Senior Housing Development in 2026: Site Criteria, Care Levels, and Where AI Fits

The most operationally complex residential asset class requires a development process to match, and AI is changing where that process starts.

Senior housing is not one product type. It is four or five, stacked together in various combinations, each with different regulatory requirements, staffing models, cost structures, and market demand curves. Institutional developers who treat it as a single underwriting exercise tend to get it wrong.

The 2024-2025 market delivered a useful correction. A wave of assisted living and memory care construction that began in 2021 came online into a market still recovering from occupancy losses. NIC MAP data (National Investment Center for Seniors Housing & Care) tracked national assisted living occupancy at roughly 84% in mid-2024, below the 87% pre-pandemic baseline. Meanwhile, the 80-plus population is projected to grow from approximately 13 million today to over 20 million by 2035.

The setup for a development cycle is clear. The question is what gets built, where, and whether development teams have the tools to underwrite it properly.

The Care Level Stack

Senior housing is typically segmented into four care levels:

• Independent living (IL): No care services included. Age-restricted apartments for active seniors, often with amenity packages. Least regulated, closest to conventional multifamily underwriting.

• Assisted living (AL): ADL support including bathing, dressing and medication management. Licensed by state. Staffing ratios and physical plant requirements vary significantly by jurisdiction.

• Memory care (MC): Secured environment for residents with dementia. Higher staffing ratios, specialized design requirements, separate licensure in most states.

• Skilled nursing (SNF): Post-acute and long-term medical care. Medicare/Medicaid reimbursement complexity. Certificate of Need restrictions apply in roughly 35 states.

Most institutional development focuses on IL and AL, often combined in a continuing care retirement community (CCRC) structure. Memory care is increasingly added as a wing or a later phase. SNF development is nearly dormant given reimbursement pressure.

Site Criteria: What Actually Matters

Demographics

Standard senior housing site analysis starts with a three- to five-mile primary trade area and an age-income overlay. The target cohort is typically adults 75-plus with household income above $50,000. Market-rate AL and IL in primary and secondary markets generally require 300 to 500 income-qualified seniors per planned unit within the primary trade area.

NIC MAP's penetration rate data, the ratio of senior housing units to the age-qualified population, is the standard benchmark. Markets below the national average of approximately 8% penetration are undersupplied relative to their demographics.

Accessibility and Visibility

Senior housing residents and their families place significant weight on drive-time accessibility. Sites on arterial roads with strong visibility outperform infill sites tucked behind other development. Proximity to retail, medical offices, and hospitals matters for family visit frequency and for operator credibility in the local market.

Physical Plant Requirements

AL facilities typically require minimum parcel sizes of 2 to 4 acres depending on density and parking structure. Single-story construction is preferred for memory care. Floor plans must accommodate wide corridors, grab bar mounting reinforcement, and centralized nurse stations. HVAC, electrical, and plumbing systems are over-specified relative to conventional multifamily.

Construction costs for AL range from $250 to $400 per square foot depending on market and care level mix, with memory care at the higher end of that range.

Regulatory Complexity

State licensure is the single biggest development risk that developers underestimate. AL licensing in states like New York, California, and Florida can involve certificate of need review, inspections, and approval timelines running 18 to 24 months before a facility opens. Developers entering a new state for the first time consistently underestimate this exposure.

Where AI Fits in the Development Workflow

Site Screening

AI can run demographic trade area analysis at scale, pulling age-income overlays, calculating penetration rates against NIC MAP benchmarks, and scoring candidate sites against operator-specified thresholds. What used to take an analyst two to three days per site now runs across a full candidate set in hours.

Regulatory Mapping

State licensing requirements, CON applicability, and zoning compatibility can be parsed and summarized by AI across jurisdictions. The outputs still require legal review, but AI screens out non-starters early and flags jurisdiction-specific risks before a team commits capital to a site.

Pro Forma and Sensitivity Analysis

Senior housing pro formas are more complex than conventional residential because revenue is a function of care level mix, occupancy ramp, and, for some operators, Medicaid reimbursement rates. AI can model multiple care mix scenarios, run occupancy ramp sensitivities, and stress-test the capital stack against construction cost variance.

The assumptions still come from the developer and operator. AI assembles them faster and makes the sensitivity range visible earlier in the process.

What Remains Judgment-Dependent

Operator selection is not an AI workflow. The performance of a senior housing asset is overwhelmingly a function of its operator. Site, design, and capital structure matter, but a well-located AL facility with a poor operator will underperform a modestly located facility with a strong one.

Due diligence on operators, reference checking on occupancy ramp assumptions, and evaluating care quality track records are human processes. AI can surface data. The judgment call on who to partner with is not automatable.

The Cycle Ahead

The 85-plus population growth curve is not a forecast that can be revised. The demand is demographically locked in. What moves is the development community's ability to underwrite it correctly, site it efficiently, and build it without the cost overruns that eroded returns in the last cycle.

AI does not solve the hard problems in senior housing development. It compresses the analysis time on the solvable ones.