The numbers are staggering. Microsoft, Google, Amazon, and Meta are set to deploy over $200 billion in combined capital expenditure in 2024 alone. The bulk of that cash is flowing into AI data centers—massive, power-hungry facilities designed to train and serve the next generation of large language models. But a new variable is entering the equation: the US midterm election cycle. And most traders are ignoring it.
I've been tracking this intersection for months. My background in cryptography taught me to look for the hidden assumptions in any system. The AI infrastructure trade assumes a stable, permissive regulatory environment. That assumption is about to be tested. The midterms are not just a political event—they are a gamma event for every portfolio long on compute.
Context: Why Now
The AI infrastructure boom is real. Training a single GPT-4 class model requires roughly 25,000 A100 GPUs running for weeks. The electricity consumption of a single hyperscale data center can rival a small city. In Ireland, data centers already consume 18% of the national grid. In the US, the demand is concentrated in states like Virginia, Ohio, and Texas—states that are also political battlegrounds.
Local opposition to data centers is growing. Residents complain about noise, water usage, and the strain on local power grids. Environmental groups are framing AI infrastructure as a climate villain. And now, with midterm elections approaching, politicians are seizing on this issue to rally voters. The question is: how far will they go?
Core: The Political Risk in the Numbers
Let's look at the data. According to the US Energy Information Administration, data center electricity consumption is projected to grow from 2% of the national total in 2022 to over 6% by 2028. That's an enormous increase in a short time. In Virginia’s Loudoun County—the data center capital of the world—over 300 data centers are already operating, and more are planned. But local opposition has stiffened. New zoning restrictions are being debated. Some supervisors are calling for a moratorium on new permits.
This is not a fringe issue. In 2023, a proposed data center in Prince William County, Virginia, was blocked after community protests. Similar battles are playing out in California, New York, and even in traditionally pro-business Texas. The midterm elections will amplify these local fights. Candidates who pledge to "protect communities from Big Tech" will win votes. The result: delays, cost overruns, and project cancellations.
For a portfolio manager long on AI infrastructure, this is a direct hit to expected returns. A 12-month delay on a $1 billion project reduces the IRR by roughly 2-3 percentage points. Multiply that across the $200 billion pipeline, and the potential value destruction is in the tens of billions.
We don't trade based on hope – we trade on data. And the data says the political risk is real, measurable, and currently underpriced. The VIX for AI infrastructure is zero. That's the opportunity.
Contrarian: The Real Opportunity Is in Decentralized Compute
Here's the angle the mainstream is missing. The political risk to centralized AI infrastructure is a massive tailwind for decentralized compute networks—projects like Filecoin, Akash, or even Bitcoin mining operations that can pivot to AI workloads. Why? Because decentralized infrastructure is inherently more resilient to local political shocks. A network of thousands of small, distributed nodes doesn't face the same permitting hurdles as a single hyperscale facility. The regulatory surface area is smaller.
Moreover, the same regulatory pushback that hurts hyperscalers can actually help crypto miners. If the US government imposes stricter energy efficiency standards on data centers, miners who use stranded energy or renewable sources gain a competitive advantage. The math is simple: the marginal cost of compute will rise for centralized players, making decentralized alternatives more attractive.
But there's a catch. The Tornado Cash sanctions set a dangerous precedent: writing code equals crime. If regulators turn their attention to decentralized compute networks, they could impose similar restrictions. The risk is not just political—it's existential. As I've argued before, the only risk is the one you didn't model. The regulatory overhang on AI infrastructure is predictable. The regulatory overhang on decentralized compute is not.
Takeaway: What to Watch Next
The midterms will be a referendum on many things, but for anyone in the crypto or AI infrastructure trade, the key signal is local. Track the zoning board meetings in Loudoun County. Watch the energy commission votes in Ohio. Monitor the statements of candidates in swing districts. The next 12 months will determine whether the US remains the world's AI infrastructure hub or whether capital flows to friendlier shores—the Middle East, Southeast Asia, or even into decentralized networks.
Speed eats strategy for breakfast. If you aren't first to the data, you're last to the exit. The political gamma in AI infrastructure is building. The question is: are you positioned for the explosion, or the collapse?
