AI Demand Is Becoming a Property Signal Beyond the Server Room
Artificial intelligence is no longer only a tenant category or a technology theme. It is becoming a measurable real estate demand signal, visible in office leasing, infrastructure services, utilities, water systems, maintenance contracts, and capital allocation. A recent Boston Globe report on earnings calls from BXP and Clean Harbors shows how quickly AI-related demand is spreading into companies that are not traditionally viewed as technology businesses.
For property intelligence readers, the useful signal is not that executives are saying “AI” more often. That is noise. The signal is where AI is turning into committed space, contracted services, and forecast revenue. BXP, the office landlord formerly known as Boston Properties, pointed to leasing activity from AI-linked companies in New York and San Francisco, while also citing Boston Dynamics’ 320,000-square-foot expansion in Waltham. That matters because it ties the AI theme to a concrete absorption event, not just investor language.
The comparison with the life sciences cycle is important. In Boston, the biotech boom encouraged speculative lab development, and some of that space is now underused. BXP’s Doug Linde suggested that AI office demand has not produced the same level of speculative overbuilding. If that holds, AI may create a different property cycle: less about building ahead of tenants, more about capturing high-value users in existing urban and innovation districts.
That distinction should be tested, not assumed. Analysts should track three data points: signed AI-related leases, sublease availability in AI-heavy markets, and the share of new development that begins without preleasing. If AI demand is real but disciplined, vacancy should tighten first in high-quality assets with power reliability, connectivity, security, and proximity to technical labor. If the market becomes speculative, the warning signs will look familiar: fast rent growth, thin tenant commitments, and development narratives that rely more on future demand than current contracts.
Clean Harbors adds a second layer to the data center story. The company expects an “integrated data center solution” to generate $200 million in annual revenue by the end of 2028, supported by $50 million of investment over three years. Its work includes water filtration, pipe debris cleaning, specialty tanks, vehicles, and long-term maintenance. This is infrastructure intelligence in plain sight. Data centers are not only power consumers. They are ongoing operational ecosystems with recurring environmental, water, waste, and compliance needs.
The strongest AI real estate signal is not executive enthusiasm. It is the conversion of AI demand into leases, service contracts, infrastructure upgrades, and recurring operating revenue.
This creates a wider map for forecasting AI’s property impact. Office landlords will watch tenant formation and expansion. Industrial owners will monitor equipment yards, service depots, and utility-adjacent logistics. Municipalities will need better models for water use, wastewater treatment, energy demand, and permitting capacity. Investors should also separate AI office demand from data center demand. They are linked by technology, but they behave differently as property markets.
The next phase of AI real estate analysis should move past keyword counting on earnings calls. The better approach is to build an exposure index: AI leases signed, megawatts under development, utility interconnection queues, service contracts awarded, water and environmental permits filed, and local infrastructure spending. That is how the market will distinguish durable demand from thematic momentum.
Source: The Boston Globe
