PwC Puts a $31.6 Trillion Price Tag on AI's Infrastructure Buildout Through 2050

The numbers have gotten big enough that billions no longer do the job. PwC released its inaugural Global Data Centre Outlook on September 2, and the headline figure is $31.6 trillion in cumulative global data center capital expenditure through 2050. According to the PwC press release, annual data center capex is forecast to rise from roughly $800 billion in 2026 to $1.8 trillion per year by 2050.

That's the central scenario. The report, built on modeling commissioned from Oxford Economics and reviewed by Data Center Frontier, puts the plausible range at roughly $22 trillion on the low end to nearly $50 trillion if AI adoption accelerates faster than expected. The downside scenario is driven almost entirely by one variable: chip export controls that disrupt semiconductor supply chains could cut cumulative global investment by nearly $6 trillion from the central forecast.

But the chip risk isn't actually the most important finding here. PwC calls power "the binding constraint" in every region it modeled. According to the report as covered by TheStreet, affordable, reliable, and increasingly low-carbon electricity at scale is the hardest requirement for most markets to meet. Capital isn't scarce. Electrons are.

That constraint explains a lot of deal activity happening right now. The Aligned Data Centers acquisition, $40 billion, closed July 21 by a consortium comprising BlackRock's Global Infrastructure Partners, Abu Dhabi's MGX, and the AI Infrastructure Partnership, was explicitly about locking down a 51-campus, 6.4-gigawatt platform before someone else did, according to a closing statement reviewed by D CEO Magazine. The consortium committed an additional $5 billion post-close to fund continued expansion. That's not a financial bet on AI software. It's a bet on power-permitted land.

The PwC report adds context to why deals at that scale are getting done. It documents a structural shift in how data center economics work: unlike roads or fiber networks, AI infrastructure doesn't depreciate quietly. GPU and server stacks require replacement every four to six years, per the PwC analysis as covered by Data Center Frontier. That turns this into a recurring capex cycle, not a one-time buildout. The report describes it as a perpetual refresh obligation layered on top of construction costs.

The geography of that spending is fairly concentrated. According to the PwC press release, the United States is projected to capture 48 percent of global investment, or $15.1 trillion, anchored by its role in the advanced chip ecosystem and its concentration of model developers and hyperscalers. The Asia-Pacific region follows at $8.2 trillion. PwC's global infrastructure leader Clara Cutajar, quoted in the report's press materials, described AI infrastructure as one of the defining capital allocation challenges of the next generation, one that cuts across technology, energy, real estate, supply chains, regulation, and financing.

That framing matters because infrastructure investors have historically kept those categories separate. Power project finance teams don't usually model GPU refresh cycles. Data center REITs don't usually run utility interconnection risk. What PwC is describing is a capital category that requires all of them simultaneously, and that's before accounting for the sovereign AI push, which the report says would redistribute investment without reducing it overall.

The $31.6 trillion figure will circulate widely, and it should be read carefully. PwC is projecting a trajectory, not a commitment. The range is enormous, and the report is explicit that the outcome depends heavily on how fast grids can be built out, how chip trade policy evolves, and how quickly enterprise demand outside the hyperscalers materializes. What the report does pin down with some precision is the sequencing: power comes first. If a region can't supply reliable electricity at scale, the capital goes somewhere that can.

Sources cited:
- PwC Global Data Centre Outlook (press release) (https://www.pwc.com/gx/en/news-room/press-releases/2026/global-investment-in-ai-infrastructure.html)
- Data Center Frontier (https://www.datacenterfrontier.com/machine-learning/article/55403120/pwc-maps-316-trillion-ai-data-center-buildout-through-2050)
- TheStreet (https://www.thestreet.com/technology/pwc-ai-data-center-capex-forecast-power-constraint)
- D CEO Magazine (https://www.dmagazine.com/business-economy/2026/07/aip-mgx-and-blackrocks-gip-close-40-billion-acquisition-of-aligned-data-centers/)
- Aligned Data Centers (closing statement) (https://aligneddc.com/press-release/aip-mgx-and-blackrocks-gip-close-acquisition-of-aligned-data-centers/)