The AI Reckoning: What It Means for Commercial Real Estate — and Why We’re Positioned for What’s Coming

Part I: The Worst-Case Scenario — What a Fully Automated World Does to CRE

Let’s start with the uncomfortable scenario — not because we think it’s inevitable, but because understanding the downside case is what separates disciplined capital allocation from wishful thinking.

The 2035–2040 Horizon: When AI Meets Physical Automation at Scale

The IMF has estimated that 60% of jobs in advanced economies involve tasks that AI can augment or automate. That number has largely played out in the white-collar world first — entry-level knowledge work, administrative roles, junior analyst positions — and the displacement is already measurable. Stanford research has documented a 13% decline in entry-level hiring across AI-exposed occupations, and unemployment rates among college graduates aged 20–24 are rising for the first time in a generation.

The current phase is still predominantly software-driven. The more disruptive phase begins when AI intelligence is paired with physical capability at commodity cost.
Humanoid robots are no longer science fiction. Figure, Agility Robotics, Tesla, and dozens of other manufacturers are in active deployment — in BMW factories, GXO warehouses, Amazon fulfillment centers. Today’s units cost $50,000 to $250,000 and require maintenance every 200–500 operating hours. But the cost curve is steep: Goldman Sachs projects units below $20,000 by the late 2020s, and with payback periods already approaching 12–18 months at today’s pricing, widespread adoption becomes economically irresistible somewhere between 2030 and 2035.

“Jobs of 15 people can be done with a chatbot that costs me $36 a month.” — Barry Sternlicht, CEO, Starwood Capital Group

When humanoid robotics reaches commodity scale — estimated around $5,000–$10,000 per unit by the early 2030s — the math changes for every sector of the physical economy. The question is no longer whether small businesses will automate; it’s how quickly.

The CRE Implications of a Worst-Case Scenario

Office: Already in Crisis, Headed for Structural Obsolescence

Office is the clearest casualty. National vacancy rates hit a record 19.6% in Q1 2025, and AI displacement of white-collar workers will compound that structural damage. The sectors that occupy the most Class A office space — technology, financial services, professional services — are precisely the sectors most exposed to AI automation. JPMorgan’s CEO has warned of thousands of roles eliminated in operations and middle-office functions. Goldman Sachs has already deployed trading AI that replaces teams of 8–12 analysts. Deloitte’s AI-assisted audit platform reduced engagement hours by 22%.

The office sector isn’t recovering. It’s repricing around a fundamentally smaller addressable market.

Mid-Tier Retail: The K-Shaped Consumer Problem

The more underappreciated risk lies in middle-market retail. If AI displaces knowledge workers faster than the economy creates replacement jobs — and the timing gap could easily be five to ten years — the result is a compression of middle-class consumer spending. High-income earners, whose wealth is tied to appreciating assets and who are actually becoming more productive with AI tools, keep spending. Lower-income households, whose jobs are more physical and therefore more durable in this cycle, hold relatively steady.

It’s the middle that hollows out. That’s a direct structural threat to mid-tier discretionary retail — apparel, department stores, lifestyle soft goods — formats that are already losing share to e-commerce and have no compelling reason to survive in a physical format at scale.

Large-Format Industrial: Overbuilt and Over-Robotized

In a worst-case scenario, large-format logistics real estate — the 300,000+ square foot distribution center — faces a paradox. Robotics makes these facilities more efficient, but that same efficiency reduces the labor force that fills them and the consumer demand that flows through them. Amazon has already scaled back its large-box leasing significantly. In markets like Charlotte, vacant 400,000 square foot facilities sit beside fully-leased small-bay product. In the most aggressive displacement scenario, bulk industrial would face a demand ceiling just as the sector was supposed to benefit from automation tailwinds.

Broader Economic Feedback Loops

The deepest risk isn’t any single sector. It’s the feedback loop. Displaced workers spend less. Less consumer spending reduces demand for retail, for housing, for the local service economy. Reduced property tax revenue stresses municipalities. And lower employment concentrations reduce the density of economic activity that justifies the location of commercial real estate in the first place. The “doom loop” thesis — already playing out in San Francisco’s downtown — becomes a national CRE risk if displacement accelerates faster than economic adaptation.

This is the scenario worth underwriting against. Not because it’s the most likely outcome, but because it’s the one most capital has not yet priced.

Part II: The Realistic View — A Transition, Not a Collapse

Here’s what history consistently tells us: transformative technologies disrupt specific sectors, eliminate specific job categories, and create enormous economic anxiety — and then they generate productivity growth that creates new demand, new industries, and new categories of real estate that didn’t exist before.

The internet didn’t eliminate retail. It eliminated certain kinds of retail and created enormous demand for industrial logistics. The smartphone didn’t eliminate office. It created entirely new categories of tech-oriented tenants. The realistic view of the AI transition follows a similar arc — just with a more severe short-term distribution shift before the long-run growth materializes.

The Realistic Timeline

The humanoid robotics market today is exactly where the EV market was in 2015 — technologically credible, economically nascent, and widely underestimated by the mainstream. IDTechEx projects widespread logistics deployment beginning in 2026–2027. Morgan Stanley’s more measured view puts true mass-market acceleration in the late 2030s. Goldman Sachs baseline: 250,000 humanoid units shipping annually by 2030.

That is not, by any measure, a labor market collapse. It’s a highly concentrated disruption in specific industries — automotive, large-format logistics, select manufacturing — followed by a gradual expansion into adjacent sectors. The 2025–2030 window is best characterized as accelerating white-collar displacement paired with early-stage physical automation piloting. The full convergence of the two — broad physical labor displacement at commodity robot cost — is a 2033–2040 story.

That runway matters for investment strategy. It means there is a meaningful window — likely a decade or more — during which the physical-world economy continues to require human presence, physical space, and the supply chains that serve both.

What Grows Alongside AI: The New Demand Drivers

The most commonly overlooked aspect of the AI transition is what it creates, not just what it destroys. Data centers are the obvious example. The physical infrastructure required to run AI models — servers, cooling systems, power transmission — is driving the largest single capital expenditure cycle in corporate history. The top five tech companies alone are projected to spend $660 billion on AI buildout in 2026.

But the second-order effects are less discussed. The entire robotics supply and service economy — manufacturing, maintenance, repair, software deployment, spare components — needs physical space. Reshoring of manufacturing, accelerated by tariffs and supply chain risk, needs industrial space. The growing trades workforce required to install, maintain, and operate the physical infrastructure of an AI economy needs the kind of space that small-bay industrial provides at the most local level.

History suggests that 60% of current employment is in occupations that didn’t exist 80 years ago. Some version of that dynamic will play out again. The question for CRE investors is which physical formats will house the new economy’s activity — and which will be stranded by the old one.

Perhaps the most honest framing we’ve heard came not from a research report, but from a colleague walking into the office recently: “What I love about AI is that it just makes those who are willing to work more efficient.” That’s the realistic version of this story — not mass unemployment, not utopian abundance, but a widening gap between operators who adapt and those who don’t. In commercial real estate, that gap shows up in who owns the right assets, in the right markets, with the right basis.

Part III: Why Excelsior Capital’s Strategy Is Built for This Environment

Excelsior Capital has never positioned itself as a macro-trend chaser. Our strategy is built on acquiring high-quality assets with strong fundamentals in supply-constrained secondary markets — and holding them with operational discipline. But the AI/robotics transition is not a macro-trend we’re watching from a distance. It’s a structural force that is directly relevant to why the asset classes we invest in are increasingly defensible, and why we remain confident in the thesis through the decade ahead.

Small Bay Industrial: The Physical-World Economy’s Backbone

• Small bay industrial — the sub-50,000 square foot, multi-tenant warehouse product that houses contractors, tradespeople, local distributors, light manufacturers, and last-mile logistics operators — sits in a uniquely privileged position in the AI economy.
• The tenant base is physically durable. The plumber, the HVAC technician, the auto restoration shop, the e-commerce seller fulfilling local orders — these businesses require a physical presence in a physical place. AI augments their work; it does not replace it. The same robots that are displacing white-collar knowledge workers are years, possibly decades, away from replacing a licensed electrician on a job site or a specialty fabricator running a three-bay workshop.
• The supply story reinforces this fundamentally. Only 0.3–0.5% of existing small bay industrial stock is under construction nationally — a number that has been falling since interest rates moved off historic lows in 2022. Construction costs have risen more than 44% since the pandemic. Land for infill small-bay development is scarce and entitlement-intensive. New supply is not coming in any meaningful quantity.
• The result is vacancy rates of 3–5% across the sub-50,000 square foot segment, compared to 7–8% for industrial broadly and nearly 20% for office. Rents in the 50 largest secondary markets — exactly where Excelsior focuses — have grown at 5.9% annually, more than triple the rate of primary markets where supply is thinner but price discovery is more mature.
• Importantly, robotics doesn’t threaten small bay industrial — it reinforces it. Even as physical automation advances, robots need space to operate, to be stored, to be maintained. The robotics-as-a-service providers, the repair depots, the component distributors — these are the next generation of small bay tenants. If anything, the humanoid robotics wave extends the runway for small bay demand rather than threatening it.

Non-Anchored Retail: The Case for Necessity and Experience

Our retail strategy has always been oriented around a simple principle: own the formats that serve daily human life rather than the formats that serve discretionary consumption of goods available online.

The AI era validates that framework with new force. As AI displaces the middle-market consumer and elevates the K-shaped economy, the retail formats that survive are the ones providing either essential services or irreplaceable experiences. Grocery-anchored centers. Service-oriented strip centers anchored by medical, fitness, dental, and personal care tenants. Food and beverage concepts. “Medtail.” These are the Internet-resistant, AI-resistant, robot-resistant formats — because they require human presence to deliver their value.

The supply picture mirrors small bay. New retail deliveries in 2025 were 63% below pre-pandemic averages. Vacancy for grocery-anchored and open-air neighborhood centers sits below 4.5%. Asking rents climbed to $25.69 per square foot nationally, with renewal spreads from five-year-old leases generating significant mark-to-market upside for patient owners.

The bifurcation we’re navigating is real and it is wide. Mid-tier discretionary retail — the format most exposed to both e-commerce substitution and the K-shaped consumer income compression that AI threatens to accelerate — is where capital should not be deployed. The necessity and services formats we target are, if anything, more essential in a world where the physical economy contracts around the things humans still need to do in person.

Secondary Markets: The Structural Advantage That Remains Intact

Our focus on high-growth Sun Belt and Midwest secondary markets — Nashville, Charlotte, Atlanta, Dallas-Fort Worth, and similar metros — is not incidental to our AI resilience thesis. It’s central to it.

The AI economy is not distributed evenly. Its first-order effects are concentrated in coastal gateway markets with heavy tech and financial services employment — San Francisco, New York, Seattle, Austin. The displacement of white-collar workers in those markets is already visible in office vacancy data and effective rent declines. The secondary markets we target have more diversified employment bases, stronger population inflows, and industrial and service-sector job growth that is additive rather than at risk.

Rent growth in secondary industrial markets has outpaced primary markets by more than 3x. Capital competition from institutional buyers — while increasing — is still meaningfully lower than in gateway markets. And the supply constraints are, if anything, more severe: smaller land bases, less infrastructure for large-scale development, and local permitting environments that favor existing owners.

Operational Discipline as a Competitive Advantage

In a market environment defined by macro uncertainty, the firms that win are the ones that can underwrite conservatively, operate efficiently, and hold through the cycle without distress. Excelsior Capital has built its portfolio around exactly this discipline.

We acquire existing buildings at a healthy basis, below replacement cost where possible. We target assets with value-add pathways — under-market rents, below-stabilized occupancy, operational inefficiencies — that allow us to generate returns through execution rather than market appreciation. And we maintain a rigorous approach to debt, avoiding the over-leveraged structures that created distress for so much commercial real estate when rates normalized.

In a world where AI and robotics will restructure the physical economy over the next ten to fifteen years, that discipline is not just a risk management posture. It is an active investment strategy. The operators who can navigate the transition — who own the right formats in the right markets with the right basis — will compound capital through a period of disruption that will strand significant amounts of poorly positioned real estate.

The Bottom Line

AI and robotics represent the most significant structural shift in the commercial real estate landscape since the rise of e-commerce. Unlike prior cycles, this one simultaneously threatens the largest CRE sector (office) while creating sustained tailwinds for the physical-world assets that cannot be digitized away.

The worst-case scenario — broad physical labor displacement, K-shaped consumer collapse, and feedback loops through the local economy — is a real risk that we believe is underpriced in most CRE underwriting today. It deserves to be modeled, stress-tested, and taken seriously.

The realistic scenario — a decade-long transition during which white-collar displacement accelerates while physical automation scales more gradually — is one that favors exactly the asset classes and markets Excelsior Capital focuses on. Small bay industrial and non-anchored service retail, held with operational discipline in supply-constrained secondary markets, are among the most defensible positions in the CRE landscape for the decade ahead.

We’re not betting against technological progress. We’re betting that people still need places to work with their hands, to buy their groceries, and to get their teeth cleaned — and that the buildings housing those activities will remain valuable long after the next wave of disruption reshapes the white-collar economy around them.