Labelbox

Labelbox is a data infrastructure company founded in 2018 by Manu Sharma, Brian Rieger, and Dan Rasmuson.

Updated 52 days ago Reviewed by 7wData

Visit Website

Profile

Labelbox provides software, expert networks, and managed services that enable AI teams to create high-quality training and evaluation data for machine learning models.

Labelbox is a data infrastructure company founded in 2018 by Manu Sharma, Brian Rieger, and Dan Rasmuson. Headquartered in San Francisco with an engineering hub in Wrocław, Poland, the company has raised $189 million across five funding rounds and reached a $1 billion valuation by late 2025. The company occupies a critical role in the AI development stack, providing software, managed services, and expert networks that power frontier AI training and evaluation.

Labelbox has shifted its positioning away from traditional data annotation toward what it calls the "data factory"—a comprehensive suite of services including reinforcement learning data generation, custom model evaluations, robotics training data, and access to its Alignerr expert network of over 1 million domain specialists vetted through rigorous screening (3% acceptance rate). The platform works behind the scenes with over 80% of leading U.S. AI labs, supporting companies including Google Cloud, Walmart, Pinterest, Etsy, Ideogram, Eleven Labs, and Suno, while maintaining partnerships with aerospace contractors, healthcare systems, and robotics companies.

In 2025, Labelbox expanded its product portfolio significantly, launching new leaderboards for multimodal reasoning, introducing Alignerr Connect for direct hiring of vetted AI trainers, and rolling out tools for LLM reasoning evaluation, agent trajectory annotation, and coding support. The company's strategic acquisitions include Upcraft in February 2026, an AI-powered sales automation startup, signaling intent to automate the scaling of its expert network. Labelbox generates an estimated $50 million in annual revenue and employs approximately 449 people as of early 2026.

The company is positioning itself against competitors like Scale AI and Hive by emphasizing the shift from simple labeling to sophisticated reinforcement learning environments and domain expertise. CEO Manu Sharma has publicly stated that "data will remain a crucial pillar in achieving artificial general intelligence," framing Labelbox's infrastructure as foundational to the post-training arms race among frontier AI labs.

Products by Labelbox

A Alignerr Connect Alignerr Connect is a specialized talent marketplace that connects organizations with expert annotators and reviewers for AI training No reviews yet Be the first to review →

Who buys this

Frontier AI labs and LLM developers (OpenAI, Anthropic, Claude-adjacent research teams)
Enterprise AI teams in finance, healthcare, and manufacturing automating computer vision and NLP
Robotics companies requiring video, trajectory, and manipulation data
Fortune 500 companies in retail, logistics, and insurance deploying custom vision models
Specialized domain experts in medicine, law, and engineering requiring structured skill-building environments

Publicly disclosed clients

Google Cloud
Walmart
Pinterest
Etsy
Ideogram
Eleven Labs
Suno
NASA Jet Propulsion Laboratory
Intuitive Surgical
Procter & Gamble

Strengths and what to watch

Strengths

Deep penetration with 80%+ of leading U.S. AI labs, providing structural advantage in understanding frontier model development needs
Differentiated expert network (Alignerr) of 1+ million vetted domain specialists with 3% acceptance rate, creating defensible supply advantage
Integrated platform spanning annotation software, managed labeling services, reinforcement learning environments, and evaluation benchmarking—not single-product competitor

Watch for

Customer concentration risk: heavy reliance on top AI labs and frontier model developers; downturns in OpenAI, Anthropic, or Meta spending could materially impact revenue
Competitive pressure from Scale AI (better-funded, larger customer base) and internal tooling by major labs; Labelbox's acquisition of Upcraft in Feb 2026 suggests scaling challenges in expert network growth
Leadership and execution questions: employee reviews on Glassdoor reference management disconnects and talent attrition, with some reports citing organizational dysfunction; growth trajectory to justify $1B valuation remains unproven outside select customer segments

Recent moves

Key Information

Industry: Data Generation & Labelling
Founded: 2018
Headquarters: San Francisco

Frequently Asked Questions

What is Labelbox?

Labelbox is a data infrastructure company that helps AI teams create high-quality training and evaluation data for machine learning models. Founded in 2018, it operates as a "data factory" offering annotation software, managed labeling services, reinforcement learning environments, and expert networks with over a million vetted domain specialists.

What services does Labelbox provide?

Labelbox provides annotation software, managed labeling services, reinforcement learning data generation, custom model evaluations, and robotics training data. The company also operates Alignerr, an expert network of over 1 million vetted domain specialists with a rigorous 3% acceptance rate, enabling AI teams to source specialized human expertise.

Who are Labelbox's main customers?

Labelbox serves frontier AI labs, enterprise AI teams, and robotics companies. Its clients include Google Cloud, Walmart, Pinterest, Etsy, Ideogram, and Eleven Labs. The company works behind the scenes with over 80% of leading U.S. AI labs, supporting companies across healthcare, finance, retail, and aerospace sectors.

What is Labelbox's Alignerr expert network?

Alignerr is Labelbox's network of over 1 million vetted domain specialists screened at 3% acceptance. It connects AI teams with experts in medicine, law, and engineering to generate training and evaluation data. Labelbox launched Alignerr Connect for direct hiring of vetted AI trainers and specialists.

How does Labelbox differ from Scale AI?

Labelbox positions itself beyond simple data annotation by emphasizing reinforcement learning environments, domain expertise, and integrated evaluation benchmarking. While Scale AI focuses on broader data services, Labelbox differentiates through its Alignerr expert network of 1+ million vetted specialists and deep penetration with 80%+ of leading U.S. AI labs.

When did Labelbox last raise funding?

Labelbox raised $110 million in Series E funding in October 2025 at a $1 billion valuation, led by SoftBank Vision Fund 2 and Andreessen Horowitz. The company has raised $189 million across five rounds. In February 2026, it acquired Upcraft, an AI automation startup, to scale its expert network.

How Labelbox compares

Direct head-to-head against 2 competitors. Picked by 7wData.

Positioning: Labelbox provides software, expert networks, and managed services that enable AI teams to create high-quality training and evaluation data for machine learning models.
Customer segments: Frontier AI labs and LLM developers (OpenAI, Anthropic, Claude-adjacent research teams)
Strengths: Deep penetration with 80%+ of leading U.S. AI labs, providing structural advantage in understanding frontier model development needs
Watch for: Customer concentration risk: heavy reliance on top AI labs and frontier model developers; downturns in OpenAI, Anthropic, or Meta spending could materially impact revenue
Recent moves: Labelbox acquires Upcraft, an AI-powered sales automation startup, to scale its Alignerr expert network

Positioning: AI training data platform targeting frontier labs and US government contracts, 49% acquired by Meta in June 2025.
Customer segments: Frontier AI labs, US Department of Defense and defense contractors, large enterprises building proprietary AI models.
Strengths: Generated over $2B revenue in 2025, the highest confirmed revenue base among dedicated AI training data providers.
Watch for: Meta's 49% ownership has prompted OpenAI and Google to reduce or exit Scale engagements over vendor neutrality.
Recent moves: Meta acquired 49% stake for $14.3B in June 2025, placing CEO Alexandr Wang at head of Meta Superintelligence division.

Positioning: Cloud AI model APIs and distributed data labeling for content moderation, trust/safety, and enterprise AI training.
Customer segments: Digital platforms needing content moderation (Reddit, Yubo), enterprise trust/safety teams, generative AI detection buyers.
Strengths: 2M+ globally distributed crowd contributors paired with pre-built API suite for NSFW, deepfake, and CSAM detection.
Watch for: Content moderation focus limits appeal for buyers seeking RLHF and domain-expert post-training pipeline support.
Recent moves: Launched Moderation 11B Vision Language Model in February 2025, built on Llama 3.2, expanding visual moderation APIs.

Sources

labelbox.com — Company website, product portfolio, customer list, mission statement
labelbox.com — Founding story, team structure, locations, mission, core values, leadership
www.crunchbase.com — Founding date 2018, founders (Manu Sharma, Brian Rieger, Dan Rasmuson), funding history, investor list, valuation
www.builtinsf.com — Series D $110M funding in January 2022, SoftBank Vision Fund 2 as lead investor, hiring plans
www.prnewswire.com — Upcraft acquisition in February 2026, strategic rationale, integration with Alignerr network
labelbox.com — Q1 2025 product launches including Multimodal Reasoning Leaderboard, Alignerr Connect, LLM evaluation tools
salestools.io — Series E $110M funding round in October 2025, $1B valuation, SoftBank and a16z as lead investors
tracxn.com — Current employee headcount ~449 as of Feb 2026, funding history, investor list