For three decades I’ve been in the business of putting up buildings. Towers in Asia. Mixed-use districts in the Southwest. Hotels and master-planned communities and innovation campuses. The work was capital-intensive, slow, and forgiving. You bought dirt, you built, you held, and the underwrite worked because demand kept compounding.
In 2026 the math is different. The dirt is more expensive. The construction is much more expensive. The cycle is longer. And the second-tier asset — the 1980s Class B office, the 1990s mall, the mid-century mixed-use destination — still produces a perfectly serviceable lease but generates almost no other line of revenue. Most of the value in those assets is sitting in a layer we haven’t priced yet.
I’m going to call that layer offline. By offline I mean the part of every commercial real estate asset that does no work between rent collections. It doesn’t show up in NOI. It doesn’t show up in the offering memorandum. It doesn’t show up in the lender’s underwrite or the LP report. And on a national basis it represents roughly a trillion dollars of trapped value waiting for the standard tools of capital markets to find it.
This piece is about how that math works, why it didn’t work in 2010 or 2015 or 2020, and what’s different now.
Real estate is a three-layer stack.
It always was. The physical layer — the dirt, the structure, the systems. The financial layer — the rent roll, the cap stack, the lease covenants, the LP report. These two layers are well-organized capital markets. We’ve spent a century learning to price them.
There’s a third layer that has never had a market.
Call it the digital layer. It’s the operating performance of the asset, expressed as data. Sensor-grade energy intensity. Occupancy efficiency. Indoor environmental quality. The engagement pattern of every tenant and visitor. The carbon footprint of every kilowatt-hour. In 2026, most institutional CRE owners produce that data as a side effect of operating the building, then throw most of it away. The data that survives goes into a quarterly utility bill spreadsheet and an ESG report.
It does not enter the cap stack. It does not appear in the underwrite. It is not priced at exit.
That is the trapped-value thesis: the digital layer is the most material un-priced asset on a commercial real estate balance sheet. And the next decade of returns in CRE comes from learning to price it.
The size of the offline layer.
The U.S. carries roughly 5 billion square feet of office inventory alone as of year-end 2025, per CoStar (2025) national commercial real-estate inventory. The Lawrence Berkeley National Laboratory (Buildings Energy Data Book) shows that the average U.S. commercial building is about 44 years old (median 36 years, average construction year 1981 per the EIA Commercial Buildings Energy Consumption Survey, 2018 cycle). Most of that inventory is functionally sound. None of it was designed with a digital revenue layer in mind.
The McKinsey Global Institute (2023, Empty Spaces and Hybrid Places) put a finer point on the office side: roughly 13 to 38 percent of office demand in major US metros may not return at all. Class B and C office is bearing the brunt. JLL Office Market Statistics (JLL Research, 2025) show Class B/C vacancy in major US markets sitting in the high teens, with some submarkets above 25 percent.
Now apply the math. A Class B office at 400,000 square feet, valued today at roughly $200 per square foot, is an $80 million asset. Its NOI runs at, say, $5.2 million. The owner has perhaps $40 million in mortgage debt against it. There is essentially no recurring digital revenue line on the asset. Less than one percent of NOI in most institutional portfolios I’ve reviewed is attributable to anything other than rent.
That number — under one percent — is the gap. Across the institutional CRE universe, it represents close to a trillion dollars of un-priced operating margin.
The trapped-value thesis lives in that gap.
Where the trapped value actually lives.
I get asked this often, usually skeptically, by allocators who’ve been pitched smart-building software for fifteen years and seen no compounding return on it. The honest answer: the trapped value lives in five places.
Connectivity as a service. The building or campus already provides Wi-Fi and structured cabling. In most leases that’s a tenant-improvement cost or a back-of-envelope amenity. In 2026 a well-instrumented destination provides carrier-grade Wi-Fi, in-building cellular coverage, an IoT mesh, and private 5G to the leased space and the public space alike. The tenant pays for the service tier. The owner books the recurring revenue at a margin meaningfully above the underlying infrastructure cost.
Tenant-experience tiers. The same destination that offers connectivity offers an engagement layer — events, programming, retail-percentage rent, virtual commerce. In master-planned mixed-use this is half the value-capture story. In a Class B office repositioning, it’s the single biggest reason a 2026 tenant signs a lease at a 2026 rent.
Operational efficiency. AI agents now run the routine optimization across the central plant. Predictive HVAC. Predictive lighting. Predictive elevator dispatch. The savings are not the headline — the headline is the labor reshape, which we’ll cover next month. But the savings compound. They show up in NOI within the first quarter of integration.
Auditable ESG. Pension funds and sovereign wealth funds increasingly screen for compliant data. GRESB, CRREM, the SEC climate rule, and EU CSRD all reward sensor-grade telemetry over reported estimates. The asset that can produce that data has access to capital that the asset that can’t no longer does.
Cap-rate compression at exit. Trophy assets in tier-one markets are already pricing the digital layer in. The arbitrage is in the next tier down. We’ll return to this. It’s the most important of the five.
Why the math has flipped.
Three things changed at roughly the same time, between 2022 and 2025.
First, construction cost. Per-square-foot delivery costs for institutional commercial product run 1.6 to 2.2 times pre-pandemic levels. The Federal Reserve Bank of St. Louis FRED PPI series on commercial-building construction makes the inflation curve uncomfortable to read. Tearing the building down and rebuilding it almost never pencils against retrofit.
Second, embodied carbon. The Carbon Leadership Forum (2017, updated 2024, Embodied Carbon Benchmark Study) and Architecture 2030 Carbon Smart Materials Palette (Architecture 2030) document the structural-and-cladding carbon penalty of new construction in terms most CFOs understand. At even a modest carbon-price assumption — fifty to two hundred dollars per ton CO2e — replace-and-rebuild becomes a markedly worse trade than instrument-and-modernize. NYC Local Law 97, California SB 253, and the EU’s CSRD make the calculus regulatory, not just rhetorical.
Third, time to revenue. A new ground-up commercial tower in a primary US market is now a five-to-seven-year endeavor from entitlement to stabilized cash flow. A modernization of an existing asset begins producing measurable performance lift within the first two quarters of integration. In a high-rate environment, that delta is the trade.
The combined effect: the underwrite no longer rewards the demolish-and-rebuild instinct. It rewards making the asset you already own do more.
A simple model.
Let me show the math on a single building. Abstractions don’t move LP committees.
Take a 400,000-square-foot 1980s Class B office in a US gateway market. Acquisition basis: $200 per square foot, so $80 million. NOI: $5.2 million. Hold-period exit cap rate assumed at 7.5 percent.
Now layer the digital infrastructure. Connectivity-as-a-service contributes a new revenue line of $0.75 to $1.10 per square foot per year, depending on tenant mix. Engagement and amenity programming add another $0.40 to $0.80 per square foot. Operational efficiency removes 8 to 15 percent of opex. Conservatively, the digital layer adds $400,000 to $700,000 of net operating margin per year.
That’s not the headline number. The headline number is at exit. The same asset, sold with documented digital-revenue lines and sensor-grade ESG telemetry attached, prices at a meaningfully tighter cap rate than the same asset sold without them. Empirically, Green Street Advisors (2024–2025, Commercial Property Price Index notes) and on data-quality pricing spread suggests 15 to 35 basis points of compression on well-documented institutional assets, with the spread widening through the decade.
At a 7.5 percent baseline and even 20 basis points of compression, the terminal value lift on an $80 million asset is roughly $4 million to $5 million on top of the NOI compounding. Across a portfolio of ten such assets, you’ve added a $45–55 million line to the fund’s terminal value that didn’t exist in the original underwrite.
This is what trapped-value capture looks like at the asset level. It compounds at the portfolio level.
This is not smart buildings v3.
I want to be honest about the prior generation. The 2010-era smart-cities pitch — IBM Smarter Planet, Cisco Smart+Connected Communities, the early Sidewalk Labs vision — promised a top-down, vendor-controlled, capital-intensive technology overlay on real estate. It didn’t work. The honest postmortem lives in Anthony Townsend’s Smart Cities (Townsend, 2013, W. W. Norton) and Shannon Mattern’s A City Is Not a Computer (Mattern, 2021, Princeton). Vendor lock-in. Misaligned incentives. No tenant economics. The model failed on every dimension where capital markets eventually grade.
The 2026 version is structurally different. Owner-controlled instead of vendor-controlled. Open data protocols instead of proprietary stacks. Composable with the BMS, the EMS, the PMS, the access control that’s already in the asset. Designed for institutional security review on day one. The economics make sense because the data lives on the asset, not on a vendor’s platform.
The difference is the same difference that separated cloud-native enterprise software from the previous generation of on-premises software. The first generation was a vendor pitch. The second generation is infrastructure.
What blocks adoption.
The honest answer: not technology, and not capital. Three things block adoption today.
Capital-allocation processes designed for buildings, not platforms. Most institutional sponsors evaluate capex one asset at a time, line-item by line-item, with no category for recurring software-margin lines. The CIO needs the LP’s permission to write the platform check, and the LP doesn’t yet have a category for it. This is changing in the largest sovereign and pension allocators.
Procurement timelines that don’t match technology cycles. A 2010-era BMS replacement cycle runs fifteen to twenty years. A 2026-era software upgrade cycle is six months. The two clocks are incompatible until the procurement organization rebuilds itself around continuous deployment. It’s already happening.
Tenant lease-rollover dynamics. The full benefit of an instrumented destination compounds across two or three lease cycles. The first signed tenant after instrumentation pays for the previous-generation experience because the lease was already drafted. The math gets better at every renewal. Patience matters here. The underwrite still holds.
What unblocks it.
The technical preconditions that didn’t exist in 2010 now do.
Open data protocols are mature. BACnet/SC for building automation. Matter for IoT. Brick Schema and Project Haystack for building data. Model Context Protocol for AI-agent interoperability. The plumbing of an institutional-grade twin exists in 2026 and didn’t in 2018.
Photoreal real-time engines are mature. The fact that Unreal Engine can run a 60-fps multi-stakeholder twin of a 1,000-acre master-planned community on a laptop is not a marketing claim. It’s the same engine running modern film virtual production. The fidelity question is settled.
Capital-markets ESG pricing is happening. CRREM 1.5°C pathway (CRREM consortium), GRESB scoring weight on data quality, and the SEC’s climate-disclosure rule (stayed since April 2024; SEC withdrew its defense in March 2025) all reward sensor-grade telemetry. The capital screens are tightening on assets that can’t produce it.
The institutional implication.
If you sit at a fund, here is the implication. An asset that earns a recurring digital yield, with auditable data, underwrites tighter than an asset that doesn’t. The pricing spread between the two will widen through the decade. By 2030, “Was the asset digitally legible?” sits in every diligence checklist. Funds that own ten digitally legible assets will have a structurally different return profile than funds that own ten that aren’t.
The arbitrage is now live. It is not a 2032 problem. The owners moving in 2026 have a five-year head start on the rest of the market.
The 2030 view.
The next decade of returns in commercial real estate will not come from building more. It will come from making what we already have legible, measurable, and monetizable.
The trapped-value thesis is the cornerstone argument for that view. The trillion-dollar offline layer is real. The tools to instrument it are mature. The capital markets that will reward it are moving — slowly at first, then quickly.
Thirty years of putting up buildings taught me that markets reward what they can price. The digital layer was un-pricable for fifteen years. It’s pricable now.
For most institutional owners reading this, the right next move is a diagnostic on a single asset. Set a baseline. Model the digital-revenue lines. Run a paid pilot. The math will tell you the rest.
Sources cited
- CoStar Group. (2025). National Commercial Real Estate Inventory. CoStar Analytics.
- Lawrence Berkeley National Laboratory. Buildings Energy Data Book. U.S. Department of Energy.
- McKinsey Global Institute. (2023). Empty Spaces and Hybrid Places: The Pandemic’s Lasting Impact on Real Estate.
- JLL Research. (2025). US Office Outlook Q1–Q4 2025. Jones Lang LaSalle.
- Federal Reserve Bank of St. Louis. Producer Price Index for Commercial Building Construction (Series WPUSI012011), FRED.
- Carbon Leadership Forum. Embodied Carbon Benchmark Study (2017, updated through 2024). University of Washington.
- Architecture 2030. Carbon Smart Materials Palette.
- Grieves, M., & Vickers, J. (2017). “Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems.” In F.-J. Kahlen, S. Flumerfelt, & A. Alves (Eds.), Transdisciplinary Perspectives on Complex Systems. Springer.
- Townsend, A. (2013). Smart Cities: Big Data, Civic Hackers, and the Quest for a New Utopia. W. W. Norton.
- Mattern, S. (2021). A City Is Not a Computer: Other Urban Intelligences. Princeton University Press.
- Green Street Advisors. Commercial Property Price Index, monthly notes 2024–2025.
- CRREM Consortium. 1.5°C Decarbonisation Pathways for Commercial Real Estate. crrem.eu.
- GRESB. Real Estate Assessment Methodology, Standard 2024–2025. gresb.com.
- ASHRAE. ANSI/ASHRAE Standard 135 (BACnet) and Standard 135-2020 (BACnet/SC).
- Connectivity Standards Alliance. Matter v1.2+ Specification. csa-iot.org.
- Anthropic. (2024). Model Context Protocol Specification. modelcontextprotocol.io.
For the full institutional framework, including the diagnostic worksheet, download The Digital Alpha Playbook (28 pages). Or talk to our team — thirty minutes, on the phone, no slides.