Google plans to back out of major AI deal

Google, the largest client of data‐labeling firm Scale AI, is planning to end its contract, according to Reuters. This shakeup came after Meta announced it would acquire a 49% stake in the startup.

Google planned to spend about $200 million on Scale this year for human-labeled training data, which is needed to train its Gemini AI model.

Reuters‘ sources familiar with the matter said Google has already begun talks with other companies this week, aiming to move much of its workload away from Scale AI. The change comes just as Meta’s investment lifts Scale’s value from roughly $14 billion to $29 billion.

Despite the upheaval, Scale AI plans to carry on. Its CEO, Alexandr Wang, and a few employees will move to Meta, while everyone else stays. But losing Google, one of its biggest clients, could seriously cut its revenue.

A spokesperson for Scale AI said that its relationships with large corporations and government agencies are solid and that it remains dedicated to safeguarding client information. They declined to address Google’s decision directly.

In 2024, Scale AI brought in $870 million in revenue. Sources say Google accounted for about $150 million of that last year.

Other leading tech firms are also cutting back

One source said Microsoft is rethinking its work with Scale, and Elon Musk’s xAI plans to leave. OpenAI reduced its use of Scale several months ago, though its finance chief recently said the company still uses Scale as one of multiple data vendors.

Rival AI developers worry that sending their data to Scale could reveal strategic plans. When companies contract Scale, they often hand over prototype tools and private datasets for labeling. With Meta now owning nearly half the business, competitors fear their research road maps and technical details could be exposed to a chief rival.

Most of Scale AI’s revenue comes from services for generative AI firms. It connects these companies with experts such as PhD-level scientists, who work with datasets to “post‐train” models.

As AI systems have grown more capable, the need for such detailed human input has soared. Scale also labels data for self-driving car makers and U.S. government agencies, clients expected to stay. Yet its largest income still comes from teaming up with generative AI developers.

Rivals gain demand as AI Firms push for neutrality after Meta deal

Sources say Google has been diversifying for over a year. Meta’s recent investment has prompted Google to switch all key agreements away from Scale. Thanks to built-in clauses in many contracts, the shift could occur rapidly.

This transition opens doors for Scale’s rivals. “The Meta-Scale deal marks a turning point,” said Jonathan Siddharth, CEO of Turing. “Leading AI labs are realizing neutrality is no longer optional, it’s essential.”

Meanwhile, some AI labs are choosing to hire in-house data-labelers to keep their information secure. “Many AI labs now want to hire in-house data-labelers, which allows their data to remain secure,” said Brendan Foody, CEO of Mercor, a startup that also offers automated tools to recruit and vet labeling experts quickly.

Cryptopolitan Academy: Want to grow your money in 2025? Learn how to do it with DeFi in our upcoming webclass. Save Your Spot