- Meta Strategic InvestmentUSD 14.3B — Amount: USD 14.3B; Post-money valuation: Over USD 29B (Scale official basis); Led by Meta Platforms | This is the most worth-studying deal in the AI industry because it simultaneously accomplished three things: ① Pricing (USD 29B)—Scale officially stated the transaction valued the company at over USD 29B. ② Acquiring talent (acquihire)—Founder Alexandr Wang joined Meta to lead Superintelligence Labs, but the company remained independent, with Wang retaining a board seat. Operational control was transferred to former Chief Strategy Officer Jason Droege (former Uber Eats founder, initially interim CEO). A USD 29B company by...
- SeriesFUSD 1B — Amount: USD 1B; Post-money valuation: USD 13.8B; Led by Accel | This round was a turning point for Scale AI, looking back full of drama. The valuation rose from USD 7.3B to USD 13.8B, and the investor list simultaneously included Amazon, Meta, and NVIDIA—interpreted at the time as a strong signal that 'the data layer gained industry-wide recognition.' However, just 13 months later, Meta bought a 49% stake for USD 14.3B, becoming the controlling shareholder. The distance between strategic investors and acquirers here is only one round apart. Officially described as a mix of primary and secondary transactions (partial cash-out of old shares). 2024 revenue approx. USD 870M. | Source: Scale AI official announcement / Sacra
- SeriesEUSD 320M — Amount: USD 325M; Post-money valuation: USD 7.3B; Led by Dragoneer Investment Group, Greenoaks Capital, Tiger Global (co-led). Valuation rose from USD 3.5B to USD 7.3B (doubled in 4 months). A typical product of peak growth stock valuations in the 2021 zero-interest-rate environment. In Nov 2021, acquired sensor data search company SiaSearch (amount undisclosed), integrating team and tech into Nucleus. By then, Scale had clearly positioned itself as a 'data-centric AI company,' covering end-to-end capabilities from annotation and automation to evaluation. | Source: Scale AI Official Announcement / Turing Post
- SeriesDUSD 160M — Amount: USD 155M; Post-money valuation: over USD 3.5B; Led by Tiger Global Management | Valuation rose from USD 1B to over USD 3.5B (3.5x in 15 months). Launched Nucleus in the same round—a dataset management and model testing product—and disclosed acquisition of Helia AI (real-time video and AI infrastructure company). This marks Scale's transition from "pure data service provider" to "data + software platform," which is also the basis for its valuation breaking away from labor outsourcing multiples. | Source: Scale AI official announcement / aiwiki
- SeriesCUSD 100M — Amount: USD 100M; Post-money Valuation: Over USD 1B; Led by: Founders Fund | First time crossing the USD 1B unicorn threshold. The Series C pitch deck for this round was later made public, becoming classic teaching material for Y Combinator—it records how Wang convinced investors during an era when 'data wasn't sexy'. The core argument was that AI bottlenecks would shift from algorithms to data, and data infrastructure would become the biggest opportunity. This judgment was fully validated after the generative AI explosion in 2022. | Source: Scale AI Official Announcement / Product Market Fit
- SeriesBUSD 18M — closed its Series B round; Amount: USD 18M; led by Index Ventures. As demand for labeled training data grew, the company focused on datasets for computer vision, autonomous driving, and machine learning applications. In the same year (2018), co-founder Lucy Guo left the company—public reports indicated disagreements on operational methods, though Guo retained equity (which later increased significantly in value as valuation soared). Source: Product Market Fit / Public Reports
- SeriesAUSD 5M — Amount: $4.5M; Led by Accel | Accel led the round and subsequently became Scale AI's most important institutional supporter—leading both the Series A in 2017 and Series F in 2024, spanning 7 years. During this period, the company began shifting from 'Human API' to autonomous driving and computer vision training data. | Source: Product Market Fit / jimmyresearch
- Y Combinator SeedUndisclosed — Amount: Undisclosed (YC standard terms); Undisclosed; Led by Y Combinator | The company was initially positioned as an 'API for human labor'—clients sent data extraction, classification, and other manual tasks via interface, with Scale handling contractor review and quality control before returning results. Alexandr Wang later recalled that early fundraising was extremely difficult: 'No VCs had trained models at the time, and data was very unsexy.' | Source: Scale AI official announcement / Product Market Fit
Physix Frontier