The AI Bubble Is Widely Misunderstood | Steve Hou

In this episode of Forward Guidance, host Felix interviews Steve Hou, Senior Quant Researcher at Bloomberg, about the misunderstood nature of the AI bubble. Hou argues that while AI is undeniably a bubble—inevitable given the massive investment and uncertainty—it doesn't diminish its transformative potential. He contrasts the AI boom with the dot-com bubble, emphasizing that AI is already being adopted at scale, driving unprecedented capital investment in data centers, chips, and infrastructure. This has become a major contributor to U.S. GDP growth, particularly through business investment and supply chain activity in countries like South Korea and Taiwan. Hou also challenges the notion that AI productivity gains are absent, suggesting they're obscured by measurement issues and time lags, and warns that the current era of cheap, subsidized AI usage is unsustainable. He envisions a future where AI pricing evolves into a tiered system, with token efficiency becoming a strategic corporate priority. On macro policy, Hou is skeptical of preemptive rate cuts based on unproven AI disinflation, arguing that the immediate inflationary impact of AI-driven demand for hardware and energy is more pressing. He also explores AI’s potential to reshape economics, enabling more realistic agent-based simulations and better policy communication, though he cautions that real-world services like plumbing and childcare remain constrained by physical bottlenecks. Overall, the conversation paints a picture of AI as a powerful, multifaceted force with long-term implications that extend far beyond hype or bubble narratives.