When a Data Center Comes to Town

Morgan Stanley strategists analyze the economics and politics of the US AI data center boom. Local resistance is growing with over 300 moratoria, but federal policy remains supportive due to US-China competition. The buildout will continue, though communities demand concessions, creating a 'conditional' expansion. Key bottlenecks are power, skilled labor, and local politics, with off-grid power seen as one solution. - AI data center investment is a major force, but local moratoria now touch 40 states, mostly as temporary pauses. - Local economic benefits from construction are significant but fade after completion, while tax revenue depends on incentive design. - Opposition focuses on electricity costs, housing prices, and environmental concerns, pushing some buildouts to rural areas and abroad. - Power, people, and politics remain the three main bottlenecks; a 38 GW power shortfall is expected through 2028. - Federal lawmakers are unlikely to impose a national ban because of the AI race with China, keeping policy supportive. - Hyperscalers will need to offer on-site power, efficiency improvements, and cost concessions to local communities. - The base case is a 'conditional buildout' with elevated AI capex and upside risk to hyperscaler spending.