PwC projects approximately $31 trillion in global data center spending over the next 25 years—an average of $1.24 trillion annually, a figure that exceeds the GDP of most developed nations.
Dallas Dolen, PwC's Tech, Media and Telco Industry Leader, framed the scale: "We looked at about 25 years of potential future spend around data center, which we forecasted to be approximately $31 trillion."
The drivers are clear: AI compute demand and hyperscale cloud expansion. Training and running complex AI models requires specialized hardware—GPUs and high-bandwidth memory in particular. Chip makers including Nvidia, Intel and AMD will capture significant share of this spending, as will memory producers Samsung and SK Hynix.
But the buildout extends beyond silicon. Data center investment encompasses land, power generation, cooling systems and networking equipment. Microsoft, Amazon and Google are leading this infrastructure race, continuously expanding global footprints at a pace smaller competitors cannot match.
This capital intensity creates durable moats. The scale required to compete in cloud infrastructure—billions annually in capex—locks out new entrants and reinforces dominance among incumbents. Established players have access to cheaper capital, existing real estate portfolios and operational expertise that smaller firms cannot replicate.
Dolen indicated a dramatic slowdown is unlikely given sustained demand for digital services. But risks lurk. Energy consumption is mounting—a critical variable as power costs and grid capacity tighten. Supply chain bottlenecks for advanced chips could delay deployments. Rising interest rates would increase the cost of capital for long-duration infrastructure projects.
Profitability in cloud and semiconductors depends less on innovation than on the ability to deploy billions in capex faster than competitors and extract returns at scale.
