Scale AI company profile: valuation, funding & revenue

This Scale AI company profile covers the company's valuation, revenue, funding rounds, investors and history. Scale AI works in artificial intelligence, with a focus on AI training data and applications. It was founded in 2016 and is headquartered in San Francisco, California, United States. Its latest valuation on record is Over US$29 billion (June 2025). Every figure links to a public source, and anything the company hasn't disclosed is marked as such.

Revenue

Funding history

Scale AI timeline

Key dates from the sources on this page, oldest first. Metric snapshots are listed after the funding events.

• 2016 — Scale AI is founded in San Francisco by Alexandr Wang and Lucy Guo.
• May 2024 — Series F: US$1B, valuing the company at Nearly US$14B. Investors: Backers including Nvidia, Amazon and Meta.
• Jun 2025 — Strategic investment (49% stake): About US$14.3B, valuing the company at Over US$29B. Investors: Meta.
• June 2025 — Meta investment: About US$14.3 billion for a 49% stake, according to sources cited by Reuters.
• July 2026 — Chief executive: Francis deSouza, formerly COO of Google Cloud and CEO of Illumina, effective August 10, 2026.

Investors

Overview

Scale AI supplies the data that artificial intelligence models learn from. When an AI lab wants to teach a model to write code, answer medical questions or follow instructions, it needs large volumes of carefully prepared examples and human feedback. Scale organises that work: it combines software with a large network of contributors, many of them specialists, to label, write and grade training data. It has also built a growing business selling finished AI applications to companies and governments.

Alexandr Wang and Lucy Guo founded Scale in San Francisco in 2016, when Wang was 19 and had left MIT. It started by labelling images and sensor data for self-driving car companies, then moved into language data as large language models took off. By 2024 it worked with most of the major frontier AI labs. In June 2025 Meta invested about US$14.3 billion for a 49% stake and hired Wang to lead a new superintelligence group, one of the largest deals in the AI industry.

What Scale does

Scale's work falls into two broad businesses:

• Data: creating and evaluating training data for AI models, including reinforcement learning from human feedback (RLHF), where people rank or correct model answers so the model learns which responses are better. It also runs evaluations and "red teaming", testing models for errors and unsafe behaviour.
• Applications: building AI systems for enterprises and governments, from analysing documents to supporting military planning. Scale's recent enterprise wins include BP and Mayo Clinic, according to the company.

Government and defense work has become a larger part of the business. Scale lists the US Department of War among its customers and has been awarded contracts with the Army and the Pentagon's Chief Digital and Artificial Intelligence Office.

The Meta deal

In June 2025 Meta invested about US$14.3 billion in Scale for a 49% stake, according to sources cited by Reuters and Fortune. Scale confirmed the investment and said it valued the company at more than US$29 billion, roughly double its valuation a year earlier. The stake was non-voting, so Meta did not take control of the company. As part of the arrangement, Wang left his role as CEO to join Meta, and he remains Scale's founder and chairman. Jason Droege, Scale's chief strategy officer and a former Uber Eats executive, became interim CEO.

The deal was unusual. It was close to an acquisition in size, yet Scale remained an independent company. Some of Scale's AI lab customers, including competitors of Meta, reportedly reduced their work with the company afterwards, worried about sharing sensitive data with a business partly owned by a rival. That pushed Scale to lean harder into enterprise, government and international customers.

Leadership change

On July 30, 2026, Scale announced that its board had appointed Francis deSouza as CEO, effective August 10, 2026. DeSouza was most recently chief operating officer and president of security products at Google Cloud, and earlier ran the genomics company Illumina, where revenue grew to more than US$4.5 billion. Droege is supporting the transition. The appointment of an experienced enterprise executive was widely read as a signal that Scale wants to grow its applications business, not just data services.

Revenue

Scale doesn't publish financial statements. Revenue figures have appeared in the press, but they vary considerably between outlets and are mostly modelled estimates or figures passed on by people familiar with the company, so we show revenue as not disclosed here. The company has said its international public-sector business and its applications business are growing, without giving exact numbers in its own announcements.

Funding history

• Early rounds: Scale went through Y Combinator and raised a series of venture rounds as it grew with the self-driving and then the language model industries.
• May 2024: US$1 billion at a valuation of nearly US$14 billion, from backers including Nvidia, Amazon and Meta.
• June 2025: Meta's roughly US$14.3 billion investment for a 49% stake, valuing Scale at over US$29 billion.

Part of Meta's investment was used to buy shares from existing holders and to pay a distribution to shareholders and vested employees, giving them liquidity without an IPO. It worked like a very large tender offer.

Restructuring

In July 2025, a few weeks after the Meta deal, Scale cut around 200 full-time employees, about 14% of staff, and ended work with a group of contractors. Droege said the company had expanded its generative AI data capacity too quickly and become too bureaucratic. The restructuring was aimed at focusing resources on the fastest-growing parts of the business.

Competitive landscape

Scale competes with other AI data companies, including Surge AI, Turing, Mercor and Invisible Technologies, several of which grew quickly after the Meta deal as AI labs spread their work among more suppliers. It also competes with labs' in-house data teams and with consulting and software firms in the enterprise and government markets. Its advantages are scale, long experience with frontier models, and security clearances for defense work.

What to watch next

Watch the strategy under the new CEO, growth in enterprise and government contracts, whether AI labs return to Scale or keep diversifying suppliers, any new funding or tender offer that updates the valuation, and how its relationship with Meta develops.

Key facts at a glance

Scale AI is a San Francisco company founded in 2016 by Alexandr Wang and Lucy Guo. It provides AI training data, evaluation and applications to AI labs, enterprises and governments. It raised US$1 billion at nearly US$14 billion in May 2024. In June 2025 Meta invested about US$14.3 billion for a 49% non-voting stake, valuing Scale at more than US$29 billion. Francis deSouza became CEO in August 2026.

Why training data became so valuable

The early large language models learned mostly from text scraped from the internet. As models got better, that wasn't enough. To make them good at reasoning, coding, medicine or law, labs needed examples written and checked by experts, and they needed people to judge model answers so the models could improve. That kind of data is slow and expensive to produce, and it requires managing large numbers of skilled contributors around the world, with strict quality control. Companies that can do this reliably became critical suppliers to the AI industry, and their valuations rose with it.

The flip side is concentration. A handful of AI labs account for most spending on frontier training data, so losing one large customer can change a data company's revenue sharply. That is one reason Scale has been diversifying into applications and government work.

Reading a "non-voting stake" deal

Meta's investment was structured as a large minority stake without voting rights. Deals like this let a strategic investor secure access to talent and capabilities without triggering a full acquisition review or taking control. For the company, they bring in a huge amount of capital and liquidity for shareholders. They also raise questions for customers and staff about independence. When you see a similar structure elsewhere, check who sits on the board, what information rights the investor has, and whether customers have reacted.

What this means if you're…

An enterprise buying AI services: ask suppliers how they protect your data and who their shareholders are, especially if a competitor is an investor.

A founder in AI infrastructure: Scale shows how fast a supplier to AI labs can grow, and how concentrated that customer base can be.

An investor: the Meta deal set a clear valuation marker, but treat press revenue figures for Scale with caution until the company publishes its own.

Latest developments (October 2026)

Scale AI is rebuilding after Meta's 2025 deal, which took a 49% stake and hired founder Alexandr Wang. Interim chief executive Jason Droege said in May 2026 that revenue this year is projected to exceed US$1 billion, with the enterprise business at about US$200 million annualised, according to KuCoin News. Government and defence work, including the Pentagon's Thunderforge programme, has become more important after some AI labs reduced their use of Scale. See our AI training data guide.

Sources

Every figure on this page comes from the public sources below. Where sources disagreed, we used the company's own announcement or a primary news report. Private-company data changes; check the date beside each figure.

1. New Straits Times (Reuters)
2. Fortune
3. Scale AI