The AI Market Is Dining on a Different Planet Than American Consumers

Somewhere in a San Jose boardroom, an AI fund manager is posting screenshots of 121-percentage-point outperformance. Somewhere in a Cleveland living room, a household that has not taken a vacation in three years is calculating whether the grocery bill will clear before the credit card minimum. These two Americans are not experiencing the same economy. They are not even experiencing the same decade.

The Polymarket wire, aggregated on 22 May 2026, captured both realities simultaneously. AI stocks have lapped the non-AI S&P 500 since the start of 2024. US consumer sentiment crashed to a record low the same week, with 57 percent of respondents telling pollsters that high prices are actively eroding their personal finances. The AI investment community is pricing in a future of agentic workflows, inference scale, and data-center saturation. The consumer class is pricing in a present of stagnant wages, housing cost burdens, and grocery lists that no longer fit the monthly budget. These are not contradictory signals to be smoothed away by aggregate statistics. They are the same economy, photographed from opposite ends of the wealth distribution.

The Zoom Anomaly Is the Rule

Zoom's reported billion-dollar paper gain on its Anthropic stake — confirmed by Polymarket on 23 May, with Zoom shares climbing 10 percent on strong AI feature demand the prior day — is presented as a curiosity. It should not be. It is the logical endpoint of a market structure that has routed AI infrastructure returns exclusively to investors who were already positioned before the 2022–2023 rate cycle turned. A video-conferencing company with a profitable core business deployed capital into a frontier AI lab, and that bet now returns more than the core business it guards. This is not innovation. This is leverage on innovation — the kind that concentrates wealth when it works and distributes losses broadly when it breaks.

The AI data-center moratorium that Polymarket puts at 92 percent probability of passing by year-end is, in this context, a regulatory pin catching a very large bubble thread. Communities from Prince William County to Mesa, Arizona, have spent two years documenting groundwater depletion, noise pollution, and grid strain from facilities that were approved fast and are now asking for more power than local utilities can allocate. The environmental opposition is not, at its core, anti-technology. It is a friction layer that the boom skipped over in its haste to build ahead of demand. That friction is now structural.

Crypto Securities and the Regulatory Theater

The SEC's decision to delay permission for "crypto versions" of US equities — also reported on 22 May — sits oddly with the broader deregulatory posture of the current commission. The delay is framed as a technical or risk-management choice. It may be that. It may also be that the commission has calculated that a retail-accessible tokenized stock product would channel speculative flows away from AI narratives and into a competing asset class that is far more exposed to the kind of leverage unwind that regulators do not want to manage before a market correction.

What the SEC is not doing is explaining the decision in those terms. The wire carries the headline; the reasoning is absent. That opacity is itself a data point. When regulatory agencies act in ways that coincidentally protect dominant market narratives — in this case, AI equity premium — without stated justification, the rational inference is that the narrative protection is the point.

The IRS Item Nobody Is Reading

Buried in the same wire batch: the IRS is reportedly considering asking taxpayers to disclose citizenship status on next year's forms. This is a tax administration item, and it will be covered as such by tax trade publications. But its implications are broader. Citizenship-status disclosure, if implemented, creates a two-tier taxpayer classification system that will surface in compliance litigation, in immigration-law practice, and in the planning documents of every multinational operating inside the US tax net. It is also, not incidentally, a mechanism for cross-referencing the informal economy — gig workers, unbanked households, cash businesses — against formal filers. The households whose consumer sentiment scores are at record lows are also, disproportionately, the households with the most complicated relationship to formal tax status. This policy, if it lands, does not help them.

What the Divergence Costs

The structural problem is not that AI is succeeding or that consumers are wrong to feel squeezed. Both things are true simultaneously, and the media's tendency to treat them as competing headlines rather than correlated phenomena is a failure of systemic analysis. When a market sector consistently outperforms by 121 percentage points over two years, it is either pricing in genuine productivity gains that will eventually filter into wages and consumption — or it is pricing in a future that requires a consumer class with no disposable income to function. The current trajectory is the latter. AI companies are selling efficiency to firms. Those firms are reducing headcount. The displaced workers are not, in aggregate, becoming prompt engineers. They are becoming price-sensitive consumers whose demand signals are already visible in the sentiment data and will eventually become visible in revenue reports for everything that is not AI infrastructure.

The data-center moratorium is the only mechanism in the current pipeline that introduces friction into this loop. It will not redirect the wealth flows. But it may slow the buildout long enough for the regulatory conversation to catch up to the fact that an economy in which AI stocks and consumer sentiment move in opposite directions is not a stable equilibrium. It is a distribution crisis that has been financialized and is currently being mistaken for a technology story.

This piece was structured around the tension between Polymarket's AI-sector performance data and its consumer sentiment wire from 22–23 May 2026. Monexus is monitoring the data-center moratorium vote and the SEC delay for tokenized equities as leading indicators of whether the regulatory layer will intervene before the market layer corrects.