The modern battlefield offers a lesson in real estate vertical integration. Look around. Rocket bought Redfin. Compass combined with Anywhere to become the country’s largest brokerage, north of 340,000 agents. Zillow is fighting over listing access. Homes.com is still buying its way to portal relevance. NAR and the MLSs are defending mandatory cooperation rules. And Bed Bath & Beyond bought Fathom, a brokerage most dismissed before finishing the headline.

On the surface these look like separate strategies. But, it’s just one strategy: Control more of the housing value chain before someone else does.

AI is about to change the battlefield

Picture the stack between a consumer and a home: discovery, engagement, the MLS that pools listings, financing, the brokerage, and the agent who does the work. Almost every move is a company grabbing the layer it owns and pushing into the others. Rocket entered at financing. Zillow owns discovery and engagement. Compass is scaling the brokerage and privatizing listings. Bed Bath & Beyond is starting from the customer and working back.

That was rational in the old world. When attention was expensive and the only way to get real answers was to call an agent and become a lead, owning more layers meant capturing the customer and protecting margin.

But AI is about to change the battlefield, and most are not prepared.

The real estate singularity is not the moment AI replaces agents. That is the lazy version. The singularity is the moment buyers and sellers can reach exactly the expertise they want, exactly when they want it, with almost no friction.

That is not transacting alone. Nobody wants a chatbot hallucinating its way through a disclosure, a title defect nor a negotiation. Judgment, local knowledge and accountability still matter. What changes is how many layers a consumer tolerates before reaching the person they need.

The future is not agentless.

AI will compress the agent population, expose weak agents and pull down margins. But information is not judgment. A buyer still needs someone who knows the neighborhood, the HOA problem and which scary-sounding inspection item is harmless and which harmless-sounding one kills the deal. That layer is not going away.

The problem: much of the industry built its economics on standing between consumers and that expertise.

Customer acquisition cost (CAC) is the hidden mechanism in real estate, not a line item. Portal and referral fees that can reach 40% are CAC. So are franchise fees, brand fees, and parts of many broker splits. The professional with the local expertise often rents back access to the consumer who wanted it in the first place.

That is the toll tower.

Not all CAC is a toll. Paying to reach a customer you would never have found is fair. Paying 40% to a layer that takes no risk and does no work is rent. AI dissolves the second kind, not the first, because a consumer’s new front door has no reason to route them through a tollbooth.

Engagement was always the real moat

Zillow did not win because consumers loved listing data. It won because they came back to look, compare and wonder what their home was worth. The Zestimate was never just a valuation. It was an engagement engine. Consumers came for an answer, the platform captured the relationship, and the industry paid to reach them later.

That is why the current AI conversation feels too small. AI search, scheduling, listing copy, CRM assistants. But most of it is heavier armor on the same tank. None of it asks the real question: how long does any of this stay yours?

The deeper pain was never search or scheduling. Buyers can see every home in their market. What they can’t see is what it means. What is this worth? Is it overpriced or just misunderstood? Explanation was always the scarce thing and that is the seam AI can exploit.

Its most important use may be assembling local expertise at a cost once impossible. Today’s defenses then look less permanent than they appear.

Portals are exposed if market understanding becomes portable, as the front door drifts to Google and general-purpose AI. Brokerages are exposed if supervision and compliance get cheaper and automated. MLSs are exposed if enough inventory moves pre-market, until the pre-market is the market. Mortgage and title are exposed if AI makes loan discovery a true shopping exercise.

None of this means vertical integration is wrong. Rocket may be the most interesting case because it connects nearly every layer, subsidizing one because it monetizes another. An incumbent with captive engagement can sacrifice any layer to defend the audience.

Owning layers is not the mistake. Believing they stay defensible because they once were expensive is the mistake. The test: A bundled layer is convenience if it survives being unbundled and price-shopped, a toll road if it survives only on lock-in.

The broker layer shows the strain. Brokers exist for real reasons: supervision, compliance, risk, and legal accountability. Those matter, and one fact does not change: under current law, AI cannot hold a license or carry the liability. Someone must. But that legal shell can be thin. Most of what justifies a heavy split is oversight labor: document review, disclosure checks, monitoring, exactly what AI makes cheaper and more consistent.

Much of the role rests on a mandate, not a market. If the law did not require a broker, some agents would still want one for genuine services. Many would not. The liability does not disappear. But it stops justifying 10 to 40 percent of a commission.

This may explain Compass’s urgency. If the brokerage layer is exposed, scale, inventory, and retention all become armor. But armor is not strategy when the threat has changed.

Consolidation deserves more attention

Bed Bath & Beyond buying Fathom deserves more attention. The roughly $53 million deal is easy to dismiss until you see where it starts: engagement, not the transaction. A home retailer with a large customer base can spot the signals that precede a move, creating listings at low incremental CAC. Homeowners stay put 11 to 12 years; a move triggers furniture, renovation, and years of spending. The transaction is not the end of the relationship. It is the start of the homeownership wallet. Even if it does not reshape the industry, the instinct is worth studying.

The future of real estate is not agentless, brokerless or MLS-less. It is thinner. The layers that survive will create trust, reduce risk, or improve outcomes. The layers that compress will mostly monetize friction.

That is the real singularity. Not the disappearance of the agent, but of everything standing between the consumer and the agent for no good reason. AI does not eliminate the need for expertise. It eliminates the excuse for making expertise hard to reach.

The winner will not be whoever owns every layer under one roof. It will be whoever delivers the right expertise, from the right person, at the right moment, with the least friction and enough trust to act.

The industry is buying bigger tanks. AI is about to change the war.

Dean DiCarlo is the CEO of Homing.

This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners.To contact the editor responsible for this piece: tracey@hwmedia.com.