June 14, 2025— Google is set to sever ties with data-labeling firm Scale AI following news that Meta is acquiring a significant 49% stake in the company, according to five sources familiar with the matter. The move could trigger a ripple effect across the artificial intelligence (AI) industry as other key players—including Microsoft, OpenAI, and Elon Musk’s xAI—also begin distancing themselves from the once-dominant AI infrastructure startup.
The decision marks a dramatic shift for Scale AI, which had built its business on serving some of the world’s most influential AI developers. Alphabet’s Google, previously Scale AI’s largest customer, had been on track to spend approximately $200 million with the company in 2025 alone, one source said. In 2024, Google reportedly spent around $150 million on Scale’s services, which are crucial for training sophisticated AI models like Gemini—Google’s answer to OpenAI’s ChatGPT.
Sources say that in response to Meta’s strategic investment, Google held discussions this week with alternative data-labeling firms in an effort to offload its contracts and minimize exposure to Scale. The concern among Google and other companies is less about the quality of Scale’s services and more about the competitive risks associated with sharing sensitive data through a firm now partially owned by one of their chief AI rivals.
“The moment Meta entered the picture, alarm bells rang across the industry,” one person close to the matter said. “These companies are sharing their crown jewels—research roadmaps, experimental data, even prototype products. Nobody wants Meta to have visibility into that.”
Meta’s Strategic Move
Meta’s acquisition of a nearly half ownership in Scale AI values the San Francisco-based company at a staggering $29 billion, more than doubling its previous valuation of $14 billion. While the financial details of the deal remain confidential, it reportedly includes a transfer of key leadership: Scale AI’s CEO Alexandr Wang and several core employees are set to join Meta as part of the arrangement.
The move has sparked widespread concerns about neutrality in the data-labeling ecosystem. Scale’s human-in-the-loop annotation services are vital for fine-tuning large language models (LLMs)—providing human-generated responses, classifications, and corrections for model outputs. These services are especially valuable for creating guardrails and improving the contextual understanding of increasingly advanced AI systems.
A spokesperson for Scale AI insisted that the business remains robust, emphasizing the company’s “commitment to customer data protection” and pointing to existing partnerships with enterprises and governments. The company declined to comment on its relationship with Google.
Competitors Reevaluate Relationships
Sources confirmed that Microsoft, another major Scale AI customer, is actively preparing to transition its contracts elsewhere. Meanwhile, OpenAI—which reportedly spends far less than Google—began pulling back from Scale months ago, according to people with knowledge of the decision. Although OpenAI’s CFO publicly affirmed that the company will continue working with Scale as “one of many data vendors,” sources say the relationship has cooled significantly.
xAI, Elon Musk’s AI company, is also planning to wind down its engagement with Scale, one source said. None of the companies—including Google, Microsoft, and OpenAI—responded to Reuters’ requests for official comment.
The exodus of top-tier customers presents a significant challenge for Scale AI. Much of the firm’s revenue comes from providing generative AI model developers with access to a network of highly skilled human annotators. These include domain experts such as historians, scientists, and PhDs, some of whom can charge up to $100 per annotation.
While Scale AI also performs data-labeling services for autonomous vehicle companies and U.S. government agencies—contracts expected to remain intact—its biggest revenue drivers are tied to generative AI. Losing Google and others could significantly impact its business trajectory.
In 2024, Scale reported revenue of $870 million, a large portion of which came from its generative AI customer base.
A Turning Point in AI Infrastructure
The situation has opened the door for rival data-labeling startups to court displaced customers. With Google actively looking for alternatives and other giants close behind, a reshaping of the AI training ecosystem appears imminent.
The incident also reflects broader trends in the AI industry, where concerns about data privacy, competitive secrecy, and model integrity are becoming increasingly central to vendor decisions.
For its part, Google has been exploring ways to diversify its data-labeling providers for over a year, sources said. Meta’s surprise move this week seems to have accelerated that process, prompting Google to cancel or reassign key contracts rapidly.
“Neutrality is non-negotiable in this space,” one senior AI executive told Reuters. “The moment a vendor looks like it might compromise that, the whole system starts to shift.”
As Meta continues its aggressive push into the AI sector, its move to secure Scale AI as a strategic partner—and partial acquisition target—signals the intensifying competition among tech giants not just to build smarter models but to control the infrastructure behind them.
In the wake of Meta’s investment, Scale’s future may hinge on how effectively it can replace departing clients and prove its ongoing independence—despite the changing face of its ownership.
