America Voted Against Data Centers: What the Midterms Mean for AI and Why It Might Be Good
Summary
This episode analyzes the impact of recent US midterm elections on the AI industry, noting that political shifts in key states like Texas and Georgia have paused or slowed nearly 45% of planned data center capacity. The host discusses how new regulations requiring fair pricing for energy and water, alongside the removal of subsidies, will likely slow the massive capex buildout but prevent overbuilding.
Topics discussed
Introduction: Data center pauses and AI capex
Resource constraints: Power, land, and water
Public sentiment and the midterm election results
New policies in Texas, Ohio, and Georgia
Shift from subsidies to fair market pricing
Bipartisan concerns and the US-China energy gap
Debunking the water consumption myth
Hyperscaler responses and stock market reaction
Revised compute projections and gigawatt estimates
New stakeholders: Local governments and legislation
Financial risks for AI labs and capex commitments
Impact on specific stocks and historical trends
The shift toward compute efficiency and open source
Elon Musk's advantage and rising AI costs
US vs China: Model access and security risks
Conclusion: Benefits of slowdown and optimization
Outro and call to action
Listen ad-free on Castria