23 Enterprise Launch Customers, 0 FQHCs: The AI Lab Gap Nobody Is Talking About

The pattern is consistent across all three labs

Anthropic's launch roster: AstraZeneca, Sanofi, Genmab, Banner Health, Stanford Healthcare, Novo Nordisk, AbbVie, Veeva, Flatiron Health. Six pharma companies, two academic medical centers, one health-tech vendor. Banner Health's deployment is impressive — 22,000+ clinical providers, BannerWise private chatbot, 85% of users report working faster. The 8-hour-per-patient pre-visit chart summarization Banner runs for oncology patients now takes minutes. The deployment proves Claude works at enterprise scale. The roster doesn't include a single Federally Qualified Health Center.

OpenAI's launch roster: AdventHealth, Baylor Scott & White Health, Boston Children's Hospital, Cedars-Sinai Medical Center, HCA Healthcare, Memorial Sloan Kettering Cancer Center, Stanford Medicine Children's Health, UCSF. Eight large hospital systems and academic medical centers. Plus a separate consumer product, ChatGPT Health, that connects personal medical records via b.well's 2.2 million-provider partnership. FQHC patients access ChatGPT Health indirectly through b.well — but FQHCs themselves have no enterprise tier deployment.

Google Cloud's launch roster: Humana, CVS Health (Health100), Highmark Health (Sidekick), Waystar (AltitudeAI), Quest Diagnostics, WellSky (20,000+ post-acute / community care sites). Four payers, one diagnostics company, one post-acute network. Highmark Sidekick grew from 1M to 6M+ prompts in 2025, delivering $27.9M in AI-enabled value. Waystar AltitudeAI prevented $15B+ in denied claims. The WellSky partnership covers 20,000+ care sites — the closest analog to OCHIN's 2,200+ FQHC site network. The architecture is built for consortium-distributed AI deployment. The FQHC consortium partnership doesn't exist.

The clinical leadership picture is uneven

Different labs have built clinical bench depth in very different ways. This matters for FQHCs evaluating which lab actually understands clinical workflow.

Google Cloud has the deepest formal clinical leadership: a Chief Health Officer (Karen DeSalvo MD MPH MSc, retiring summer 2025), a Chief Clinical Officer (Michael Howell MD, succeeding DeSalvo), and a PhD-led Health AI division (Greg Corrado PhD, who co-founded Google Brain). Plus the Med-PaLM legacy of peer-reviewed clinical AI papers. This is the most credentialed clinical bench across the three labs.

OpenAI's clinical leadership is anchored on Nate Gross MD — co-founder of Doximity, the largest US physician network. Practicing-physician background. Plus 'hundreds of physician advisors' informed ChatGPT for Clinicians, which is FREE for individual physicians, NPs, and pharmacists. That's the differentiator: any FQHC clinician can pilot ChatGPT for Clinicians personally without procurement. No other AI lab offers a free clinician tier.

Anthropic has the thinnest publicly named clinical bench. CEO Dario Amodei holds a PhD in physics; co-founders are technologists and policy professionals. There is no publicly named Chief Medical Officer. Anthropic builds healthcare on top of the general Claude API rather than deploying a healthcare-specific clinical leadership function. This isn't necessarily disqualifying — Banner Health's 22,000-clinical-provider deployment proves Claude scales — but it is a notable gap relative to OpenAI and Google.

Wardly AI, the early-stage pre-visit intake startup, names two clinical advisors / customers publicly: Dr. Prince Shah MD FACP FACG (founder of Gilroy Gastroenterology, California) and Adam Tilson PA-C (Director of Adult Medicine at Aizer Health). Academic logos displayed include UC Berkeley, Thomas Jefferson University, and Penn State. For an early-stage company, this is meaningful clinical bench depth — but no MD/NP/PhD is on the founding team.

Why the FQHC absence matters strategically

First, FQHC populations are not represented in the launch-customer alignment work. Foundation model AI trained predominantly on academic medical center data and pharma research data underperforms on safety-net panels — Spanish-first, multi-comorbidity, fragmented prior records from migration patterns, low health literacy. The same equity question that applies to Epic Cosmos / CoMET training data applies to all three AI lab healthcare products. The launch customers shape what 'works' looks like.

Second, FQHCs absorb the patient-facing externality without partnership benefit. ChatGPT Health connects FQHC patients (via b.well) to AI-mediated health navigation. FQHC providers will increasingly encounter 'I asked ChatGPT and it said...' visits. Anthropic's HealthEx partnership creates the same dynamic for Claude Pro users. FQHCs handle the operational consequences — patient confusion, AI hallucinations, expectation management — without an enterprise relationship that would let them shape patient-facing AI behavior.

Third, the partnership architecture exists. WellSky's 20,000+ post-acute site partnership with Google Cloud is functionally equivalent to what OCHIN's 2,200+ site FQHC network could be with any of the three labs. The technology integration model is proven. The bargaining-leverage scale is there. What's missing is the consortium-broker willing to pull the partnership across the line.

The verified case study outcomes are striking

Anthropic Claude case studies (verified): Banner Health 22,000+ clinical providers, 85% report working faster, 1,400+ oncology notes processed, 8-hour-to-minutes chart summarization. Stanford Health Care built ChatEHR on Claude. Elation Health reports a 61% drop in documentation time. Carta Healthcare reports 66% reduction in clinical data processing time with 99% accuracy.

OpenAI ChatGPT case studies (verified): launch customers AdventHealth (51 hospitals), HCA Healthcare (188 hospitals — largest of the launch roster), UCSF, Memorial Sloan Kettering. Specific outcome metrics from these enterprise pilots have not yet been publicly disclosed at the depth Anthropic / Google have published.

Google Cloud case studies (verified): Highmark Sidekick — 1M to 6M+ prompts, $27.9M AI-enabled value 2025. Waystar AltitudeAI — $15B+ in denied claims prevented, 90% reduction in appeal/documentation workflows. Humana Agent Assist — Gemini Enterprise CX background support. Merck — $1B+ multi-year agentic AI deal. WellSky — Gemini multimodal models across 20,000+ care sites.

Wardly AI case studies (verified): Gilroy Gastroenterology (Dr. Prince Shah MD founder testimonial — pre-visit AI calls capture symptoms + prior evaluation history before GI visits), Aizer Health (Adam Tilson PA-C testimonial — adult medicine pre-visit intake). Specific metrics not publicly disclosed at the depth larger competitors have published.

What FQHC executives should do this quarter

Three concrete moves before your next executive team meeting.

• Send a partnership inquiry to all three AI labs. Anthropic does not name a Chief Medical Officer publicly — escalate to your Anthropic enterprise account manager or via the Claude for Healthcare contact form. For OpenAI, escalate to Nate Gross's team via the ChatGPT for Healthcare page. For Google Cloud, engage Michael Howell MD's incoming Chief Health Officer team. Use the 6-question OCHIN account manager letter as a template — adapt it for AI lab outreach.
• Pilot ChatGPT for Clinicians personally. It's free for verified physicians, NPs, and pharmacists. Have 3-5 of your providers try it for 30 days on documentation, medical research, and clinical task assistance. Compare against Claude Pro ($20/mo) and Gemini Advanced ($20/mo) for the same providers. The cheapest, most reversible test you can run.
• Verify your Google Workspace contract. Many FQHCs are already on Google Workspace for back-office. The Gemini Enterprise tier may already be available with HIPAA BAA — check with your Google Cloud contact before clinical use. If you have it, you may already have AI access most FQHCs are paying separately for. If you don't, the upgrade economics may surprise you.

The bigger move: NACHC or OCHIN as broker

Individual FQHCs have low bargaining leverage with $850-900B AI labs. NACHC has 1,400+ member health centers. OCHIN has 44,000+ providers across 2,200+ sites. Either organization could broker an FQHC consortium partnership with one or more of the AI labs that would benefit the entire safety-net sector — analogous to WellSky's 20,000-site partnership with Google Cloud or Abridge's FQHC cohort report.

The negotiation asks should include: (1) FQHC-tier pricing analogous to NACHC Select bundled programs, (2) Spanish-language parity at launch (not staged) for any patient-facing AI, (3) FQHC patient population representation in alignment / training data, (4) a published clinical advisor cohort that includes safety-net providers (not just academic medical centers), and (5) free or deeply discounted pilots for the 6-12 month proof-of-value window each lab needs to demonstrate community-health outcomes.

The opportunity is short-lived. Once each lab has its 50 named enterprise customers across hospital systems and pharma, the FQHC market becomes an afterthought — accessible only via SMB API contracts at AMC-tier volume pricing. The window is now, while each lab is hungry to demonstrate clinical impact at meaningful scale. NACHC's P&I Forum, OCHIN's annual meeting, and CHCF's strategic priorities for 2026-2028 all converge on this. Someone has to make the move.