AI is filling some buildings, emptying others, and changing how rent may be set.

Real estate makes money from rent, from what buildings sell for, and from the fees earned leasing and running them. AI now reaches all three. It is changing who needs space, what it costs to run a building, and how rent may be set.

It is changing who needs space. AI companies are hiring and leasing fast, but only in a few cities. In San Francisco they signed 58% of all office leases in the first half of 2026. Across the country, though, about a fifth of office space is still empty. So AI is moving demand for space from some places to others. It is not simply adding to it.

Data centers are the clearest winner. The buildings that run AI are in short supply. Only 1.4% of data center space in the main markets was empty in the first half of 2026. Most new ones are leased before they are finished.

It is cutting the cost of leasing and running buildings. Software now answers people looking for a home at any hour. It books tours and sorts routine repair requests. Nearly every large owner and tenant is testing AI. But only 5% say they have reached all their goals.

It has brought new legal risk to how rents are set. The US Justice Department sued RealPage. Its software recommended rents to competing landlords using their private data. In a 2025 settlement, the company agreed to stop using competitors' current private data. Greystar, the largest apartment operator, paid $50 million to settle a private suit. New York banned software that sets rents this way.

It is changing what brokers and managers are paid for. Brokers and property managers bill for some work that AI can now do, such as summarizing leases and answering first questions. Clients will notice, and some will ask for lower fees. Firms that use AI to run more buildings per person can earn more. Firms that bill for routine hours will be asked to charge less.

Time saved is not money saved. When AI answers the phones, the owner saves money only by hiring fewer people or moving them to other work. Otherwise the owner has bought free time, and the software is just another bill.

Here is what that looks like in practice. An owner of 4,000 apartments adds an AI assistant. It answers most first questions from people looking for a home and books their tours. It costs about $100,000 a year. The owner decides not to replace three leasing agents when they leave. That saves about $210,000 in pay, so the owner is about $110,000 a year better off. Without that hiring decision, the owner is simply $100,000 a year worse off, with a leasing team that has more free time.

What should a real estate leader do first? Three plain things.

1. Find out what data every pricing tool you use is built on. Ask whether it includes competitors' private numbers, and get the answer in writing.
2. Test an AI assistant on leasing and routine requests, with a person watching. Decide in advance which jobs it will change.
3. Look at your buildings by who leases them. Some markets are being filled by AI companies. Others may lose tenants whose staff AI makes smaller.

None of this needs a new team or a large budget. It needs one leader who owns each answer.

The full guide shows where AI touches the money, what has actually happened, who gains and who loses, and what to ask before you spend.