A Necessary Recalibration of AI Expectations

Measured regulation of US data center construction could improve the economics of AI investment and make equities more attractive after the recent sell-off.

• The seven US states leading data center development are expected to have Democratic governors in 2027, raising the risk of state-level restrictions and a patchwork regulatory regime.
• The Philadelphia Semiconductor Index fell 15% in two days after the election as investors marked down demand, prices, margins, and earnings expectations across the AI supply chain.
• More than a year of aggressive AI spending has created bottlenecks, shortages, and sharp price increases for semiconductors and other inputs, weakening future returns for customers.
• Longer project timelines could allow construction, energy, and semiconductor suppliers to expand capacity, easing shortages and lowering costs for hyperscalers and AI model developers.
• Projects already approved are generally expected to proceed, leaving developments planned for 2028 and later more exposed to the new regulatory regime.

Takeaway: The sell-off represents a necessary reset in AI expectations; with the S&P 500 at about 18x forward earnings versus five-year and ten-year averages of 19.8x and 19.1x, US equity risk has eased at the margin, while emerging-market and Japanese equities remain more attractive to the author.