JLL: Global Data Center Capacity To Double By 2030 In $3T 'Supercycle'

The continued surge in digital infrastructure investment is projected to double global data center capacity with nearly 100 gigawatts of new supply by the end of the decade, according to a JLL report published this week.

JLL’s baseline projections, in which global data center capacity would reach 200 GW within 6 years, would represent a 14% compound annual growth rate for the data center sector.

That new capacity equates to $1.2T in real estate asset value creation, with spending on chips and other information technology equipment hosted in these facilities bringing total digital infrastructure investment to $3T. 

Roughly $870B of new debt financing would be required to support this build-out, according to the report, which calls the expected spending surge an investment “supercycle” — the largest in modern history. 

“The underlying fundamentals remain quite strong,” JLL Global Head of Data Center Research Andrew Batson told Bisnow. “How we use technology and data centers and how we power them, that's all up for debate over the next few years. But generally we see continued growth in the sector, we see strong investor appetite … and there's plenty of money to finance it.”

The Americas, particularly the U.S., will remain the largest data center region by a significant margin, according to JLL. The region accounts for roughly half of the world’s data center capacity, and it is projected to add inventory at the fastest clip through the end of the decade with a 17% annual growth rate.

Close to 90% of that capacity is expected to be in the U.S. 

Hyperscalers — tech giants like Amazon, Microsoft, Google and Meta — will continue to account for the largest share of demand, according to the report. While this capacity will be used both for artificial intelligence and traditional cloud workloads, AI’s share is climbing rapidly.

Today, AI accounts for one-quarter of all data center workloads. By 2030, according to JLL's projections, it will be half. 

Although hyperscalers will continue to pursue strategies that include self-developed data centers and leasing capacity from data center providers, the majority of the 97 GW of new capacity expected by the end of the decade will be built by these third-party developers, according to JLL. The leased data center segment, which includes colocation and build-to-suit assets, is expected to add 62 GW by 2030. 

Still, hyperscalers could build around 41 GW of owner-occupied capacity during that same period, which would more than double the size of that segment to reach 70 GW. 

Beyond surging data center growth and investment, the report also anticipates inference overtaking training as the dominant AI requirement as soon as 2027.

While this projection assumes accelerated creation and adoption of AI applications well beyond what exists today, the shift would mark an inflection point for the data center sector. 

As inference accounts for a greater share of demand, development is going to become increasingly dispersed, spreading toward population centers to reduce latency and serve users directly. According to Batson, secondary and tertiary markets will account for “a growing slice of the pie,” although not to the detriment of the industry’s traditional core markets. 

“Inference is going to take up a larger and larger percentage of data center workloads over the next decade,” Batson said. “There are significant benefits to that being close to population centers where the demand is being derived from.”

The report’s authors acknowledged that JLL's projection of a 14% compound annual growth rate for this nascent and unpredictable sector is contingent upon several assumptions that could be undermined by evolving technology or economic forces.

This optimistic outlook assumes innovations will successfully mitigate the energy constraints that have previously limited growth and that demand will keep up with supply, thereby maintaining a global vacancy rate below 10%.

But it is far from certain that these assumptions will hold true.

Concerns exist about a potential “AI bubble,” in which revenue generated from AI products and services may fail to justify the massive scale of infrastructure investment or the high valuations assigned to various AI-focused companies.

Given this inherent unpredictability, JLL also published a downside scenario, projecting a 7% CAGR through 2030. This could be triggered by a substantial reduction in AI investment or by persistent supply-side issues, such as continuing energy and supply chain challenges.

Still, while 7% annual growth would represent a slowdown for the sector, it would be considered robust growth for most other asset classes and would outpace the recent expansion of the commercial real estate sector as a whole, according to most analyses. 

Additionally, the report’s authors remain skeptical that the ongoing wave of investment represents a bubble for the data center sector. They highlighted the fact that the supply of data center capacity trails well behind the demand that already exists, with a 97% data center occupancy rate at the end of 2025 and 77% of capacity under construction already committed to tenants. 

“Property metrics do not point to a bubble,” the authors wrote. “These metrics are hardly a sign of froth or overbuilding. Additionally, power constraints and long project lead times will help keep the sector in balance.”

Conversely, an upside scenario outlined in the report projects a 20% annual growth rate. This accelerated growth assumes AI acts as a potent additive driver through the rapid adoption of autonomous driving, robotics and other AI-powered emerging technologies.