Google, Scale AI plan to split after Meta deal
SAN FRANCISCO, California— Alphabet's Google, the largest customer of Scale AI, plans to cut ties with Scale after news broke that rival Meta is taking a 49-percent stake in the artificial intelligence (AI) data-labeling startup, as reported by Reuters.
Google had intended to pay Scale AI approximately $200 million this year for human-labeled training data essential for technological development. This data is crucial for the sophisticated AI models that power Gemini, Google's ChatGPT competitor.
Shift in Business Dynamics
The search giant has already initiated discussions with several of Scale AI's competitors this week as it aims to transition a substantial portion of its workload away from Scale, according to inside sources. Meta's sizable investment in Scale AI, valuing the company at $29 billion, has prompted Google's move. This is a significant increase from Scale AI's previous valuation of $14 billion.
Scale AI, despite facing the loss of a major client like Google, plans to continue its operations. The company's CEO, Alexandr Wang, along with a select group of employees, will be moving to Meta as part of the deal. Given that Scale AI's business heavily relies on a few key clients, losing a major customer like Google could have substantial implications.
Concerns and Reactions
Other tech giants such as Microsoft are also distancing themselves from Scale AI. Elon Musk's xAI is reportedly looking to exit its relationship with Scale AI as well. OpenAI had already scaled back its involvement with Scale several months ago.
Companies in the AI sector are apprehensive about sharing their research priorities and roadmaps with a potential competitor like Meta through their engagement with Scale AI for data-labeling services. With Meta acquiring a significant stake in Scale AI, there are heightened concerns about sensitive business information being exposed.
Impact on Revenue and Operations
Scale AI generates a substantial portion of its revenue by providing access to a network of human trainers specialized in annotating complex datasets for generative AI model makers. While the company also caters to enterprises like self-driving car manufacturers and government agencies, its primary source of income stems from collaborations with generative AI model developers.
Recognizing the potential risks associated with the Meta-Scale deal, various AI labs are considering in-house data-labeling solutions to maintain data security. This shift in strategy is expected to expedite the scaling up of their data labeling operations.
The Meta deal is poised to benefit Scale AI's investors, including Accel and Index Ventures, as well as its current and former employees. Meanwhile, Meta is striving to dispel concerns about lagging behind in the AI arena following the underwhelming performance of its initial Llama 4 large language models released in April.
