Google plans to split from Scale AI after Meta's deal

Alphabet’s Google is preparing to end its long-standing relationship with data-labelling startup Scale AI after Meta Platforms acquired a 49 per cent stake in the company, Reutersreported.

The move marks a major shakeup in the AI ecosystem, with several top artificial intelligence (AI) labs now seeking alternative or in-house solutions to protect their proprietary data.

Meta’s investment in Scale AI values the startup at $29 billion—double its previous valuation from just a year ago. But the eye-catching deal has spurred immediate concern from competitors in the AI space, who fear their confidential data, research directions, and prototype tools could now be indirectly exposed to a rival.

Scale AI specialises in providing expertly labelled datasets which are essential for training generative AI models such as Google’s Gemini or OpenAI’s GPT series.

Google alone was slated to spend around $200 million on Scale’s services this year and had paid $150 million in 2024.

But insiders say Meta’s new stake and the appointment of Scale CEO Alexandr Wang to lead Meta’s superintelligence team was the tipping point for Google, Microsoft, Elon Musk’s xAI, and even OpenAI, all of whom are now scaling back or exiting their contracts with Scale.

Data privacy, competition fuel concerns

At the core of the backlash is a shared worry over data security and competitive leakage. Companies like Google and Microsoft rely on data-labelling partners to handle sensitive materials—including experimental products and training data—which may now be accessible to a firm like Meta through its close ties to Scale.

Though Meta has not taken a board seat at Scale AI and insists the two companies will continue to operate independently, AI labs are sceptical. Wang’s move to Meta, along with a few Scale employees, only intensifies fears that Scale can no longer maintain the neutrality necessary to work with competing AI developers.

Scale’s rivals gain ground

The fallout is rapidly benefiting Scale AI’s competitors. Firms like Labelbox, Turing, Handshake, and Mercor are seeing a surge in demand from AI labs eager to pivot away from Scale. Handshake, which specialises in a vetted network of PhDs and technical experts, reported its client workload tripled overnight after news of the Meta deal broke.

Labelbox, another major rival, is reportedly on track to gain “hundreds of millions” in new revenue this year as customers flee Scale. Several companies are also taking the in-house route, hiring their own data labellers to ensure tighter security and control.

Meta’s strategic move and Wang’s role

Meta’s acquisition appears driven not just by data but by talent. The deal gives Meta access to Alexandr Wang, the 28-year-old tech prodigy who co-founded Scale in 2016. Wang, who dropped out of MIT to build Scale, is now tasked with leading Meta’s “superintelligence” team—a move that underscores CEO Mark Zuckerberg’s push to catch up in the AI race.

Meta has faced delays in launching new AI models and has lost key staff in recent months. By bringing Wang onboard, the company is betting that a sharp business mind, rather than a research-heavy background, can restore its competitive edge—much like OpenAI's Sam Altman.

Scale AI will continue to operate independently and serve its government and enterprise clients, but with its largest customers heading for the exit, the company’s growth trajectory is uncertain. The deal is also likely to invite regulatory scrutiny, especially given Meta’s existing antitrust battles in the US.

(With inputs from agencies)