At the AIF Investment Roundtable, a panel of investors and operators argued that the split between behavioral and physical health was always a billing artifact, and that the most durable behavioral health companies now treat autism, mental health, and medical complexity as one problem.
Key Takeaways
- The standalone category is dissolving. Panelists from Autism Impact Fund, Oak HC/FT, and Andreessen Horowitz argued that behavioral health is no longer an investable silo on its own, because the conditions it treats almost never arrive alone. Capital is shifting toward integrated models that address behavioral, physical, social, and environmental needs together.
- Co-occurrence is the core financial fact. Federal data and operator experience converge on the same point: most children with autism carry additional diagnoses, and most medically complex children carry behavioral ones. Companies built around a single condition increasingly look fragile to investors who have watched the co-morbidities drive cost and utilization.
- Whole-person models are producing durable businesses. Imagine Pediatrics and value-based Medicaid platforms were cited as evidence that integrated care can be profitable and clinically effective at once. The pattern that investors reward is operational depth across constituents, not scale at any cost.
- Technology is the connective tissue, not the thesis. AI and data infrastructure earn their place by coordinating fragmented care and measuring outcomes, the panel said, not by replacing clinicians. The firms positioning for the next decade are funding the plumbing that lets whole-person models actually run.
The pitch that defined a decade of behavioral health investing went something like this: pick a condition, build a provider platform around it, and scale. Autism therapy became the marquee example, with applied behavior analysis (ABA) platforms commanding the highest multiples in the sector and private equity assembling regional providers into multistate networks. That logic is now being openly questioned by some of the people who helped fund it.
At the Autism Impact Fund Investment Roundtable, a webinar panel convened under the title “Where Is Real Innovation Happening in Behavioral Health?” and moderated by Kelly Evans, anchor of CNBC’s “The Exchange” and co-anchor of “Power Lunch,” four investors and operators made a strikingly unified argument: the boundary between behavioral and physical health was never clinically real, and the companies that treat it as real are building on sand. “The bifurcation has always been a billing artifact, not clinical reality,” said Chris Male, co-founder and managing partner of the Autism Impact Fund (AIF). He predicted “the end of behavioral health as a standalone category,” with the most interesting companies treating the whole individual: behavioral, physical, social, and environmental.
It is a notable thing to hear from a fund built by and for the autism community. But Male framed it as the lesson of deploying AIF’s first fund, which closed at $60 million in 2024. The firm reached the first close of its second fund this spring with roughly $150 million in assets under management across 17 portfolio companies and two exits, and Male said the broader mandate reflects what the team kept seeing. “Over 90% of individuals with an autism diagnosis have at least one other co-occurring condition, and as many as 50% have four or more,” he said. The companies creating the most durable value, in his telling, “aren’t autism companies, they’re healthcare companies that happen to be transformative for those with disabilities.”
Co-Occurring Conditions Break the Single-Condition ABA Model
The clinical case for an integrated approach rests on numbers that have only grown harder to ignore. The Centers for Disease Control and Prevention’s most recent surveillance, released in April 2025, identified autism in about 1 in 31 eight-year-olds, up from 1 in 36 two years earlier. Among those with cognitive testing on record, close to 40% also have a co-occurring intellectual disability, and a long tail of medical and psychiatric conditions, including ADHD, anxiety, gastrointestinal disorders, and feeding difficulties, tends to travel with an autism diagnosis. The billing system, by contrast, still treats those conditions as separate line items, a fragmentation that operators say works against the patient.
George Boghos, CEO of Imagine Pediatrics, has watched the same pattern from the opposite direction. His company started as a medical group for children with medical complexity, the kind leaving the NICU dependent on feeding tubes, ventilators, and tracheostomies. “What we found very quickly is 70% of those kids, if not more, have at least one mental or behavioral health comorbidity,” Boghos said. “We weren’t selecting for that.” ADHD, depression, anxiety, suicidal ideation, trauma, and serious mental illness presented alongside the medical conditions, and underneath both sat social factors: food insecurity, housing instability, transportation gaps. “Children that we serve have on average three lifelong chronic conditions,” he said. The question of whether a child can “grow out” of autism, he argued, misses the structure of the problem. “You’ve got to come at it from 10 different angles.”
That structure is what makes single-condition platforms look exposed. A provider optimized to bill one service line can be clinically excellent at that service and still leave most of a patient’s needs unaddressed, which is precisely where cost accumulates. Boghos noted that children with special needs may be one in four of the pediatric population but account for north of half of pediatric spending. When a feeding tube fails at two in the morning and a family ends up in a rural emergency room, the expensive event is rarely the one the provider was contracted to manage.
Why Whole-Person Care Is the More Durable Behavioral Health Business
The panel’s second argument was financial, and it cuts against a common read of the market. Annie Lamont, founder and managing partner of Oak HC/FT, pushed back on the idea that value-based care has stalled. “People think of value-based almost as dead because they think of it all as ChenMed, Oak Street, some of the other examples,” she said, referring to the primary-care-risk model built around Medicare Advantage. “But the reality is some of the most successful value-based care companies have been like CareBridge and Medicaid and Imagine and Medicaid.”
The CareBridge example is instructive. The value-based manager of home- and community-based services for complex Medicaid and dual-eligible members was acquired by Elevance Health’s Carelon division for a reported $2.7 billion, a deal completed in December 2024, after reaching more than 115,000 patients across 17 states and Washington, D.C. It is the kind of exit that has been scarce in pure behavioral health, where Lamont noted that even profitable assets can be hard to sell. An inpatient psychiatric company Oak HC/FT built over a decade is “very effective” and “very profitable,” she said, “but getting out of it, very difficult,” because headline risk around seriously mentally ill populations scares off buyers.
Imagine Pediatrics is the panel’s clearest proof of concept for the integrated alternative. The Nashville-based group delivers 24/7 virtual and in-home care through value-based arrangements, primarily in Medicaid, and expanded into four new states in January 2026 (Georgia, Missouri, North Carolina, and New York) on the heels of a $67 million Series B that included Oak HC/FT and AIF among its backers. Boghos said the company is profitable, maintains a net promoter score above 90 with families, and reduces total healthcare spend for its population by roughly 10% while improving outcomes. “The only thing we take away is time spent in the hospital unnecessarily,” he said.
Male said the lived-experience test AIF applies to deals that tends to surface the same distinction. “Would I put my child in this program?” he asks, noting that his son uses Cortica, an AIF portfolio company. The funds the panel described walking away from were not the unprofitable ones. “We’ve passed on companies that have great unit economics, look good on paper, but the businesses aren’t bringing those constituents together,” Male said. “They’re optimizing around reimbursement, not the family experience.” In behavioral health, he argued, you can build a profitable company “that’s quietly making lives for families harder,” through no-show rates, skills that don’t transfer to real life, and models that cherry-pick away from the most complex cases. The integrated operators, by contrast, “have better retention, better word of mouth and better relationships over time.”
AI in Behavioral Health: Connective Tissue, Not the Thesis
If the thesis is whole-person care, the enabling layer is data and AI, and here the panel was careful to define the role narrowly. “We don’t view AI as a standalone theme,” Male has said of AIF’s approach; the panel treated it as infrastructure for connecting fragmented information and measuring what happens to patients over time. David Haber, a general partner at Andreessen Horowitz who helped open the firm’s New York office and co-leads its AI applications fund, framed his interest as living “between fields of expertise.”
Haber’s investments illustrate the pattern. He was an early backer of Camber, an AI revenue cycle management company that began in the autism community and now runs one of the larger claims databases across ABA, speech-language pathology, occupational therapy, and physical therapy. “People ask, is that a healthcare company, a FinTech company, or an AI company?” he said. “The answer is just yes.” Camber automates claims submission and appeals to move providers’ first-pass collection rates from the mid-80s toward the mid-90s, recovering revenue for services already delivered. He also sits on the board of Tennr, which automates the pre-visit referral and intake work that, left manual, is a leading cause of the denials Camber later has to fight.
Boghos described using the same toolkit to make a multidisciplinary model run. Imagine layers third-party claims, pharmacy, and hospitalization data with ZIP-code-level social data and its own record of patient interactions to build predictive models, including one that flags which children are most likely to land in the hospital for a mental health issue in the next 90 days. He put its accuracy above 80%. The point, he stressed, is not to replace clinicians. “When it’s two in the morning and something happens, you don’t want AI telling you what to do. You want a human being that you trust.” The technology’s job is to tell that human who needs help and when, and to strip away the administrative load, turning an hour of chart review into seconds so the hour can go to the family instead.
That distinction, between technology that displaces care and technology that lets care scale, is where the panel located the next decade of value. Haber argued the best AI companies are not just cost-reduction stories but growth enablers: moving a provider’s collection rate up by 15 points of revenue, or accelerating the working capital that leveraged, private-equity-owned providers need to reinvest. The through line is the same one Male returned to repeatedly. “Using tech and AI to accelerate this, not by replacing care, but giving us the infrastructure to manage and improve outcomes for this complex population, is where I see it going.”
The implication for operators and acquirers is concrete. The market is beginning to distinguish between platforms that scale a billing code and platforms that can coordinate a whole patient, and the panel’s bet is that the second kind will prove both more clinically defensible and more valuable. For a sector that spent a decade building single-condition platforms, that is less a tweak to the thesis than a replacement of it.
