According to The Information, Meta ( Facebook's parent company) is planning to reform efforts to develop artificial intelligence (AI) for the fourth time in just six months. This move comes as CEO Mark Zuckerberg wants to accelerate the development of general artificial intelligence (AGI) to catch up with the fierce race in Silicon Valley.
Accordingly, the Superintelligence Labs research unit ( Meta's new AI division) will be divided into four small groups: a TBD lab; a product group focusing on Meta AI assistants; an infrastructure group; and a Basic AI Research lab (FAIR) in charge of long-term research. However, Meta has not yet made an official comment on this information.
Meta's continued restructuring comes with many risks, especially after some senior personnel left and the recent launch of the open source model Llama 4 was not as well received as expected.
However, Zuckerberg has affirmed his determination to invest heavily in AI, considering this a strategic step to open a new revenue line in the future.
In parallel with organizational changes, Meta is boosting infrastructure investment. Earlier this month, Reuters said the company had partnered with PIMCO Bank and asset management firm Blue Owl Capital to lead a $29 billion funding package to expand data centers in rural Louisiana.
In July, Mr. Zuckerberg revealed that Meta plans to spend hundreds of billions of dollars to build a series of large-scale AI data centers. In the latest financial report, the company also raised its capital expenditure forecast for this year by 2 billion USD, to 66 - 72 billion USD.
Meta said that the sharp increase in costs will mainly come from investing in data centers and paying high salaries to attract AI researchers, expected to make the cost growth rate in 2026 far exceed that of 2025.
With rapid expansion and ambitious vision, Meta is betting heavily on AI, but whether these risky steps will help the company reverse the situation in competition with OpenAI, Google or Microsoft is still an open question.