The Rust Belt conjures images of shuttered steel mills and hollowed-out factory towns across the industrial Midwest. But a new geography of economic vulnerability is forming in the United States, and it doesn’t look anything like the old one. This time, the threat isn’t globalization or automation of manual labor. It’s artificial intelligence. And the communities most exposed aren’t blue-collar strongholds — they’re the white-collar, digitally connected metro areas that were supposed to be the winners of the knowledge economy.
Researchers at Tufts University’s Digital Planet initiative have coined a term for these newly vulnerable regions: “Wired Belts.” The phrase is deliberately provocative. It’s meant to echo the Rust Belt, but with a twist — the very connectivity and digital sophistication that made these places prosperous now makes them sitting ducks for AI-driven displacement.
The research, led by Bhaskar Chakravorti, Dean of Global Business at Tufts University’s Fletcher School, along with Ravi Shankar Chaturvedi and Christina Filipovic, maps AI job risk across all 50 states and more than 380 metropolitan statistical areas. Their findings, published by Tufts Digital Planet, upend the conventional wisdom about which American workers should be worried.
“The places that rode the wave of the internet economy are the ones most exposed to AI disruption,” Chakravorti told Tufts Fletcher School. The logic is straightforward once you see it. AI doesn’t replace the work of hands. It replaces the work of minds — data analysis, report generation, routine coding, financial modeling, administrative coordination, content creation. The jobs that cluster in prosperous, tech-forward metros.
The states most at risk read like a roll call of American economic success: Washington, D.C., sits at the very top of the vulnerability index, followed by states like New Jersey, Maryland, Massachusetts, Connecticut, and Virginia. California, home to Silicon Valley, ranks high as well. These aren’t backwaters. They’re the places with the highest concentrations of college-educated workers, the highest median incomes, the deepest integration into the digital economy.
That’s precisely the problem.
The Tufts researchers constructed their analysis by examining the occupational composition of each metro area and state, then cross-referencing those profiles against assessments of how susceptible each occupation is to AI automation. They drew on data from the Bureau of Labor Statistics, the U.S. Census, and established AI exposure indices developed by researchers at institutions including OpenAI and the University of Pennsylvania. The resulting maps don’t just show which jobs are at risk — they show where those jobs are concentrated geographically, and therefore which local economies face the greatest aggregate threat.
Metro areas dominated by government, professional services, finance, and information technology scored highest on the AI risk index. Washington, D.C.–Arlington–Alexandria topped the metro rankings, which shouldn’t surprise anyone who’s spent time in the capital region. The federal workforce and its vast contractor ecosystem are heavy with the kind of cognitive, routine-adjacent tasks that large language models and AI agents are already beginning to perform. Think of the policy analyst who synthesizes reports, the procurement specialist who processes contracts, the paralegal who reviews regulatory filings. All of them face some degree of exposure.
But this isn’t just a D.C. story. San Jose–Sunnyvale–Santa Clara, the heart of Silicon Valley, ranks among the most exposed metros. So do Boston, San Francisco, Seattle, and Austin — the very cities that have attracted billions in venture capital and tech talent over the past two decades. The irony is thick. The places that built AI are the places most likely to be reshaped by it.
By contrast, the metros least exposed to AI disruption tend to be smaller, more rural, and more reliant on physical labor — agriculture, construction, extraction, and hands-on services. Places like Yuma, Arizona. El Centro, California. Dalton, Georgia. These communities have their own economic challenges, to be sure, but AI isn’t about to replace the work of fruit pickers, roofers, or home health aides. Not yet, anyway.
The Tufts team is careful to distinguish between AI “exposure” and outright job elimination. Exposure means a significant portion of a job’s tasks could be performed or augmented by AI systems. That doesn’t automatically mean the job disappears. It might mean the job changes — fewer people needed, different skills required, altered workflows. But the distinction between transformation and elimination offers cold comfort to a metro area where 60% or more of the workforce holds positions with high AI exposure. Even partial displacement, spread across an entire regional economy, can trigger serious disruptions in housing markets, tax bases, consumer spending, and public services.
This matters for policy in ways that are only starting to become clear. The federal government, state legislatures, and city councils have spent decades designing economic development strategies around attracting knowledge workers and tech companies. Tax incentives, broadband buildouts, university partnerships, innovation districts — all of it predicated on the idea that digital connectivity and cognitive labor are the path to prosperity. The Wired Belt thesis suggests that strategy may now carry a hidden downside. Communities that went all-in on the knowledge economy have, in effect, concentrated their risk.
Chakravorti and his colleagues argue that policymakers need to think about AI resilience the way they think about climate resilience — as a structural challenge requiring proactive planning, not just reactive cleanup. That means investing in workforce retraining programs tailored to AI-adjacent skills, diversifying local economies so they aren’t overly dependent on one type of cognitive labor, and building social safety nets that can absorb the shock of rapid occupational change.
The comparison to the Rust Belt isn’t just rhetorical. It carries a warning. When manufacturing left the Midwest, the political and social consequences were enormous — rising opioid addiction, population decline, the fracturing of civic institutions, and a lasting sense of betrayal that reshaped American politics for a generation. If AI displacement hits knowledge-economy metros with similar force, the fallout could be equally profound, though it would look different. Instead of empty factories, think half-occupied office towers. Instead of laid-off steelworkers, think underemployed MBAs and software engineers competing for a shrinking pool of positions that still require a human touch.
And there’s another dimension the research highlights: inequality within metros. AI exposure isn’t evenly distributed across income levels or demographic groups. Workers in mid-tier cognitive roles — think bookkeepers, administrative assistants, customer service representatives, junior analysts — face the highest immediate risk. These are often middle-class jobs held by workers without advanced degrees or specialized technical skills. The highly paid executives and senior engineers who design AI systems are, for now, less exposed. The result could be a hollowing out of the middle of the labor market, widening the gap between the top and the bottom.
The racial and geographic equity implications are significant too. The Tufts data shows that certain states with large minority populations in white-collar roles — Maryland and Virginia, for instance, where the federal government has long been a major employer of Black professionals — face disproportionate exposure. If AI reduces headcount in federal agencies or government contracting firms, the impact on Black middle-class wealth could be severe.
So what are companies and governments actually doing about this? The honest answer: not nearly enough. Most corporate AI adoption strategies focus on productivity gains and cost reduction, not workforce transition. Most state workforce development programs are still oriented around the last disruption — manufacturing automation and the shift to services — rather than the next one. A few forward-looking initiatives exist. Singapore, for instance, has launched national AI upskilling programs. Some U.S. states, including Colorado and Virginia, have begun pilot programs for AI literacy training. But these efforts are small relative to the scale of the challenge.
The tech industry itself sends mixed signals. On one hand, companies like Microsoft, Google, and OpenAI talk constantly about AI as a tool that augments human workers rather than replacing them. On the other hand, those same companies are laying off thousands of employees while investing billions in AI infrastructure. The actions and the rhetoric don’t always align.
There’s a temporal dimension here that deserves attention. The Rust Belt didn’t collapse overnight. It eroded over decades, as trade agreements, automation, and corporate offshoring gradually ate away at manufacturing employment. The Wired Belt disruption could move faster. AI capabilities are improving at a pace that far exceeds the speed of industrial automation in the 20th century. Large language models that could barely write a coherent paragraph three years ago can now draft legal briefs, generate code, and analyze complex datasets. The window for proactive policy intervention may be narrower than anyone thinks.
The Tufts researchers also note that AI risk doesn’t respect state boundaries in neat ways. Supply chains of cognitive labor — outsourced accounting, remote customer service, distributed software development — mean that disruption in one metro can ripple outward. A law firm in New York that adopts AI to handle document review might reduce its need for paralegals in New Jersey. A tech company in San Francisco that automates QA testing might cut contractors in Austin. The interconnectedness that defines the modern economy ensures that AI displacement will be a national phenomenon, even if its intensity varies by geography.
None of this means the knowledge economy is doomed. Far from it. AI will create new jobs, new industries, new forms of value. It always has, with every major technological shift. But the transition costs are real, and they fall unevenly. The Wired Belt framework is valuable precisely because it forces a reckoning with geography — with the fact that economic disruption isn’t abstract, it’s local. It hits specific towns, specific neighborhoods, specific families.
The old Rust Belt never fully recovered. Trillions of dollars in wealth, decades of social cohesion, and millions of livelihoods were lost before anyone in Washington took the problem seriously. The question now is whether America will repeat that mistake with its Wired Belts — whether policymakers, business leaders, and communities will act before the map of AI vulnerability becomes a map of economic decline. The research from Tufts suggests the clock is already ticking. The geography of American prosperity is being redrawn, and the pen is in the hands of algorithms.
WebProNews is an iEntry Publication