PodcastExcess ReturnsBarry Ritholtz

AI Productivity Is Real. 83% of Stocks Are 10% Off Highs. Why Does Barry Ritholtz Like Bonds Again?

The discussion weighs AI’s potential to improve productivity against risks from concentrated investment and a buildout that could prove excessive if more efficient chips reduce data-center needs. Speakers also consider practical AI applications in advisory work and price tracking, while one argues that the economy is slightly contracting outside AI and data-center spending. Barry Ritholtz points to weak market breadth and says current Treasury and municipal yields make bonds attractive, while cautioning against reaching for yield. A speaker also compares the potential scale of Anthropic and OpenAI IPOs with the historical total for U.S. tech IPOs.

Artificial IntelligenceAI InfrastructureMarket BreadthFixed IncomeInterest RatesEconomic Growth

Preview · 16 more points in the app

Read the Full Summary

Key Takeaways

Speakers describe AI as more likely to change how people work than to eliminate analyst roles: one says widespread access to similar tools may make any investment edge small and temporary, while Bob Michael’s advice, relayed by a speaker, is that workers may be replaced by people who use AI better. A separate speaker expects another advisory productivity step-change within a year or sooner, though compliance requirements could slow firms’ progress.

Speakers describe AI-assisted client records and accessible meeting notes as ways to improve continuity and make planning more responsive: teams can stay informed about relationships and connect past life events to future funding needs, such as a potential $500,000 need for a child’s first years of medical school.

Barry describes Claude automating two distinct administrative tasks: filling out forms and producing printable PDFs, and reducing the time to format and organize his InstaPaper reading list from hours of assistant work to five minutes. He presents these as personal workflow examples, not general productivity estimates.

Barry questions whether projected AI buildout spending of roughly $4 trillion to $7 trillion will remain justified if more-efficient semiconductors and GPUs reduce data-center needs, potentially leaving expensive facilities underused. Separately, speakers warn that AI’s dominance of capital allocation could deepen if large AI and technology IPOs draw capital from other businesses; one speaker’s $5.2 trillion combined valuation scenario for Anthropic, OpenAI, and SpaceX is conditional on currently discussed valuations and could be substantially lower.

GEO-POLITICS

Dynamic Pricing — Barry says digital shelf displays could let supermarkets change prices for individual shoppers. He says states have already taken action and early lawsuits over this kind of dynamic pricing have emerged.

“The Kroger's and some of the other big supermarkets, they don't want to put prices stuck on a can…”—

TECHNOLOGY

AI Advantage — A speaker says AI has not put analysts out of work or produced the anticipated wave of outperforming black-box funds. Because everyone received similar tools at roughly the same time, any edge from using them well may be small and temporary.

“I mean, there was a lot of hype maybe three years ago when we were really just kind of getting vibed up on LLMs that this was going to put…”—

AI Skills — A speaker relays JPMorgan fixed-income head Bob Michael’s advice that AI skills could distinguish workers as spreadsheet skills once did. He says workers are more likely to be replaced by someone who uses AI better than by AI itself.

“And he gives the same speech to all his interns which says those of you who look to us who who've come before you, the people in my generation who…”—

Infrastructure Risk — Barry questions whether projected AI buildout spending of roughly $4 trillion to $7 trillion will remain justified as semiconductors and GPUs become more efficient. If data-center requirements shrink sharply, he warns, expensive facilities could be left underused.

“And so the question is the half a trillion dollars scaling up to depending on who you believe four 567 trillion in AI buildout is that really A-rated?…”—

ECONOMY

Underlying Contraction — A speaker argues that the economy is slightly contracting when AI capital expenditure and data-center spending are excluded.

“However, back out AI capex and data center spending and the economy is slightly contracting.”—

MARKETS

AI Concentration — A speaker worries that passive-market dynamics and the AI theme have left capital markets heavily exposed to AI. The theme dominates capital allocation and attracts substantial profits and capital expenditure, the speaker says.

“One of the places that I think I'm seeing it the most or feeling it the most. It's just this is the market cap story and the S&P 500…”—

Narrow Breadth — Barry Ritholtz says market breadth was narrow even as the S&P 500 reached all-time highs. In Q3, 83% of its stocks were down 10% from their 52-week highs, and 42% were down 20%. He calls this a warning that fewer companies are participating, not proof that the bull market is over.

“What I do care about is that at all-time highs, we have 83% of the S&P 500 in Q3 down 10% from 52- week highs and about half as many…”—

IPO Concentration — A speaker argues that a wave of large AI and technology IPOs could deepen market concentration and draw capital away from other businesses. IPO valuations can also serve as collateral for borrowing and as anchors for large indexes, amplifying their influence beyond the initial share sale.

“The bigger issue to me is the volume of massive IPOs this market is supposed to absorb, all of which are on that one side, right?…”—

FORECASTS

IPO Valuations — The speaker says that if Anthropic and OpenAI go public at their currently discussed valuations, their combined value with SpaceX would be $5.2 trillion. That would be about 20% more than the $4.1 trillion total for all U.S. tech IPOs from 1980 to 2025. He cautions that the discussed valuations could be substantially lower.

“SpaceX, Anthropic, and Open AI…”—

Meet Click Beta: AI and the stock-picking question

The speakers question whether AI can reliably pick stocks, finding little evidence that AI-branded funds have produced lasting outperformance, while arguing that skilled AI users may gain a temporary work advantage.

AI concentration and market breadth

They describe heavy AI-related market concentration and narrowing S&P 500 breadth, noting that many stocks are well below their highs and that activity outside AI and data-center investment is slightly contracting.

Tech IPO valuations and data center financing

They compare prospective SpaceX, Anthropic, and OpenAI valuations with decades of U.S. tech IPOs, and warn that data-center debt could be risky if efficiency gains reduce future infrastructure needs.

Barry Ritholtz: AI and portfolio rebalancing analysis

Barry Ritholtz explains how a secure data lake and Claude helped his firm identify portfolio rebalancing needs, household tax constraints, and clients requiring advisor conversations far faster than manual analysis.

Can AI improve the human side of financial advice?

The advisors say AI meeting notes can help teams remember clients’ family circumstances and personal concerns, making follow-up more thoughtful and strengthening relationships rather than replacing human judgment.

Better client questions and better prompts

They value clients arriving with AI-informed questions and describe using prompts to guide unsuitable podcast pitches toward better research, helping people ask more specific and useful questions.

The AI imagination gap: collaboration versus search

The speakers argue that AI’s value depends on collaborative iteration, not merely using it as a search engine, and share examples ranging from custom baseball scorecards to completing complex forms.

Dynamic pricing and the consumer-retailer arms race

They discuss AI-enabled price tracking alongside retailers’ dynamic pricing, describing a developing consumer-retailer arms race and potential public backlash against personalized or opaque prices.

Consumer data and the incentives behind AI agents

The group questions whether AI agents will serve consumers when platforms make money from advertising and customer data, and considers privacy, corporate incentives, and the possibility of political scrutiny.

Astronomy, music, and testing your assumptions

Barry describes using AI to test assumptions about the Drake equation and intelligent life, and to explore music and cultural lists by applying explicit criteria rather than relying on popularity rankings.

Automating busywork to make room for curiosity

Barry says automating podcast transcripts and reading-list organization frees time for creative exploration, while warning that AI can also open tempting rabbit holes involving expensive interests.

Creative experiments and choosing what to automate

The speakers share playful AI experiments and argue that people should automate work they dislike while preserving activities they enjoy, such as the strategic choices in a role-playing game.

Why Barry Ritholtz is excited about bonds

Barry says bonds have become attractive again, citing Treasury and municipal yields that can offer strong tax-equivalent returns, while recommending investment-grade quality and cautioning against reaching for yield.

Bonds versus equity income and closing thoughts

Matt agrees that bonds now compete with equities without complicated strategies, and the hosts close by contrasting bond income with equity-income substitutes before sharing where listeners can find them.

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal