Meta is assembling a team to develop a «super intelligence». Will it become the leader of the AI race?

Meta Platforms CEO Mark Zuckerberg has begun personally assembling a new team of experts to create what he calls general artificial intelligence (AGI). Zuckerberg wants Meta to be the first to develop an AI capable of performing a wide range of tasks at a level comparable to humans, write Bloomberg.

Zuckerberg is personally overseeing the selection of specialists for the new team, which Meta informally calls the Superintelligence group, sources told Bloomberg. It plans to hire about 50 people in total. According to the agency's interlocutors, Zuckerberg has created a group chat room on WhatsApp called Recruiting Party with other top managers where candidates for the team are discussed around the clock, and also ordered the rearrangement of tables at Meta's headquarters so that members of the Superintelligence group sit next to him.  

The group's goal, according to the agency, is to get ahead of the competition in creating Artificial General Intelligence (AGI), a level of AI at which machines will be able to perform a wide range of tasks on par with humans. The company plans to build AGI technologies not only into Meta's social networks and other platforms, but also into a broader line of AI products - from chatbots Meta AI to Ray-Ban glasses.

Meta is also preparing for a multi-billion dollar investment in Scale AI, a company that provides data for training language models and develops custom AI solutions for businesses and government agencies, Bloomberg reported. Scale AI founder Alexander Wang is expected to join the Superintelligence group after the deal closes. This will be Meta's largest outside investment in its history, the agency noted. Scale AI's valuation will be about $28 billion, including funds raised, wrote The Information.

Zuckerberg has openly stated in recent months that artificial intelligence is Meta's priority. His desire to personally oversee the recruitment of staff for the new project is due, among other things, to disappointment with the quality of the Llama 4 language model released in April, which turned out to be worse than Zuckerberg's expectations, Bloomberg writes. That AI was criticized both internally and by outside developers who felt the model's capabilities were overrated, the agency claims.

Subsequently, Meta postponed the launch of its largest model at the time, called Behemoth, despite public claims of its superiority over solutions from OpenAI, Google, and Anthropic. According to data from The Wall Street Journal, Meta executives concluded that the model's actual progress compared to previous versions was too insignificant.

Therefore, Zuckerberg decided to get personally involved in the process of creating the Superintelligence group. He expects that the new team will significantly improve the quality of Llama models, as well as advance the development of AI tools, especially for voice and personalized functions.

According to the agency's sources, Zuckerberg has held lunches and dinners at his homes in California in recent months, where he personally persuaded AI researchers and engineers to join his project. He reasoned that unlike competitors dependent on funding rounds, Meta has steady advertising revenue and is able to fund the team with its own money. He emphasized that Meta already has enough money to build a multi-gigawatt data center, one of the most powerful in the world.

Meta's strategic investments in long-term growth - especially in artificial intelligence - as well as its dominant position in the social networking market are key drivers of growth in the company's stock, JPMorgan analysts wrote on June 5. They raised their target price on Meta shares to $735 from $675, up 4.6 percent from the closing price on June 10, and reiterated a «buy» recommendation on the stock, wrote GuruFocus.

Although Meta may spend up to $65 billion on AI infrastructure this year, its gross margin remains one of the highest in the industry at 82% and revenue has grown nearly 20% over the past 12 months, JPMorgan noted. The investment bank forecasts steady growth in revenue and earnings per share through at least 2026. The bank stressed that Meta's focus on AI, including the development of generative models and the use of AI in targeted advertising, should support future monetization and growth in user engagement. AI is the main growth driver for Meta.

«We believe Meta's efforts to build a more powerful AI infrastructure and develop large language models (LLMs) to incorporate AI into products create excellent long-term monetization opportunities at lower costs,» Phillip Capital analyst Serena Lim wrote in a recent research note that cites Barron's.