In the 21st century, artificial intelligence can be viewed as a form of infrastructure – a general-purpose capacity that undergirds productivity and innovation across industries1, much like transportation networks or the electric grid. This section explores what it means to treat AI infrastructure as an economic development tool and why doing so is critical for extending the benefits of the AI revolution to America’s smaller communities. It also outlines the core components of AI infrastructure (physical, digital, and human) and how investing in these components in under-resourced areas can create a foundation for economic stability and growth.
AI as the “New Electricity.” AI has been likened to electricity for its economy-wide impact. Just as electrification transformed agriculture, manufacturing, and daily life in the early 20th century, AI stands to transform virtually every sector today – from logistics and health care to retail and farming. But electricity required massive infrastructure (power plants, transmission lines) to deliver its benefits broadly. Similarly, AI’s transformational potential depends on infrastructure: data centers housing the servers that train and run AI models, high-speed internet/fiber connecting users to AI services, and an array of software frameworks and cloud platforms that serve as digital infrastructure for AI development. There is also a less tangible infrastructure of standards, protocols, and pre-trained models that make it easier to deploy AI solutions. For small towns and rural regions, the challenge is to build or attract enough of this infrastructure to be relevant in the AI economy.
Historically, both rural America and inner ring suburbs have lagged in infrastructure deployment – whether it was rail lines in the 1800s or broadband internet in the 2000s. A concerted push was needed in each case (e.g., rural electrification programs in the 1930s, federal broadband grants in recent years) to close the gap. A similar push is needed now for AI infrastructure. This means extending fiber networks to every community (as envisioned in recent infrastructure legislation), supporting the construction of regional cloud computing nodes and edge data centers, and perhaps even treating computing power as a public utility. Some economists argue that widespread access to computing and AI tools will determine the next generation’s economic winners, much as access to reliable electricity did a century ago.
Physical Infrastructure Demands of AI. Unlike purely digital innovations, cutting-edge AI (especially generative AI and large language models) comes with heavy physical infrastructure needs. Training a single state‑of‑the‑art AI model can consume megawatt‑hours of electricity and require specialized infrastructure 2. As a result, there companies are building new data centers at a rapid pace. These facilities often locate where power is cheap and land is available – conditions abundant in many small towns. For example, the agricultural town of Quincy, Washington (population ~7,000) attracted major data centers by offering inexpensive hydropower from the Columbia River dam system. Similarly, rural communities in Oregon’s Columbia Gorge and Northern Virginia’s Loudoun County have become data center hubs, benefiting from tax incentives and existing transmission lines.
The expansion is set to continue. A Department of Energy report in late 2024 found that data centers consumed about 4.4% of all U.S. electricity in 2023 and could draw 6.7% to 12% by 20283 In absolute terms, data center electricity use in the U.S. tripled from 2014 to 2023 (58 TWh to 176 TWh) and may triple again by 2028 . This growth is being driven by AI workloads and the need for ever-larger computing capacity. Such figures underscore that AI infrastructure is not just a tech industry concern but a national infrastructure concern – with parallels to the growth in electricity demand during the industrial expansion of the early 1900s. For small towns, it presents an opening: hosting AI infrastructure can become a new economic base, much as hosting a factory or power plant was in earlier eras. A single large data center can inject tens of millions of dollars in local construction activity and create hundreds of permanent jobs.
However, to truly leverage AI infrastructure for local revitalization, communities must treat it as part of an ecosystem. A data center by itself provides property tax revenue and some jobs, but its broader impact is magnified when paired with other elements: workforce training centers, tech incubators, broadband for local businesses, and education pipelines. For example, if a community college creates a program to train data center technicians and AI model operators, local residents can fill those jobs rather than outside hires. If a region has robust broadband, secondary businesses (like software startups or AI-enabled farming cooperatives) can spring up to utilize the computing power. In this way, a physical investment (like a server farm) can anchor a wider innovation ecosystem.
General-Purpose Technology, Local Impact. Economists classify AI as a general-purpose technology (GPT), meaning it has broad applicability and can spur complementary innovations. Historically, GPTs like the steam engine or the semiconductor have also created geographic concentrations of wealth – think of Manchester in the industrial revolution or Silicon Valley in the digital age. But GPTs do not have to be geographically concentrated; policy and timing matter. The Interstate Highway System, for instance, spread the gains of automobile technology across the country by connecting remote areas to markets. With AI, we are at an early stage where interventions can influence its geography. If left purely to market forces, AI activity might remain clustered in a few knowledge hubs (due to network effects and talent pools). But deliberate investments in AI infrastructure as a public good can counteract that. Initiatives like the National Science Foundation’s funding of AI research institutes in various states, or the Department of Defense’s creation of AI testbed sites around the country, are examples of seeding capacity …beyond Silicon Valley4.
Already, there are signs of shifting geography. Major cloud providers (Amazon, Microsoft, Google) have announced new cloud regions and data centers in states like Alabama, Ohio, Iowa, and Oregon5 (Figure 1), not just in California or New York. As noted, states like Texas and Pennsylvania have gigawatts of data center capacity in the pipeline. This diffusion is partly driven by pragmatic needs (power availability, lower cost) and partly by incentives (many states now offer tax exemptions for data center investment). It presents an opportunity for distressed areas high tax burdens to claim a share of the digital economy’s infrastructure.
Importantly, AI infrastructure is more than hardware. It also includes soft infrastructure: open-source AI frameworks, large datasets, and cloud platforms that communities can leverage. For instance, a small manufacturing town could tap into an AI cloud service to implement predictive maintenance in its factories without developing AI from scratch. The availability of such cloud-based AI (if accessible via good internet) essentially brings sophisticated capabilities to any corner of the country. The key is ensuring communities have the connectivity and know-how to utilize it.
In summary, treating AI infrastructure as an economic development tool means recognizing that government and industry must collaborate to build out the “AI backbone” of computing power and connectivity, much as we built highways and power grids. For small towns, this approach shifts the narrative from being victims of automation to being hosts of automation – i.e., owning the infrastructure that powers AI and thereby sharing in its economic returns. In the next sections, we explore how small towns can capitalize on the AI moment—not just through infrastructure and training, but by unlocking the often-overlooked backbone of revival: civic capacity, local leadership, and aligned incentives. When policy is designed to share both risk and reward, it can unlock trillions in idle private capital—from pension funds to corporate reserves—and redirect it into building a distributed AI economy. That alignment is especially critical in places like Dolton, a historically overlooked inner-ring suburb with the assets, urgency, and potential to lead.
-
McKinsey & Company. What’s next for AI in infrastructure? August 2023.
https://www.mckinsey.com/industries/public-and-social-sector/our-insights/whats-next-for-ai-in-infrastructure -
NVIDIA. The Energy Cost of AI: Scaling Deep Learning in the Data Center, 2023.
https://blogs.nvidia.com/blog/ai-energy-consumption/ -
U.S. Department of Energy. Data Center Energy Use Report, December 2024.
https://www.energy.gov/articles/ai-and-data-centers-energy-demand-outlook -
National Science Foundation. NSF Expands AI Research Institutes Across U.S. 2023.
https://beta.nsf.gov/news/nsf-announces-expansion-national-artificial-intelligence-research-institutes -
Microsoft. Building the Infrastructure for AI: New Data Centers in Underserved Regions, 2024.
https://news.microsoft.com/datacenters-and-ai-investment/