Data-labelling start-up Surge AI is investigating raising up to $1 billion (£730m) in its first capital raising as it seeks to capitalise on investor interest in the sector after a massive investment in competitor Scale AI by Facebook parent Meta Platforms last month, Reuters reported.
Surge has hired advisors to conduct the investment round and is seeking a valuation of more than $15bn, but the reports source’s said that the talks are still in their early stages and the figure could be higher.
The funding would be a mix of primary and secondary capital that provides liquidity for Surge’s employees, the report said.
Funding round
Meta invested $14.3bn in Scale AI last month to acquire a 49 percent stake and hired former Scale chief Alexandr Wang to co-lead Meta’s AI efforts.
The deal valued Scale at more than $29bn, following a previous funding round in 2024 that valued it at $14bn.
The deal, which was designed to boost Meta’s AI plans in the face of competition from the likes of Google and OpenAI, highlighted the data-labelling sector and has also prompted an exodus of customers from Scale that has benefited Surge and other data-labelling firms.
Surge, Scale and other such companies provide nuanced labelling for AI training data that improves the quality of the resulting models.
Google, which was Scale’s largest customer, along with OpenAI and others, are reportedly planning to move away from Scale over concerns that working with it could reveal their research priorities to Meta.
Scale has said that its business remains strong and that it protects user data.
Human intelligence
Founded in 2020 by Edwin Chen, a former Google and Meta engineer, Surge has been bootstrapped by Chen and has not previously turned to capital markets for funding.
With customers including Google, OpenAI and Anthropic, it reportedly brought in more than $1bn in revenue last year, more than Scale’s $870m.
The large funding round is a test of investor interest in the data-labelling sector, which some feel will become obsolete due to its reliance on human skills that could be susceptible to automation.