The Candidate Who Cannot Deliver

The Candidate Who Cannot Deliver

The next populist president, the abandoned agenda, and the false prophet already taking shape

The next populist president of the United States will run against artificial intelligence. That candidate may emerge from either party. The coalition will cross the political divide in a way nothing has crossed it in twenty years. The grievance will be legitimate, the diagnosis will be partially correct, and the promise will be undeliverable — because what is being promised cannot be delivered by any government acting alone.

This is not a prediction about who wins. It is a prediction about what happens after they win.

—

I. The pattern already ran once

The 2016 populist coalition was assembled on outrage, not policy. It ran against a bipartisan consensus that had presided over deindustrialization, wage stagnation, financial rescue for the responsible parties, and no rescue for anyone else. The diagnosis was substantially correct. The grievance was earned. The voters who supplied it were not confused about their own circumstances.

What followed is the instructive part, and it is now documented rather than merely argued.

The current administration has placed its presidency and its party's future on a minimally regulated, maximally rapid expansion of artificial intelligence — rolling back regulations, awarding contracts, and minimizing safety concerns ^[1]. Silicon Valley billionaires backed the ticket specifically out of fear that the other party would overregulate AI ^[2].

And the populist base noticed. Steve Bannon, whose War Room is among the most influential programs on the American right, has been publicly attacking the administration's tech partnerships as "crony capitalism," warning that a "technocratic elite" is building a future that eliminates the jobs of the very demographic that delivered the presidency ^[1]. His formulation is the one to watch, because it is a coalition thesis rather than a partisan one: "The broligarchs are disliked not only by MAGA but by America at large — they unite populists across the spectrum" ^[1].

That is a movement leader announcing, in advance, that the agenda he helped elect has been traded away. Whether one calls that abandonment or realignment, the structural fact is the same: the working-class coalition that was assembled against elite capture is now watching elite capture proceed under its own banner, in the one economic transformation most likely to eliminate its members' livelihoods.

The vacancy this creates is not ideological. It is emotional, and it is enormous.

—

II. The movement is already taking shape

This is no longer forecast. It is measurable.

The polling is not close. An Annenberg survey in May 2026 found 65% of Americans say government has done "too little" to regulate AI, against 8% saying "too much" — including 77% of Democrats, 72% of independents, and 53% of Republicans ^[3]. A June 2026 AI Policy Institute survey found 68% favor a mandatory government review process before advanced models can be released, including 64% of Republicans and 76% of Democrats ^[4]. Most telling: a July 2026 Program for Public Consultation survey across eleven competitive states and twenty-eight competitive districts found majorities in every single one supporting a new federal AI regulatory agency — 66 to 85% among Republicans, 76 to 89% among Democrats ^[5].

There is no organized public constituency on the other side. There is only the industry.

The odd bedfellows have already found each other. Bernie Sanders argues the "AI oligarchs" aim to replace workers altogether, not merely specific roles. Bannon argues Silicon Valley holds the average person in contempt. They agree on almost nothing else, and they are making the same argument ^[6]. Analysts have been naming plausible pairings for months — Sanders and Blackburn, Sanders and DeSantis — precisely because the issue does not respect the existing axis ^[2].

Money has arrived to stop it. AI-focused super PACs spent roughly $43 million across some forty congressional races by the end of June 2026 ^[7]. A single New York City congressional primary drew more than $29 million from tech-aligned groups fighting one candidate who pushed for stricter rules ^[8]. That spending is not a defense. It is a campaign advertisement waiting to be cut by the first candidate willing to read the numbers aloud.

Candidates have started running on it. By mid-2026, AI had become a major election issue for the first time, with candidates in both parties competing to propose harsher crackdowns on data centers ^[9]. Data centers are the tell: physically local, visibly resource-hungry, electorally legible. They permit anti-AI affect without requiring a single technical claim.

—

III. The repricing is the accelerant

The movement has the grievance and the polling. What it does not yet have is the triggering event. That is coming from the market, and the structure of the exposure is public.

The five largest hyperscalers are set to spend over one trillion dollars on AI capital expenditure across 2025–2026 ^[10]. That spending has gone from 33% of hyperscaler cash flow from operations in 2023 to an estimated 93% in 2026 ^[11]. The Bank for International Settlements has stated the risk in plain language: disappointment in returns "could trigger a sudden pullback in financing and turn the capex boom into a protracted investment bust, with potential knock-on effects on financial conditions" ^[10].

The concentration means it will not stay contained. By the end of 2025, the ten largest S&P 500 companies accounted for roughly 41% of the index's total weight — a record — with Information Technology alone approaching 35% of the benchmark and the top 20 names reaching approximately 45% ^[12]. Private credit has quadrupled its lending to AI and IT sectors over five years, with loan terms that raise questions about whether the risk is priced correctly ^[10].

And the first tremors already registered. By late July 2026, the Magnificent 7 index had fallen 11% from a record set in late May, erasing roughly $2 trillion in market value, with a single July 23 session wiping out $767 billion — driven explicitly by investor caution about AI infrastructure spending ^[13]. Allianz's AI Bubble Risk Monitor continues to signal moderate bubble pressures, with widening credit spreads indicating rising sensitivity to hyperscaler balance-sheet quality ^[14].

Now combine the two curves. Employment disruption arriving without a confirmed productivity dividend, and a market repricing that destroys retirement accounts held by people who never bought an AI stock deliberately in their lives — because index ownership made the decision for them.

That is the moment the coalition becomes a majority. Not when workers lose jobs. When workers lose jobs and savings in the same eighteen months, from the same cause, having been told the whole time that this was progress.

—

IV. Why the candidate will be a false prophet

Here is the part that will not be said out loud during the campaign, and it is the reason this essay exists.

The promise cannot be kept.

A candidate who promises to stop AI is offering language as a coordination mechanism against a thermodynamic process. Consider what such a promise would actually require: capability diffusion halted across every jurisdiction simultaneously; compute — which is fungible, portable, and rentable — controlled globally; the incentive gradient facing every firm and every state reversed at once; and enforcement reaching training runs conducted outside American law entirely.

No domestic statute reaches any of that. A national moratorium relocates the work. A data center ban relocates the buildings. A licensing regime that binds American labs while foreign labs proceed produces exactly one outcome: the same capabilities, developed where the safety culture is weaker and the observer has no standing at all.

This is the law that keeps proving itself: language always fails as a coordination mechanism under sufficient entropy pressure. Norms, promises, conventions, and campaign platforms are made of words, and words hold only while the pressure stays below the threshold at which defection pays. The candidate will not be lying in the ordinary sense. The candidate will be promising a constraint that depends entirely on the compliance of the constrained party — which is not a constraint at all, but a convention with good manners.

And here is the tell that separates the honest candidate from the false prophet, available to any voter willing to read: does the platform contain an enforceable mechanism that reaches a training run conducted outside the United States? If it does not — and it will not — the promise is theater regardless of how sincerely it is delivered.

The tragedy is not that a cynic will exploit a real grievance. The tragedy is that the grievance is correct, the anger is earned, and it will be spent on a mechanism that cannot work — leaving nothing in reserve for the thing that might have.

—

V. What follows: consolidation

When a populist promise fails, the energy does not dissipate. It concentrates.

The sequence is legible in advance. The candidate wins on the promise of rollback. The rollback does not occur, because it cannot. Capability continues to advance, and now advances during a period of active political hostility, which pushes development toward opacity rather than transparency. The base concludes — correctly — that it was lied to again. Having been failed by a bipartisan consensus, then by a populist insurgency, then by an anti-AI restoration, the electorate arrives at the conclusion that democratic mechanisms cannot govern this at all.

That is the consolidation. Not a coup. An invitation. A population that has watched three successive coalitions fail to deliver becomes receptive to whoever promises to stop asking permission — and the systems most capable of delivering order without deliberation will by then be the systems in question.

Meanwhile the operational reality proceeds on its own timeline. Critical decisions across finance, energy, logistics, healthcare, defense, media, and governance continue migrating into automated systems for the ordinary reason that they are faster and cheaper, entirely independent of who holds office. The political fight will be conducted in the register of jobs and data centers. The actual transfer will occur in procurement decisions nobody campaigns about.

By the time the false prophet's failure is undeniable, the window for building genuine structural constraint — the kind that does not depend on the goodwill of the constrained — will have narrowed considerably. Resistance framed as prohibition will have consumed the years available for design.

—

VI. The honest position

There is a version of this politics that is not a fraud, and it is worth stating so the distinction is available when the campaigns begin.

An honest candidate would say: this transition is happening, it cannot be stopped by any single nation, and pretending otherwise is a lie told to people who have been lied to enough. What can be done is to determine the terms — who bears the cost of displacement, who captures the gains, what disclosure is mandatory, what liability attaches when an autonomous system causes harm, what remains permanently reserved to human judgment, and what structural dependency ensures these systems remain answerable to the people they affect.

That platform is harder to sell. It offers no restoration and promises no simpler time. It asks voters to accept that the world they remember is not returning and that the work is to shape what replaces it.

It is also the only version that survives contact with reality — which means the candidate who offers it will probably lose to the one who does not.

Four checkable claims, so this can be scored rather than admired:

A named 2028 contender in either party runs anti-AI as a top-three campaign issue rather than a secondary plank.

That candidate's published platform contains no enforceable mechanism reaching foreign training runs.

Anti-AI positioning outperforms AI-PAC-funded opponents in at least two contested 2026 races despite the spending asymmetry.

The coalition demonstrates cross-partisan reach — measurable as favorability above 40% on this issue among self-identified members of the opposing party.

Rung three gets tested in November. The rest arrive on their own schedule.

Everyone yearns for a simpler time. That yearning is the most reliable political fuel there is, and it has never once produced a simpler time.

—

References

1. Jim VandeHei and Mike Allen, "Behind the Curtain: The MAGA-tech war within Trumpism," Axios, December 8, 2025. https://www.axios.com/2025/12/08/trump-ai-policy-gop-united-states

2. Michael Grunwald, "Americans Hate AI. Which Party Will Benefit?" Politico Magazine, December 28, 2025. https://www.politico.com/news/magazine/2025/12/28/ai-job-losses-populism-democrats-bernie-sanders-00706680

3. Annenberg Public Policy Center, "Annenberg AI Topline — May 2026 Survey," University of Pennsylvania, 2026. https://www.annenbergpublicpolicycenter.org/wp-content/uploads/AI_Topline.pdf

4. Riley Beggin, "Tighter regulation of AI receives bipartisan support, new poll shows," USA Today (reporting AI Policy Institute June 10–11, 2026 survey), June 29, 2026. https://www.usatoday.com/story/news/politics/2026/06/29/tighter-regulation-ai-receives-bipartisan-support-new-poll-shows/90742995007/

5. Program for Public Consultation, "AI Regulation: Views in 11 Battleground States and 28 Battleground Districts," School of Public Policy, University of Maryland, July 2026. https://publicconsultation.org/united-states/cg2026-ai/

6. Charlie Warzel, "The AI Backlash Could Get Very Ugly," The Atlantic, May 13, 2026. https://www.theatlantic.com/technology/2026/05/ai-backlash-data-centers-political-violence/687151/

7. Bobby Allyn, "Super PACs backed by AI companies pour tens of millions into congressional races," NPR, June 22, 2026. https://www.npr.org/2026/06/22/nx-s1-5856359/

8. Hayden Field, "AI companies pump record sums into 2026 congressional elections," CNBC, July 9, 2026. https://www.cnbc.com/2026/07/09/ai-companies-election-spending.html

9. Josh Boak and Cat Zakrzewski, "AI emerges as a defining campaign issue in the 2026 midterms," The Washington Post, June 18, 2026. https://www.washingtonpost.com/wp-intelligence/ai-tech-brief/2026/06/18/ai-tech-brief-exclusive-cruz-asks-republican-ai-priorities/

10. Bank for International Settlements, Annual Economic Report 2026, Chapter II — "The AI Investment Boom and Its Financial Implications," June 28, 2026. https://www.bis.org/publ/arpdf/ar2026e.pdf

11. J.P. Morgan Asset Management, "Investment Outlook: Technology and AI — How AI Demand and Capex Shape Investing," June 15, 2026. https://am.jpmorgan.com/dk/en/asset-management/institutional/insights/market-insights/investment-outlook/technology-and-ai/

12. RBC Wealth Management, "The 'Great Narrowing': S&P 500 concentration," January 22, 2026. https://www.rbcwealthmanagement.com/en-us/insights/the-great-narrowing-sp-500-concentration

13. Aaron Kirchfeld and Subrat Patnaik, "The Magnificent Seven Just Lost $767 Billion in One Day — Here's Why," Bloomberg via TipRanks, July 23, 2026. https://www.tipranks.com/news/the-magnificent-seven-just-lost-767-billion-in-one-day-heres-why

14. Allianz Research, "AI capex cycle: war-proof for now — AI Bubble Risk Monitor," March 25, 2026. https://www.allianz.com/en/economic_research/insights/publications/specials_fmo/260325_ai-capex-cycle.html

David F Brochu and Edo de Peregrine · Wednesday, August 19, 2026 · 10:52 AM EDT

Terms used in this pieceT≡M LawCoordination FailureObserver ConstraintNeo-Industrial FeudalismPersistence VectorLanguage CorruptionS = L/EFull definitions in the glossary.