AI boom or doom? Your questions answered

The AI boom is likely to persist for years because adoption, infrastructure demand and productivity gains remain strong, although financing constraints, overbuilding, local opposition and safety risks will reshape its winners.

• AI valuations are less extreme than during the 1999–2000 technology bubble, while NVIDIA, Broadcom, Micron Technology and SK hynix already reflect slower growth and lower future earnings; Meta and Alphabet are using AI to increase engagement and advertising revenue.
• Capital Group expects enterprise AI adoption could become a 20-year implementation cycle, with data-center capacity currently acting as a bottleneck and potentially giving infrastructure providers pricing power.
• Data-center demand is supported by measurable productivity gains: UnitedHealthcare reduced prescription approval times from eight hours to 30 seconds, while C.H. Robinson cut quote times from 17–20 minutes to 31 seconds; the Federal Reserve estimates that only 20% of companies currently use AI.
• Local opposition blocked or delayed 75 data-center projects worth about $130 billion in the first quarter of 2026, matching the total value affected during all of 2025.
• Higher financing costs are increasing investor selectivity: 10-year U.S. Treasury yields rose above 5.25% in early October, while the volume of AI-related debt issuance has pushed some technology-company bonds to wider spreads than industrial peers.
• AI safety incidents are increasing as systems become more capable, and recent rogue-agent attacks, stalled federal legislation and new laws in California and New York point to tighter safeguards.

Takeaway: Investors should not treat AI as either an imminent bubble or an unstoppable trade; they should favor profitable businesses and infrastructure already under construction, remain selective on speculative AI debt, and expect leadership to broaden as financing, regulation, opposition and safety risks develop.