Google Plans To Cut Ties With Scale AI After Meta Acquires Major Stake

Google is planning to sever its relationship with Scale AI, the data-labeling startup that has been one of its largest AI partners, following news that Meta has acquired a 49% stake in the company. The move, confirmed by five sources familiar with the matter, could unravel up to $200 million in planned spending and reshape a critical piece of the AI training pipeline.

Google was previously Scale AI’s largest customer, relying on its network of human annotators to label the high-quality training data used to develop large language models, including those behind Google’s Gemini. But Meta’s entry has triggered concerns that working with Scale could now expose Google’s research priorities or intellectual property to a rival.

The search giant is already in talks with Scale AI competitors and is expected to reallocate much of its data-labeling workload to other vendors, sources said.

Meta's Entry Raises Security and Trust Concerns

Meta’s investment values Scale AI at $29 billion, up from $14 billion prior to the deal. As part of the agreement, Scale’s CEO Alexandr Wang will take a senior position leading Meta’s AI work, while a small group of employees will also move to Meta.

That shift has raised alarm among AI companies that compete with Meta. Data-labeling contracts often require companies to share sensitive training materials, early prototypes, or even strategic roadmaps with vendors. With Meta now a major stakeholder in Scale, those companies fear potential exposure of competitive data—even unintentionally.

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

Scale Faces Broad Retraction from AI Labs

Google’s departure is just one piece of a larger exodus. Other top customers, including Microsoft and Elon Musk’s xAI, are also pulling back. OpenAI reportedly scaled down its usage months ago, though the company said last week it would continue working with Scale as one of several vendors.

Scale’s business is heavily concentrated. The company generated $870 million in revenue in 2024, and Google alone spent roughly $150 million last year. That reliance on a few major clients makes the firm especially vulnerable to customer churn—especially when competitors are already circling.

Labelbox CEO Manu Sharma told Reuters that his company may “generate hundreds of millions in new revenue” by year’s end as AI labs seek alternatives. Handshake, another rival, saw demand triple after news of the Meta deal broke.

Some labs are now moving to build in-house labeling teams, preferring to retain tighter control over sensitive data. Brendan Foody, CEO of Mercor, said his company is helping labs rapidly scale internal operations to reduce reliance on external vendors altogether.

Human Labeling Still Crucial in an AI-First Market

Despite advances in automated model training, the final stages of developing high-quality generative AI still depend heavily on expert human input. Scale AI’s value has long come from its access to trained annotators—including historians, scientists, and PhDs—who can craft nuanced examples for post-training and reinforcement learning.

These human annotations can cost as much as $100 each, and demand has surged alongside interest in powerful language models. Beyond tech firms, Scale also provides data services to clients in sectors like autonomous vehicles and the U.S. government—business that is likely to remain stable, according to sources.

What This Means

The Meta-Scale deal has triggered a reckoning in the AI vendor landscape. As companies race to develop the next generation of models, trust in infrastructure and supply chains is becoming just as important as the models themselves.

For AI labs, neutrality in data-labeling relationships is no longer a luxury—it’s a strategic necessity. And for Scale AI, Meta’s backing may bring long-term capital and leadership, but in the short term, it’s reshaping the company's customer base and market position.

As the lines between partners and competitors blur, AI companies are being forced to rethink not just what they build, but who they trust to help build it.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.