A funding boom that looks like a new era - and a familiar warning
Another week, another billion-dollar stack aimed at AI and healthtech. US startups raised more than $1.5 billion between November 24-30, led by Physical Intelligence ($600M), Kalshi ($300M), and Function Health ($298M). Most of that money is flowing into AI-driven platforms - and healthcare is right in the center of it.
For healthcare leaders, this isn't abstract. Capital is concentrating in vendors that want to automate revenue, reduce admin burden, and support clinical decision-making. The upside is real. So is the exposure if you back the wrong partner or buy hype instead of outcomes.
What's new: speed and size
- Valuations are moving in months, not years. Multiple AI startups have doubled or tripled their private valuations within a single quarter.
- At least 49 US AI companies raised $100M+ rounds in 2025, with many returning for second or third mega-rounds.
- Cursor (Anysphere) is the poster child: from a ~$9.9B valuation in June to ~$29.3B in November, with reports of $1B+ in annualized revenue.
- Healthcare AI is a standout. Investors highlighted 18 "most promising" health startups this month, and AI-led companies captured the majority of digital health funding in H1 2025.
Where healthcare AI is gaining traction
- Revenue cycle, payments, and data plumbing: Akasa, Inbox Health, RapidClaims, Stedi, Courier Health, PayZen. These tools create cash-on-cash returns by reducing denials, speeding collections, and improving claims accuracy.
- Workforce, hiring, and training: Carefam, Clasp, Stepful. Focused on staffing gaps, faster onboarding, and targeted upskilling with AI coaching and screening.
- New care models and diagnostics: Cadence (chronic care at scale), Knownwell (obesity/metabolic), Diana Health (women's care), Teal Health (at-home cervical screening), Twin Health (metabolic "digital twins"), Nudge (ultrasound-based neuro interfaces), Wellsheet (EHR prioritization), Infinitus (voice agents for prior auths). These are not pilots on the sidelines - they are shipping into real workflows.
Why this cycle feels different - and riskier
- Some vendors have real revenue. A subset is reporting nine- and even ten-figure run rates with enterprise adoption.
- Capital intensity is off the charts. Model training and inference capacity require billions in GPUs, data centers, and energy. Risk is spilling into credit markets through partner financing structures tied to AI build-outs.
- Capital is concentrated. Fewer companies, bigger checks. That lowers noise, but raises systemic exposure if a "winner" stumbles.
Bottom line: the opportunity is massive, and so are the assumptions priced into it - on performance, margins, and regulatory stability.
What healthcare leaders should do right now
- Anchor on use cases with hard ROI. RCM automation, clinical documentation, prior auths, and patient access typically pay for themselves faster than net-new clinical tools.
- Demand verifiable outcomes. Ask for site-level baselines, matched-control studies, and clinician-level metrics (time saved per note, denial reduction, days in A/R, throughput).
- Treat model quality as a living system. Require drift monitoring, bias testing, and ongoing calibration. No "set it and forget it."
- Line up governance. Establish an AI review board with compliance, privacy, clinical leadership, and IT. Approve claims, datasets, and integration patterns before go-live.
- Use contracts to de-risk. Insist on audit logs, uptime SLAs, PHI encryption defaults, BAA, indemnity for model errors, and exit/portability clauses for your data and fine-tuned models.
- Budget beyond licenses. Include implementation, change management, EHR build, GPU/compute pass-through, and support. Many "cheap" tools shift costs to your teams.
- Stay FDA-aware. If a product influences clinical decisions, confirm regulatory pathway, intended use, and change-control processes for model updates.
A quick vendor checklist
- Clear clinical and operational claims? Documented impact at systems like yours?
- Evidence package: study design, sample size, confidence intervals, and external validity?
- Data sources and permissions: training, fine-tuning, and your PHI segregation?
- Security: encryption in transit/at rest, SSO, RBAC, audit trails, breach response in hours, not days?
- EHR integration: native support, write-back behavior, versioning, rollback plan?
- Model lifecycle: drift detection, monitoring dashboards, and human-in-the-loop controls?
- Regulatory status: FDA classification (if applicable), QMS, and change management?
- Commercials: all-in TCO, compute surcharges, SLA credits, and termination rights with data export?
The next 90 days: a focused plan
- Pick two high-leverage pilots (e.g., AI scribe + RCM automation). Define success with 3-5 measurable KPIs each.
- Stand up an AI governance huddle that meets biweekly. Keep decisions visible and fast.
- Map PHI data flows for each vendor. Close gaps before deployment.
- Negotiate model update controls and rollback rights. Require release notes with measurable changes.
- Train the frontline: 60-minute playbooks for clinicians and revenue teams. Track adoption and feedback weekly.
- Publish a one-page "AI safety and use" guide for staff. Simple rules beat dense PDFs.
- Set quarterly vendor reviews: ROI check, incident log review, and equity/algorithmic bias audit.
Why the funding surge matters to your budget
Big rounds can buy stability, faster product delivery, and better support. They can also push vendors to grow faster than they can execute. Your protection is evidence, integration maturity, and contract discipline.
Expect more secondary rounds, higher valuations, and pressure to sign multi-year deals tied to compute. Move fast on pilots, slow on lock-ins.
Where to skill up your team
If your clinicians and ops leads need practical AI fluency (prompting, safety, evaluation), a short course can pay for itself in weeks. See curated options by job role here: AI courses by job.
The bottom line
AI funding is breaking records. Healthcare is getting a big share - from revenue cycle to bedside support. The upside is meaningful: fewer clicks, faster claims, better access, and more time with patients.
The risk isn't hype. The risk is buying speed without guardrails. Keep your bar on evidence, integration, security, and contracts. If the current crop of vendors converts capital into real outcomes, 2025 won't look like 2021 - it will look like progress you can measure on your balance sheet and at the bedside.
Your membership also unlocks:
