A report published today by JLL, a provider of commercial real estate services, projects that data center capacity will nearly double from 103 GW to 200 GW by 2030, requiring as much as $3 trillion in total investment over the next five years.
The report also noted that achieving that goal will require approximately $870 billion in new debt financing to create real estate assets that will be valued at about $1.2 trillion.
Most of that growth will be driven by a sharp rise in artificial intelligence (AI) workloads, which are expected to account for half of all data center capacity by 2030, according to the report.
Daniel Thorpe, head of data center research for EMEA at JLL, said that the number one obstacle to achieving that goal will be ensuring there are sufficient energy resources available to make 200 GW of data center capacity available by 2030, representing a 14% compound annual growth rate (CAGR).
That issue will also lead to demarcation in where different classes of AI workloads will be deployed. Use cases involving training AI models will need to be deployed in locations where energy resources are less costly, while AI inference engines will be deployed closer to where applications are being accessed to reduce latency, noted Thorpe.
Average grid connection lead times are exceeding four years in primary markets. Due to utility interconnection delays and mounting pressure from rising grid electricity costs, the report notes some operators of data centers are also moving to directly fund their own energy generation, and several markets have implemented de facto “bring your own power” mandates, including Dublin and Texas.
Data centers are also adopting diverse regional energy strategies to address grid constraints. Natural gas, for example, is projected to play a major role in alleviating grid constraints in the U.S., while the four primary providers of cloud computing services are all investing in renewable energy projects. Battery energy storage systems (BESS) are gaining momentum, enabling cost-effective handling of short-duration outages and positioning the technology as a dynamic grid asset to speed up interconnection timelines.
Additionally, the report notes that solar-plus-storage will become a key component of global data center energy strategies by 2030, with renewable energy costs projected to outcompete fossil fuels across all major regions.
According to the report, the Americas will maintain its position as the largest data center region, representing about 50% of global capacity and achieving the fastest growth rate through 2030. The Asia-Pacific region is projected to expand from 32 GW to 57 GW, while Europe, the Middle East and Africa will add 13 GW of capacity.
Much of that pipeline (77%) has commitments from tenants, many of which actually own the data center itself, which suggests that there is no AI investment bubble that is about to burst, said Thorpe. Existing globe data center capacity consumption rates are already at 97%, the report finds. Funding, however, is also becoming more challenging to secure as investors become more wary of the return on data center investments as more capacity comes online, he noted.
Builders of data centers are now also preordering materials up to 24 months in advance, with more than half of projects experiencing construction delays of three months or more. The average equipment lead time globally is now 33 weeks. Builders of data centers are now focusing on more modular approaches to construction, with annual sales of modular systems and micro data centers projected to reach $48 billion by 2030, the report noted.
Exactly how much data center capacity will be required is hard to ascertain at this point, especially as the way applications are built becomes more efficient and investments are made in making the electrical grid more efficient. The one thing that is for certain is that there will soon be a lot more options available than any time before.
