Last updated: 2026-09-28

U
Undergraduate level

Democracy and the AI Build-Out: Testing a Political-Economy Argument

28 September 2026

Eduardo Porter's Guardian article of 27 September 2026 argues that public backlash against AI data centres has caught the technology industry's leaders off guard, because their outlook leaves little room for democratic friction [1]. It is a well-sourced argument with a clear thesis, which makes it a good case for practising a particular reading skill: separating what a piece reports from what it infers, tracing each claim to its source, and testing the inferences against evidence the piece does not discuss. This page does that in three steps. It summarises the argument, checks the article's facts and links, and then asks five questions about its underlying assumptions, setting the article's text beside the wider evidence.

What the Article Argues Ephemeral / ToolingKnowledge that evolves in months to a year — check for updates

Porter opens with an anecdote from Rob Reich's book System Error. At a dinner, a technology figure describing a hypothetical state built to optimise science said that democracy "is too slow" and that a "beneficent technocrat" should be in charge [2]. He then sets polling on American attitudes to AI and data centres against the pace of the build-out, and reads the backlash, with the political scientist Henry Farrell, as frustration among people who feel they have little say over the choices that shape their lives.

Porter also gives structural reasons to expect the build-out to continue regardless: the sector's profitability and global dominance, a bipartisan preference for beating China, and an administration that backs data centres, including an EPA proposal on public input for air permits. The core of the argument is historical. The historian Quinn Slobodian's account of neoliberalism, set out in Globalists, describes a project that sought to shield markets from democratic pressure while acknowledging democracy's legitimacy [3]. Porter contrasts that with a Silicon Valley "programmers' mentality" in which democratic messiness is "bad code", and offers Peter Thiel's 2009 essay [4], remarks attributed to Sam Altman and Elon Musk, and the industry's move from a self-image as liberalising force to alliance with the Trump administration as evidence. He closes by suggesting that alliance is fragile. Slobodian's later book on market radicals and democracy covers related ground [5].shielding markets from democratic pressure

The article's figures and most of its quotations hold up, with a small number of qualifications. Two of the weaker links, the Altman and Musk items, matter because they are the named evidence for the industry's outlook, whereas the Thiel quotation and the polling are solid.

ClaimWhat the source showsStatus
73% say data-centre costs outweigh benefits; 70% say AI is bad for societyMarquette Law School national survey, fielded 22–29 July 2026, n=1,076: 73% and 70% [6]Matches. The AI question is forced-choice, with no middle option. A later Marquette survey (2–9 September, n=1,023) gives 71% and 64% [7].
Concern about AI has risen from 37% (2021) to 52%; 55% of young adultsPew survey, 22–28 June 2026, n=3,488: 52% of US adults more concerned than excited; 55% of under-30s [8]Matches.
Americans are more mistrustful of AI than people elsewherePew's 25-country report: about half of US adults more concerned than excited, against a median of 34%; on trusting their own country to regulate AI, 44% trust and 47% distrust [9]Concern is supported. "Mistrustful" is looser than what was measured.
Thiel said in 2009 that freedom and democracy are incompatibleCato Unbound, 13 April 2009: "I no longer believe that freedom and democracy are compatible" [4]Accurate. Thiel goes on to propose technological escape: cyberspace, outer space, seasteading.
The Reich dinner anecdoteReported in System Error [2]Plausible. One anecdote about one unnamed host.
The article's account of Altman and an AI replacing human presidentsThe article links a Joe Rogan Experience episode of 6 October 2023 [10]. Notes on that episode record Altman calling the phrase "do the greater good for humanity" Orwellian.Not verified. The passage could not be located, and the strongest reading looks overstated. Check the recording.
Musk is described as "answerable to no one"The phrase comes from Tim O'Reilly's guest essay in The Economist, 12 July 2026, describing structures Musk has assembled [11]A writer's characterisation of corporate structures, not a statement by Musk.
Trump called data centres "beautiful"Axios interview, 23 August 2026; the available excerpt applies "beautiful" to power plants [12]Partly verified. Full text not checked.
The EPA plans to end public input on air permitsA proposal to end the federal requirement that states publicise air-permit applications and take public input, including for data centres and power plants. States could still offer comment, and nearly 200 groups and over a dozen states object [13].A proposal and not yet a rule.
Section 230 gave social media a liability shield in 199647 U.S.C. §230, part of the Communications Decency Act, which is Title V of the Telecommunications Act of 1996 [14]Correct.
Quotations from Farrell and Rumman ChowdhuryFarrell is a political scientist at Johns Hopkins. The quotations appear in the article and its syndicated copies.Not independently traceable.

Five Questions for the Argument Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

Every argument rests on assumptions, and the assumptions are where disagreement usually lives. Five are worth testing here. For each, the article's text comes first, then the wider evidence, then an assessment.

1. Does "Silicon Valley" name a single outlook?

The article names individuals, and it relies on Slobodian's description of a "leadership class" for the generalisation. The evidence behind that generalisation is a set of individual statements and one dinner anecdote, which is thin for a claim about an industry. And the industry is divided in public. Altman told a US Senate subcommittee in May 2023 that regulatory intervention would be critical, and proposed licensing the most powerful systems [15]. Andreessen's Techno-Optimist Manifesto of October 2023 attacks what it calls "vetocracy" and rejects existential-risk arguments [16]. Zuckerberg argued in July 2024 that open-source AI is safer and spreads power away from a few closed labs [17].

Disagreement about regulation doesn't settle the article's claim, though. That claim concerns attitudes to democratic constraint on whether and where to build, and supporting a licensing regime, or opposing one, is a different position from that. Assessment: the generalisation rests on few voices and the public divisions are real, though they don't refute the thesis on their own.

2. Is there a competing explanation for local opposition?

The article doesn't discuss self-interest explanations such as NIMBYism ("not in my back yard"), and it reads the backlash through Farrell's account of frustration at having no say. The research on siting conflicts suggests the NIMBY label explains less than it appears to. Wolsink's study of wind-farm siting argues that institutional capacity shapes outcomes more than public attitudes do and questions the NIMBY explanation [18]. Devine-Wright's review of six research strands criticises NIMBY as an explanation [19], and Schively's review finds that responses vary with the facility, the participants and how the facility is described [20]. All three concern wind energy or siting in general, so applying them to data centres is by analogy.

Data-centre polling points the same way. A Gallup survey fielded in March 2026 found 71% opposed to an AI data centre in their area, 48% of them strongly. Opponents' reasons were mostly concrete: resource use (50%, mainly water and energy), pollution, and utility bills. Supporters cited jobs [21]. Project-tracking figures need care. Data Center Watch counted $18bn of projects blocked and $46bn delayed between May 2024 and March 2025, but it is funded by a firm that serves AI companies, and its own report warns that attributing delays to opposition oversimplifies [22].

One further point is our own argument and not a finding from those sources: self-interest and democratic voice are not opposites. Residents objecting to a facility that affects their water or bills are also exercising the voice the article says the process denies them. Assessment: the article leaves the self-interest reading unexamined, and the research suggests that reading explains less than the label implies.

3. What kind of worry do the polls measure?

The article reports concern and distrust, and it doesn't say the public wants AI stopped. The wider evidence shows what kind of worry this is. In Pew's April 2025 survey, 56% of the US public were extremely or very concerned about AI eliminating jobs, against 25% of AI experts [23]. In a Gallup and SCSP survey fielded in spring 2025, 80% wanted safety rules maintained even if AI developed more slowly, and 9% prioritised speed; the figures were 88% among Democrats and 79% among Republicans and independents [24]. That is specific worry combined with support for rules that could slow development, which cuts against the idea that most people would trade oversight for speed.

The polls carry their own caveats. Marquette's AI question is forced-choice, and its figure moved from 70% to 64% between July and September [6][7]. A single reading of opinion is a snapshot. Assessment: the polls support a picture of specific worries alongside support for oversight, which fits the article's account, with a general caution about forced-choice questions.

4. What does the article leave out about benefits?

The article is an argument about power and consent, and it gives little space to what the technology delivers. The benefit evidence exists but is local. Brynjolfsson, Li and Raymond studied 5,179 customer-support agents at one firm and found productivity up 14% on average and 34% for novice and low-skilled workers, with minimal gains for experienced workers [25]. That is one firm and one task. The labour-market side, including early employment declines among young workers in exposed occupations, is set out in Who Buys What the Machines Make?.

Our own reading is that benefits and democratic control are not alternatives. Whether documented gains justify a given build-out is the kind of question a democratic process exists to weigh. Assessment: the article underplays benefits, though its subject is who decides, and it doesn't claim to be a cost-benefit account.

5. How new is the anti-democratic strand?

The article's title and framing suggest a recent collision, but its sources reach further back. Porter cites Thiel's 2009 essay [4] and draws on Slobodian, whose book on market radicals traces hostility to democracy well before the current alliance [5]. The "Good Guys" passage reports the industry's self-image as a liberalising force, and whether that image was accurate is left open.

The wider history supports the article's implicit view that two strands coexisted. John Perry Barlow's 1996 Declaration told governments they had no sovereignty in cyberspace [26]. Barbrook and Cameron's "Californian Ideology" argued that Silicon Valley fused countercultural and libertarian, market-driven thinking [27], and Fred Turner traced the route from 1960s counterculture to cyberculture [28]. Assessment: the article itself documents the older strand. What remains arguable is its framing of a recent rupture, since the earlier material it cites shows the roots were there all along.

What Stands Up Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice

The article's strongest elements are its polling, which checks out against the source, and its institutional analysis of why a build-out with public opposition can still proceed. Its weakest element is the use of individual statements and one anecdote as a proxy for an industry's outlook, and two of those attributions, Altman's and Musk's, need qualifying. None of the five questions overturns the thesis. Three expose gaps in what the article covers: the generalisation from a few voices, the unexamined self-interest reading of local opposition, and the thin treatment of benefits. The other two, on public opinion and on history, are largely answered by the article's own sources and by the polling.

Reading Political Commentary Critically FoundationalKnowledge that endures for decades — core principles

  • Separate three kinds of statement. Reported facts can be checked, inferences can be tested against other evidence, and values shape which inferences seem natural. A difference in values is not an error.
  • Follow each link and check its date. A figure can match a source and still be superseded by a newer release, and a link can lead to a different version of the number.
  • Treat quotations and anecdotes as illustration. One dinner, or one podcast remark, says little about an industry.
  • Check that a quotation belongs to the person it is attributed to. A phrase from a commentator's essay is not a statement by its subject.
  • Read what a piece says before deciding what it assumes. An assumption attributed to an article without a passage to point to is a guess about it, and criticism needs the same checking as the piece it criticises.

References

  1. Porter, E. (2026, 27 September). Democracy blindsides Silicon Valley's power players looking to transform the world with AI. The Guardian. https://www.theguardian.com/technology/ng-interactive/2026/sep/27/democracy-ai-datacenters-power. Republished by the author as "Silicon Valley and democracy do not get along": https://eduardoelreportero.substack.com/p/silicon-valley-and-democracy-do-not
  2. Reich, R., Sahami, M., & Weinstein, J. M. (2021). System Error: Where Big Tech Went Wrong and How We Can Reboot. HarperCollins.
  3. Slobodian, Q. (2018). Globalists: The End of Empire and the Birth of Neoliberalism. Harvard University Press.
  4. Thiel, P. (2009, 13 April). The Education of a Libertarian. Cato Unbound. https://www.cato-unbound.org/2009/04/13/peter-thiel/education-libertarian/
  5. Slobodian, Q. (2023). Crack-Up Capitalism: Market Radicals and the Dream of a World Without Democracy. Metropolitan Books.
  6. Marquette Law School Poll (2026, 5 August). New Marquette Law School Poll national survey finds inflation and cost of living the most important issue; economy next most important. https://law.marquette.edu/poll/2026/08/05/new-marquette-law-school-poll-national-survey-finds-inflation-and-cost-of-living-the-most-important-issue-economy-next-most-important/
  7. Marquette Law School Poll (2026, 16 September). National survey release, fielded 2–9 September 2026. https://law.marquette.edu/poll/2026/09/16/mlspsc35-1/
  8. McClain, C., & Park, E. (2026, 18 August). Young adults in the U.S. are increasingly wary of AI, concerned it will take jobs. Pew Research Center. https://www.pewresearch.org/short-reads/2026/08/18/young-adults-in-the-us-are-increasingly-wary-of-ai-concerned-it-will-take-jobs/
  9. Pew Research Center (2025, 15 October). How People Around the World View AI. https://www.pewresearch.org/global/2025/10/15/how-people-around-the-world-view-ai/
  10. Rogan, J. (Host). (2023, 6 October). #2044 – Sam Altman [Audio podcast episode]. The Joe Rogan Experience.
  11. O'Reilly, T. (2026, 12 July). Elon Musk is building a form of capitalism that Adam Smith would loathe. The Economist (By Invitation). https://www.economist.com/by-invitation/2026/07/12/elon-musk-is-building-a-form-of-capitalism-that-adam-smith-would-loathe
  12. Axios (2026, 23 August). Trump defends AI data centers amid growing backlash. https://www.axios.com/2026/08/23/trump-data-centers-michael-cohen-interview
  13. PBS News. EPA proposal could leave the public in the dark on data center plans. https://www.pbs.org/newshour/politics/epa-proposal-could-leave-the-public-in-the-dark-on-data-center-plans
  14. Congressional Research Service. Section 230: A Brief Overview (In Focus IF12584). https://www.congress.gov/crs-product/IF12584
  15. Altman, S. (2023, 16 May). Written testimony, "Oversight of A.I.: Rules for Artificial Intelligence." US Senate Judiciary Subcommittee on Privacy, Technology, and the Law. https://www.judiciary.senate.gov/download/2023-05-16-testimony-altman
  16. Andreessen, M. (2023, 16 October). The Techno-Optimist Manifesto. https://a16z.com/the-techno-optimist-manifesto/
  17. Zuckerberg, M. (2024, 23 July). Open Source AI Is the Path Forward. Meta. https://about.fb.com/news/2024/07/open-source-ai-is-the-path-forward/
  18. Wolsink, M. (2000). Wind power and the NIMBY-myth: institutional capacity and the limited significance of public support. Renewable Energy, 21(1), 49–64. https://doi.org/10.1016/S0960-1481(99)00130-5
  19. Devine-Wright, P. (2005). Beyond NIMBYism: towards an integrated framework for understanding public perceptions of wind energy. Wind Energy, 8(2), 125–139. https://doi.org/10.1002/we.124
  20. Schively, C. (2007). Understanding the NIMBY and LULU phenomena: Reassessing our knowledge base and informing future research. Journal of Planning Literature, 21(3), 255–266. https://doi.org/10.1177/0885412206295845
  21. Gallup (2026, 13 May). Americans oppose data centers in their area. Survey fielded 2–18 March 2026, n=1,000. https://news.gallup.com/poll/709772/americans-oppose-data-centers-area.aspx
  22. Data Center Watch. Report on data centre projects blocked or delayed by local opposition, May 2024–March 2025. https://www.datacenterwatch.org/report
  23. Pew Research Center (2025, 3 April). Views of risks, opportunities and regulation of AI. https://www.pewresearch.org/internet/2025/04/03/views-of-risks-opportunities-and-regulation-of-ai/
  24. Gallup and SCSP (2025). Americans prioritize AI safety and data security. Survey fielded 25 April–5 May 2025, n=3,128. https://news.gallup.com/poll/694685/americans-prioritize-safety-data-security.aspx
  25. Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at Work. The Quarterly Journal of Economics, 140(2), 889–942. https://doi.org/10.1093/qje/qjae044
  26. Barlow, J. P. (1996, 8 February). A Declaration of the Independence of Cyberspace. Electronic Frontier Foundation. https://www.eff.org/cyberspace-independence
  27. Barbrook, R., & Cameron, A. (1996). The Californian ideology. Science as Culture, 6(1), 44–72. https://doi.org/10.1080/09505439609526455
  28. Turner, F. (2006). From Counterculture to Cyberculture: Stewart Brand, the Whole Earth Network, and the Rise of Digital Utopianism. University of Chicago Press.