'Democracy Is Too Slow': Silicon Valley's AI Power Grab Meets Public Resistance

A Stanford ethicist was once told point-blank that democracy is an obstacle to technological progress. "Democracy is too slow, and it holds science back," Rob Reich recalls being told at a dinner organized by a Silicon Valley mogul. "To optimize for science, we need a beneficent technocrat in charge."

The dinner, as Reich recounts in the book "System Error: Where Big Tech Went Wrong and How We Can Reboot It," gathered technologists to discuss what a state designed to maximize science and technology — powered by commercial models — might look like. The assembled guests were drawn to a concrete idea: building a small nation-state dedicated to that endeavor.

When Reich asked the group whether this imagined state would be a democracy, the answer was a flat no.

The anecdote, resurfacing now as the Guardian examines the collision between Silicon Valley's AI ambitions and democratic oversight, frames a question with real stakes for the industry's build-out. The companies racing to deploy AI at scale are discovering that voters, regulators and local communities get a say — and often use it to say no.

The datacenter backlash

The Guardian's reporting centers on a wave of public pushback against the datacenters that AI development requires. "The push against datacenters speaks to the frustration of people who don't feel they are making the choices in their lives," the piece argues in its opening line.

That framing matters for the AI industry. Datacenters are the physical substrate of the current AI boom — the training and inference infrastructure behind every large model. They demand enormous amounts of land, water and electricity, and they are increasingly being sited in communities that did not ask for them. When residents object, they are not merely protesting construction. They are contesting who decides how the AI economy gets built, and who bears its costs.

This is the tension the Guardian identifies at the heart of the story: a cohort of technology leaders who, by Reich's account, have contemplated political structures explicitly designed to route around democratic deliberation — now confronting democratic processes that are, in fact, slowing them down.

What Reich's anecdote reveals

Reich is a Stanford ethicist and co-author of "System Error," a book examining where big tech went wrong. The dinner he describes was organized by a Silicon Valley mogul whose identity he does not disclose in the quoted passage. The purpose was speculative but serious: to work through the design of a state that would maximize science and technology using commercial models as its engine.

The group's attraction to the nation-state concept is the telling detail. Rather than asking how existing democracies could better support scientific progress, the gathered technologists gravitated toward a blank-slate polity — one where the friction of elections, public comment periods and contested politics could be engineered away.

Reich's question — would this state be a democracy? — and the cool negative it received, distill an attitude that critics have long suspected animates parts of the industry. The reply he quotes treats democratic consent as a performance tax on optimization. The preferred alternative, as stated to him, is "a beneficent technocrat in charge."

Why this matters now

The collision course the Guardian describes has sharpened as AI has become a capital-intensive, infrastructure-heavy industry. The moguls and labs driving it need physical assets — datacenters, power generation, transmission — that cannot be built in a political vacuum. Zoning boards, utility commissions, environmental reviews and ballot initiatives all sit between an AI roadmap and its realization.

Public frustration, as the Guardian's own summary puts it, stems from people "who don't feel they are making the choices in their lives." Datacenter disputes become flashpoints for that broader grievance. Residents see decisions about their energy rates, their water supplies and their local environment being shaped by distant actors pursuing goals they never voted on.

For an industry whose leaders have, at least in the setting Reich describes, openly mused that democracy is a brake on progress, the resistance is not an anomaly. It is the system working as designed — deliberation, contestation and consent requirements doing exactly what the dinner guests found so inconvenient.

The stakes run in both directions. If democratic processes block or delay the infrastructure AI companies say they need, the pace and geography of AI development will be shaped by politics rather than by capital plans alone. If companies succeed in overriding that resistance — through lobbying, special permitting regimes or siting in jurisdictions with fewer checks — they will prove the dinner-party premise correct in practice, if not in name.

Reich's account, published in "System Error," suggests the impulse is not hypothetical. The question he posed at that dinner table — whether a technology-maximizing state would remain democratic — is now being asked, in effect, by the communities standing in the path of the AI build-out. Their answer, like his hosts', will shape how the next decade of AI infrastructure actually gets built.