Scale

Machine Learning & Artificial Intelligence

Scale is a San Francisco-based AI infrastructure company founded in 2016 by Alexandr Wang and Lucy Guo, focused on building the operational backbone for enterprise AI deployment.

Reviewed by 7wData
Share Email

On this page

Profile

Scale provides AI training data, model evaluation, and enterprise deployment infrastructure for large AI model developers and government agencies.

Scale is a San Francisco-based AI infrastructure company founded in 2016 by Alexandr Wang and Lucy Guo, focused on building the operational backbone for enterprise AI deployment. The company emerged from Y Combinator and achieved unicorn status in August 2019 after Peter Thiel's Founders Fund led a $100 million round. Scale's platform evolved from data labeling into a full stack spanning data sourcing, model evaluation, and enterprise deployment infrastructure.

In June 2025, Meta invested $14.3 billion for a 49% non-voting stake, valuing Scale at $29 billion and triggering Wang's departure to join Meta's AI efforts. Jason Droege, a former Uber executive and Scale's Chief Strategy Officer, assumed the CEO role. The deal immediately precipitated departures from Scale's largest customers: Google—which had spent roughly $150 million on Scale's services in 2024—ceased its contract in June 2025, followed by OpenAI, Microsoft, and xAI, each citing concerns about data security and competitive neutrality.

Scale reported 2025 as its strongest financial year, with over $1 billion in new business signed and its data business achieving profitability; Sacra estimates 2025 revenue at $2 billion, up from $870 million in 2024. In March 2026, Scale launched Scale Labs, an expanded research division building on its Safety, Evaluation, and Alignment Lab (SEAL) to advance post-training evaluation and agentic AI systems. The company operates through subsidiaries Remotasks (computer vision and autonomous vehicle labeling with over 240,000 contractors) and Outlier (LLM annotation with specialized contractors), though both have faced lawsuits alleging wage theft, worker misclassification, and psychological harm from exposure to disturbing content.

Track Scale and 240+ vendors.

335k+ subscribers read the daily AI & data note. One email, both newsletters. Unsubscribe anytime.

Who buys this

  • Foundation model developers and AI labs training large language models
  • Autonomous vehicle and robotics companies requiring computer vision data
  • Healthcare systems deploying clinical AI and reducing administrative burden
  • US and international government agencies on defense and social program optimization
  • Enterprise organizations implementing agentic AI systems

Publicly disclosed clients

  • Mayo Clinic
  • General Motors
  • Time Magazine

Strengths and what to watch

Strengths

  • Scaled to $2 billion annual revenue (2025) by building two-sided markets for data annotation and securing government contracts.
  • Diversified beyond labeling into model evaluation and enterprise deployment via Scale Labs research (March 2026) and RL Environments (February 2026) for agent training.
  • Achieved 49% ownership stake from Meta—largest AI investor—validating strategic value despite triggering customer exodus.

Watch for

  • Customer concentration risk: loss of Google, OpenAI, Microsoft, and xAI in H2 2025 due to Meta deal concerns; repeat attrition would accelerate runway pressure.
  • Contractor labor and data security lawsuits filed 2024–2025 alleging wage theft, misclassification, and psychological harm; June 2025 data exposure via public Google Docs compounded reputational damage.
  • CEO succession under duress: Jason Droege assumed role during customer exodus and has limited public track record; scaling the business under investor scrutiny will test new leadership.

Recent moves

Key Information

Industry
Machine Learning & Artificial Intelligence
Founded
2016
Headquarters
United States

Frequently Asked Questions

What is Scale AI?

Scale is a San Francisco-based AI infrastructure company founded in 2016 by Alexandr Wang and Lucy Guo. It provides AI training data, model evaluation, and enterprise deployment infrastructure for foundation model developers, autonomous vehicle companies, healthcare systems, and government agencies worldwide.

What services does Scale AI provide?

Scale provides AI training data sourcing and labeling, model evaluation for safety and alignment, and enterprise deployment infrastructure. It operates subsidiaries Remotasks for computer vision data and Outlier for LLM annotation. In 2026, Scale launched Scale Labs for agentic AI evaluation and RL Environments for agent training.

Who are Scale AI's customers?

Scale serves foundation model developers and AI labs, autonomous vehicle and robotics companies, healthcare systems deploying clinical AI, US and international government agencies on defense and social programs, and enterprises implementing agentic AI. Notable clients include Mayo Clinic, General Motors, and Time Magazine.

Why did Google leave Scale AI?

Google ceased its contract with Scale in June 2025 after Meta invested $14.3 billion for a 49% stake, valuing Scale at $29 billion. Google—Scale's largest customer, spending $150 million in 2024—cited competitive neutrality and data security concerns. OpenAI, Microsoft, and xAI also ended partnerships citing similar concerns.

What are Scale Labs and RL Environments?

Scale Labs, launched March 2026, expands Scale's research division building on its Safety, Evaluation, and Alignment Lab to advance post-training evaluation and agentic AI systems. RL Environments, launched February 2026, allows companies to train and evaluate AI agents in simulated workflows. Half of Scale's new projects use RL Environments.

How much is Scale AI worth?

Meta's $14.3 billion investment in June 2025 valued Scale at $29 billion, with Meta acquiring a 49% non-voting stake. Scale reported 2025 as its strongest financial year, with over $1 billion in new business signed. Sacra estimates 2025 revenue at $2 billion, up from $870 million in 2024.

How Scale compares

Direct head-to-head against 3 competitors. Picked by 7wData.

This company

Scale

Positioning
Scale provides AI training data, model evaluation, and enterprise deployment infrastructure for large AI model developers and government agencies.
Customer segments
Foundation model developers and AI labs training large language models
Strengths
Scaled to $2 billion annual revenue (2025) by building two-sided markets for data annotation and securing government contracts.
Watch for
Customer concentration risk: loss of Google, OpenAI, Microsoft, and xAI in H2 2025 due to Meta deal concerns; repeat attrition would accelerate runway pressure.
Recent moves
Scale AI launches Scale Labs, expanded research hub for agentic and multimodal AI evaluation

Snorkel AI

Positioning
Programmatic training data and LLM evaluation layer for regulated enterprises building production AI systems at scale.
Customer segments
Fortune 500 enterprises in financial services, insurance, and government. Buyer is the ML engineering or AI platform team lead.
Strengths
Programmatic data labeling claimed 10-100x faster than manual annotation, with a track record across five of the top ten U.S. banks.
Watch for
Repeated strategic pivots (labeling to LLM eval to expert data) and a September 2025 layoff signal organizational instability and unclear product focus.
Recent moves
May 2025: $100M Series D at $1.3B valuation led by Addition, total $237M raised. Accenture joined as strategic investor August 2025.

Appen

Positioning
Specialist in bespoke human-annotated training data for LLMs and multimodal AI, repositioning away from commodity annotation volume.
Customer segments
Enterprise AI labs, big-tech LLM teams, automotive and AV builders needing LiDAR and sensor fusion annotation.
Strengths
1M+ annotators across 170+ countries enabling multilingual and multimodal data at scale including LiDAR point cloud labeling.
Watch for
Single-client revenue concentration exposed by Google contract loss, which drove a 10% YoY revenue drop in H1 FY25.
Recent moves
Ryan Kolln confirmed CEO via shareholder vote at May 2026 AGM, following 48 layoffs in February 2025 during CEO transition.

Labelbox

Positioning
Scale AI alternative for enterprise AI teams needing a full-stack data factory: labeling, RLHF pipelines, and model evaluation under one platform.
Customer segments
Fortune 500 ML engineers and AI platform teams (Walmart, Adobe, P&G) plus AI labs building or fine-tuning foundation models.
Strengths
Operates Alignerr, a 1M-plus expert annotator network, enabling enterprises to staff, operate, or self-serve data labeling at scale.
Watch for
Default probability rose from 0.49 to 0.59 between 2024 and 2025 with no new institutional funding since the 2022 Series C.
Recent moves
Acquired Upcraft (February 2026) to embed agentic automation into its Alignerr annotator network and expand expert-workforce sourcing.

Sources

  1. scale.com — Current product positioning (Data Engine, GenAI Platform), customer segments, mission statement, and claim of serving 90% of leading AI model builders
  2. en.wikipedia.org — Company founding (2016 by Wang and Guo), Y Combinator background, unicorn status (August 2019), valuation timeline, founders' Quora background, SEAL founding (2023), Remotasks and Outlier subsidiaries
  3. techcrunch.com — Meta's $14.3 billion investment details, Wang's departure as CEO, Jason Droege becoming interim CEO, company independence claim
  4. www.cnbc.com — Google's $150M spending in 2024, planned $200M 2025 spending, departure decision post-Meta deal, customer competitive concerns
  5. winbuzzer.com — June 2025 data exposure incident via public Google Docs, affected customers (Google, Meta, xAI), contractor infrastructure concerns
  6. scale.com — March 2026 Scale Labs launch, research focus on agentic and multimodal systems, evaluation methods, government collaboration
  7. scale.com — February 2026 RL Environments announcement, agent training in simulated workflows, 50% of new projects using RL environments