Entering the largest private financing in history! It is reported that Meta plans to invest billions of dollars in Scale AI to strengthen the AI data arms race

According to reports, Meta (META.US) is in negotiations to invest billions of dollars in Scale AI. This financing could be valued at over $10 billion, making it one of the largest private financing events in history. In 2024, Scale AI was valued at approximately $14 billion in a funding round that included participation from Meta.

Scale's CEO, Alexandr Wang, may not be as well-known as OpenAI's Sam Altman, but his company has become the absolute leader in the data sector among the three pillars of AI—chips, talent, and data. This startup provides data labeling services necessary for AI model training to tech companies like Meta and OpenAI through a large outsourced team and assists in developing customized AI applications. According to insiders, Scale is increasingly recruiting highly educated experts such as PhDs and nurses to participate in the development of complex models.

For Meta, deepening cooperation with Scale may help it keep pace with AI competitors like Google (GOOGL.US) and OpenAI, as well as establish closer ties with the U.S. government as it becomes more involved in defense technology. For Scale, collaborating with Meta will bring a powerful and financially strong ally. This will also be a significant "closed-loop moment" for Wang—shortly after launching Scale, Wang mentioned that a venture capitalist asked him when he decided to start a startup, to which he "naively replied that he was inspired by 'The Social Network'"—the film that tells the story of Facebook's founding.

Scale's Growing Status

Three months after DeepSeek released models comparable to the top U.S. technologies, the 28-year-old Wang proposed the establishment of a "national AI data reserve" and ensuring power supply to data centers during a congressional hearing, which received bipartisan support. Additionally, Scale has deepened its cooperation with the government through defense contracts, and its former executive Michael Kratsios has now become a core technology advisor to Trump.

In many ways, Scale's rise mirrors that of OpenAI. Both companies were founded about a decade ago and bet on the industry approaching what Wang calls the "AI turning point." The two CEOs are friends and briefly lived together, both skilled at networking and representing the AI industry before Congress. OpenAI has also received multi-billion dollar investments from large tech companies.

Scale's development trajectory has been influenced by the AI boom sparked by OpenAI, while also reacting to this trend. Initially, Scale focused more on labeling images of cars, traffic lights, and road signs to help train models for building autonomous vehicles. However, it later began to assist in annotating and managing the massive amounts of text data required to build the so-called large language models that support chatbots like ChatGPT. These models learn by extracting patterns from the data and their respective labels Despite facing allegations of psychological trauma related to overseas cheap labor (the U.S. Department of Labor has terminated the relevant investigation), Scale continues to evolve. In the context of tech companies attempting to synthesize data to reduce traditional labeling demands, its focus has shifted to specialized fields such as medical law, for example, enhancing AI's ability to handle differences in tax laws across countries.

To meet this demand, Scale is increasingly turning to hiring higher-paid graduate-level contractors to optimize its AI systems. These experts participate in a process known as reinforcement learning—where correct answers are rewarded and incorrect responses are penalized. According to an informed source, experts working with Scale are responsible for designing tricky questions (essentially tests) for the model to solve. The source stated that by early 2025, among the contributors involved in the model optimization process, 12% hold PhDs in fields such as molecular biology, and over 40% have master's degrees, law degrees, or MBAs in their respective fields.

Such bets have driven significant growth for the company. According to reports from April, Scale's revenue for 2024 is approximately $870 million, with expectations of reaching $2 billion this year. Informed sources indicate that following the rise of DeepSeek, Scale's demand for expert networks has increased as more companies invest in models that mimic human reasoning and execute more complex tasks