Scale AI
Scale AI provides data labeling and evaluation services for training AI models.
Scale AI is an AI infrastructure company providing data annotation, RLHF services, and LLM evaluation through a global workforce of 240K labelers. The company offers an API-first platform for collecting, curating, and annotating unstructured data—images, videos, documents, audio, LiDAR, and 3D point clouds—optimized for model training. Scale serves self-driving, robotics, and AI teams at major enterprises like Google, OpenAI, and Meta. The company's core differentiator is cost-effective human-in-the-loop labeling at scale, enabling continuous data pipelines and real-time model feedback loops.
Problem solved
AI teams need large volumes of precisely labeled, production-ready training data but lack efficient, cost-effective mechanisms to collect, curate, and annotate unstructured data at scale.
Target customer
Enterprise AI teams and model developers at companies building self-driving vehicles, robotics, LLMs, and computer vision applications. Includes Fortune 500 tech, automotive, and defense organizations.
Founders
A
Alexandr Wang
Co-founder, Former CEO
MIT dropout who interned at Hudson River Trading and worked as a software engineer at Quora before co-founding Scale AI in 2016. As of June 2025, he joined Meta as a strategic advisor while remaining on Scale's board.
L
Lucy Guo
Co-founder
Product designer who met Wang at Quora and co-founded Scale AI in 2016. Departed in 2018.
Funding history
Seed
Unknown
2016
Led by Y Combinator
· Unknown
Series B
Unknown
August 2018
Led by Accel
· Unknown
Series C
Unknown
August 2019
Led by Multiple
· Founders Fund, Coatue, Thrive Capital, Spark Capital
Series E
$325M
2021
Led by Dragoneer, Wellington
· Unknown
Series F
$1B
May 2024
Led by Accel
· Cisco Investments, DFJ Growth, NVIDIA, Intel Capital, ServiceNow Ventures, AMD Ventures, WCM, Amazon, Elad Gil, Meta
Series G
$14.3B
June 2025
Led by Meta (49% stake acquisition)
· Unknown
Total raised:
$15.9B
Industries
Pricing
Usage-based pay-as-you-go model starting at $0.02–$0.10 per image label. Expert RLHF annotation at $40+ per hour. Enterprise contracts range from $50,000 to $400,000+ annually with volume commitments and discounts. Average annual cost approximately $93,000.
Notable customers
Google, Microsoft, Meta, OpenAI, General Motors, Waymo, Lyft, Zoox, Cruise, Toyota Research Institute, Pinterest, Airbnb, U.S. Government (military projects), Qatar (social programs)
Integrations
API-first platform; integrations with major cloud providers and ML frameworks (specific integrations not detailed in research)
Tech stack
React (JavaScript frameworks)
Next.js (Web servers)
GSAP (JavaScript frameworks)
Lodash (JavaScript libraries)
Howler.js (JavaScript libraries)
core-js (JavaScript libraries)
Keen-Slider (JavaScript libraries)
Microsoft ASP.NET (Web frameworks)
Headless UI (UI frameworks)
three.js (JavaScript graphics)
Webpack
PWA
Open Graph
Module Federation
DocuSign
Sitecore (CMS)
Magento (Ecommerce)
Vercel Analytics (Analytics)
Atlassian Statuspage (PaaS)
HSTS (Security)
Node.js (Programming languages)
PHP (Programming languages)
Apple iCloud Mail (Webmail)
Google Workspace (Email)
Cloudflare (CDN)
HubSpot (Marketing automation)
MySQL (Databases)
Plasmic (Page builders)
Vercel (PaaS)
Priority Hints (Performance)
Segment (Customer data platform)
Website
Competitors
Labelbox
AI data factory platform for generating and labeling training data; broader feature set but higher costs than Scale.
SuperAnnotate
Data labeling and annotation platform with focus on computer vision; smaller scale and lower enterprise penetration.
V7
Annotation tool with integrated machine learning; more developer-focused, smaller customer base.
Snorkel AI
Programmatic data labeling platform; different approach emphasizing data programming over human labeling.
Encord
Video and image annotation platform; narrower focus on computer vision compared to Scale's multi-modal data support.
Why this matters: Scale AI raised $14.3B in June 2025 when Meta acquired 49%, making it one of the most valuable AI infrastructure companies. However, the deal triggered immediate customer attrition—Google ($150–200M annually) and OpenAI both announced reduced engagement, signaling major market uncertainty. The company faces a critical inflection point balancing Meta's strategic interests with maintaining vendor neutrality for competitors.
Best for: Enterprise AI development teams building self-driving vehicles, robotics, LLMs, and computer vision models who need production-grade labeled datasets at scale without building internal annotation infrastructure.
Use cases
Self-Driving Vehicle Development
Autonomous vehicle teams at companies like Waymo, Lyft, and Cruise use Scale's LiDAR and video annotation APIs to label real-world driving data, enabling safe street navigation without building costly internal labeling operations.
LLM Training and Evaluation
AI labs like OpenAI and Meta use Scale's RLHF services and LLM evaluation to fine-tune language models with human feedback at scale, significantly reducing the time and cost of iterative model improvements.
Continuous Model Improvement
ML teams leverage Scale's API-first platform for continuous data labeling pipelines, enabling real-time feedback loops that identify model failures in production and rapidly generate training data to address them.
Multi-Modal Data Labeling
Computer vision and robotics teams annotate diverse data types—images, 3D point clouds, video—through a single platform, centralizing data management and reducing integration overhead across labeling workflows.
Alternatives
Labelbox
Choose if you need broader data factory capabilities including programmatic labeling and model evaluation in one platform, though at higher cost than Scale.
Snorkel AI
Choose if you prefer programmatic labeling approaches and want to minimize human annotation costs, rather than large-scale human-in-the-loop workflows.
SuperAnnotate
Choose if you're focused on computer vision annotation with mid-market pricing and don't need the enterprise scale or multi-modal data support of Scale.
FAQ
What does Scale AI do? +
Scale AI is an AI infrastructure platform that provides data annotation, RLHF services, and LLM evaluation through a global workforce. It enables enterprises to collect, curate, and annotate unstructured data—images, video, audio, LiDAR, 3D point clouds—optimized for model training with an API-first approach.
How much does Scale AI cost? +
Scale offers pay-as-you-go pricing starting at $0.02–$0.10 per image label, with expert RLHF annotation at $40+ per hour. Enterprise contracts range from $50,000 to $400,000+ annually with volume discounts. Average annual cost is approximately $93,000.
What are alternatives to Scale AI? +
Labelbox (broader data factory platform), Snorkel AI (programmatic labeling focus), SuperAnnotate (computer vision focus), V7 (developer-focused annotation), and Encord (video/image annotation). Scale differentiates on cost efficiency and multi-modal data support at enterprise scale.
Who uses Scale AI? +
Enterprise AI development teams at companies like Google, Microsoft, Meta, OpenAI, General Motors, Waymo, Lyft, Cruise, and Toyota Research Institute. Customers span autonomous vehicles, robotics, LLM development, and computer vision applications.
How does Scale AI compare to Labelbox? +
Scale emphasizes cost-effective human-in-the-loop labeling at massive scale (240K global labelers), while Labelbox offers a broader data factory platform with integrated programmatic labeling and model evaluation. Scale is preferred for highest-volume, cost-sensitive projects; Labelbox for teams seeking all-in-one data infrastructure.
Tags
data annotation
data labeling
RLHF
LLM evaluation
AI infrastructure
machine learning
autonomous vehicles
computer vision
data pipeline
human-in-the-loop