Autistic adulthood as infrastructure: employment, housing, and supported autonomy

When an autistic person turns 21 — or ages out of the school system, whichever comes first in their jurisdiction — the infrastructure disappears. Not thins out. Not transitions. Disappears. The IEP that organized their educational supports, the developmental pediatrician who tracked their care, the school-based therapist, the aide who knew their sensory profile, the structured day with predictable demands — all of it ends. What replaces it is a landscape so sparse and fragmented that calling it a “system” is charitable. This is the transition cliff, and the word “transition” is doing dishonest work, because a transition implies continuity. What actually happens is a discontinuity. Autistic adults don’t transition into a thinner version of the pediatric system. They fall out of one and land in nothing much.

The autistic-led research community has named autistic adulthood as a top priority. In what autistic people actually want researched, I covered AutINSAR’s community priorities, where autistic adulthood — employment, housing, quality of life, aging — consistently ranks at or near the top. The researchers and advocates closest to this work are not treating adulthood as a “life stage” topic, the way a developmental psychology textbook might. They are treating it as an infrastructure problem. That framing is correct, and most of the policy world hasn’t caught up to it.

Here is the gap I want to name: the entire body of “transition” writing and programming treats autistic adulthood as a life-stage program to graduate into. You complete a transition plan, you exit school, you enter a program. The program ends. Adulthood continues. Nobody is treating autistic adulthood as an infrastructure build where autonomy is the design goal. The difference matters. A program is something you finish. Infrastructure is something you build and maintain, and it either exists or it doesn’t.

The transition cliff is an infrastructure cliff

The pediatric autism system — for all its real failures and inequities — is at least a system. There are protocols, funding streams, dedicated professionals, institutional homes. Developmental pediatrics, special education law, early intervention, school-based therapy, case management tied to school enrollment: these are infrastructure, unevenly distributed and often inadequate, but present. They form a scaffolding around the autistic child and their family.

At adulthood, the scaffolding comes down. Adult autism services, where they exist at all, are means-tested, siloed by funding source, gated by diagnosis and eligibility determinations that favor people diagnosed in childhood with specific support need profiles, and organized around crisis response rather than sustained support. Developmental pediatricians don’t typically see adults. The adult mental health system is built for different referral questions and often doesn’t know what to do with autistic adults who aren’t in acute psychiatric crisis. Vocational rehabilitation services exist but are underfunded and time-limited. Housing waitlists are years long. Case management, if it exists, is patchy.

For autistic adults diagnosed late — and this includes a large cohort of people who are now being identified in their thirties, forties, fifties — there was never a pediatric scaffold to begin with. They hit the same sparse landscape with no prior infrastructure, no accumulated support history, and often a trail of misdiagnoses and wrong accommodations behind them.

The result is not a transition. It is an infrastructure cliff. The pediatric side had infrastructure; the adult side largely doesn’t. And “transition programming” that spends a few years preparing an autistic young person to navigate a world without infrastructure is not a substitute for building the infrastructure.

Autonomy is the design goal

If you reframe autistic adulthood as an infrastructure problem, the design goal becomes clear: autonomy. Not independence in the libertarian sense — the ableist fantasy of doing everything alone — but autonomy in the sense of agency over one’s own life, with the supports that make agency possible. Autonomy means having work that is real, housing that is yours and in your community, and legal agency over your own decisions. It means the built environment — employment programs, housing stock and support staff, legal frameworks — is designed to support your agency rather than substitute for it.

This is where the infrastructure framing does real work. A transition program asks: is this person ready for adulthood? An infrastructure framing asks: is the environment ready for this adult? The first question locates the problem in the individual. The second locates it in the system. The second question is the right one.

Supported employment: place-then-train

The dominant disability-employment model for autistic adults is still the train-then-place approach: pre-vocational training, sheltered workshops, segregated settings, repetitive tasks at subminimum wage, with the theoretical promise of eventual placement in competitive employment that often doesn’t materialize. This model sorts autistic people into tracks — and as I argued in what neurodiversity at work programs get wrong, the sorting logic itself is the problem. It assumes autistic people need to be prepared, remediated, and contained before they can be allowed into the real labor market, and it produces a parallel employment world that is segregated, low-wage, and low-expectation.

Individual Placement and Support (IPS) inverts this. Originally developed for people with serious mental illness, IPS is a place-then-train model: the person is placed in competitive employment first, and support is provided on the job, ongoingly, by an employment specialist who works with both the employee and the employer. The core principles are competitive employment as the goal, rapid job search rather than prolonged pre-vocational preparation, integration of employment support with clinical care, and zero exclusion — nobody is deemed too disabled to try.

The fit with autistic adulthood is strong. IPS assumes the person can work and builds the support around that assumption rather than gating entry behind readiness assessments. It normalizes real employment rather than segregating. It provides ongoing support rather than time-limited job coaching that ends at an arbitrary milestone.

I need to be honest about the evidence. IPS has a robust evidence base in serious mental illness populations — this is one of the most well-supported supported-employment models in the psychiatric rehabilitation literature. Its adaptation to autistic adults is promising but less established. The studies that exist are smaller, fewer, and more recent. The core principles transfer well conceptually, but autism-specific IPS implementation — including how to handle sensory accommodations, social communication differences, and the particular patterns of autistic burnout in employment — is still being worked out. I am arguing that IPS-style supported employment is the better model and that the evidence supports taking it seriously for autistic adults, not that the autism-specific evidence base is settled. It isn’t. That’s a reason to build it, not to wait.

Housing: in the community, with the supports actually funded

Housing for autistic adults spans a spectrum: independent living with supports (a person lives in their own home or apartment with services brought in), supported living (a person lives in their own place with more intensive on-site support), intentional communities and farmstead models (designed communities with shared supports), and congregate care (institutional or facility-based settings with high staffing).

The principle that should govern this spectrum is community integration. The Olmstead decision established that disabled people have the right to receive services in the most integrated setting appropriate to their needs — and that unjustified segregation is discrimination under the Americans with Disabilities Act. That principle should govern how we build autistic adult housing: community integration is the default, not the exception.

But I want to be honest here, because honesty is what separates advocacy from salesmanship. Some autistic people with high support needs — including people with co-occurring intellectual disability, complex medical needs, or self-injurious behavior that requires constant supervision — may genuinely need intensive or congregate models. “Independent living with supports” is not a universal answer, and pretending it is makes the rest of the argument less credible, not more. There are autistic adults for whom a small, well-run congregate setting with trained staff is the most appropriate environment, and the Olmstead principle itself says “most integrated setting appropriate” — appropriate is doing real work in that phrase.

The problem is not that congregate settings exist. The problem is that they are the only option in most regions, and that they function as the default rather than as one point on a spectrum. The spectrum should run from independent living with drop-in supports, to shared housing with live-in or rotating staff, to small congregate settings for those who need them — with community integration as the organizing principle at every level. And the second problem — the one that gets less attention — is that “independent living with supports” only works if the supports actually exist and are funded. A housing model without funded supports is just isolation with a lease. An autistic adult in their own apartment with no staff, no check-ins, and no crisis line is not “independent”; they are unsupported, and the difference matters enormously.

Housing is infrastructure: the units, the support staff, the funding. You have to build all three. A unit without staff is warehousing in a nicer package. Staff without stable funding is a program that evaporates at the next budget cycle. And funding without units is a voucher with nowhere to go. The infrastructure framing forces you to ask whether all three layers are in place — because if any one is missing, the system fails.

Supported decision-making vs guardianship

The same framing applies to legal infrastructure, and specifically to the question of who makes decisions in an autistic adult’s life.

Guardianship, in its most common form, strips an autistic adult of legal agency. A parent, guardian, or the state is empowered to make decisions for them — about money, healthcare, where to live, who to see. In many jurisdictions, guardianship is still the default at age 18: if no one objects and no alternative is in place, the court appoints a guardian, and the autistic person’s legal capacity is removed, often wholesale and often permanently.

Supported decision-making (SDM) is the alternative framework. It keeps legal agency with the autistic adult and builds a support network — trusted people, formal or informal — who help the person understand options, weigh tradeoffs, and communicate their choice. The person retains the legal right to decide; the supports make the deciding possible. This is the autonomy-as-design-goal principle applied to law: the question is not “can this person decide alone?” but “can this person decide with the right supports?”

The default presumption at age 18 should be capacity with support — SDM, not guardianship. Guardianship should be the last resort, reserved for situations where the person genuinely cannot decide even with robust supports, and even then, it should be tailored and limited rather than plenary. A person who needs help managing complex financial decisions may not need help deciding what to eat or who to visit. Full guardianship for a person who needs partial support is overreach dressed as protection.

I want to be honest again. SDM is not a panacea. There are situations — acute psychiatric crisis, severe cognitive impairment, exploitation risk — where substituted decision-making is necessary and where SDM, applied rigidly, could leave someone unprotected. The framework works when the support network is real, trained, and accountable. It fails when “supported decision-making” becomes a label pasted over a system that provides no actual supports — the same trap as “independent living” with no staff. But the default and the effort should go to preserving agency, not stripping it by routine. Guardianship at 18 as a paperwork default is a failure of legal infrastructure, not a protection.

The infrastructure gap is the point

Here is the structural pattern across all three domains. Employment, housing, decision-making: each is an infrastructure problem. The autistic adult’s autonomy depends on whether supported-employment programs exist and are funded, whether community housing with real supports exists and is funded, and whether SDM legal frameworks exist and are used. The “transition cliff” that families experience at age 22 — the sudden disappearance of the pediatric system’s scaffolding — is really an infrastructure cliff. The pediatric system had infrastructure: IEPs, school-based services, dedicated staff, legal entitlements. The adult system has aspirations and waiting lists. Building autistic adulthood means building the infrastructure — not running a “transition program” that ends at 22 and calls the rest living.

The long-haul build

This is a long-haul build. It is partly policy: Olmstead enforcement, Medicaid-funded supported employment, SDM statutes. It is partly funding: the supports have to be paid for, and not by families alone. It is partly community: the employers, landlords, neighbors, and support networks that make integration real. The autistic-led research priority — adulthood as a top-tier concern — was correct, and the reason is structural. Autistic adulthood is the underbuilt layer. The pediatric layer has decades of law, funding, and professional infrastructure behind it. The adult layer has a cliff.

Treat autonomy as the design goal and build the supports under it. That is the entire thesis.

Listen first

One final honesty, because it matters. I am a parent of an autistic child. I can see the structural pattern, and I can argue for infrastructure, and I can read the research. But I am not an autistic adult, and autistic adults are the authorities on their own adulthood — what they need, what failed them, what worked. This essay frames the pattern; it does not narrate the experience. If you are building anything — a program, a policy, a housing development, a legal framework — listen to autistic adult advocates and communities before you build. They have been describing the infrastructure gap for years. The gap is not their failure to adapt. It is ours to close.