The a16z Show: Healthcare 2026 — AI Doctors, GLP-1s, and Insurance Defection
Podcast: The a16z Show
Episode Date: January 27, 2026
Host: Jay Rugani (a16z Health & Bio Partner)
Guest: Nikhil Krishnan (Founder, Out Of Pocket)
Episode Overview
In this episode, Jay Rugani sits down with Nikhil Krishnan to dissect the rapidly shifting tectonics of U.S. healthcare. The discussion zeroes in on three major trends: defection from traditional health insurance, the consumerization and technology-driven transformation of care (from AI doctors to diagnostics and GLP-1 drugs), and the regulatory, economic, and ethical ripples of these shifts. The episode is an energetic, witty debate over predictions for 2026—and a candid analysis of both opportunities and perils.
Key Discussion Points & Insights
1. Insurance Defection and a Fragmenting System
-
The Surge in the Uninsured ([04:29]-[08:17])
- Prediction: Uninsured rates will spike, possibly reaching 15%—undoing progress made since the ACA era.
- Drivers:
- Skyrocketing premiums, high deductibles, and low perceived value for the healthy.
- Employers opting out, especially small, healthier groups, to avoid subsidizing sicker pools.
- Cultural shift: Regular Reddit threads debating whether to forgo insurance for cost reasons.
“If I were really rational about this, and I’m a relatively healthy person, no expected procedures, and my premiums are $550 a month with a $5–6k deductible... you just do the math. Now suddenly you’re paying $10–15 grand before even anything is covered.”
—Nikhil ([07:13]) -
The Pros and Cons of Defection ([08:22]-[11:04])
- Cons: People may misjudge risk and face catastrophic costs; fragments risk pools; could create tiers of care.
- Pros: Greater consumer agency, price transparency, and pressure for providers to improve experience and efficiency.
"If you believe more in consumer-directed healthcare... people get money and use it to pay for services they want. If people feel the costs, they’ll be better shoppers.”
—Nikhil ([09:42]) -
Parallel Systems Emerging ([14:19]-[23:40])
- High-income “proactive health” crowd layering cash-pay diagnostics/wellness on top of insurance.
- Lower-income and newly uninsured facing access, price, and transparency hurdles.
- Novel middle-ground: Health sharing ministries—crowdfunding, community-vetted alternatives to insurance.
“I don’t want our healthcare system to splinter into one group that can afford to defect, run their own stack, and those relying on charity care cycling through the system.”
—Jay ([19:59])
2. The Cash-Pay Boom, Diagnostics, and Proactive Care
-
Product Opportunities in a Cash-Pay World ([24:07]-[27:08])
- Care navigation, AI triage, hyper-low-cost services (e.g., $4 AI prescription refills in Utah).
- Local “medical tourism” pricing—bundled, transparent, upfront rates for procedures.
-
Diagnostics & At-home Testing Go Mainstream ([27:27]-[33:05])
- Skyrocketing consumer demand for monitoring, screening, and self-ordered diagnostics as proactive health gains favor.
- Cash-pay, direct-to-consumer diagnostics companies blossom, especially for those who don’t fit standard care guidelines.
“Diagnostics is one of the areas that’s very underinvested. A lot of people just want to be monitored—even if it says nothing... the appetite is really high.”
—Nikhil ([27:48])- Regulatory Lag: Existing system aims for median, guideline-driven care, but consumers increasingly want personal agency.
“People just want to feel like they’re in control, and getting a test done... is just going to be a part of healthcare going forward.”
—Nikhil ([32:24])
3. Wearables, AI, and Regulatory Battles
-
The Rise of LLMs and Wearable Data ([33:34]-[40:42])
- OpenAI, Anthropic, and others report hundred-million user engagement with health-related LLMs (ChatGPT, etc.).
- Wearables (Oura, Whoop) and digital tools provide real-time health signals but face a regulatory gray zone.
- Regulatory gap: “Wellness” vs “medical device”—need for a hybrid “digital health screener” category.
“FDA either says it’s a wellness product (no claims) or a medical device (go through years of approval). There’s not that many in between, but there should be.”
—Jay ([35:18])- Need for real-world data (RWD): Wearable-generated “ground truth” for subtyping conditions like diabetes, sleep disorders, etc. Untapped patient-driven “biobanks.”
4. AI in Clinical Workflows, State Policy, and Regulation
-
States vs. Feds: Patchwork AI Regulation ([41:16]-[47:19])
- States running pilots (AI prescribing in Utah), others banning AI therapy/chatbots; federal government promising a national framework.
- Arguments for letting states experiment, then “speedrun” best practices nationwide if proven highly effective.
“If we can do something like... this pilot looks so good, we really just gotta boost it everywhere. That could be a better federal approach.”
—Nikhil ([46:37]) -
Healthcare AI IP and Licensing Wars ([47:55]-[53:18])
- Licensing and copyright fights coming for foundational medical content—disease scales, CPT codes, clinical protocols, and even influencer "protocol IP."
- Net-new data sets created by new tools—wearables, AI scribes, patient-uploaded records—will matter more than legacy claims data.
“Protocols are IP... The Peter Attia longevity protocol—make it available with an AI agent. That’s very interesting.”
—Nikhil ([50:22])
5. Populist Backlash & Ethical Tensions in AI Adoption
-
Three Major Populist Fears ([55:25]-[66:11])
-
Jobs: AI will destroy healthcare employment (23M jobs at risk)
- Counter: U.S. healthcare is “super supply constrained”—admin jobs may shrink but productive patient-facing/community roles may grow.
“Everyone always complains about the administrative bloat... We should want to decrease that!”
—Nikhil ([59:20]) -
Mistakes: AI will harm or kill patients via errors
- Counter: All systems (human or AI) make errors; AI can raise the median care bar, and errors are often more transparent and fixable at scale.
“AI is going to make a mistake... but how many net people did it help? The median AI implementation will raise the bar.”
—Nikhil ([61:12]) -
Autonomy: AI removes patient choice, or dehumanizes interactions
- Counter: System already suffers from bureaucracy; AI might provide faster answers and more standardization.
- Higher-quality AI bots still needed; current phone AI reps are “bad.”
“If you can get answers faster, I’d argue it’s better. No one wants to wait to hear if prior auth went through.”
—Nikhil ([65:31])
-
6. Drug Market Disruption: GLP-1s and Peptides
-
GLP-1 Drugs: Mainstream Breakthrough and Beyond ([67:05]-[74:41])
- Fast-falling prices, oral formulations, direct-to-consumer digital channels, and sky-high demand (“infinite demand”).
- Use will likely surpass 30% of U.S. adults.
- Experimental use in mental health and inflammatory conditions—less promise for addiction or behavioral health (so far).
“It might be like the next statin—or just put it in the water supply.”
—Jay ([71:08]) -
Peptides & the Gray Market ([74:47]-[83:00])
- Unregulated peptides procured from online sources and private parties—used for pain, anti-aging, etc.
- FDA predicted to crack down, push for more accredited compounding pharmacies, but faces ideological “freedom vs. public health” dilemma.
7. U.S. Healthcare Economic Futures
-
Will Tech Slow Healthcare Spending Growth? ([85:56]-[89:57])
- Nikhil: Sees a path to “deflationary or flat spend” via AI, GLP-1s, policy shifts.
- Jay: Expects U.S. will simply spend more on (improving) healthcare as GDP and prosperity rise—health and education absorb slack from decreased costs elsewhere.
“Are we going to be spending more on health care? I think way more, but we’ll just get a lot more for it.”
—Jay ([87:38]) -
Healthcare as the Bedrock for the Next Economic Phase ([88:32]-[89:27])
- Nikhil speculates on a new "community/care economy"—as AI automates white collar and service work, healthcare provides the rails for paid caregiver/friendship roles to absorb displaced labor.
Notable Quotes & Memorable Moments
-
On the Reddit insurance exodus:
“If you go on the health insurance subreddit, extremely common question now... should I just not get health insurance at all?... And for a lot of people that answer is actually probably no.”
—Nikhil ([00:00]) -
On the fragmentation dilemma:
“Choice is like the worst way. It’s choosing between the worst options now.”
—Nikhil ([21:30]) -
On doctor-patient tension:
“The existing healthcare system wants standardized care... but people want more agency in their healthcare. No one wants to be told ‘just wait and see.’”
—Nikhil ([01:02], [32:24]) -
GLP-1s Zeitgeist:
“Everyone wants to not be—like—obese, right? Near infinite demand here.”
—Nikhil ([68:10])
“It might be like the next statin... just put it in the water supply.”
—Jay ([71:08]) -
On AI slop in health content:
“I’m just not having fun on the internet anymore. AI has really distorted how content is created. We’re all in on [real life] events now.”
—Nikhil ([84:39])
Segment Timestamps
- Insurance Defection, Why Now? – [04:29]-[11:04]
- Cash Pay, Parallel Systems, & Health Sharing – [14:19]-[23:40]
- Startups: Opportunity in Navigation & Triage – [24:07]-[27:08]
- Diagnostics, Screening, & Proactive Health – [27:27]-[33:05]
- Wearables, LLMs, & Regulatory Challenges – [33:34]-[40:42]
- State vs. Federal Regulation in AI Health – [41:16]-[47:19]
- Healthcare AI: IP, New Data, Protocols – [47:55]-[53:18]
- Populist Backlash: Jobs, Errors, Autonomy – [55:25]-[66:11]
- GLP-1s, Drugs as Consumer Products – [67:05]-[74:41]
- Peptide Gray Market & FDA Regulation – [74:47]-[83:00]
- U.S. Healthcare Economics: Predicting the Future – [85:56]-[89:57]
- Content & Learning: Combatting AI Slop – [90:54]-[92:54]
Additional Insights
- The “Peter Attia Protocol” as IP: Individual doctors/influencers creating branded, possibly licensed protocols implemented via AI.
- Net-new biobanks: Untapped opportunity for startups to collect and leverage patient-uploaded real-world data for research and personalized medicine.
- The Event-ification of Healthcare Content: Digital content quality declining; shift toward in-person hackathons, workshops (Out of Pocket events) for genuine learning and engagement.
Closing Thoughts
Both Jay and Nikhil agree: U.S. healthcare is experiencing a once-in-a-generation inflection. The confluence of consumer-out migration from insurance, GLP-1s and experimental drugs, and the juggernaut of AI is blowing up old models, creating new opportunities—and new risks—for innovators, clinicians, regulators, and patients alike.
“People working in technology are coming into healthcare... there’s cool stuff happening. Come learn about it, come make an impact.”
—Jay ([90:25])
Summary compiled in the energetic, candid, and insightful style of the original discussion.