70% of Empty Offices Can't Become Apartments. The Floor Plate Won't Allow It.

Stand at the center of a typical Class B office floor in downtown Denver, Chicago, or Philadelphia. Look toward the windows. They are 35 feet away. Between you and the glass sits a corridor of fluorescent-lit space that has functioned perfectly well as a bullpen for insurance adjusters or a call center for a regional bank, where daylight was a perk rather than a building code requirement. Now imagine placing a bedroom here, in this sunless interior zone where the nearest window is farther away than the length of most studio apartments. You cannot. Not because of money, not because of zoning, not because of political will, but because the International Residential Code requires every habitable room to have a window with glazed area equal to at least 8 percent of the floor area, and no amount of tax incentives or adaptive reuse ordinances can repeal geometry.

This is the problem that nobody mentions at city council meetings when mayors announce office-to-residential conversion programs.

A Geometry Problem Disguised as a Housing Solution

Office vacancy in the United States hit 17.8 percent in March 2026, according to CommercialCafe, with CBRE reporting 18.6 percent for the first quarter. Remote work hollowed out downtowns that were already struggling with post-pandemic occupancy declines, and $150 billion in office mortgages now sit under pressure as buildings trade at fractions of their pre-2020 valuations. Cities responded predictably: San Francisco launched its Downtown Adaptive Reuse Program in July 2023, waiving planning code requirements for conversions in commercial districts. Chicago, Los Angeles, and Washington followed with their own incentive packages, mixing tax waivers with streamlined permitting and zoning relief.

RentCafe's 2026 analysis projects 90,300 office-to-apartment conversions this year, up from 71,000 in 2025 and part of a fourfold surge since 2021. New York leads with 16,358 units in the pipeline, followed by Washington, Chicago, and Los Angeles. Politicians love these numbers because they suggest a neat solution: empty offices become needed housing, tax bases recover, downtowns revive. Clean narrative. Terrible architecture.

Because the fundamental constraint is not economic. It is spatial.

35 Feet From the Window

A standard American office building has a floor plate depth of 60 to 80 feet, measured from the exterior wall to the structural core. Offices tolerate this because workers do not have a legal right to natural light at their desks, and because open floor plans push workstations into interior zones that rely entirely on artificial lighting and mechanical ventilation. Residential occupancy inverts these assumptions completely. Every bedroom, every living room, every habitable space must have direct access to a window. In New York City, the Multiple Dwelling Law goes further, requiring that windows in habitable rooms open onto a street, yard, or court of minimum dimensions, not merely onto a light well or setback.

Run the numbers on a center-core building with two exposures, which describes the majority of the American office stock built between 1960 and 2000. A 70-foot floor plate with a central core means each side extends roughly 35 feet from the exterior wall to the core. Residential units need to keep habitable rooms within approximately 30 feet of the window wall to meet natural light requirements and to create spaces where someone would actually want to live rather than merely survive. That extra five feet, multiplied across thousands of square feet of floor area, produces a belt of interior dead zone that cannot legally or practically contain bedrooms, living rooms, or kitchens.

Corner buildings with four exposures fare better, because light penetrates from two directions and corridors can run through the core, allowing units to pull daylight from multiple orientations. Side-core buildings, where the elevator and mechanical shafts sit against one wall rather than in the center, leave the majority of the floor plate within reach of the opposite facade. Pearl House at 160 Water Street in Manhattan converted successfully in part because it had three exposures and a side core, allowing unit layouts that placed every habitable room within 25 feet of glass.

Most buildings do not share Pearl House's structural luck. Industry estimates suggest roughly 30 percent of the American office stock has geometry favorable to residential conversion, meaning narrow floor plates, side cores, multiple exposures, or some combination that keeps habitable space within reach of exterior walls. Seventy percent are architecturally stranded assets, buildings whose physical form precludes residential use regardless of what incentives the city offers, what tax breaks the developer secures, or how desperately the housing market needs supply.

Where AI Enters the Floor Plan

A traditional feasibility study for an office-to-residential conversion costs $50,000 to $100,000 and takes three to six months. An architecture firm measures the building, models potential unit layouts, checks structural grids against residential planning requirements, traces plumbing risers, evaluates ceiling heights, and produces a report that tells the developer whether the project is worth pursuing. For the 70 percent of buildings that fail on geometry alone, that study is an expensive confirmation of what a 30-second glance at the floor plate could have revealed.

Autodesk and Arcadis launched what they call the "X-Ray Vision" initiative, combining AI with laser scanning, sensor data, legacy floor plans, and GIS information to create predictive 3D models of existing buildings. David Benjamin, Autodesk's Director of AEC Industry Futures, describes the approach as triangulating "multiple data sets that are limited instead of one data set that is comprehensive," producing inferences about hidden building elements such as pipe positions, material conditions, and structural health from imperfect but overlapping information. It can estimate whether a building's core placement and floor plate depth are conversion-compatible without a full measured survey, collapsing months of preliminary investigation into a computational exercise.

Gensler developed its own screening tool, ConversionScore, which evaluates office buildings across eight to ten criteria including floor plate depth, structural grid spacing, window-to-core distance, ceiling height, and facade condition. Buildings receive a green, yellow, or red rating, allowing portfolio owners to triage hundreds of properties before commissioning a single feasibility study. CBRE and other major brokerages have built similar analytics platforms to screen their managed portfolios for conversion candidates at scale.

How Deep Is Too Deep

Nobody publishes this table, so we built it. A geometric model estimating the percentage of gross floor area that remains usable for residential conversion at various floor plate depths, assuming a center-core layout with two exposures and a maximum habitable room depth of 30 feet from the exterior wall.

Inputs: core width assumed at 20 percent of floor plate, corridor width at 5 feet, habitable room maximum depth 30 feet from exterior glass. Calculation: total floor plate area minus core minus corridor minus any zone beyond 30 feet from the nearest window equals usable residential area. At 70 feet, you lose approximately 14 percent of your gross area to dead zones adjacent to the core. At 80 feet, you lose a quarter. At 100 feet, 40 percent of the floor is unbuildable for residential, and no unit plan, no matter how inventive, can place a legal bedroom in the dark interior belt that wraps around the mechanical core like a moat.

AI screening automates this calculation across entire building portfolios by ingesting publicly available data: tax assessor records for gross floor area, GIS footprints for building dimensions, permit records for structural modifications, and satellite imagery for facade orientation and solar exposure. A tool running these inputs can flag the 30 percent of buildings worth investigating further and reject the 70 percent that fail on first principles, saving developers the cumulative cost of commissioning feasibility studies on buildings that were never going to work.

What the Plumbing Tells You After the Geometry Passes

Even buildings that clear the floor plate screening face a second barrier. Office buildings concentrate plumbing in the core: a cluster of restrooms near the elevators, fed by a small number of vertical risers running through structural slabs. Residential conversion requires distributing plumbing across every unit. Each apartment needs its own kitchen and bathroom connections, and running new waste lines through reinforced concrete floor plates costs $40 to $80 per square foot. Each new plumbing riser requires cutting through structural slabs that were designed to carry office live loads, not to be perforated after the fact. Some developers use above-floor pump systems to avoid cutting concrete entirely, routing waste upward to a macerating pump that connects to an existing drain line. It works, but adds mechanical complexity, maintenance, and noise to every unit.

Strongest Counterargument

A gifted architect can make buildings work that an algorithm would reject. Light wells cut through the center of a deep floor plate create interior courtyards that bring daylight to units the geometry says are unbuildable. Double-height spaces carved from two floors of a deep-plate building produce loft apartments with windows at the perimeter and dramatic volume compensating for horizontal distance from the glass. Japanese micro-apartment design has placed habitable rooms 40 feet or more from windows for decades, using borrowed light, translucent partitions, and spatial compression that Western building codes do not contemplate. An AI screening tool that scores buildings red based on floor plate depth alone might reject the very buildings where creative design intervention could produce the most architecturally interesting housing, and where the cost savings from distressed acquisition make ambitious design affordable.

AI screening also risks creating a self-fulfilling prophecy: if every developer and every lender uses the same algorithmic criteria to filter conversion candidates, the 30 percent of buildings that score green will attract competitive bidding that erases the acquisition discount that made conversion financially viable, while the 70 percent scored red will languish regardless of whether a more creative approach could unlock them. Algorithmic consensus narrows the field, and narrowed fields produce inflated prices.

If You Are Evaluating a Conversion

Before hiring an architect, before commissioning a feasibility study, measure the floor plate depth from the exterior wall to the structural core. If it exceeds 65 feet in a center-core building with two exposures, the conversion will require either cutting light wells through the structure, losing significant floor area to corridors, or accepting that a substantial percentage of your gross square footage will become storage, corridors, or amenity space rather than revenue-generating residential units. Do this measurement first. It costs nothing and eliminates the majority of unsuitable candidates.

If the geometry passes, check the plumbing. Count the existing vertical risers and estimate the number of residential units per floor. Each riser typically serves four to six units. If your target density requires more risers than the existing structure provides, budget $40 to $80 per square foot for new plumbing penetrations through structural slabs, and add four to six weeks to the construction timeline per floor for the concrete cutting and structural reinforcement work.

Run the acquisition math against conversion costs of $300 to $400 per square foot, the range Vanbarton Group reports for recent New York conversions, versus $500 to $600 per square foot for ground-up residential construction. Conversion is cheaper per square foot. But only if the geometry cooperates.

At 80 feet of depth, the math zeroes out. You are paying the same per usable square foot as building new, with fewer design options and a structural grid you cannot move. AI screening catches this arithmetic before you spend a year discovering it.

Brookings found that without public policy intervention, most conversions are economically unviable even when the geometry cooperates, because conversion costs plus acquisition still exceed the achievable rents in many downtown markets. AI screening cannot fix that arithmetic, but it can prevent you from spending six months and $100,000 learning that the arithmetic was impossible from day one.

Limitations

Industry estimates that 30 percent of office stock is geometrically convertible come from practitioner consensus and Gensler's screening work, not from a peer-reviewed nationwide survey of building geometries. No such survey exists. Conversion cost data relies heavily on New York City projects, where labor costs, regulatory requirements, and building typologies differ substantially from Sunbelt and Midwest markets. RentCafe's 90,300-unit projection includes planned and pipeline units alongside completed conversions, and historical completion rates for announced conversion projects are not publicly tracked. Autodesk and Arcadis's X-Ray Vision initiative is experimental and not commercially available as of this writing. Our floor plate depth analysis uses simplified geometric assumptions, including uniform core placement and rectangular floor plans, that do not account for the irregular geometries of many real office buildings, particularly pre-war structures with setbacks, L-shaped floor plates, or non-central cores. Post-occupancy satisfaction data comparing converted office apartments to purpose-built residential units does not exist in any published research we could locate.

← Back to AI Home Building