Four Acres and a License: What Actually Decides a Senior Housing Deal

• 

  Three-storey senior living building with landscaped courtyard and accessible entry

Occupancy is at 89.9 percent and has risen for twenty consecutive quarters. Inventory is growing at four tenths of one percent. Fewer than 16,000 units are under construction in the entire country. This should be the easiest development case in American real estate, and it is one of the hardest.

Start with the fundamentals, because they are almost absurd.

In the second quarter of 2026, senior housing occupancy across the 31 primary markets tracked by NIC MAP reached 89.9 percent, up four tenths of a point from the prior quarter and marking the twentieth consecutive quarterly increase. Fifteen of those 31 markets sat at or above 90 percent. Boston hit 93.3 percent. San Francisco reached 92.7. Baltimore 91.8. Even the weakest markets, Miami at 86.2 and Atlanta at 86.5, would be considered healthy in most other asset classes.

Against that, supply did essentially nothing. Year over year inventory growth ran 0.4 percent. Occupied units totaled 639,650, an increase of roughly 3,700 for the quarter. Fewer than 16,000 units were under construction nationally, and NIC's Lisa McCracken observed that high occupancy paired with static construction simply means fewer housing options for older adults. Quarterly construction starts had already fallen to 1,076 units in the primary markets during the first quarter of 2025, the lowest reading since the second quarter of 2009.

Now the demand side. The population aged 65 and over reached 61.2 million in 2024, up 3.1 percent in a single year, and is projected to grow from 58 million in 2022 to 82 million by 2050. The first baby boomers turned 80 in 2026, which matters because 80 and above, not 65 and above, is the cohort that actually fills assisted living. The national median rate for assisted living reached 6,200 dollars a month in the 2025 CareScout cost of care survey, a five percent increase, drawn from more than 25,000 rate collections at the metro level.

Rising rates, rising demand, near-record occupancy, and a pipeline that has effectively stopped.

There are two reasons for that gap, and neither of them is capital. Both are decided on a plan set, months before anyone breaks ground.

You build 889 square feet to rent 490

Here is the arithmetic that has quietly moved against developers, and it is not on most pro formas.

CBRE's 2026 senior housing development cost study, covering 36 projects between the third quarter of 2023 and the second quarter of 2026, found total development cost at 388,830 dollars per revenue unit, or 364 dollars per square foot. That is a 23.6 percent increase since 2023, when the same measure sat near 317,400 dollars per unit. Hard costs made up 72.5 percent of the total, soft costs 16.2 percent, and site acquisition 8.1 percent, with furniture, fixtures and equipment adding 11,900 dollars per unit and site costs running 16,000 to 36,600 dollars per unit.

Those numbers get quoted. This next one does not.

Net rentable area has fallen to 55.5 percent of gross building area, down from 59.1 percent in 2023.

Read that as a building rather than a ratio. An assisted living apartment runs roughly 320 to 650 net square feet for single occupancy with a kitchenette. A memory care studio runs 260 to 500. But the gross area required per unit, once you add corridors, dining rooms, activity space, commercial kitchen, mechanical, storage and staff areas, lands somewhere between 740 and 930 gross square feet. CBRE's own figures imply about 889 gross square feet per unit for a typical 90-unit building of roughly 80,000 gross square feet.

So on a 490 square foot apartment you are constructing roughly 889 square feet of building. Nearly half of everything you build, heat, insure and maintain generates no rent directly.

And that ratio got worse by 3.6 percentage points in three years. Some of that is code, some is competitive amenity creep, some is the shift toward higher-acuity residents who need more clinical and common space. Whatever the cause, the practical effect is that a developer using a 2022 efficiency assumption will underbuild their gross area by roughly six percent and discover it during design development, when the fix is expensive.

The site planning consequence follows directly. Building footprint, parking and setback all scale off gross area, not off rentable units. A 90-unit building at 59 percent efficiency and the same building at 55.5 percent are different buildings on the site plan, with different massing, different parking counts and potentially different height. Size the parcel from gross square footage and current efficiency, not from unit count times a remembered ratio.

Memory care wants to be single-story, which is a land problem

The second structural fact compounds the first.

The prevailing design pattern in this asset class, described by David Dillard of D2 Architecture, pairs a multi-story assisted living building with a single-story memory care building attached to it. Single-story memory care wins on three grounds. Some states permit single-story wood frame construction, which he estimated saves ten to twenty dollars per square foot. Lower walls let more sunlight into secured courtyards. And it keeps assisted living residents from looking directly at what he called their personal future.

All three arguments are sound. All three cost land.

A standalone single-story memory care building typically needs one to two acres. In San Ramon, California, a 54-bed memory care facility of roughly 22,691 gross square feet was approved on a 1.53 acre lot. Assisted living, which can go vertical, commonly plans against two to four acres, though real projects range enormously: multi-story infill has been built on parcels under two acres, while campus scale continuing care communities consume far more. One Erickson community in Roseville, California carries roughly 1,200 independent living units plus 200 assisted living and memory care beds on 55.6 acres.

Memory care also carries physical requirements that assisted living does not, and several are set by state license rather than by local code. Washington requires memory care facilities to provide common areas including at least one resident-accessible outdoor area, paired with a plan to prevent wandering and exit-seeking, and prohibits eliminating outdoor space that already exists on a floor. Florida requires new assisted living facilities and additions above 16 licensed beds to comply with the Facility Guidelines Institute standards for residential health and care facilities as referenced in the state building code. Missouri sets community living and dining minimums of 25 square feet per resident for facilities licensed after 1987 and caps rooms at four beds.

Then there is the household model. Memory care is increasingly organized into households of ten to twenty residents sharing dining and living space, because smaller groupings reduce noise and agitation. That is good clinical design and it is inefficient floor plate design, since each household needs its own dining and living area. It is another draw on the net-to-gross ratio.

Add it up and memory care is the most land-hungry and least efficient product in senior

housing, requiring the most secure and specialized site, at exactly the moment when large parcels adjacent to residential neighborhoods are the hardest thing in America to entitle.

Which brings us to the license, and then to the hearing.

The license nobody models

Senior housing carries a permitting track that most commercial developers never encounter, running through a state health or social services agency and entirely separate from local zoning.

The useful cross-state reference is the National Center for Assisted Living's annual state regulatory review, which documents each state's licensing agency and category names. It is a regulatory summary rather than a bed database, but it is the fastest way to learn what a given state actually licenses.

What matters for anyone trying to measure a market is that some states publish their licensed capacity and some do not. This is genuinely uneven, and it determines whether you can build a beds-per-capita analysis for free or whether you are filing records requests.

State	Agency and registry	Capacity published	Format
California	Dept. of Social Services, Community Care Licensing; RCFE registry	Yes, licensed capacity	Public search plus downloadable CSV
Florida	Agency for Health Care Administration, ALF locator	Yes, with bed-type detail	Searchable with download function
Oregon	Dept. of Human Services, Licensed Long-Term Care Settings Search	Yes	Search, sort, filter and export
Texas	Health and Human Services Commission, LTC Provider Search	Yes, Type A and Type B	Searchable, no account required
Minnesota	Dept. of Health, provider directory	Yes, licensed beds by type	Downloadable file
Massachusetts	Dept. of Public Health facility lists	Yes, monthly	Downloadable
Kentucky	Div. of Health Care certified ALC list	Units per facility	County-organized PDF, not searchable
Missouri	Dept. of Health and Senior Services	Aggregate only	Summary counts

California alone licenses more than 7,800 residential care facilities for the elderly serving over 210,000 residents, published at address level with capacity. That is a market study you can build in an afternoon.

Two warnings before anyone builds a national map from this.

The first is that memory care is frequently not a separate licensed category. Many states license assisted living as one thing and treat dementia care as an endorsement or add-on to that license. So a state registry may show you total licensed beds while telling you nothing about how many of them are secured memory care beds, which is the segment with the tightest supply. Massachusetts adds a related wrinkle: its nursing home bed totals may include beds that are out of service.

The second is the certificate of need question, which is widely misunderstood in this sector.

Thirty-five states and the District of Columbia maintain certificate of need programs, but they overwhelmingly target hospital beds, nursing homes and major medical equipment. Conventional assisted living is generally not subject to certificate of need. Kentucky states the exemption explicitly in its regulations. New York is the meaningful exception, requiring a certificate of need from the state health department to operate an Assisted Living Program.

Where certificate of need does bite is at the skilled nursing edge. If a program contemplates a nursing wing, Minnesota's statutory moratorium on new nursing home beds and Nebraska's moratorium under its certificate of need act become hard constraints, and the practical answer is usually to drop the skilled component or acquire transferable beds rather than to fight for new ones.

The word in the zoning code that decides the deal

Now the constraint that actually stops projects.

Senior housing is most often approved by conditional use permit or special use permit rather than by right, which means a public hearing and discretionary findings. That is normal enough for a large building. What is less obvious, and more consequential, is how the code classifies the use.

Assisted living is frequently classified as institutional or healthcare rather than residential. Somerville, Massachusetts places assisted living facilities under an institutional housing principal use requiring a special permit with suitability findings. Under the International Building Code, larger assisted living and memory care buildings fall into Institutional occupancy I-1 Condition 2, which carries its own requirements including twenty-minute rated self-closing doors.

That classification is not a technicality. It determines which districts the use is even eligible for, whether the application goes to a staff-level review or to a board, and, critically, what the neighbors think they are being asked to approve. A building described as residential reads differently at a public hearing than a building described as an institution, even when the drawings are identical.

The denial record shows what happens next.

In Roswell, Georgia, the planning commission denied and the city council repeatedly deferred a rezoning for a roughly 275,000 square foot independent, assisted living and memory care project housing around 200 residents on 13 acres. The application conflicted with the comprehensive plan, and it sat on a corridor where three previous senior housing proposals had already been turned down.

In Edison, New Jersey, the zoning board unanimously rejected a 122-bed memory care proposal in a single-family zone. That denial came after a federal court ruling in March 2024 had already found that an earlier denial violated federal law and discriminated against elderly and disabled residents. The applicant is pursuing roughly 40 million dollars in damages. Read that sequence carefully, because it establishes the outer edge of entitlement risk in this asset class: a board can deny a senior housing application even under an active federal finding of discrimination, and the developer's remedy is years of litigation rather than a building.

In Northfield, Illinois, more than 300 neighbors petitioned against a 44-unit conversion comprising 33 assisted living and 11 memory care units. In the Black Mountain Ranch area of San Diego, residents petitioned against a 74-bed assisted living and memory care facility that required both a land use plan amendment and a conditional use permit.

Approvals do happen. Clermont, Florida approved an 8.44 acre independent, assisted living and memory care project on a three to two vote over vocal traffic-based opposition. But a three to two vote is not a business plan.

The pattern across all of these is that the demographic argument, which is overwhelming, loses at the hearing to traffic, scale and neighborhood character, which are local and immediate. Demand does not get you entitled.

What actually gets approved

Three things move the odds, and all of them are decided at the site planning stage.

Parking, correctly sized. Senior housing generates far less parking demand than its square footage suggests, and codes are increasingly catching up. The Institute of Transportation Engineers reports parking demand for assisted living and memory care between 0.24 and 0.74 spaces per dwelling unit, with an eighty-fifth percentile design rate of 0.58. Jurisdictional requirements vary widely around that. Baltimore County requires one space per three beds for small assisted living facilities. Ewing Township in New Jersey uses the same ratio. Harlingen, Texas requires two spaces per living unit, which is roughly three times observed demand. New York City eliminated off-street parking minimums for senior and subsidized housing within its transit zone, requires none in the Manhattan Core, and lowered minimums outside the transit zone to ten percent for new affordable senior housing.

The gap between 0.58 observed and 2.0 required is the difference between a site that works and one that does not, and it is worth a parking demand study early rather than a variance request late.

Density bonuses, where they exist. California's state density bonus law names senior housing explicitly, and AB 2345 raised the maximum bonus to 50 percent while reducing required parking. Los Angeles layers on more: senior housing setting aside ten percent of units for lower income or five percent for very low income seniors qualifies for an additional 15 percent bonus, up to 35 percent total, and the city maintains a dedicated eldercare facility ordinance allowing the use across a wide range of zones subject to zoning administrator approval.

State preemption, where it has arrived. Arizona's SB 1473 declares the occupancy and zoning regulation of assisted living to be a matter of statewide concern, preempts inconsistent municipal ordinances, and bars municipalities from restricting assisted living in residentially zoned areas based on resident count. That is the single most developer-favorable statutory framework identified in this research, and it is worth watching whether other states follow, because it would materially change where this product can be built.

The order of operations

For anyone underwriting a senior housing site, the sequence that matters is not the sequence most pro formas follow.

Confirm the classification first. Find out whether the local code treats assisted living as residential, institutional or healthcare, and whether it is permitted by right or requires a discretionary permit. That single answer sets your timeline, your risk and your probability of approval more than any market datapoint.

Check the corridor's history second. Roswell's application failed partly because three prior senior housing proposals on the same road had already been denied. A record of denials in a submarket is a leading indicator, and it is free to research.

Then size the parcel from gross area at current efficiency. Two to four acres remains a reasonable planning default for a stacked assisted living building, and one to two acres for single-story memory care, but the honest number comes from gross square footage at 55.5 percent net rentable, plus parking at the ratio the code actually requires rather than the ratio demand justifies, plus the secured courtyard the state license will insist on.

Then check the license. In California, Florida, Oregon, Texas, Minnesota and Massachusetts you can pull licensed capacity yourself and build a beds-per-capita picture for the trade area in a day. Elsewhere, budget time for a records request, and assume memory care capacity will not be separable from assisted living in the state's numbers.

Only after all of that does the demand case matter, and it will be strong, because it is strong nearly everywhere.

The gap will not close by itself

Twenty consecutive quarters of occupancy gains against four tenths of a percent inventory growth is not a market signal that developers are ignoring. It is a market signal that developers are receiving clearly and cannot act on.

The obstacles are specific and they are all upstream of construction. Development cost rose almost a quarter in three years while the share of that building you can actually rent fell by nearly four points. The most needed product, secured memory care, is the most land-hungry and the least efficient to build. The license is granted by an agency that never talks to the zoning board. And in most American jurisdictions the use is classified as an institution and sent to a hearing where the demographic case is not the argument being had.

None of that gets solved by a better rent comp. It gets solved on a site plan that anticipates the classification, sizes to the real efficiency, parks to the real demand, and puts the courtyard where the state inspector will want it.

The first boomers turned 80 this year. There are eighteen years of that cohort behind them, and fewer than 16,000 units under construction to meet it. The developers who solve the entitlement problem will not need to worry about the demand one.

Sources:

• Supply and demandNIC MAP senior housing occupancy, inventory growth, units under construction and construction starts.

• US Census Bureau population estimates.

• Population Reference Bureau projections.

• CareScout and Genworth Cost of Care Survey for national median rates.

• CBRE Senior Housing Development Costs, 2026 edition, for per-unit cost, cost composition and net rentable area as a share of gross.

• American Seniors Housing Association and Weitz for construction cost by product tier.