Meta to Invest Hundreds of Billions on AI Data Centres

Mark Zuckerberg, the chief executive of Meta Platforms, has revealed plans to invest “hundreds of billions of dollars” in the construction of data centres that will power artificial intelligence capabilities. The company’s first “multi-gigawatt” data centre, named Prometheus, is expected to go live next year. Another facility, Hyperion, is designed to eventually scale up to 5 gigawatts over several years. Both projects represent Meta’s commitment to establishing a leading position in AI infrastructure development.

According to Zuckerberg, Prometheus will span an area comparable to significant parts of Manhattan. Writing on Threads, Meta’s social media platform, he announced that the company is also constructing multiple titan clusters – supercomputing centres capable of handling vast computational tasks. SemiAnalysis, an industry publication, has indicated Meta could become the first AI lab to launch a supercluster with a power capacity exceeding one gigawatt.

Meta is heavily investing in talent acquisition to accelerate its AI ambitions. Zuckerberg noted that the firm is working to assemble the “most elite and talent-dense team in the industry.” This strategy involves offering lucrative multimillion-dollar pay packages to top AI engineers. Reports have surfaced that Meta has extended lucrative offers to staff currently employed by OpenAI, further intensifying competition in the sector.

These announcements come as the industry faces growing demand for regulatory reforms to facilitate accelerated project development. President Trump recently attended a summit with leading tech executives to discuss measures to streamline connections to the power grid and provide federal land for building data centres that support AI advancements. Meta’s massive investment strategy is another signal of how central AI is becoming to the business strategies of major technology players.

While Meta Platforms possesses substantial financial resources, with a market value of $1.8 trillion, the company faces increasing pressure from investors over the high costs of AI infrastructure projects. Earlier this year, the DeepSeek episode, wherein advancements by a Chinese AI start-up caused turmoil, negatively impacted the stock valuations of Meta, Amazon, and the chipmaker Nvidia, sparking questions about the long-term payouts of major AI investments.

Meta’s history as an early leader in open-source AI has not shielded it from these competitive pressures. The performance of its Llama 4 large language models, released earlier this year, fell short of expectations. Additionally, delays in rolling out its flagship AI model, Behemoth, have compounded scepticism about the company’s ability to compete effectively in the AI race.