Is AI a Mistake? (EP. 452)

Episode Summary

Executive Summary: The episode centers on the rapid rise of AI and its impact on software, labor, investing, and prediction markets. The hosts argue AI is real and disruptive but caution against extreme narratives: some stocks and business models deserve re-ratings, yet many fears are overextended. They also discuss market volatility, international outperformance, Gen Z investing, and the growing legitimacy—and risks—of prediction markets.

Main Topics: AI disruption: real, fast, but not apocalypse (Priority: 5/5): A long debate on whether AI is vaporware or an immediate job-destroying force. The hosts settle on a middle-ground view: AI is powerful, especially for software/coding, but widespread job replacement is likely overstated in the near term. Software stock selloff and valuation re-rating (Priority: 5/5): They discuss sharp declines in SaaS and wealth-tech names, arguing the market is quickly repricing software moats and long-duration earnings rather than pricing a full collapse in revenue. AI capex, hyperscalers, and business economics (Priority: 4/5): The show examines massive AI infrastructure spending, the economics of training/running models, and why large tech firms are pouring money into AI despite uncertain ROI. Prediction markets vs. gambling and information discovery (Priority: 4/5): The hosts explore the growth of prediction markets, their potential to improve forecasting, and the need for guardrails to avoid manipulation and erosion of trust. Broad market volatility and international leadership (Priority: 3/5): They highlight unusual single-stock volatility, strong earnings moves, and the relative strength of Japanese and non-U.S. equities versus U.S. markets. Personal finance, frugality, and generational behavior (Priority: 3/5): They briefly revisit 'The Millionaire Next Door,' arguing that higher income matters more than frugality alone, and discuss how Gen Z is investing more while homebuying declines.

Key Arguments: AI is not fake; it is clearly capable and improving fast, but near-term forecasts that it will replace most white-collar work are likely too extreme. Software is especially vulnerable because coding is one of the easiest tasks for AI to automate, so SaaS valuations can compress even without revenue collapsing. The market may already understand the risks to software moats, but is overreacting in some cases by treating all AI exposure as equally threatened. Tech leaders may not fully appreciate second-order consequences such as inequality, labor displacement, or misuse in biological/ethical contexts. Hyperscalers’ massive AI capex is rational because AI is shrinking task completion time and headcount needs, changing the economics of software development. Prediction markets could become a meaningful forecasting tool because participants are incentivized for accuracy, but sports-like or leakable contracts need restrictions. Enterprise software remains sticky because buyers optimize for career risk, support, continuity, and blame allocation—not just cheapest features. International stocks and Japanese equities are benefiting from a regime shift, including higher yields, less financial repression, and catch-up valuation dynamics. Gen Z appears increasingly willing to invest in markets rather than rely on homeownership as the default wealth path. The labor market is not yet showing AI-driven layoffs broadly; instead, companies are slowing hiring, which supports the idea that disruption is still emerging rather than fully here.

Data Points: AI task completion horizon: Doubling every 7 months - Listener email describing how AI can handle increasingly larger work blocks over time. Hyperscaler AI headcount cost: $1 million per headcount per year - Listener email estimating the cost of one engineering employee at a hyperscaler, including compensation and overhead. Anthropic revenue growth: $0 to $100M to $1B to a $14B run rate - Referenced to show the speed of Anthropic’s business expansion. Business adoption of Anthropic on Ramp: 1 in 5 businesses - Ramp data cited showing rapidly increasing usage of Anthropic among businesses. Business adoption of Anthropic one year earlier: 1 in 25 businesses - Compared with current adoption to show the pace of change. Anthropic share on Ramp AI index: 19.5% - Cited as Anthropic’s latest share of business AI usage on Ramp. OpenAI share on Ramp AI index: 35.9% - Cited as OpenAI’s latest share of business AI usage on Ramp. Average post-earnings stock move: 5.2% - Wall Street Journal stat showing the biggest average earnings-day move since at least 2012. Stocks moving 15%+ on earnings: A bunch of stocks - Used to illustrate unusually high single-name volatility this earnings season. S&P 500 revenue growth expected next quarter: 9% - Mentioned as being strong despite market fear and stock volatility. S&P 500 revenue growth currently: 8.3% - Current quarter revenue growth cited as unexpectedly strong. U.S. prime-age labor force participation: ~84% - Labor force participation rate for ages 25 to 54 near all-time highs. Gen Z annual transfers to investment accounts: 14.4% - Share of people 25 to 39 making annual transfers, more than tripled from 2013 to 2023. 26-year-olds investing since age 22: 40% - Share of 26-year-olds who transferred funds into investment accounts since turning 22, as of May 2025. 26-year-olds investing in 2015: 8% - Historical comparison showing a large rise in youth investing. Homebuyer share age 18-39 in 1999: 51% - Redfin data used to discuss young-adult homebuying trends. Homebuyer share age 18-39 in 2025: 40% - Shows young adults’ reduced share of home purchases today. MSCI Japan 12-month return: Over 40% - Used to show Japan’s strong performance amid rising yields and inflation expectations. S&P 500 12-month return: About 16% - Comparison point against MSCI Japan’s stronger recent performance. AI-related stock move: Schwab down 10% in a day - Illustrates how AI fear and recency bias can violently reprice adjacent sectors. Prediction markets share at DraftKings: 1% - Prediction markets remain small relative to sports betting despite market reaction. DraftKings revenue growth: 43% year over year - Strong fundamentals contrasted with a sharp stock selloff. Business adoption of private AI tools: One in 5 vs one in 25 - Repeated on Ramp data to emphasize Anthropic’s acceleration.

Pivotal Quotes: "The technology is incredible and it keeps getting more and more incredible." — Michael Batnick / Ben Carlson discussing Derek Thompson’s framing: A rebuttal to claims that AI is merely hype or vaporware. "When you buy software for your company, you're not just buying features, you're buying someone to blame when things go wrong." — Referenced from an essay on SaaS defensibility: Explains why enterprise software can remain sticky despite AI feature commoditization. "Both can't be true at the same time." — Vivek Ari (quoted by the hosts): On the contradiction between AI capex being uneconomic and AI simultaneously being powerful enough to wipe out SaaS.

Implications: Listeners should expect continued AI-driven volatility in software and adjacent sectors, but not assume every incumbent is broken. For investors, moats, regulation, switching costs, and business model structure matter more than hype. Prediction markets may grow, but will need guardrails.