The AI bubble is 17 times the size of the dot-com frenzy, analyst says (www.marketwatch.com)

Independent research firm MacroStrategy Partnership warned that the current AI investment boom is a bubble roughly 17 times larger than the dot‑com era and about four times the size of the U.S. subprime bubble. The note — authored by analysts including Julien Garran, formerly of UBS’s commodities strategy team — argues that years of artificially low interest rates pumped cheap capital into AI startups and projects, flooding markets with funding that now confronts hard scaling limits. For the AI/ML community this is a cautionary signal: oversized valuations could prompt a rapid funding correction, shifting incentives from growth-at-all-costs to efficiency and measurable ROI. The note highlights technical implications behind the “scaling limits” — rising compute and energy costs, data availability bottlenecks, and diminishing returns from ever‑larger models — which could accelerate demand for model compression, algorithmic efficiency, hardware innovation, and domain‑specific solutions. Practitioners and investors should expect tighter capital, greater scrutiny of deployment economics, and more emphasis on practical gains (latency, cost per inference, data efficiency) rather than headline model size.