For much of its history, real estate has been defined by the ownership of physical assets. Land, buildings and location determined where value accumulated and who benefited from it. That remains the case, but a second layer of ownership is emerging around the information used to value, market, finance and manage those assets.

Listing records, transaction histories, construction data and automated valuations now influence decisions across the property market. The companies assembling that information are no longer simply selling software. They are building the infrastructure through which developers, investors, brokers and owners understand the market - and decide where money goes.

A market still difficult to measure

The market for property data and the technology used to collect, analyse and sell it is growing, although estimates of its value vary considerably.

The Business Research Company values the global AI-in-real-estate market at $301.58bn in 2025, rising to $404.9bn this year and $1.3 trillion by 2030. Grand View Research, by contrast, puts the wider global property technology market at $50.1bn in 2026 and the market for real estate software at $14.3bn.

These estimates are not directly comparable. They cover different products, services and definitions of artificial intelligence. But when one valuation of AI in real estate substantially exceeds another estimate for the wider property technology sector, it becomes difficult to treat any single forecast as a dependable measure of the market.

More revealing, then, is what companies are actually buying - listing platforms, transaction records, valuation tools and construction databases. These purchases show who is acquiring the information that determines how property is priced, marketed and sold.

The concentration of commercial property data

CoStar offers one of the clearest examples of how property information is becoming concentrated. Its platforms bring together commercial listings, transaction records, market analysis and pricing data used by brokers, investors, lenders and owners.

An antitrust class action filed in April alleges that CoStar controls approximately 80% of the American market for online commercial property listings and information services. The complaint raises concerns about long-term agreements with major brokerages and restrictions on sharing data with competing platforms. CoStar disputes the allegations and maintains that its contracts protect information developed through substantial investment.

The case reflects a broader question for the property market. A platform becomes more useful as more participants contribute listings and transaction records. But once enough information is concentrated within one system, competing services become harder to establish and customers have fewer alternatives.

CoStar’s recent acquisitions show how that position can extend into other parts of the industry. On 21 August, it completed its $800m acquisition of Zonda , a residential construction data business covering land development, building activity and housebuilder operations.

The acquisition extends CoStar’s reach beyond completed properties into land development, construction and housebuilder operations.

Where data meets its limits

The residential market presents a different question: how much of an advantage does owning property data actually provide?

Zillow’s Zestimate, one of the best-known automated property valuation tools in the United States, has a median error rate of approximately 1.8% for homes on the market and 7.2% for those that are not. The difference reflects the information available: listed properties provide asking prices and current market signals that off-market homes do not.

The limitations became more apparent through Zillow Offers, the company’s attempt to buy and resell homes directly. Its 2021 financial results show that its Homes division recorded an $881m pre-tax loss as the business was wound down. Estimating a property’s value and making a profitable investment decision, it turned out, were not the same thing.

More recently, questions about the company’s position have extended beyond valuation. On 24 August, the Federal Trade Commission announced a settlement concerning an agreement between Zillow and Redfin . The agency said Zillow had paid Redfin $100m under an arrangement involving the closure of Redfin’s apartment-listing advertising business and restrictions on its future participation in the market. Zillow denies wrongdoing, and the settlement requires Redfin to resume competing.

The two examples raise different questions, but both concern the limits of control. Access to more information does not guarantee better investment decisions, whilst a stronger market position can bring greater scrutiny of how that information is used.

Access, ownership and the role of regulation

The question of who controls property information is also emerging through the multiple listing services used across the American housing market.

The National Association of Realtors agreed to pay $418m to settle the Sitzer/Burnett commission litigation and amended rules governing compensation for buyer brokers. On 19 August, the Eighth Circuit upheld the wider settlement . Separately, NAR approved 18 changes to its MLS handbook , giving individual listing services greater discretion over who can access their platforms.

Changing the rules has not necessarily made the underlying information easier to obtain. Financial Business Systems, which operates the Flexmls platform, declined to release listing and commission data requested in related litigation, arguing that it needed authorisation from the individual listing services involved. A judge subsequently approved a request to remind those organisations of their obligations under the settlement, although some objected to disclosure .

The dispute demonstrates how ownership of a property, control of its listing and access to the data behind it can sit with different organisations. A change in regulation, therefore, does not automatically resolve those competing interests.

The physical cost of the data economy

The expansion of property data also depends on infrastructure that occupies land, consumes electricity and requires investment of its own.

Alphabet, Amazon, Microsoft and Meta are expected to spend approximately $725bn on capital expenditure in 2026 , much of it linked to cloud computing and artificial intelligence. In the United States, annualised data centre construction spending had reached $50.7bn by April , compared with $43.8bn for general offices.

In Europe, electricity costs, limited land and delays securing grid connections are pushing projects further from established urban markets. JLL figures reported by Reuters suggest hyperscale data centres planned between 2026 and 2028 will be located an average of 175 kilometres from major cities, compared with 46 kilometres for projects developed between 2022 and 2025.

What makes a site valuable is changing. For data centre developers, access to electricity, grid connections and available land may matter more than an address in an established market.

The same principle applies to property information. Its value lies increasingly with the companies that control how it is collected, shared and used, rather than necessarily with those who own the underlying assets.

For now, the advantage rests with the platforms that have already accumulated the data and the users. Whether it stays there will depend on how regulators respond, how easily information can move between systems and whether owners, developers and brokers retain meaningful control over the data their properties generate.