified government environments with human-in-the-loop oversight and end-to-end ownership of outcomes.
Technology
Core technology comprises a proprietary Data Engine for producing high-quality training data, annotation tooling for images, semantic segmentation, 3D point clouds, LiDAR and RADAR, RLHF pipelines, an evaluation and red-teaming platform with public and private leaderboards, and the Scale GenerativeAI Platform. Public sector capabilities include a scalable multi-agent architecture with decomposable reasoning and external system integration, multi-classification access (CUI, SIPR, TS), and an end-to-end ML Ops platform for computer vision, all combined with human-in-the-loop subject matter experts.
Go-to-market
Direct enterprise and government sales supported by a "book a demo" motion on its website, partnerships with frontier AI labs, public sector contracting with the Department of Defense, Intelligence Community and federal civilian agencies, availability through AWS Marketplace for the U.S. National Security Community, and public benchmarking/leaderboard and policy-advocacy activity that supports positioning with AI developers and governments.
Frontier AI model developers and technology companies (Meta, Microsoft, Google, OpenAI), large enterprises across automotive, energy, healthcare, media, education, real estate and financial services (General Motors, Toyota Research Institute, BP, Cengage, Time, Mayo Clinic, Howard Hughes, Morgan Stanley, Brex, Instacart, Flexport, Shore Capital), robotics companies (Physical Intelligence, Universal Robots), research organizations (Center for AI Safety), and governments including the U.S. Department of Defense, Intelligence Community, U.S. Army, Defense Innovation Unit, CDAO and the Qatari government.
Geography
Headquartered in San Francisco, California, with an office opened in St. Louis in 2022 and job listings across New York, Seattle, Washington DC and remote roles in the US, Europe and Asia. Contributors are sourced globally, and the company works with governments including the United States and Qatar.
History
Scale was founded in 2016 by Alexandr Wang and Lucy Guo, going through Y Combinator's Summer 2016 batch; the two had previously worked together at Quora. Guo was fired in 2018. In August 2019 a $100 million investment led by Founders Fund pushed the valuation above $1 billion. Scale contracted with the U.S. Department of Defense in 2020, and in May 2021 former U.S. Chief Technology Officer Michael Kratsios joined as managing director and head of strategy. By July 2021 the company was valued at $7 billion following a financing led by Greenoaks, Dragoneer and Tiger Global. In January 2022 it won a $250 million contract providing federal agencies access to its tool suite, and that February it built an Automated Damage Identification Service analyzing satellite imagery of building damage after the Russian invasion of Ukraine for humanitarian groups. It opened a St. Louis office in 2022 and was named to Time's Best Inventions of 2022 list. Scale laid off 20% of its workforce in January 2023. Later milestones include the May 2023 XVIII Airborne Corps deal, becoming OpenAI's "preferred partner" for fine-tuning GPT-3.5 in August 2023, participation in Meta's Purple Llama initiative in December 2023, a Department of Defense LLM test-and-evaluation contract in February 2024, an Accel-led round at nearly $13 billion in March 2024 and a $1 billion round at a $14 billion valuation in May 2024, and an August 2024 agreement with the U.S. AI Safety Institute. In 2025 it released Humanity's Last Exam with the Center for AI Safety, signed a five-year partnership with the Qatari government, became a third-party model evaluator for the U.S. AI Safety Institute, and won a multimillion-dollar DoD deal for the Thunderforge military planning project. In June 2025 Meta Platforms paid more than $14 billion for a 49% non-voting stake and Alexandr Wang left to join Meta, with Jason Droege becoming CEO; the company subsequently announced the appointment of Francis deSouza, previously a Google Cloud executive/COO, as CEO.
Risks & controversies
In December 2024 a former employee sued Scale alleging wage theft and worker misclassification, and a second employee filed a similar suit the following month. In January 2025 several contractors sued the company alleging psychological harm from exposure to disturbing content. The company laid off 20% of its workforce in January 2023. Its defense work, including Defense Llama and Thunderforge, and reporting on U.S. military automation have drawn public scrutiny. Meta's 49% ownership stake and the departure of the founder-CEO, followed by further CEO turnover, represent governance and continuity considerations.
Compiled by commissioned research from 13 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Founder mafia
The Scale AI mafia →47 people who came through Scale AI went on to found or lead other companies.
+ 35 more
Competitors · 7
by search overlapCompanies competing with Scale AI for the same Google search keywords, organic and paid, via search-intersection analysis.
Non-dilutive funding · 1 SBIR/STTR award
Federal grants — no equity taken| Agency | Phase | Year | Amount |
|---|---|---|---|
| U.S. Air ForceAir Force | Phase II | 2021 | $749.9K |
Source: SBIR.gov award data (U.S. Small Business Administration). SBIR/STTR awards are competitive federal R&D grants and contracts — non-dilutive capital alongside any venture rounds above.
Acquisitions · 2
Early investors' stakes continue via these dealsSymplex conducts research and development in reinforcement learning applications. The company was acquired by Scale AI.
Data analytics platform (acquired by Scale AI)
In the news
▸Research sources · 13
primary sources listed
- Scale AIscale.com · web
13 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Scale AI do?
- Scale AI supplies training data, model evaluations and full-stack AI systems to enterprises, AI labs and governments.
- Who founded Scale AI?
- Scale AI was founded by Alexandr Wang, Lucy Guo in 2016.
- Who are Scale AI's investors?
- Scale AI's investors include Accel, Adit Ventures Management, Llc, Beyond Capital, Cheyenne Ventures, Coatue Management, Elad Gil, Felicis Ventures, Founders Fund and 32 more.
- How much funding has Scale AI raised?
- Scale AI has disclosed $1B raised across 1 of its 16 known rounds.
- Where is Scale AI headquartered?
- Scale AI is headquartered in San Francisco, US.









.png)









.jpg)






