Facebook owner Mark Zuckerberg's BIG step to develop AI models, to invest ... not in Open AI, or DeepSeek, but in...

Frustrated at Meta's slow and unsatisfactory progress in training and developing Artificial Intelligence (AI) models, CEO Mark Zuckerberg has decided to rope in talent from outside and spend billions of dollars to poach Scale AI CEO Alexander Wang. According to CNBC, the parent company of Facebook is in the final stage of the talks to seal a deal for investing $14 billion into Scale AI. Wang has not only established the most ambitious AI startup, he has built a reputation as a leader who understands AI’s technical complexities. He is also known as a person who knows how to build a business.

Why will Mark Zuckerberg not buy Scale AI?

Taking cues from other giants like Google and Microsoft, who have roped in outside talent to develop AI, Mark Zuckerberg decided not to acquire Scale AI. Instead, he has decided to buy significant stakes in the AI startup. Meta has taken this decision at a time when it is facing the Federal Trade Commission trial for antitrust claims. He does not want to worsen the case by sending signals that he is acquiring Scale AI. 

Meta's stake in Scale AI?

According to 'The New York Times', Meta will take a 49% stake in the data-labeling startup and Wang will help lead a new AI research lab at the social networking company. He will be joined by some of his colleagues. Wang and Scale AI made their reputation by helping major tech companies like OpenAI, Google and Microsoft in preparing data they use to train cutting-edge AI models. Scale AI has helped Meta earlier too.

What is value of Scale AI?

Established in 2016, Scale AI is valued at $14 billion and ranked 28 on CNBC’s Disruptor 50 list. Soon, it made its presence felt and hit the headlines in 2024 when it signed one of the biggest recent commercial leases in San Francisco. It bought 180,000 square feet of space in a downtown building that had been occupied by Airbnb.  It announced a multi-million-dollar deal with the US Department of Defence in March this year after collaborating with Meta on Defense Llama. It is a custom version of Meta’s open-source Llama foundation model designed to "support American national security missions".