The AI-driven scramble for data center space is putting unprecedented strain on North America’s colocation market, driving vacancy rates to near zero, according to a new report by JLL.
Vacancy across the market has declined to a record low of 2.3% and is expected to remain tight for the next few years as the construction pipeline of 8 GW is 73% preleased, the brokerage’s North America Data Center Report – Midyear 2025 report found.
The combination of AI adoption, cloud migration, and digital transformation has formed a “perfect storm” of data center demand, leading to a massive supply crunch that is expected to persist for a while. As a result, markets like Dallas-Forth Worth are seeing unprecedented competition for scarce capacity and major cloud providers are locking down power reservations years in advance.
North America is poised to see up to $1 trillion in data center development through 2030, according to the report. JLL projects that more than 100 GW of colocation and hyperscale capacity could break ground or deliver in that time, not accounting for the potential of quantum computing to further accelerate the sector.
“The colocation market is experiencing unprecedented demand pressure under an increasingly stressful environment,” said Andy Cvengros, JLL’s executive managing director and co-lead of U.S. data center markets in a news release. “The first half of the year was riddled with disruptions, including the DeepSeek news at the beginning of the year and the potential impact of tariffs on demand and construction. Despite the turbulence, the sector posted another record-shattering performance.”
Adding to the pressure on companies trying to expand their data center operations, commercial electricity rates have jumped by nearly 30% in the past five years and the average wait time for a grid connection across the U.S. has increased to four years. Three-quarters of development activity is concentrated in markets with low electricity costs, according to JLL.
A study by McKinsey & Company found that demand for AI-ready data center capacity will grow at an average rate of 33% per year between 2023 and 2030 in a midrange scenario, meaning that by 2030, AI will account for about 70% of total demand. Generative AI specifically (think ChatGPT) is expected to account for about 40% of the total.
Running advanced AI workloads requires enormous amounts of computational power and energy consumption. AI training demands advanced computational resources, extensive storage capacity, and high-speed interconnectivity, whereas “inference” – applying the trained model – typically places less strain on computational resources but requires low latency and scalabilty.