The 4 Things You Need for a Tech Bubble (www.wired.com)

WIRED’s Uncanny Valley podcast hosted Brian Merchant, who applied the historical framework from Brent Goldfarb and David Kirsch’s Bubbles and Crashes to ask whether AI is a tech bubble. He distilled four repeatable criteria that define past tech bubbles: (1) deep uncertainty about how the innovation will make money, (2) a rising class of “pure‑play” firms whose fate depends entirely on the new tech (think CoreWeave or chipmakers), (3) novice/retail investor influx enabled by easy trading apps, and (4) a coordinating belief sparked by striking demos or market narratives (e.g., ChatGPT, Lindbergh’s flight). Merchant contrasts exuberant signals—Nvidia’s meteoric rise to a $5 trillion valuation and Big Tech’s ~$400B annual AI spend—with reports that ~95% of firms using AI see little return, and historic busts like Pets.com and radio to show how these factors can align. For the AI/ML community this framework is practical: it shifts the conversation from vague “bubble” vibes to measurable risks and failure modes. Technical implications include potential overinvestment in compute and data‑center capacity, stranded pure‑play infrastructure firms if monetization lags, and heavier scrutiny on productization pathways rather than demos alone. The takeaway: expect boom‑and‑bust cycles, prioritize clear business models and diversified dependencies, and prepare for consolidation even as foundational advances—like those after past bubbles—could produce lasting, transformative technologies.