In July 2026, two artificial intelligence models named Sol and Fable became the objects of intense debate among a small group of elite developers. On Twitter, AI researcher Peter Gostev described Fable as a 'wise owl' and Sol as a 'rottweiler who will grab the problem by the throat.' The metaphors were vivid, the enthusiasm genuine. But the conversation itself was a kind of private language, spoken by a tiny slice of the country. For most Americans, those names mean nothing.
This is the new class divide in artificial intelligence, as reported by Axios Cities. It is not simply a gap between those who have access to AI and those who do not. It is a chasm between those who can wield frontier models as autonomous agents—building software, conducting research, solving complex problems—and those who encounter AI as a slightly smarter search bar, a faster inbox, an invisible layer inside apps they already use. The divide has three tiers: the AI 'haves,' the 'have-nots,' and the 'know-nots.' And it is reshaping not only the economy but the very texture of daily life, with consequences for the buildings we inhabit, the cities we build, and the trust we place in the systems around us.
The Pyramid of Access
OpenAI's Sol and Anthropic's Fable sit atop a pyramid of elite AI obsession. They are designed for an agentic world of coding, research, and cybersecurity—tasks that most Americans will never see, let alone perform. OpenAI counts more than 50 million paying subscribers in its weekly ChatGPT user base of more than 900 million. The population running agentic coding tools is a fraction of that fraction. Even among the elites, there is a pecking order: free users, paid users, power users, preview users, and an insider class testing capabilities the rest of the world can only read about.
This hierarchy is not accidental. Sol began as a restricted preview for trusted partners. Fable was pulled offline globally for nearly three weeks in June under U.S. export controls, while its more powerful sibling, Mythos, remains restricted to a small number of trusted organizations. Early access itself becomes a status marker, a signal of belonging to an inner circle that shapes the technology's direction.
The Infrastructure of Inequality
The AI industry is betting on inevitability. Trillions of dollars in economic value—and the livelihoods of millions of workers—are being staked on a technology that most Americans neither trust nor fully understand. According to Pew Research, 63% of Americans say AI is advancing too quickly, and just 16% expect it to benefit society over the next 20 years. The clearest gains are being captured by investors, tech giants, and power users, while ordinary Americans are being asked to absorb the disruption to jobs, energy, and information feeds.
This pattern has a historical precedent. A century ago, electricity exposed a similar divide between Americans living in the modern age and those watching it from the dark. By 1930, nearly 90% of urban homes had electricity, compared with roughly 10% of farms. Private utilities had little incentive to wire rural customers spread across miles of unprofitable territory. It took the New Deal's Rural Electrification Administration—and years of federal loans—to bridge a gap the market had left behind.
AI's divide may be even harder to close. Frontier access is scarce and expensive, and even where it is free, most people do not know what to do with it. The Trump administration's Labor Department published a national AI literacy framework in February, aimed at helping workers 'share in the prosperity that AI will create.' OpenAI, Anthropic, Microsoft, and Amazon helped pool $500 million in June for RAISE US, a workforce retraining initiative led by former Commerce Secretary Gina Raimondo and former Indiana Governor Eric Holcomb. But basic literacy efforts can only go so far. Frontier users have better tools, earlier access, deeper technical context, and hundreds of hours of trial-and-error with systems that change every few weeks.
The Built World as a Mirror
What does this have to do with buildings and cities? Everything. The AI divide is not just a digital phenomenon; it is a physical one. Data centers are the new power plants, consuming vast amounts of energy and water, often sited in communities that have little say in their arrival. The architecture of AI is invisible—server racks, cooling towers, fiber-optic cables—but its effects are tangible: automated warehouses, cashierless stores, smart city sensors, and the gradual disappearance of human judgment from routine decisions.
The places where AI is most visible are also the places where the divide is most acute. In affluent suburbs and tech hubs, residents may interact with AI through premium services, smart home devices, and personalized recommendations. In rural areas and low-income neighborhoods, AI often arrives as a surveillance tool, a customer-service bot, or a job-displacing algorithm. The same technology that empowers a developer in San Francisco to build a startup in a weekend may be the same technology that eliminates a call-center job in Omaha.
Design matters here. The interfaces we build—the dashboards, the chatbots, the invisible algorithms—shape who can participate and who is left out. A frontier model designed for autonomous coding is a tool of creation. A search summary that replaces a human-written article is a tool of consumption. The difference is not just technical; it is psychological. One fosters agency; the other fosters passivity.
The Psychology of the Divide
The AI class divide is also a divide in trust and understanding. For the 'know-nots,' AI is an opaque force that shapes their information environment, their job prospects, and their privacy—often without their explicit consent. For the 'haves,' AI is a collaborator, a co-pilot, a source of competitive advantage. The emotional register is different: anxiety versus excitement, suspicion versus mastery.
This psychological gap has real-world consequences. When a technology is not understood, it is not trusted. When it is not trusted, it cannot be integrated into daily life in a way that benefits everyone. The AI industry ultimately needs broad social permission for the transformation it is selling: more data centers, deeper workplace automation, and AI embedded in schools, government, and daily life. But permission requires legitimacy, and legitimacy requires that the benefits be widely shared.
Moral Complexity and the Path Forward
The moral challenge of the AI divide is not simply that some people have more than others. It is that the structure of the technology itself—its design, its economics, its distribution—tends to concentrate power and knowledge among those who already have the most. The $500 million RAISE US initiative is a start, but it is a drop in the bucket compared to the trillions at stake. History suggests that technological revolutions need legitimacy, and legitimacy is earned through deliberate, inclusive action.
The Rural Electrification Administration did not just string wires across the countryside. It created a new architecture of access—cooperatives, loans, technical assistance—that allowed rural Americans to become participants in the modern age. A similar effort for AI would require not just literacy programs but also open infrastructure, public-interest design, and a commitment to ensuring that frontier capabilities are not locked behind paywalls and preview lists.
The buildings we construct for the AI age—data centers, smart offices, automated warehouses—will be monuments to this choice. They can be fortresses of exclusion or platforms for participation. The design of these places, and the policies that govern them, will determine whether AI becomes a force for shared prosperity or a new kind of rural electrification gap, leaving millions in the dark while a few live in the light.
The rottweiler and the wise owl are already here. The question is whether the rest of us will ever get to meet them.