Unsustainably High - Tech Billionaires vs. Voters

Data Centers Become a Major Election Issue

The panel reported an unusually broad, bipartisan backlash against data-center projects in the 2026 elections. Of 41 local ballot measures across 10 states, 37 were reported to have rejected data-center construction. Most of the remaining measures also favored restrictions, while one Ohio vote was still uncertain. The results were preliminary at the time of recording.

The hosts argued that data centers have become a visible stand-in for wider frustration with AI companies and their wealthy leaders. Voters have heard promises of job disruption and economic upheaval, while local communities face potential noise, water and power demands, pollution, and limited long-term employment. The panel emphasized that local opposition can matter even if companies eventually move projects elsewhere. They said the issue is not simply about data centers, but also about corporate power and accountability.

One proposal in Perry, Ohio, drew particular praise: voters approved a requirement that developers seeking tax breaks back their promised jobs, tax revenue, and other benefits with a financial guarantee. The panel said this could help communities avoid accepting unfulfilled promises, as has happened with other development projects.

AI-Generated Voter Outreach

The panel discussed a West Virginia candidate who used chatbots to send targeted messages about data-center impacts. He lost, but the hosts noted that the campaign operated on a very small budget and saw the approach as a likely template for future political outreach.

They were unsure whether chatbot messages would persuade voters and questioned whether people can reliably tell when a political text is automated.

Data Centers’ Energy and Environmental Costs

The show examined proposals to repurpose old military jet engines as generators for data centers. The hosts warned that these engines are inefficient and could bring persistent noise and pollution to nearby communities.

They also criticized the industry’s earlier environmental promises, arguing that the rush to secure power is encouraging polluting stopgap solutions.

Anthropic’s Biological Research and AI-Safety Claims

Anthropic said its biology lab used Claude to identify a previously unknown enzyme system resembling CRISPR. The hosts described the reported discovery as potentially significant, but stressed that it still needed independent verification. They raised questions about who should own discoveries made using AI trained on publicly available research, whether such findings should be made openly available, and how gene-editing methods could be misused.

The panel was more skeptical of Anthropic’s claim in an IPO filing that AI-assisted research could advance faster than the company’s ability to develop safety measures. They read the language as less proof of imminent recursive self-improvement than a warning about a possible mismatch between capabilities and safeguards. They also questioned whether companies can use uncertainty as an excuse to avoid responsibility.

Hardware Shortages and Financial Risks

Sam Altman’s comments that personal computers’ RAM is underused compared with data-center memory prompted criticism from the hosts, who said the remark ignored how AI companies have contributed to hardware shortages and rising prices. They also noted the conflict of interest in encouraging people to rely on cloud services rather than run models locally. Elon Musk mocked Altman over the remarks, despite operating a competing AI company.

The panel also warned that AI companies’ heavy spending and debt could create broader financial risks. They discussed lenders packaging higher-risk AI-related debt, comparing the practice to the bundling of risky mortgages before the 2008 financial crisis.

The hosts stressed that AI may have valuable applications, but said the industry has yet to demonstrate that the economic returns will justify the scale of its investments.

AI in the Outdoors, Schools, and Daily Work

The hosts cited several incidents in which hikers followed chatbot advice that led them into dangerous situations, including inadequate routes or provisions. They framed these cases as examples of “cognitive surrender”: users assuming a chatbot has reliable, optimized information without checking it against authoritative sources.

That concern led to a broader discussion about education. The panel worried that students may become overly dependent on AI and lose practice in reasoning and writing. They discussed teaching students to compare AI responses, assess their reliability, and explain their reasoning, rather than simply banning the tools or allowing them to complete assignments unchecked.

Google’s decision to default AI tools off for younger students was welcomed, but the company’s claim that disabling AI also disables Search drew criticism. The hosts argued that conventional web results can still be accessed and questioned whether Google was using product integration to discourage schools from turning AI off.

Other Technology Stories

• Google’s experimental data-center satellite reportedly generated just 1 kilowatt and could power only one of its four processors for about 20 minutes at a time. It produced a haiku, but the panel stressed the system’s severe power and heat constraints. Radiation was also said to make space a poor environment for training models.
• OpenAI’s plan to call data centers “AI compute campuses” was ridiculed as a rebranding effort that does nothing to address power use, water consumption, noise, or local opposition.
• Elon Musk said Grok’s operating costs were “unsustainably high,” prompting the panel to question how that fits with earlier claims that heavy usage would drive the product’s success. They also discussed Grok’s difficulty competing for enterprise customers.
• A University of Kentucky employee was granted a religious exemption from a workplace push to use AI. The panel connected the case to the principle that people should not abdicate their moral agency to automated systems.

Picks

• Paris Martineau recommended a seasonal Ethiopian coffee from PERC.
• Father Robert Ballecer recommended a portable espresso maker.
• Jeff Jarvis recommended the television series Cupertino.

Presenters and Contributors

Father Robert Ballecer, SJ (guest host); Jeff Jarvis; Paris Martineau; Leo Laporte (recorded sponsor segments).