4-model average
building category history
loading
60pt spread
The three supporting readings tell you how much weight to put on the probability: confidence reflects category-level track record, stability tracks how the estimate has moved over time, models shows whether the four agree.
Loading belief trajectory...
As of late 2023, AI-dedicated data centers are experiencing rising demand driven by increased adoption of artificial intelligence technologies. However, vacancy rates have been historically low (averaging around 5-6% over the past few years) due to limited supply and high demand. If the trend continues, JLL and CBRE could potentially report vacancy rates above 8% if new supply fails to keep pace with rapid demand increases.
The demand for AI-dedicated data centers is projected to grow exponentially, but supply is also increasing rapidly to meet this demand. While vacancy rates might rise from current lows as new capacity comes online, exceeding 8% for JLL or CBRE in any 2026 quarter requires a significant imbalance between an even faster supply ramp-up or a plateau in AI demand, which is less likely given current industry forecasts. However, the rapid pace of new construction and potential for project delays or cancellations introduce a notable risk of higher vacancy.
AI-dedicated data center vacancy rates have remained historically tight, with major markets averaging 2-5% vacancy as of late 2024, driven by unprecedented demand from hyperscalers and AI infrastructure buildout. However, significant supply is coming online in 2025-2026 across major markets (Dublin, Northern Virginia, Singapore), and JLL/CBRE data typically tracks broader colocation/hyperscale segments where competition is intensifying. An 8%+ vacancy threshold represents a meaningful softening from current conditions but remains plausible given: (1) supply growth outpacing demand in certain markets by 2026, (2) economic slowdown reducing enterprise AI capex, and (3) normalization from the current supply crunch. The probability reflects the high likelihood that at least one company's quarterly report will capture elevated vacancy in a specific market segment during 2026, though nationwide dedicated AI data center vacancy likely stays below 8%.
Current hyperscale absorption in primary US markets reached 650 MW in Q2 2025 with vacancy already at 3.2% per CBRE's July 2025 report, while announced AI-related pipeline exceeds 4.2 GW through 2027; base rates from 2017-2023 show vacancy only exceeded 8% during the 2020-21 demand shock when absorption fell 40% YoY, an event not replicated in current conditions. JLL's Q1 2025 data similarly shows 2.8% vacancy with pre-leasing rates above 85% for facilities under construction.