Mark Zuckerberg is rapidly expanding Meta’s AI infrastructure with new data centers, a move aimed at catching up with rival companies in the generative AI race. The company has lagged behind in AI this year, prompting Zuck to recruit top talent and invest heavily in infrastructure.
Meta has recently made significant investments in AI, including a $14 billion deal to buy Scale AI and its leader Alex Wang. However, the company’s Llama 4 model was not up to par with rival offerings from DeepSeek, OpenAI, Anthropic, and Google. To address this, Zuckerberg is personally recruiting top AI talent with pay packages in the $100 million range.
To succeed in generative AI, Meta needs three key ingredients: high-quality data, infrastructure, and a talented team. The company’s new AI infrastructure will include multiple data centers that use over 1 gigawatt of power each, making them massive facilities.
In a surprising twist, some of these data centers are being built in tents, according to SemiAnalysis. This approach prioritizes speed above all else, allowing Meta to add capacity quickly and respond to the rapidly evolving AI landscape. While this method presents challenges, such as overheating issues during hot summer days, it is essential for Meta’s short-term goals.
Zuckerberg has confirmed some of these details in a Facebook post, stating that Meta will build several new AI data centers with multiple gigawatt capacity clusters, including the Prometheus and Hyperion clusters. The company aims to establish itself as a leader in superintelligence research and is committed to investing hundreds of billions of dollars into compute.
As the AI landscape continues to evolve, Meta’s aggressive expansion plan sets an ambitious tone for the industry. With its massive investments in infrastructure and talent, the company is well-positioned to stay ahead of the competition.
Source: https://www.businessinsider.com/meta-mark-zuckerberg-building-ai-data-centers-tents-catch-up-2025-7