The Ezra Klein Show: The A.I. Revolt Is Here
Summary
Jasmine Sun discusses her reporting on the growing bipartisan backlash against AI data centers in the Midwest, where opposition has surged due to noise, energy costs, and distrust of tech companies. She analyzes how non-disclosure agreements and the lack of a clear public benefit for AI fuel "AI populism," contrasting the situation with more accepted infrastructure like solar farms.
Topics discussed
Sponsorship and introduction to the AI debate
Introducing Jasmine Sun and her reporting on data centers
Political reactions: Desantis, Sanders, and data center moratoriums
Sponsorship and transition to the interview
The physical scale and aesthetic impact of hyperscale data centers
Comparing data centers to other industrial installations like fracking
Three layers of concern: process, direct impacts, and AI itself
The role of NDAs in fueling local opposition and rumors
Water usage myths and the Great Lakes context
Electricity concerns and the promise of new power plants
Property tax benefits and skepticism about job creation claims
Skepticism toward utility rate promises and job longevity
Shifting power dynamics and community benefits agreements
The AI bubble fear and the Foxconn precedent in small towns
Stranded assets and the Janesville GM plant analogy
Risk of obsolete infrastructure if AI demand or companies fail
XAI's Colossus and the reality of compute capacity sales
NIMBYism vs. essential infrastructure: housing and solar comparisons
Tactics of opposition: Facebook groups, town halls, and zoning fights
Public perception of AI utility: memes vs. essential services
Defining 'AI Populism' as resistance to an elite political project
Water misinformation and the 'beef vs. data center' comparison
Personal conflict and echoes of populist messaging in Michigan
Exclusionary zoning lawsuits and the settlement of small towns
The gap between executive rhetoric on democracy and actual governance
Political suspicion of money and Anthropic's labor market warnings
Sponsorship break
The Silicon Valley belief in creating an AI underclass
Skepticism of the 'inequality reduction' vision and time to adjust
Software engineering job displacement and the limits of reskilling
The contradiction of building what you fear while racing to do so
Reconciling executive fears with the drive to accelerate AI
Three rationales for building superintelligence: determinism, money, and curiosity
Risk appetite and the 'flip the coin' mentality in Silicon Valley
The superintelligence bet: acceptable levels of extinction risk
Why tech elites don't build data centers in their own backyards
The 'race' metaphor: finish lines and recursive self-improvement
Defining the finish line and the perpetual feeling of falling behind
The race as a central dividing line in AI policy debates
Alliances between AI safety experts and populist movements
Cross-ideological alliances: social conservatives and labor left
Strategic questions for AI safety: building broad coalitions
Offshoring data centers to Texas, Louisiana, and space
Economic benefits vs. public backlash: doing the deals right
Compute demand vs. supply and the risk of constraining supply
AI infrastructure as geopolitical leverage for states and countries
Inequality fears: who benefits from superintelligence?
Sponsorship: NYT Games
Public sentiment in China: less fear of AI than in the US
Chinese regulation of companion chatbots and social concerns
Debunking the 'Chinese PsyOp' theory of data center backlash
AI executives taking public backlash seriously in 2026
Marketing problems and the need to stop fear-mongering
Transition to discussing industry beliefs and IPOs
Do industry actors believe their own warnings? The culture of internal debate
Technological determinism and the inevitability of AGI
Losing control of the AI narrative and the need for tangible benefits
Public opinion on UBI and immortality: not as popular as hoped
AI's role in scientific progress and the friction of real-world implementation
Which jobs are most susceptible to AI automation? IC roles vs. relational work
The 'lazy' approach to AI solutions in climate and robotics
The difficulty of building physical infrastructure vs. software innovation
Political failures of AI-adjacent figures and Anthropic's White House issues
Conclusion: The Technology Trap and sharing the benefits of AI
Listen ad-free on Castria