I spent the summer of 2017 auditing ICO smart contracts in a Seattle basement, and I learned something that has stayed with me through every cycle since: the infrastructure nobody sees is always the first thing that breaks. Back then, it was reentrancy bugs in token contracts. Today, it is the physical infrastructure of artificial intelligence โ the data centers, the power lines, the water-cooling systems โ and the political environment that surrounds them. The US midterm elections have arrived, and the market is focused on polling numbers, but I have been listening to the silence between market cycles. And the silence tells me the real risk to AI infrastructure trade is not technological. It is the political temperature of a local zoning board meeting.
The macro backdrop is staggering. Microsoft, Google, Amazon, and Meta are projected to spend a combined $200 billion on capital expenditures in 2024, with most of that pouring into AI data centers. This is not a discretionary bet. This is the physical foundation of the entire AI commercialization chain, from model training to inference services. But here is the tension I keep coming back to: a data center is not a software deployment. It is a land acquisition. It is a water rights negotiation. It is a grid connection contract. It is a permitting process that involves local politicians, environmental groups, and community members who may not care about the difference between a large language model and a large language problem. The AI industry has built its future on a physical substrate that is highly vulnerable to local political dynamics.
Let me walk through the mechanics. A single hyperscale data center can consume hundreds of megawatts annually. In Ireland, data centers already account for over 18% of national electricity consumption. In Northern Virginia, the largest data center market in the world, the power demand is straining the grid. And when you build these facilities, you are not just building a building; you are making a claim on resources that communities also need. The fight is not about AI, at least not on the surface. It is about water rights, noise pollution, visual blight, and whether the local school district will see any tax benefit. During my 2024 ETF regulatory impact study, I noticed that the same pattern appeared in crypto mining. The political opposition to Bitcoin mining was rarely about cryptography. It was about electricity prices and noise. The same dynamic is now being replicated at a larger scale for AI.
Here is the core insight I want to stress: AI infrastructure investment has shifted from a pure technology bet to a policy-driven trade, and the market is only beginning to price this in. The valuation logic for data center projects now has to include the probability of regulatory friction, the speed of local approval, and the stability of the political environment. You cannot model a delay from community opposition the same way you would model a GPU failure. But investors are trying to, and they are discovering that the political risk premium is something new. I remember co-authoring a "DeFi for Beginners" guide in 2020, and the core lesson was that liquidity is not just about token flows โ it's about confidence. The same logic applies here. If the political environment in a key state turns hostile, the confidence in the entire AI infrastructure trade will erode, and that will ripple through to the cost of capital for every project.
The market is currently in a bull phase, and that euphoria can mask these technical flaws. Just as I used to audit ICO contracts for reentrancy, I am now looking at the structure of these AI infrastructure deals, and the reentrancy risk is political. The capital expenditure cycle is in a prisoner's dilemma: the tech giants know that cutting spending would sacrifice their competitive position, so they continue to invest even when facing political headwinds. This investment rigidity increases their exposure to political risk, because they cannot easily pause. If a local community successfully blocks a data center in one state, the capital does not simply vanish; it moves to another state, or another country. This is where the global competitive landscape becomes interesting.
Let me add a contrarian angle: The political risk to AI infrastructure is not a drag on the entire trade โ it is a catalyst for its own version of creative destruction, and the winners will be those who can turn community resistance into a moat. I remember the 2022 bear market, where the panic led to community-led educational initiatives, and the projects that survived were the ones that had built real community alignment. The same will happen with AI infrastructure. The data center operators that invest in community relations, that offer local jobs, that partner with municipal utilities, and that commit to carbon neutrality, will become the safe havens for institutional capital. This is not just a risk to manage; it is an opportunity to create differentiation. The companies that treat the local community as a stakeholder, rather than an externality, will be the ones that secure the prime locations and the long-term stability. The market is already starting to see this in the valuation gap between data center projects in different regions.
But I worry about the blind spots. The opposition to AI data centers is not a single movement; it is a coalition of environmentalists, local residents, and people who fear job displacement. The narrative is shifting from "not in my backyard" to "why should your compute take our power?" This is a resource allocation argument, and it is a harder one to counter. The industry has relied on the old playbook of promising tax revenue and short-term construction jobs, but that is not enough to counter the long-term anxiety about the environmental cost and the cultural change. In my 2026 AI-crypto symbiosis study, I wrote about the need for a "human-in-the-loop" consensus model for AI-driven economic activities. I think the same principle applies to data center siting: we need to design a process that brings the community into the loop, not just the engineering team.
Looking at the data from my 2024 ETF regulatory study, I can see a pattern: the flow of institutional capital into crypto assets accelerated after regulatory clarity, but the flow was also concentrated in regions with predictable regulatory environments. The same will happen with AI infrastructure. The capital will flow to places with predictable policy, not necessarily the lowest energy cost. Saudi Arabia, the UAE, and Southeast Asia are becoming attractive because they offer a more stable political environment for infrastructure. This is not about escaping regulation; it is about finding the regulatory certainty that is necessary for 20-year capital commitments. The US is still the leader, but the political uncertainty is a discount factor on its future infrastructure growth.
The midterms will be a stress test. If the election results in a divided government, the level of political noise will increase, and infrastructure projects may face more scrutiny. But even if the results are clear, the underlying local resistance will remain. The AI infrastructure trade is not a pure software trade; it is a physical asset trade that must be priced in the local political climate. I see this as the new frontier of the market. The people who understand the politics of the physical infrastructure will have an edge. The people who just look at the compute capacity will miss the whole picture.
So, as the dust settles on the midterms, my takeaway is not about which party won. It is about the fact that AI infrastructure has become a political asset. The next bull run in AI will not be determined by the size of the GPU cluster, but by the depth of the political goodwill. The companies that treat the community as a balance sheet item, rather than a line item, will be the ones that secure the long-term licenses to operate. As we watch the election returns, I am not looking at the balance of power in Washington. I am looking at the local elections in the data center corridors โ and the signals from the zoning boards. The silence between market cycles is where the real political risk is building.
The question is not whether AI will be a dominant technology; that is already clear. The question is whether we can build the infrastructure to support it without a social cost that becomes a political backlash. It is a question of design, and it is not just a technical design. It is a question of community design. And I have not seen a clean answer yet, but I am confident that the companies that ask this question honestly will be the ones that lead the next era. The ones that ignore it will be the cautionary tales. I have been in this industry long enough to see the pattern. The technical road is never the hard part; the human road is. And we are now entering a phase of the human road that is political. Let's be prepared for it.