The Two-State Problem

Updated September 2026

America is not entering the AI economy as one country. It is entering as two.

One economy is concentrated in a handful of metropolitan corridors where AI capital, talent, research, infrastructure and policy influence reinforce one another. The other includes rural towns, small cities, tribal nations, agricultural communities and resource regions that remain largely outside the institutions shaping AI's future.

We can now see that divide in actual AI use. Microsoft's first county-level estimates of AI adoption found that 32.9% of working-age people in metropolitan counties use AI, compared with just 16.2% in rural counties. The rural adoption rate is roughly half the metropolitan rate. Microsoft also found that connectivity explains only part of the difference; education, age, occupations and the presence of institutions such as colleges and universities matter too. (Microsoft)

That gap matters, but it is not the whole problem. The deeper divide is about who captures the value: who builds the companies, owns the intellectual property, develops the workforce, governs the data and has a voice in how AI is deployed.

Rural America is already part of the AI economy. The question is what part it gets to play.

The Geography of AI Is Still Highly Concentrated

The AI economy is growing extraordinarily quickly, but geographically it remains concentrated.

U.S. private AI investment reached $285.9 billion in 2025. California alone accounted for $218 billion—more than three-quarters of the national total. More than half of U.S. states received less than $100 million in mapped private AI investment. (Stanford AI Index)

Brookings found a similar pattern in talent and employment. The San Francisco and San Jose regions remain in a class of their own, and the Bay Area alone accounts for roughly 13% of U.S. AI-related job postings. Most regions have much thinner concentrations of AI talent, research and enterprise activity. (Brookings)

There is another divide inside those numbers: organizational capacity. Census Bureau data from 2026 shows that 37% of firms with 250 or more employees were using AI, compared with less than 20% of the smallest businesses. Across all businesses, usage was still only around 17% to 20%. (U.S. Census Bureau)

That matters in rural America, where small businesses, local governments, schools, hospitals and nonprofits carry much of the economic and civic load. Giving those organizations access to an AI tool is not the same thing as giving them the capacity to use it well.

Broadband matters. But broadband alone does not provide training, technical talent, trusted institutions, data governance, procurement expertise or the ability to turn AI into new businesses and jobs.

Rural America Is Already Supplying the AI Economy

The AI economy also depends on enormous amounts of land, energy, water and physical infrastructure, and data-center development is increasingly moving into rural communities where those resources are available. That can bring significant investment, but it also raises questions about water, electricity, land use, public infrastructure and how much lasting local economic value these projects actually create. (Brookings)

The risk is familiar: rural America could supply essential inputs to the AI economy while much of the capital, intellectual property and high-value economic activity accrues elsewhere.

There is a similar issue with data and knowledge. Agriculture, energy, wildfire management, natural resources, logistics and manufacturing all generate valuable information rooted in specific places and industries. Communities should have a meaningful role in deciding how that knowledge is used and how the benefits are shared.

The issue is not whether data centers, AI companies or new technologies are inherently good or bad for rural communities. It is whether those communities have enough knowledge, negotiating power and technical capacity to participate on their own terms.

Why AI Is Different

Rural America has been on the wrong side of technology transitions before. AI is different because of its speed and reach.

It changes the economics of expertise. A small business, school district or county government can suddenly access capabilities that once required teams of specialists. It also raises the value of local data and knowledge because AI systems become more useful when they understand the environments where they are actually being deployed.

At the same time, AI is beginning to influence decisions in healthcare, lending, insurance, education, public benefits, agriculture and government. If rural realities are absent when those systems are designed and tested, urban assumptions can quietly become national defaults.

There is also less time to catch up. Technology adoption once unfolded over years or decades. AI capabilities are advancing in months. Communities that sit out the early stages risk missing the period when use cases are defined, data practices are established, workforce pathways are built and new companies are created.

About 46.2 million Americans—roughly 14% of the population—live in nonmetropolitan America, spread across most of the country's land area. (USDA Economic Research Service)

They cannot be an afterthought in a transformation this large.

Colorado Shows Both the Opportunity and the Gap

Colorado is a useful example. The state has a legitimate AI ecosystem: federal laboratories, leading universities, major technology companies, startups and an increasingly sophisticated state-government AI program.

Colorado has made real progress since this article was first written. By February 2026, state agencies had identified 281 generative-AI use cases, 220 of which had been approved, and by May more than 6,400 state employees were using Gemini after completing required training. The state is now moving toward a broader strategy built around technical infrastructure, employee education, governance, performance measurement and partnerships. (State of Colorado)

That is significant progress, but it solves a different problem. A strategy for using AI inside state government is not the same as an economic strategy for helping rural Colorado build businesses, develop talent, govern data and participate in the AI economy.

Most of Colorado's AI assets are still concentrated along the Front Range. Fort Lewis College's AI Institute in Durango is an important exception: a regional institution serving the Four Corners with direct relevance to rural, Indigenous and place-based AI.

The Eastern Plains, San Luis Valley, Four Corners, Western Slope and mountain communities need more than access to AI. They need the institutions, skills and relationships that let them build with it. The same is true for tribal communities, where connectivity alone does not answer questions about Indigenous data sovereignty, ownership of local knowledge or who benefits when AI systems are trained on or deployed within tribal communities.

Those are no longer abstract questions.

We Have Seen This Movie Before

There are two useful historical lessons.

In 1935, only about one in ten rural homes had electricity. Private utilities had little economic incentive to serve sparsely populated areas. Rural electrification succeeded because the country deliberately created institutions, financing mechanisms and local cooperatives capable of changing the economics.

Broadband offers the cautionary version of the story. Billions of dollars have expanded rural connectivity, and that investment has been essential. But a connection by itself does not create businesses, train workers or build local institutions.

AI requires both lessons at once: the urgency and institution-building of rural electrification, without assuming—as we too often did with broadband—that infrastructure alone equals participation.

The Decision Point

Rural America does not need to be “brought into” the AI economy. It is already in it.

The real question is whether rural communities will have the capacity to shape how AI is used, build companies and careers around it, govern the data and knowledge that come from their communities, and retain a meaningful share of the value created.

That is a different future from one in which AI is largely designed elsewhere and delivered to rural America as a finished product.

And that is the two-state problem.

Unlike some of the geographic divides created by earlier technology waves, this one is not yet settled. We still have time to design against it.

Andrew Aitken is the Founder and Executive Director of the Center for Rural AI (ruralai.org), a fiscally sponsored project of SW Community Foundation based in Durango, Colorado. CRAI is partnered with the AI Institute at Fort Lewis College.