Video · 2026-07-17 · 1h 27m · 6 moments

You’re About to See the Real AI Winners Stand Up

✦ AI generated

timeline · colored by role

Prediction

The asset management industry is in the early stages of a third major relationship shift, where investing platforms — including AI-assisted financial assistants — are increasingly owning the client relationship, following earlier eras when asset managers themselves and then financial advisors owned it.

Jonathan Thomas traces asset management's relationship ownership through three eras — direct-to-consumer, advisor-led, and now an emerging platform-led model — and warns the shift could hurt clients because platforms lack the trust needed for the behavioral side of investing.

transcript

Jonathan Thomas: So first the asset manager owned the relationship, then the advisor owned the relationship and we had to pivot to support the adviser. And now increasingly you see platforms owning those relationships whether retirement platforms or AI assisted financial assistance and so it's just shifting.

Data◆

Avantis's first five ETFs, launched in 2019, have outperformed their benchmarks by an average of 280 basis points annually while charging only about 7 basis points more in fees than comparable passive index funds.

Thomas cites concrete performance data showing Avantis's low-cost, near-passive-fee active ETFs have delivered substantial benchmark-beating returns since 2019, explaining much of the brand's rapid growth to $150 billion in assets.

transcript

Jonathan Thomas: The average fee on those is 26 pips. The average looks more like passive than active. And in fact, if you look at their passive equivalent indexes, the average of that fee is 19. So you're paying seven more bips for the opportunity to outperform. And for the products that we launched, those first five, on average, we've outperformed the bench by 280 bips.

Claim

Nvidia's stock is not in a bubble because its valuation multiple has been compressing even as its earnings and guidance keep beating expectations, unlike a true bubble where the multiple would expand dramatically.

Josh Brown argues that Nvidia's falling P/E multiple despite strong earnings growth signals the opposite of a bubble — the market is discounting future competition rather than irrationally inflating the stock.

transcript

Josh Brown: This is not a bubble. This would be a bubble if Nvidia's multiple had gone from 30 to 90 times earnings. Didn't happen. This is a discount to the market... it's a huge discount.

extends · 1

Mechanism◆

The real risk in AI infrastructure spending isn't hyperscalers cutting capex — it's the circular financing web of Neoclouds, GPU makers backstopping purchases, and private-equity-backed data center buildouts, which could unravel and leave many funding commitments to vanish when the cycle turns.

Josh Brown lays out how GPU makers like Nvidia effectively backstop purchases from unproven 'Neocloud' data-center startups funded by sovereign wealth and private equity, warning many of these interlocking commitments could evaporate when the cycle turns — a dynamic Thomas confirms is the real mechanism behind an 'earnings bubble.'

transcript

Josh Brown: Take that example times a thousand and everybody's lending and/or equity investing in everyone else's projects and the concern is when the music stops all these people go for the chairs and a lot of people going to be left without a chair because some of those funding commitments are going to vanish as though they were a vapor.

extends · 1

Mechanism◆

AI investments should be evaluated according to which of five layers of Jensen Huang's 'AI cake' framework a company occupies — base infrastructure/energy, chips, data center infrastructure, models, and applications — since capital intensity and defensibility differ sharply by layer.

Jonathan Thomas explains Nvidia CEO Jensen Huang's 'AI cake' analogy — five layers from raw infrastructure up to consumer apps — as the right framework for judging whether a given AI-exposed stock's market position is sustainable, expandable, and defensible.

transcript

Jonathan Thomas: There's the base level which is really the ground, the cooling, the electricity. You've got the chip level, right? The Nvidia of the world. You got the infrastructure level which are the data centers themselves. You have the model level which are the LLMs and then sitting on top of all of that is the app level where the actual value's created.

explains mechanism · 2

Fact

The Stowers Institute deliberately studies why certain animals evolved to avoid diseases like cancer, diabetes, and osteoporosis, and publishes both its successful discoveries and its failed hypotheses for free, so other researchers don't waste resources repeating failed experiments.

Jonathan Thomas describes the Stowers Institute's unusual research model — focusing on animals that evolved resistance to diseases humans suffer from, and openly publishing failed hypotheses alongside successes so the broader scientific community doesn't duplicate wasted effort.

transcript

Jonathan Thomas: Almost everybody who makes a discovery loves to tell the world about it and publish a paper, but we also talk about our failures. If we have a hypothesis that doesn't pan out, we actually publish that as well... we don't want the world to waste resources replicating a hypothesis that we've tested and learned that it did not work.

Highlight slides

Avantis ETFs Beat Benchmarks at Near-Passive Cost✦ from: Avantis's first five ETFs, launched in 2019, have outperformed their benchmarks by an average of 280 basis points annually while charging only about 7 basis points more in fees than comparable passive index funds.Fees: Active Pricing, Passive-Like Cost✦ from: Avantis's first five ETFs, launched in 2019, have outperformed their benchmarks by an average of 280 basis points annually while charging only about 7 basis points more in fees than comparable passive index funds.Result: Rapid Asset Growth✦ from: Avantis's first five ETFs, launched in 2019, have outperformed their benchmarks by an average of 280 basis points annually while charging only about 7 basis points more in fees than comparable passive index funds.The Real AI Infra Risk Isn't Hyperscaler Capex Cuts✦ from: The real risk in AI infrastructure spending isn't hyperscalers cutting capex — it's the circular financing web of Neoclouds, GPU makers backstopping purchases, and private-equity-backed data center buildouts, which could unravel and leave many funding commitments to vanish when the cycle turns.When the Music Stops✦ from: The real risk in AI infrastructure spending isn't hyperscalers cutting capex — it's the circular financing web of Neoclouds, GPU makers backstopping purchases, and private-equity-backed data center buildouts, which could unravel and leave many funding commitments to vanish when the cycle turns.The 'AI Cake': Five Layers of AI Investment✦ from: AI investments should be evaluated according to which of five layers of Jensen Huang's 'AI cake' framework a company occupies — base infrastructure/energy, chips, data center infrastructure, models, and applications — since capital intensity and defensibility differ sharply by layer.Where Value Gets Created✦ from: AI investments should be evaluated according to which of five layers of Jensen Huang's 'AI cake' framework a company occupies — base infrastructure/energy, chips, data center infrastructure, models, and applications — since capital intensity and defensibility differ sharply by layer.

Related episodes

Audio · We Study Billionaires

TIP825: Meta, Adobe, Booking Holdings w/ Stig Brodersen, Tobias Carlisle & Hari Ramachandra

→

Video · All-In Podcast

OpenAI vs Anthropic IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts

→

Audio · All-In Podcast

Elon's Anthropic Deal, The Next AI Monopoly?, "FDA for AI" Panic, Trading the AI Boom

→