Bruno S. Sergi states that halting new data centers will not prevent artificial intelligence gains from concentrating in a few firms, urging Congress to instead spread adoption across the economy. Following the November 3 midterms, Democrats secured both chambers of Congress, prompting broader policy scrutiny regarding data center permitting, environmental disclosures, and economic distribution.
Data center politics influenced races in Texas, Georgia, and Ohio, though over-focusing on geographic location risks missing systemic adoption barriers. A March 2026 Gallup survey revealed that 71 percent of Americans opposed local AI data centers compared to 53 percent opposing nuclear plants, and dozens of restrictive referendums passed during the midterms. Goldman Sachs warned that state-level moratoriums could affect 45 percent of planned US data center capacity expansion, driving investments toward Southeast Asia where companies like Microsoft and Google expand capacity in Malaysia and Indonesia. Big Tech still owns frontier models despite infrastructure shifts, necessitating greater transparency regarding electricity, water, and grid costs.
Regulatory friction targeting physical infrastructure such as the server farms evaluated by Goldman Sachs displaces capital expenditure rather than altering the underlying ownership of frontier models. Hyperscalers maintain control over proprietary architectures independently of where heavy compute clusters draw power, shifting capacity toward jurisdictions with lower regulatory resistance like Malaysia and Indonesia. Consequently, local zoning referendums merely redirect the geographical footprint of energy-intensive computing without diffusing corporate consolidation.
No comments:
Post a Comment