America’s next major economic disruption may not hit factory floors first. It could arrive in office towers, university hubs, and urban business districts where white-collar workers have long viewed themselves as insulated from automation.
A new study from Tufts University warns that AI job displacement could become heavily concentrated in knowledge-driven cities across the United States, potentially creating economic and political fallout comparable to the manufacturing decline tied to the so-called China shock of the early 2000s.
Bhaskar Chakravorti, dean of global business at Tufts University, developed the American AI Jobs Risk Index to map which regions and occupations face the greatest exposure to AI automation. The model assesses 784 occupations and estimates that 9.3 million American jobs could be vulnerable, representing roughly $200bn in lost income under moderate assumptions.
The projections become significantly larger under more aggressive automation scenarios. Chakravorti estimates income losses could rise to $1.5tn if companies adopt AI systems capable of replacing a broader share of white-collar labour.
Silicon Valley and university hubs face highest exposure
Unlike the manufacturing downturn that hit industrial cities such as Detroit and Cleveland, the AI disruption appears concentrated in metropolitan areas built around technology, research, and professional services.
Chakravorti said cities including San Jose, Seattle, Boston, New York, and the Raleigh-Durham corridor face disproportionately high exposure because of their concentration of office-based knowledge work.
“There are 14 knowledge-driven metros,” Chakravorti told Fortune, adding that these regions could experience job losses and income declines far greater than traditional manufacturing centres.
The findings arrive as companies continue integrating generative AI tools into customer service, coding, finance, legal research, marketing, and administrative work. While the broader US labour market remains relatively resilient, layoffs linked to AI are becoming more visible across the technology sector.
Outplacement firm Challenger, Gray & Christmas recently reported more than 49,000 layoffs associated with AI automation so far this year, approaching the total recorded during all of 2025. Some of those reductions also reflect cost-cutting measures by technology groups redirecting spending toward AI infrastructure and data centres.
The unemployment rate within the technology sector rose to 3.8% last month, according to recent labour market data, although it remains below the broader national unemployment rate of 4.3%.
Why economists see echoes of the China shock
The comparison to the China shock reflects more than employment losses alone. Economists David Autor, David Dorn, and Gordon Hanson used the term to describe the long-term impact of Chinese import competition on American manufacturing communities after China joined the World Trade Organization in 2001.
Research from Autor and his colleagues found that many affected regions struggled for years with lower wages, weaker labour participation, and political instability after factories closed or moved overseas.
Some business leaders now fear a similar adjustment could unfold in white-collar industries. Microsoft AI chief Mustafa Suleyman recently suggested that a large share of entry-level office jobs may disappear within the next two years as AI systems improve.
Even JPMorgan chief executive Jamie Dimon acknowledged recently that earlier promises to retrain workers affected by free trade agreements such as NAFTA largely failed. Speaking alongside Anthropic chief executive Dario Amodei, Dimon said reskilling programmes “didn’t work” because they were not designed effectively.
However, several economists argue the comparison may ultimately prove positive for productivity growth. Apollo Global Management chief economist Torsten Slok recently noted that lower-cost Chinese imports eventually helped improve US manufacturing output and efficiency, despite painful regional disruptions.
US real manufacturing output increased roughly 50% between 2001 and 2024, according to Slok’s analysis. He believes AI could produce a comparable productivity surge across service industries by reducing operating costs and accelerating business creation.
McKinsey estimated in a 2023 report that generative AI could add between $2.6tn and $4.4tn annually to the global economy if adoption continues across sectors ranging from banking to healthcare.
Political consequences may extend beyond the labour market
Chakravorti believes the political consequences of AI job displacement could become especially significant because the affected workers differ sharply from those impacted during the manufacturing downturn.
“These are people who are on LinkedIn,” he said. “They know their congressman’s phone number.”
Unlike many industrial communities affected by factory closures, displaced white-collar workers may possess stronger digital networks, communication skills, and political influence. That could intensify pressure on lawmakers to respond with new labour protections, retraining programmes, or limits on AI deployment.
For now, the broader economy continues to absorb AI adoption without a major employment shock. But the concentration of risk in high-income urban regions suggests the next phase of automation may reshape parts of corporate America that previously appeared economically secure.
