California has become the first state where more than half of autism services are delivered in clinical settings rather than in homes. It’s a threshold that seemed distant just a few years ago—and one that raises questions about what gets gained, and what gets lost, as the industry’s center of gravity shifts.

The milestone, captured in CentralReach’s 2025 Autism and IDD Market Report, isn’t an anomaly. It’s the leading edge of a national transformation. Since 2020, multidisciplinary care models—ABA integrated with speech, occupational therapy, and mental health services—have grown at a 30% compound annual rate. States like Virginia, Alabama, and New Mexico posted year-over-year growth in multidisciplinary services ranging from 163% to 368%. Clinic-based care is no longer a niche preference. It’s becoming the default.

The forces driving the shift are practical, if not always patient-centered. Clinics offer operational efficiency: predictable schedules, centralized supervision, easier coordination across disciplines. Staffing is simpler when therapists aren’t crisscrossing cities between homes. And families, particularly those juggling work schedules and multiple children, often prefer the structure of a dedicated facility over the logistical puzzle of in-home visits.

But efficiency has tradeoffs. In-home therapy meets children in their natural environment—where the behaviors actually occur, where parents can observe and participate in real time. For families in rural areas, or those without reliable transportation, a clinic thirty miles away might as well be on the moon. The question isn’t whether clinic-based care works—it does—but whether the industry is building toward a model that inadvertently prices out or excludes the families who need the most flexibility.

The access gap no one’s solving

The CentralReach data reveals a parallel problem: authorized hours remain dramatically underutilized. Overall utilization hovers between 35% and 44%, depending on provider size. Caregiver training—widely considered one of the most impactful components of ABA—saw only 20% of authorized hours used in early 2025, down from 26% the year before. Roughly 608,000 caregiver training hours go unused each month across the sector.

This isn’t a supply issue. The hours are approved. The services exist. Something in the system—scheduling friction, staffing constraints, family availability, or simple operational neglect—is leaving value on the table. And the irony is sharp: organizations that do prioritize caregiver training see up to 35% lower client churn. The ROI is there. The execution isn’t.

AI enters the infrastructure

Meanwhile, technology is quietly becoming load-bearing. Seventy-five percent of organizations purchased AI-enabled tools in 2025, and the use cases have moved well beyond novelty. Providers are deploying automation for note auditing, claims review, scheduling optimization, and compliance monitoring. What started as experimentation has calcified into expectation.

The promise is efficiency at scale—fewer hours lost to administrative friction, faster identification of quality issues, more consistent documentation. The risk is that AI becomes a crutch for understaffed organizations rather than a complement to clinical judgment. The technology can surface patterns, but it can’t replace the therapist watching a child’s face.

Quality is the balance sheet

If there was one theme Molko kept circling back to, it was clinical integrity. Families vote with their feet; when quality slips, revenue follows. Serious buyers now diligence training systems, outcomes posture, and whether programs actually align when you roll in acquisitions. If a buyer isn’t asking hard questions about clinical quality early in the process, that’s a red flag for sellers—and a landmine for buyers post-close.

What California’s plan does—and doesn’t—address

California, for its part, has responded with a 10-year Master Plan for Developmental Services, released in March 2025. The document includes 167 recommendations spanning access, workforce development, and system modernization. It calls for replacing aging case management software, ending sub-minimum wages, and simplifying wait times—all worthy goals.

But the plan is a framework, not a solution. It establishes workgroups and listening sessions. It commits to revisiting progress each March. What it doesn’t do is answer the harder questions: How do you maintain access for rural families as clinics concentrate in population centers? How do you ensure multidisciplinary integration doesn’t become a billing strategy rather than a clinical one? How do you keep caregiver training from being the first thing cut when schedules get tight?

Where this leaves the industry

The structural shift underway isn’t inherently good or bad—it’s a response to real constraints. Clinics make sense for many families. Multidisciplinary models can deliver better-coordinated care. AI can free clinicians to focus on what matters.

But the industry is scaling faster than it’s reckoning with the gaps. Utilization rates suggest a system that’s better at authorizing care than delivering it. The migration to clinics risks creating access deserts for the families least equipped to adapt. And the workforce challenges—turnover among RBTs remains endemic—aren’t solved by floor plans or software.

California crossing 50% clinic-based is a milestone. Whether it’s a model worth following depends on what the state, and the industry, does next.