Scale AI: Here’s what Mark Zuckerberg’s $15B ‘superintelligence’ investment is

Amid reports that Mark Zuckerberg’s Meta is betting a $15 billion stake in Scale AI to supercharge its ‘superintelligence’ ambitions, many are wondering just what this data-labeling startup does and what made it worthy of the tech giant’s attention. 

As it happens, Meta has made plans to invest about $15 billion in Scale AI and hire its co-founder Alexandr Wang alongside key researchers, which would provide the company with a 49% stake in the artificial intelligence (AI) startup and value Scale at approximately $28 billion, per a recent report.

What is Scale AI?

Headquartered in San Francisco, Scale AI’s core business involves manually labeling the data used in training AI models and its mission is to “help deliver value from AI investments faster with better data by providing an end-to-end solution to manage the entire ML lifecycle.”

According to its website, Scale’s Generative AI Data Engine lends its hand to the most advanced large language models (LLMs) and generative models. It does this through world-class reinforcement learning from human feedback (RLHF), data generation, model evaluation, safety, and alignment.

Furthermore, the startup’s GenAI platform builds, tests, and optimizes generative AI applications to “unlock the value of your data.” At the same time, Scale Donovan allows its clients to customize, evaluate, and deploy specialized, mission-tailored AI agents “from concept to combat.”

Partnering with Scale AI is part of Zuckerberg’s efforts to build a ‘superintelligence’ lab that would outdo OpenAI, Anthropic, and Google. As a reminder, all of them have claimed to be developing products exceeding human intelligence, turning the AI industry into a highly competitive landscape.

Indeed, the establishment of this lab would represent an important milestone for Meta’s AI division, as it attempts to address the limitations of its underwhelming Llama 4 model. Notably, this new open-source AI platform has faced issues with core reasoning and coding benchmarks. 

Meanwhile, Meta has earlier this year announced a major breakthrough with ‘mind-reading’ AI. As the company said, the AI can decode and reconstruct sentences from the mind solely by reading brain signals, getting closer to realizing advanced machine intelligence (AMI).