Funding timelineScale AI is the San Francisco data infrastructure company behind the AI boom's labeling and evaluation layer, founded in 2016 by Alexandr Wang and Lucy Guo.
Every disclosed equity round for Scale AI, newest to oldest, with lead and participating investors where reported. Data is stored oldest-first; the timeline is displayed latest-first.
Meta invested ~US$14.3B for a ~49% non-voting stake at a ~US$29B implied valuation — one of Meta's largest deals since WhatsApp, structured to avoid regulatory scrutiny. Founder-CEO Alexandr Wang joined Meta to lead its superintelligence division (remaining on Scale's board); Jason Droege became interim CEO. The deal triggered customer flight: Google (~US$150M in 2024 spend), OpenAI and Microsoft cut ties, with xAI exploring alternatives.
Accel led; Amazon and Meta participated. Doubled the Series E valuation after a 20% layoff year.
Doubled the Series D valuation in four months amid the 2021 AI funding surge.
Accel's first check — the beginning of its lead relationship.
Scale turned data labeling — the unglamorous work of teaching models what the world looks like — into a mission-critical business for every frontier lab, from autonomous vehicles to foundation models. Its customers included OpenAI, Google, Microsoft and Meta, making it the shared data layer beneath the competing model stacks.
A US$120K YC seed in 2016 became a US$13.8B Series F in 2024 and a ~US$29B valuation by 2025. The valuation ladder — US$1B (2019), US$3.5B (2020), US$7.3B (2021), US$13.8B (2024), ~US$29B (2025) — tracks the AI boom almost exactly.
Meta's June 2025 investment of ~US$14.3B for ~49% non-voting control was a strategic masterstroke and a commercial earthquake: it removed the conflict of one AI giant owning a stake in the shared data layer, and it sent Scale's other hyperscaler customers to competitors. Wang's departure to Meta — while retaining his board seat — blurred the line between Scale and its largest investor.
Scale insists it remains independent — Meta holds no voting power, and Droege's interim CEO appointment kept an operating executive in charge. But with its founder at Meta, ~49% of the company owned by Meta, and its former customers building in-house alternatives, the ~US$29B valuation is a bet on Scale's continued neutrality that the 2025 churn put in question.
Scale AI's lead investors are profiled in the Investor Atlas when their firms join the directory. Track who backs every frontier AI round on the Funding Leaderboard.
To stay under the regulatory line a controlling stake would cross: ~US$14.3B for ~49% with no voting power — one of Meta's largest deals since WhatsApp — engineered to avoid scrutiny. The structure also preserved the commercial fiction that mattered: Scale's customers were Meta's rivals. xAI's all-stock absorptions of X Corp and then SpaceX made control the point; here, avoiding it was the deal.
Because the shared data layer stopped looking neutral: Google alone was worth roughly US$150M of 2024 spend, and with Meta holding ~49% and founder Alexandr Wang moving to lead Meta's superintelligence division — board seat retained — every rival read the vendor as an arm of a competitor. That churn is why the ~US$29B mark is really a bet on neutrality the deal itself destabilized.
The boom's phases: US$1B (2019) on autonomous-vehicle labeling, US$3.5B (2020), US$7.3B (2021) in the surge, US$13.8B (2024) on model-era data demand — each step a distinct industry moment. The last step is the outlier: ~US$29B priced Meta's strategic need rather than the standalone market, about 33x Scale's ~US$870M of 2024 revenue, set by a buyer who wanted exclusivity more than returns.
No — since June 2025 he has led Meta's superintelligence division while keeping his Scale board seat, with Jason Droege serving as interim CEO. That triangle — founder employed by the largest shareholder, an operating executive in the top job, Meta at ~49% non-voting — is why the status line reads 'independent,' and why that single word now carries so much of the investment case.
By catching each wave's data problem in turn: autonomous vehicles (the 2019 unicorn round), then government and enterprise contracts, then frontier-model labeling — with Amazon and Meta joining the 2024 Series F syndicate as customer-investors. The 2025 irony closed the arc: those customer-investors became customers-lost, flipping the story from picks-and-shovels to contested neutrality.