Within a span of just a few days, two governments on opposite sides of the world announced major data center policies. One bold and forward. The other, measured and safe.
Australia chose not to slow development of hyperscalers that will inevitably power our AI future. But they emphatically demand that the private industry behind these data centers must underwrite the energy this infrastructure requires. Those costs cannot be passed on to local residents. In New York, state leaders were more measured. They moved to pause the development of new hyperscale data centers while the state evaluates the long-term implications for electricity, water resources, land use, and local communities.
Other WRAL Top Stories
Through one lens, these appear to be stories about energy policy. Or environmental policy. Or economic development. But in reality, they are stories about artificial intelligence.
For the last three or four years, AI discussions have been primarily centered on software. We debated algorithms, datasets, model architectures, safety, bias, copyright, and regulation. In the last year, the conversation has shifted toward substations, transmission lines, cooling systems, water rights, tax districts, and zoning hearings.
Artificial intelligence has become physical. More importantly, it now has an address.
Although New York and Australia have captured the headlines, the future of AI infrastructure is unlikely to be decided primarily by governors, presidents, or members of Congress. More often than not, it will be determined in county commission meetings, planning boards, utility authorities, and zoning hearings, many of them taking place in rural communities that never expected to find themselves debating the future of the global AI economy.
I was reminded of that reality yesterday while speaking at the Newsmakers event at Fayetteville State University. Also on the program was North Carolina's Chief Deputy Secretary of Commerce, Kenny Flowers, who was asked about the state's position on data centers. His response struck me as both measured and insightful. Rather than advocating for a blanket statewide policy, he acknowledged the complexity of the issue and suggested that these decisions are often best informed by local priorities and local voices.
I believe that is almost certainly the right answer. And for Secretary Flowers, it happens to also be the politically convenient answer.
Local governments, local infrastructure, local AI policies
Local governments have responsibilities that technology companies do not. Their first obligation is to the people who already live there. They must provide reliable electricity, protect water supplies, preserve public safety, steward the environment, and ensure that growth benefits the community rather than simply passing through it. In summary, your county commissioners have a primary responsibility to provide the infrastructure for a thriving economy and high quality place to live.
Those responsibilities should not become secondary simply because the latest technological revolution has arrived. Yet there is another reality quietly emerging.
Local planning commissions are becoming AI policymakers.
Few county commissioners would describe themselves as experts in large language models or inference optimization. They shouldn't have to. But by deciding where data centers may be built, how much electrical capacity should be expanded, whether additional water infrastructure is
justified, or how quickly projects move through permitting, they are making decisions that will influence where intelligence can exist for decades to come.
There is an irony in that observation. They are not debating transformers in the machine learning sense. They are debating transformers in the electrical sense. Yet both may ultimately determine the future availability of intelligence. And that is core to the Data Economy.
This raises a larger question. If AI increasingly becomes part of the infrastructure that powers modern economies, are we beginning to regulate technology differently than we have for the past half century?
Historically (and still today), governments regulated the applications of technology. They debated privacy laws, telecommunications policy, cybersecurity, antitrust, and online content. Today, local governments are increasingly regulating the infrastructure that makes those applications possible.
That represents an entirely new layer of AI policy, which I believe is one that few people anticipated. Significantly, it forces us to think differently about geography. For much of the Internet era, distance steadily became less important. Cloud computing concentrated enormous amounts of computing power into hyperscale facilities, allowing organizations almost anywhere to access nearly limitless resources. Data moved to wherever computing happened to be the cheapest.
Artificial intelligence may reverse part of that equation. Today's AI assistants can tolerate a second or two of delay before responding. Tomorrow's AI systems increasingly will not.
The rise of physical AI
There is a new wave of innovation rising in what I’ll call “physical AI”. I’ll write much more about this in the weeks to come, but in short, you can think of these cyber-AI-physical systems as collaborative robots operating on factory floors. Autonomous agricultural equipment. Military systems. Smart electrical grids. Intelligent logistics networks. Autonomous cars, drones and delivery robots. Medical devices. Eventually, perhaps even robotic surgery.
These systems are not simply generating information. They are making decisions that interact with the physical world. In those environments, latency is no longer a technical specification buried in an engineering manual. It becomes an economic constraint, and sometimes a safety constraint.
For decades, computing optimized for centralization. The AI economy may increasingly optimize for proximity.
As intelligence moves closer to where decisions are made, geography once again becomes a competitive advantage. Suddenly, the location of data centers, edge computing facilities, substations, and electrical capacity matters not only to technology companies, but to manufacturers, hospitals, research institutions, defense installations, and entire regional economies. It should also matter to county commissioners and the general public.
Why? Because we've seen this unfold before. When broadband infrastructure was first deployed, some communities invested aggressively while others waited. Then the dot-com bubble burst. Investment slowed dramatically. The places that already possessed robust digital infrastructure continued attracting employers, entrepreneurs, startups, remote workers, and investment.
Many communities that missed that first wave spent decades trying to catch up. Today, virtually no one argues that broadband was a mistake. Instead, we ask why everyone didn't receive it sooner.
I wonder if future generations will ask similar questions about AI infrastructure. There have been well-documented debates about whether we are in an AI infrastructure bubble as the Magnificent 7 invest billions of dollars into hyperscale data centers. There is a pretty significant chance that we hit a speed bump in deployment at some point in the near to mid-term. Will places that didn’t invest now, while the investment is hot, regret that they must rely on the infrastructure of another region (or simply miss out on those low latency use cases)?
Are data center moratoriums a mistake?
To be clear, this is not an argument that every proposed data center deserves automatic approval. Communities have every right and every responsibility to ask difficult questions about water consumption, electrical reliability, environmental impacts, and quality of life. Those concerns are legitimate and should remain central to the conversation.
Personally, I find myself somewhere between the growing calls for moratoriums and the assumption that every project should receive a green light. If you read my article last week, I advocated that I believe that big tech should shoulder much more responsibility for educating and preparing our communities to take advantage of AI. My thinking follows the same path here.
In broad strokes, history has favored places that invested in infrastructure abundance over places defined by infrastructure scarcity. Railroads, ports, interstate highways, airports, industrial parks, and fiber networks all required communities to think beyond immediate costs toward long-term opportunity. Prosperity has rarely belonged to the regions that rationed infrastructure. More often, it has belonged to those that figured out how to build it responsibly. That is why I find Australia's emerging approach particularly interesting.
Rather than rejecting AI infrastructure, it asks a different question: who should pay for the abundance required to support it?
If hyperscale AI companies require enormous new investments in electrical generation, transmission, substations, cooling systems, or water infrastructure, those costs should not simply be shifted onto rural taxpayers. Communities should negotiate from a position of strength, ensuring that private investment expands public infrastructure rather than consuming scarce public resources. The objective should not be scarcity. It should be abundance financed by those who stand to benefit most from it.
The stakes are higher than they may first appear.
Infrastructure is rarely built in a smooth, continuous progression. Railroads, interstate highways, cellular networks, and broadband all experienced bursts of extraordinary investment followed by years of consolidation. AI infrastructure is likely to follow a similar pattern. If that happens, the communities that establish themselves during this first wave may enjoy advantages long after the cranes disappear and the headlines move on.
Ultimately, this debate is about far more than where to place a data center. It is about where intelligence will live. The decisions made today in county commission meetings and planning departments may determine which regions become producers of intelligence, which become consumers of it, and which find themselves waiting for the next investment cycle.
Of course, infrastructure has never been enough by itself. Railroads did not automatically create thriving businesses. Fiber optic cable did not automatically produce successful software companies. Infrastructure creates opportunity; people create value.
If this first chapter of the AI economy is about deciding the street address where intelligence will live, the next chapter will be about an even more important question. Who will know how to use it and create the jobs of the future from it? Stay tuned for my thoughts on this in next week’s article. Thanks for reading.
