Real estate: the full guide

An owner earns rent and gains or loses when a building's value changes. A manager or broker earns fees for leasing, running and selling buildings.

Six pools of money decide the year.

AI touches three of these most. Demand for space changes rent and occupancy. Pricing software changes how rent is set. And software that talks to prospects and residents changes operating and leasing costs.

Adoption is broad but early. Nearly all large owners and tenants are testing AI. Only 5% say they have met all their goals, and more than 60% of investors say they are not ready.

Rent and occupancy: demand is moving, not simply rising. AI companies are taking large blocks of office space in San Francisco, New York, Boston and Seattle. Elsewhere, national office vacancy stays near 20%. Data centers are the clearest winner: record construction, most of it leased before completion. Over time, AI may also shrink the staff of some office tenants, and with it their space.

Operating and leasing costs: lower, if staffing changes. AI assistants answer inquiries, book tours, chase late rent and sort maintenance requests. Most apartment operators using AI say it cut their operating costs. The saving is real only when the owner hires less or reorganizes staff; otherwise it is freed time.

How rent is set: the rules have changed. Pricing software is legal. Sharing competitors' private, current rent data through it is where the Justice Department drew the line. The department said landlords followed the software's recommendations 80% to 90% of the time, which it treated as evidence of coordination.

Fees: pressure on routine work. Brokers and managers who bill for work AI can now do, such as summarizing leases or first responses to prospects, will face clients asking for lower fees.

Value: set by who leases. A building's price follows its tenants. Buildings leased to fast-growing AI firms, or data centers with long leases, gain; buildings in markets AI is not reaching do not.

Two patterns stand out. AI is moving demand for space into a few cities and into data centers, rather than lifting demand everywhere. And the most concrete change in how buildings are run so far is legal: limits on the data rent-pricing software may use.

Scenario 1: AI office demand spreads, or stays in a few cities. AI companies keep leasing, and the growth reaches more markets. Assumes: AI firms keep hiring engineers who work in offices. Signal to watch: AI leasing beyond San Francisco, New York, Boston and Seattle, and national office vacancy moving from about 20%. Last checked: September 2026. Demand is still concentrated, and national vacancy has barely moved. The opposite risk is tenants in other industries needing less space as AI shrinks their staff.

Scenario 2: More states restrict rent-pricing software. States and cities write their own rules after the federal settlement. Assumes: pressure over housing costs keeps rising. Signal to watch: new state or city bans, and private suits against landlords, not only software makers. Last checked: September 2026. New York has banned algorithmic rent-setting, RealPage has sued to overturn the law, and private suits continue.

Scenario 3: Fewer people run each building. Operators move leasing and resident service to central teams supported by AI. Assumes: AI handles routine requests well enough to trust. Signal to watch: operators reporting units per on-site employee, or announcing central leasing teams. Last checked: September 2026. Operators report changing their staffing, but public figures are scarce.

All three reward the same preparation: knowing which tenants AI will grow or shrink, written proof of what data your pricing tools use, and a staffing plan for the work AI takes on.

Gain: owners of data centers and the land and power to build them. Demand exceeds supply, and most new capacity is leased before completion.

Gain: owners of offices in AI hubs. In San Francisco, AI firms are the main source of new leasing.

Gain: large apartment operators. They can spread the cost of AI leasing and resident tools across many buildings and centralize the work.

Lose: owners of older offices outside AI hubs. National vacancy remains near 20%, and some tenants may need less space as AI changes their staffing.

Lose: landlords relying on shared competitor data. Settlements and state bans have made this a legal and financial risk.

Mixed: brokers and property managers. Routine work gets cheaper, which pressures fees, but firms that use AI to handle more properties per person can grow margins.

Mixed: renters. Faster answers and service at any hour, and, under the settlement, rents set with less use of competitors' private data.

The cost ranges are illustrative and will be replaced with reviewed figures. The order matters more than the amounts: check pricing data first, because it removes the largest risk for the least money.

One test before any contract. Ask the vendor to run its tool on your own past leases or inquiries, and compare its output with what your team produced.

Tenants. Every tenant's industry is changing with AI, and with it how many people they employ and how much space they need. Office tenants in AI hubs are growing; others may shrink. The guides for those industries show how.

September 29, 2026: first published.

August 2026 (reflected at launch): data center vacancy at a record 1.4%, with record construction more than 80% preleased. Affects sections 3 and 5.

First half of 2026 (reflected at launch): AI companies signed 58% of San Francisco office leases. Affects sections 2, 3 and 4.

November 2025 (reflected at launch): the Justice Department settled with RealPage over rent-pricing software. Affects sections 2, 3, 4 and 6, and changes the guide's conclusion: checking the data behind pricing tools comes first.