Joe Lonsdale on AI, Defense, and American Optimism
Joe Lonsdale (8VC, Palantir/Addepar/OpenGov co-founder) on Uncapped with Jack Altman. The frame: **'we're taking the 2030s and 2040s and packing them into the next two or three years.'** Three substantive threads. **(1) The AI-acceptance / marketing problem** — *'America does not love AI right now. China is 75-80% positive on AI, the US is ~30%.'* Lonsdale's read: people are scared (jobs), and AI is inheriting social media's reputational poison — *'this is the next consumer experience after social media, of course they're afraid.'* **(2) Healthcare as the killer middle-class application** — *'the number-one thing AI could do for 350 million Americans is make healthcare half-cost, fast.'* He's pushing **healthcare-AI-sandbox legislation** via the Cicero Institute (active in 20 states) to legalise AI-led primary-care workflows (Utah already allows AI re-prescription for chronic-disease patients). **(3) Defense as a live, working proof point** — *'Palantir and Anthropic are being used right now and it's working really well — one of the most accurately fought wars with the least civilian deaths and the most bad guys falling fast.'* On productivity/wealth: the 1870-1900 productivity boom made the median working-class person 2.4x wealthier; he predicts **4-5x over the next 30 years 'if we allow it.'** On building: AI-native founders are *'tripling revenue in a quarter, hitting $1B revenue in year three'* — a different universe from the OpenGov-era grind.
Key points
- **The AI-acceptance gap is a strategic risk, not just sentiment: US ~30% positive vs China 75-80%.** *'America does not love AI right now and not nearly enough people do. They're scared — there's a threat for jobs.'* Lonsdale's diagnosis: AI is inheriting social media's reputational damage (*'social media has had a terrible impact on our democracy... this is the next consumer experience after social media to them, of course they're afraid'*). **Directly parallels [the AI-backlash / Pope-encyclical thread in this week's All-In](/issues/2026-05-31)** — the societal-acceptance constraint is becoming a cross-cohort theme, and a genuine regulatory-risk input to the whole AI trade.
- **Healthcare is the named middle-class killer-app — and the regulatory unlock is specific and trackable.** *'Healthcare should cost half as much and be way better. We need primary care to work with deterministic AI — start a business with one doctor, a bunch of nurses, optimised through AI on proven-safe workflows.'* The vehicle: **healthcare-AI-sandbox legislation via the Cicero Institute (gets things done in 20 states).** Utah already allows AI (not a doctor) to handle chronic-disease re-prescription. **A concrete, state-level catalyst to watch for the AI-healthcare-services cohort** — sandbox-bill passage is the leading indicator.
- **Defense is the live proof point: Palantir + Anthropic in active use, working.** *'Palantir and Anthropic are being used right now and it's working really, really well — one of the most accurately fought wars with the least civilian deaths and the most bad guys falling so fast.'* **Triangulates with [the commercial-tech-into-warfare thesis in this week's ILTB DoD-advisor episode](/issues/2026-05-31)** — two independent defense-insider sources confirming the software-defined-warfare shift, and a direct read-through to the Palantir/Anduril/defense-tech cohort.
- **The productivity → working-class-wealth thesis: 4-5x over 30 years 'if we allow it.'** *'Last time productivity went up a lot over 30 years (1870-1900), the median working-class person was 2.4x wealthier in real terms. I think it's going to be more than that — 4 or 5x — over the next 30 years if we allow us to rise.'* The conditional is the whole bet: *'if we don't make it illegal.'* **The optimistic mirror of the doom narrative** — and a framing worth holding against the [job-loss / AI-washing debate in this week's All-In and 20VC](/issues/2026-05-31).
- **The AI-native founder velocity is real and re-rating what 'good' looks like.** *'We're next to this kid who's tripling revenue this quarter. One of these kids who worked for me has a billion revenue in year three. Compared to OpenGov — 14 years, sold for $1.8 billion, palling through the dirt.'* **A venture-side data point on why the IPO-cohort multiples (debated in [this week's 20VC roundtable](/issues/2026-05-31)) look the way they do** — the underlying growth curves genuinely have changed shape, even if the prices are stretched.
- **The bifurcation risk: AI-resistance as a wealth-inequality driver.** *'You have a lot of people resistant to AI — that sounds like the beginning of wealth inequality.'* Lonsdale's answer is to make the middle-class applications (healthcare, primary care) *'win faster and make sure they're legal'* — i.e. the policy fight over which AI applications are permitted IS the inequality fight. **Connects the societal-acceptance theme to a concrete investable: the states/sectors that legalise AI-led services first capture the productivity dividend.**
Notable quotes
We're taking the 2030s and 2040s and packing them into the next two or three years because so much is happening right now. And that means we're taking the 2050s and going to put them into the end of the decade.
America does not love AI right now. Some surveys show China is like 75 or 80% positive and we're like 30% positive. I think they're just more optimistic about how they could build with it and create wealth with it.
The number one thing AI could do for 350 million people in the US is we can make healthcare like half cost, like really fast. Healthcare should cost half as much and it should be way better.
Palantir and Anthropic, it's working like really, really well. This is one of the most accurately fought wars with least civilian deaths and the most bad guys falling so fast.
Last time productivity went up a lot over 30 years, 1870 to 1900, the median working class person was 2.4 times wealthier. I think it's going to be four or five times wealthier next 30 years if we allow us to rise.
Themes
- AI societal acceptance & regulatory risk
- Healthcare-AI as middle-class application
- Defense-tech proof points
- Productivity & working-class wealth
- American optimism & state-level policy