Delays Become Permanent Damage
AI is splitting the country by zip code & the gap is solidifying
Thirty metro areas now generate two-thirds of all AI-related job postings in the United States. Washington D.C. uses AI nearly four times the national average while Mississippi uses one-fifth the average.[1] These differences will not play out in tech. AI is already driving credit, labor, and inequality differentials.
Investment in AI has now surpassed the GDP contribution of IT during the entire dot-com boom, according to the St. Louis Fed.[2] This is not about forecasting a technology cycle. It is about where wealth is going to be created and where it is not.
This also has important regional implications in both the short and long run. For public finance, this concentration of AI use will drive long-term credit profiles, even if the AI data isn’t apparent now. Municipal bond analysts should be watching this. The divergence between high-adoption and low-adoption regions will show up in tax bases, employment levels, and debt-service capacity within a few years.
AI Power Users are concentrated in high tech communities like San Francisco, New York City and Boston. And lower AI adoption rates are in places like the Industrial Midwest or Rural South which are likely to have lower concentrations of Power Users. The variation in AI attitude can be a prominent issue in inequality. This is not a casual observation. Anthropic’s Claude usage is 3.82 times the norm in Washington D.C. and 0.21 times it in Mississippi.[3] That eighteen-to-one ratio is a leading indicator of the AI induced problems to come.
Attitudes Drive AI Demand
Who uses AI and who does not is an economy-wide issue, extending beyond the individual. Attitudes toward AI appear in data center protests, labor market unrest, and new state regulations on AI. These are educated responses. This is rational behavior from people who have seen technology fail them before. AI is in many ways, though, a different technology, which deserves a more nuanced response. AI comes with problems, but also a wellspring of opportunity.
There is no simple taxonomy of AI users. Andrej Karpathy, the ex-Tesla AI director, and OpenAI founding member, has identified a clear grouping of attitudes about AI.[4] He has warned of a widening gap between AI Power Users and Skeptics. Power Users invest time learning AI systems and get strong, profitable results. Doubters want to see the data and get further evaluation before they commit serious time and resources. Deniers have analyzed AI's potential and concluded it is overhyped. Resisters, including Neo-Luddites and vested institutions, are not opposed to all technology, just AI. Power Users and Skeptics, as Karpathy says, seem to be “speaking past each other.” His framework is useful but incomplete. He describes individuals. He leaves out the divide playing out in cities, regions, and industries.
Half of American adults are currently more concerned about AI than excited about it, according to Pew Research.[5] It is a number that has grown since 2021. These are not technophobes. Generally, they want more proof. The missing element, though, is that there is a high cost to themselves and their communities from spreading unnecessary uncertainty about AI. Karpathy’s framework names the poles. It misses the majority and it misses the geography.
eBooleant’s research across domestic and international professionals finds something the aggregate data misses: adoption is happening function by function, not attitude by attitude. We find that professionals are not dividing into power users and skeptics. They are integrating AI where it fits their lives and declining it where it doesn’t. We explore that function-level decision-making, which is lost in the aggregate data. But it explains why blanket AI mandates fail.
At Tech Gather NYC this past Thursday, a professional said two of his friends had recently been let go and replaced by AI. It happened suddenly, not warned, not transitioned, but replaced. His shock is being mirrored across the country as the reality of AI comes into force. AI is here and needs to be reckoned with, not just criticized.
The Gap That Locks In
You, the reader, need to decide which side of that divide you are on and whether you can afford to wait. The question is whether the regions, firms, and individuals that are on AI’s wrong side can close the gap before it is permanent.
Philip J. Fischer is a finance PhD and former Wall Street Managing Director. He is the author of Investing in Municipal Bonds (McGraw-Hill) and a forthcoming book on AI.
[1] Mapping the AI Economy: Which Regions Are Ready for the Next Technology Leap?, Mark Muro and Shriya Methkupally, October 2025. https://www.brookings.edu/articles/mapping-the-ai-economy-which-regions-are-ready-for-the-next-technology-leap/
[2] Federal Reserve Bank of St. Louis, Tracking AI's Contribution to GDP Growth, Hannah Rubinton and Bontu Ankit Patro, January 26, 2026. https://www.stlouisfed.org/on-the-economy/2026/jan/tracking-ai-contribution-gdp-growth
[3] Mapping the AI Economy: Which Regions Are Ready for the Next Technology Leap?, Mark Muro and Shriya Methkupally, October 2025. https://www.brookings.edu/articles/mapping-the-ai-economy-which-regions-are-ready-for-the-next-technology-leap/; Anthropic Economic Index Report: Uneven Geographic and Enterprise AI Adoption, Ruth Appel, Peter McCrory, Alex Tamkin et al., September 15, 2025.
https://www.anthropic.com/research/anthropic-economic-index-september-2025-report
[4] Karpathy, Andrej (@karpathy). Post on X (formerly Twitter), April 2026. https://x.com/karpathy/status/2042334451611693415
[5] Pew Research Center, How Americans View AI and Its Impact on People and Society, survey conducted June 9–15, 2025, published September 17, 2025. https://www.pewresearch.org/science/2025/09/17/how-americans-view-ai-and-its-impact-on-people-and-society/
