Meta plans to invest $15 billion in Scale AI to expand its AI infrastructure footprint

Meta Platforms plans to invest approximately $15 billion to acquire approximately 49% of the shares of Scale AI, an artificial intelligence training platform. The transaction is a non-controlling investment and is expected to promote long-term cooperation between the two parties in data annotation, model training and algorithm infrastructure.

According to multiple media reports, the transaction has entered the final stage. After the transaction is completed, Alexandr Wang, founder of Scale AI, will join Meta to lead the newly established "Super Intelligence Laboratory", which will focus on promoting the research and development of large language models and general artificial intelligence (AGI).

Scale AI is a company that mainly provides high-quality data annotation services, and its customers include OpenAI, Microsoft, Google and US government agencies. The company has industry advantages in data annotation efficiency, automated processing and security compliance, and is widely used in AI model training and precision optimization.

Currently, Meta is continuing to increase its investment in artificial intelligence technology. This investment is seen as an important supplement to its AI training. Previously, the company has launched the Llama series of open source large models and developed a new generation of models called "Behemoth" internally, intending to improve the model's reasoning ability and performance level.

Scale AI's latest round of financing is valued at approximately $14 billion. The company's revenue in 2023 is approximately $870 million, and is expected to grow further in 2025. According to the transaction structure, Meta's investment in this round will not change Scale's independent operating status, but the two parties will establish a deeper data and technology synergy mechanism.

This transaction is also Meta's largest AI-related investment in recent years, which complements its investment in multiple technical routes such as the metaverse, AI hardware and language models. Analysts believe that the addition of Scale will help Meta obtain a more stable data foundation and a more efficient development process in the training and deployment of large models.

At present, there is a clear trend of cooperation and competition among artificial intelligence technology companies. Microsoft and OpenAI, Google and Anthropic, Amazon and Stability AI have established multi-level cooperative relationships. The cooperation between Meta and Scale is expected to trigger a new round of adjustments in the layout of the infrastructure level within the industry.