The CRE AI Conversation Just Shifted From Which Model to Who Owns the Rollout

AI Readiness · Data Ownership · Vendor Governance

The CRE AI Conversation Just Shifted From Which Model to Who Owns the Rollout

Every story I read this week framed AI in CRE as a tool decision. The actual decision in front of asset managers is who governs the agent that touches buildings, tenants, and money — and whether that intelligence runs on an owner-controlled data plane or a vendor’s.

TL;DR: Microsoft on the IBcon stage, Yardi’s five-pillars rollout framing, and Memoori’s consolidation math are all real — and they all skip the governing question. If you don’t own your data & digital infrastructure, your vendors do. The next 90 days belong to a written Coordinate baseline, a pilot review, and a portability test — not another model bake-off.

Every story I read this week framed AI in CRE as a tool decision. Microsoft on the IBcon main stage. Yardi publishing “Five Pillars of AI Rollout.” Memoori reporting $22 billion in smart-building AI consolidation across roughly 35 deals in 27 months. None of them named the actual decision in front of asset managers right now. The decision is not which AI to buy. The decision is who governs the agent that touches your buildings, your tenants, and your money. That conversation is already happening in boardrooms outside CRE. Most CRE owners are still on the previous question.

A new AI Leadership newsletter from Geoff Woods this week put it cleanly: “Six months ago it was which model. Today it’s who owns the rollout, and what gets redesigned first.” He cited the same week’s signals from ServiceNow, IBM, and CNN. ServiceNow President Amit Zavery: “Advisory AI has run its course. Enterprises need AI that senses, decides, and securely acts.” IBM CEO Arvind Krishna at Think 2026: “The enterprises pulling ahead are not deploying more AI. They’re redesigning how their business operates.” Inside CRE the same week, Yardi published a partner piece in Realcomm titled “Rolling Out AI in Commercial Real Estate: The Five Pillars,” opening with the same observation. A year ago the question was what AI would do to CRE. Today it is how to actually roll it out.

Consolidation tightens the vendor window

The vendor side is moving fast. OpenAI launched the OpenAI Deployment Company this week, explicitly to help organizations build and deploy AI agents at scale. Realcomm announced Microsoft’s Mo Benhalim on the IBcon 2026 main stage for a fireside on “the long-term path from connected and smart buildings toward autonomous real estate operations.” Memoori’s research, also surfaced in Realcomm Weekly this week, tracked roughly $22 billion of disclosed M&A across approximately 35 smart-building AI transactions in the 27 months ending April 2026 — and only 34 new entrants since 2022. Independent smart-building AI vendors, in Memoori’s framing, are an “endangered species.”

What that means for an asset manager: the universe of vendors who will operate AI inside your buildings is collapsing fast. The choice is narrowing while the conversation is still being shaped. And the framing your vendors will offer is the easy one — “which AI tool would you like.” The harder question, the one Geoff Woods named in the boardrooms, is the one your vendors do not want you to ask. Who owns the rollout. Who governs the agent. Whose data plane does the agent run on. Whose decisions does the agent make first.

The boardroom math is running ahead of the procurement deck

The boardroom math outside CRE is not encouraging. Writer’s 2026 Enterprise AI Adoption survey reports that 60% of companies plan layoffs for employees who don’t adopt AI, and 92% are cultivating an “AI elite.” Meanwhile, Deloitte’s State of AI 2026 finds only 29% of organizations report meaningful ROI. The Conference Board reports 75% of executives admit their AI strategy is “more for show.” The industries leading the AI rollout are spending hard and underperforming. The ones pulling ahead are not buying more tools. They are redesigning the operating layer underneath.

The CRE-specific version of that math is uglier. MLQ’s State of AI in Business 2025 reports 95% of AI pilots fail across industries. JLL’s 2025 Global Real Estate Technology Survey found that while 90% of CRE companies are now piloting AI, only 5% have hit all program goals. The bottleneck is not the model. The bottleneck is whether the firm has the data & digital infrastructure to make the model useful — and the governance to let agents act safely once it does.

Capital is back. Discipline is not.

The capital markets context sharpens the stakes. CRE mortgage volume is up 52% with lending at a five-year high (Connect CRE, May 11). Brookfield deployed roughly $20 billion into CRE in two months (Financial Post on Brookfield’s near-term transaction pace, May 12). And yet apartment rents posted a 33rd consecutive month of decline, producer prices jumped 6% year over year (Connect CRE coverage, May 13), and multifamily CMBS maturities climbed to 7.71% in April per Trepp (Multifamily Dive, May 13). Capital is back. Discipline is not. Asset managers who can defend NOI with a credible operations story will refi at materially better terms than asset managers who can’t. AI is now part of that operations story, whether the asset manager is ready for it or not.

The wedge strategy nobody says out loud

Here is the part nobody is saying out loud. The vendor strategy in CRE right now is the same strategy enterprise software ran a decade ago: get the tool installed inside the operating workflow before the owner figures out who governs the agent and where the data goes. Apartment List is selling AI leasing into properties. BirdEye is selling AI tenant retention. AMC just rolled out a residential portal platform. Each one is a wedge into a workflow that touches tenants and operating data. Each one defaults to vendor-controlled data flowing back to the vendor. Each one ships before the owner has named who owns the rollout.

The cross-industry signal is identical to the CRE signal. The companies winning this cycle are not the ones running the biggest pilots. They are the ones that have already decided who owns the rollout, what gets governed, which workflow goes first, and what gets retired. That is an operating decision, not a procurement decision. Tools cannot answer it. Owners can.

Stanford’s 2026 AI Index gave the broader market the punchline. The performance gap between open-source and frontier AI models has collapsed, with cost to hit benchmark performance falling roughly 5x to 10x annually. Most enterprises now run three or more model families at once. Multi-model is the default. Translation: the model is becoming a commodity. What differentiates one portfolio from another is no longer the AI tool the owner bought. It is whether the owner controls the layer above it.

Mapping the trend to PPP 5C™

Mapping this to the Peak Property Performance® plan makes it concrete. Clarify still sets the measurable operating bar before anyone accelerates throughput with agents. The CRE AI rollout question is a Coordinate-and-Control question, not just a Connect-or-Collect question. Most owners have spent the last 18 months on Connect (network) and Collect (some data, some dashboards). The new boardroom question is whether identity, access, lineage, and the rules of agent action are governed by the owner — that’s PPP 5C™ Coordinate. And whether decision engines and AI workflows act under owner permissions, regardless of which vendor’s agent is in the seat — that’s Control. Together those two stages are what Property Brain™ was built to deliver at the single-property level, and what Portfolio Brain™ scales across the asset book.

Operationally, that looks like three things in the next 90 days. First, an asset-manager-led data & digital infrastructure review at one pilot property, sized to surface where the agents are going to want to act and what the owner does or does not currently govern. Second, a Coordinate baseline — identity, access, lineage, retention, and rules of agent use — written down before any new AI agent gets installed. Third, an orchestration plane test. Plug in two decision engines from two vendors against the same data, under the same governance, and confirm both can be swapped without rewiring the building. That last test is what proves the building is portable. Vendor- and LLM-agnostic by design is not a phrase. It is a property-level capability that has to be built and tested.

The 5S® user experience — Seamless Mobility, Security, Stability (resilience), Speed, Service — sits underneath all of it as the operating standard the AI agents have to honor. Without it, the agents act fast in environments that aren’t built to absorb fast actions safely. With it, autonomous starts to mean autonomous-and-safe, not autonomous-and-someone-else’s-problem.

The asset-manager translation

Here is the asset-manager translation. Every dollar of recoverable NOI is worth roughly fifteen to twenty-five dollars of capitalized value at typical cap rates. The owners who win this cycle will not be the ones with the most AI tools installed. They will be the ones who can show an underwriter, on the next refi or exit, that their buildings run on an owner-controlled data plane and a governed trust plane — that the AI lives inside their architecture, not inside someone else’s. That is the redesign question Krishna was naming. Translated to CRE, it is the difference between a building whose intelligence belongs to the owner and a building whose intelligence belongs to the vendor.

The model is the commodity. The moat is the layer above it. If you don’t own your data & digital infrastructure, your vendors do — and inside the next twelve months, the agent that operates your building will belong to whoever owns that layer.

If your team is still arguing about which AI is best for CRE, the conversation is two quarters behind. Start a data & digital infrastructure review at one property, draft the Coordinate baseline before the next agent goes in, and prove portability with a two-vendor orchestration test. That is the rollout decision. The vendors will not make it for you in your favor.

Own your data & digital infrastructure. Operate with strategic foresight. Build for the long game.

— Bill