Google is preparing to cut ties with Scale AI, the artificial intelligence data labeling company that has been its biggest customer to date, right after it was revealed that Meta had acquired a 49 percent stake. According to five different sources, Google was preparing to give Scale AI around $200 million worth of work this year. However, Meta’s move seems to have seriously shaken the balance in the sector.
Google parts ways with Scale AI after Meta’s stake
Google started talks with alternative data labeling companies that could replace Scale AI this week after this major partnership with Meta. The company spent $150 million on Scale AI services last year.
Scale AI, which generated a total of $870 million in revenue in 2024, was getting most of its revenue from large language model developers like Google. Now it seems likely that these sources will start to dry up rapidly.
Meta’s partnership with the company with a valuation of $29 billion has mobilized not only Google, but also other large customers such as Microsoft, xAI and OpenAI. According to sources, these companies are also in the process of severing their ties with Scale AI. Although OpenAI has stated that it has not completely ended the relationship by saying “we work with multiple data providers,” it has significantly reduced its business volume in recent months.
Why did it get so much backlash?
Scale AI provides complex human-labeled data required for training artificial intelligence models. In this process, companies’ prototype systems, proprietary data, and strategic plans are shared with Scale AI for labeling. Meta’s 49 percent stake brings with it the risk of gaining insights into the technical infrastructure and roadmaps of its direct competitors.
Companies competing with Meta, such as Google and Microsoft, in particular, argue that the principle of neutrality is no longer a luxury, but a necessity. The fact that Scale AI CEO Alexandr Wang will move to Meta and start managing the company’s artificial intelligence unit also reinforces these concerns.
Competitors who want to fill the gap between companies that have parted ways with Scale AI are quickly turning the process into an advantage. While one of its competitors, Labelbox, announced that it expects “hundreds of millions of dollars in new revenue” by the end of the year, expert network-focused companies such as Handshake say they are experiencing an explosion of orders. “Our demand tripled overnight after the news,” said the CEO of Handshake.
Another alternative, Mercor, is also competing directly with Scale AI, and is developing automation technologies that will allow companies to build their own labeling teams, keeping data security completely in-house.
