Google to sever ties with Scale AI following Meta's significant stake
Google had intended to pay Scale AI approximately $200 million this year for human-labeled training data, crucial for developing sophisticated AI models like its ChatGPT competitor, Gemini. Last year, Google spent around $150 million on Scale AI's services, as per reports.
ADVERTISEMENT
Google, the largest customer of AI data-labeling startup Scale AI, is reportedly planning to cut ties with the company after rival Meta acquired a 49% stake. This move by Meta, which values Scale AI at $29 billion (up from $14 billion pre-deal), has sent ripples through the AI industry, raising concerns among other major tech companies about potential exposure of their research priorities and roadmaps to a competitor, as reported by Reuters.
As per reports, Google had intended to pay Scale AI approximately $200 million this year for human-labeled training data, crucial for developing sophisticated AI models like its ChatGPT competitor, Gemini. Last year, Google spent around $150 million on Scale AI's services.
The decision by Google to seek new data service providers comes as other prominent Scale AI customers, including Microsoft (MSFT.O) and Elon Musk's xAI, are also reportedly re-evaluating their partnerships. OpenAI, while spending less than Google, reportedly scaled back its engagement with Scale AI several months ago, though its CFO stated that the company will continue to work with Scale AI as one of its many data vendors.
Scale AI, founded in 2016, specializes in providing vast amounts of labeled data, which is essential for training advanced AI tools. In 2024, Scale AI reported $870 million in revenue. The bulk of its revenue is generated by charging generative AI model makers for access to a network of human trainers with specialized knowledge, some holding doctorate degrees, who annotate complex datasets for "post-training" AI models. These sophisticated annotations can cost as much as $100 per instance.
The departure of key customers like Google could significantly impact Scale AI's core business, which has been concentrated around a few major clients. As part of the Meta deal, Scale AI's CEO, Alexandr Wang, along with a few employees, is reportedly moving to Meta to lead its AI efforts.
This development presents a significant opportunity for Scale AI's competitors. Jonathan Siddharth, CEO of Turing, a rival company, noted that "The Meta-Scale deal marks a turning point. Leading AI labs are realizing neutrality is no longer optional, it's essential." Labelbox CEO Manu Sharma anticipates "hundreds of millions of new revenue" by the end of the year from customers seeking alternatives to Scale. Similarly, Garrett Lord, CEO of Handshake, a competitor specializing in a network of PhDs and experts, stated that their "demand has tripled overnight after the news."
Furthermore, some AI labs are now considering hiring in-house data-labelers to ensure data security. Brendan Foody, CEO of Mercor, a startup that recruits and vets candidates for in-house data labeling operations, suggests this trend will enable AI labs to scale their data labeling quickly while maintaining control over their proprietary information.
The Meta deal is expected to be a boon for Scale AI's investors, including Accel and Index Ventures, as well as its current and former employees. For Meta, this strategic investment appears to be a move to accelerate its AI ambitions and address perceptions that it may have fallen behind in the AI race, particularly after the performance of its initial set of Llama 4 large language models in April.
