Meta Seeks Data Centre Growth for AI Superintelligence Goals

Meta has reported strong quarterly earnings as it deepens its investment in the infrastructure needed to support large-scale AI systems. 

The social media giant posted revenue of US$47.5bn for the three months ending June 2025, up 22% year-on-year. Net profit climbed 36% to US$18.3bn.

While much of that success is underpinned by advertising, the company is accelerating data centre development and compute capacity in pursuit of what CEO Mark Zuckerberg calls “AI Superintelligence”.

Meta’s infrastructure spend is rising sharply, driven by demand for high-density compute and the need to host its expanding suite of large language models.

Infrastructure takes priority in AI expansion

Meta’s expenses rose 12% to US$27bn in the quarter, with infrastructure investment emerging as the dominant line item. This includes:

• Billions of dollars allocated to new servers, high-density racks and AI-optimised data centres
  
• Upgrades to facilities capable of supporting large-scale training workloads
  
• Investments in custom chips and hybrid cooling to power performance at scale

"We're building AI that can help with everything from scientific breakthroughs to personal assistance," says Mark. 

The scale of this ambition is driving new demand for hyperscale-ready facilities.

Strategic acquisitions and talent reshaping Meta’s AI unit

Alongside infrastructure, Meta is also investing in talent and acquisitions:​​​​

The aim is to position Meta as a credible contender in foundational AI, after criticism of the Llama 4 model prompted a more intensive R&D push.

Dual focus: enterprise-grade AI and personal assistants

In a video posted before the earnings release, Zuckerberg outlined two goals for Meta’s AI roadmap:

• AI Superintelligence: A general-purpose platform for global-scale problems
  
• Personal Superintelligence: AI agents embedded into everyday tasks and applications

This strategy is a major shift from Meta’s earlier AI work, which primarily supported content curation and ad targeting.

Realising these ambitions will require robust infrastructure, with a focus on performance, energy efficiency and operational scale.

Data centres support Meta's future-facing bets

With 3.4 billion daily active users across its platforms, Meta has the scale to deploy AI across massive infrastructure environments. 

This user base generates both data and revenue that can be reinvested into compute facilities.

• Meta is using its social media cash flow to build and upgrade data centre assets
  
• The company is increasing support for high-density GPU clusters used in model training
  
• Hybrid cloud infrastructure is being designed to support internal and external workloads

Zuckerberg’s strategy relies on this cycle: ad revenue funds AI investment, which in turn enhances services and user engagement, boosting revenue further.

Analysts divided on cost versus future growth

The market has reacted positively to Meta’s earnings.

Shares jumped more than 10% in extended trading, with many investors encouraged by the firm’s ability to balance profitability with long-term investment.

"AI-driven investments into Meta's advertising business continue to pay off, bolstering its revenue as the company pours billions of dollars into AI ambitions like superintelligence," says Minda Smiley, Senior Analyst at Emarketer.

Still, some remain cautious about the long-term viability of such high expenditure.

Mike Proulx, VP and CMO at Forrester, believes Meta is “future-proofing itself as a growth company” amid changing social media trends and competition from rivals like TikTok and X.

Meta’s approach demonstrates how large digital firms are using their cash reserves to invest in next-generation infrastructure, laying the foundation for future AI platforms.

"Meta's exorbitant spending on its AI visions will continue to draw questions and scrutiny from investors who are eager to see returns," Minda adds.