Forget models and GPUs. The future of AI belongs to those who control the megawatts, minerals, and infrastructure behind the scenes.

On Friday, the U.S. government quietly did something historic: it officially tied the future of artificial intelligence to the future of nuclear power.

Buried in the executive order were provisions that designate AI data centers as critical defense facilities, call for the rapid deployment of advanced nuclear reactors to power them, and mandate accelerated approval timelines for new nuclear builds — all under the banner of national security.

This wasn’t just about safety frameworks or model governance. It was about energy sovereignty — and who gets to scale AI in the years ahead.

If you’ve followed my writing, you know I’ve been preparing for this moment for years. From Uranium Energy Corp and Energy Fuels, to ASP Isotopes and the isotope bottleneck no one’s talking about. From Applied Digital and energy-optimized data centers, to the thesis that infrastructure — not algorithms — will define the next tech era.

Friday’s order didn’t change the game. It confirmed it.

The AI Hunger Games are officially underway. And the winners won’t be the ones with the biggest models.

They’ll be the ones who can power them.

The AI Hunger Games Are Real

Everyone’s chasing AI dominance right now — building bigger models, buying more GPUs, acquiring more data. But few are talking about the real bottleneck: electricity.

Training a single large language model can consume more energy than 100 American homes use in a year. Inference at scale? Even worse. And it’s not just about how much power — it’s about where that power comes from, how stable it is, and who controls it.

This is the part of the AI arms race most people miss.

We’re not just dealing with an innovation race. We’re in a logistics war — one that stretches from uranium mines to power substations to server racks.

Data centers are popping up like mushrooms, and the grid can’t keep up. That’s why the executive order didn’t just authorize reactor builds — it told the Department of Energy to treat AI data centers as defense-critical infrastructure, and to accelerate nuclear deployment timelines to keep pace.

It’s a signal that Washington sees what early investors already knew: whoever solves the energy bottleneck wins the AI war.

This is why I called it The AI Hunger Games in my original piece months ago. The tech giants aren’t just racing for language dominance. They’re stockpiling energy like it’s oil in the 1970s. And just like oil, nuclear offers the trifecta: it’s dense, scalable, and politically sovereign.

This isn’t theoretical anymore. It’s playing out in real time — with governments and private companies racing to lock up power before someone else does.

Everyone’s Focused on the Models — But It’s a Real Estate & Energy Play

When people talk about AI, they talk about models — GPT-4, Gemini, Claude, Mistral.

But they should be talking about megawatts, land rights, and cooling systems.

Because here’s the truth: the future of AI won’t be determined in boardrooms or research labs. It’ll be decided by who controls the physical infrastructure — the energy sources, the compute clusters, and the places to put them.

Data centers are the new oil fields. And they’re running out of space.

You can’t drop a hyperscale cluster just anywhere. You need access to water, a stable grid, proximity to fiber, and — most critically — cheap, uninterrupted power. The kind of power that doesn’t fluctuate with weather patterns or geopolitics.

That’s where nuclear — especially small modular reactors (SMRs) — comes in. SMRs can sit closer to where data is needed, offer constant base-load power, and scale alongside demand. They don’t require thousands of acres, and they don’t crash the grid. They’re the energy equivalent of edge compute — dense, portable, and local.

Companies like Applied Digital already get this. Their entire pitch is built around energy-optimized infrastructure. They’re not just selling compute — they’re selling efficiency per kilowatt, designed for the era of AI workloads.

This is why Friday’s executive order matters so much. It’s the first real acknowledgment that we can’t scale AI unless we rethink the ground it runs on — literally.

Forget just chasing better prompts. The real winners are chasing better power purchase agreements.

I Didn’t Invest in AI — I Invested in What AI Needs

AI isn’t just about data and algorithms. It’s about infrastructure — land, energy, and the industrial supply chains no one talks about until they break.

That’s why, for the past six years, I’ve been investing not in the AI boom itself, but in the undercarriage that makes it possible. The companies I’ve backed may not make headlines, but they’re solving the silent bottlenecks AI can’t grow without:

• Uranium Energy Corp (UEC) — Domestic uranium production is a national security issue. UEC is restoring U.S. control over the nuclear fuel cycle — without it, there’s no sovereign energy future.
• Energy Fuels (UUUU) — A dual play in uranium and rare earths. They’re strategically positioned to support both clean energy and tech supply chains — and they’re doing it from American soil.
• ASP Isotopes (ASPI) — Most people overlook isotopes until they become critical. ASPI is pioneering next-gen isotope production for medicine, fuel, and industrial use — right as global demand spikes.
• Applied Digital (APLD) — A data center company with a power-first mindset. They understand that in the AI era, compute is constrained by kilowatts. APLD designs for that reality.

My thesis has always been simple: AI is compute-hungry, compute is energy-hungry, and energy is geopolitically fragile. The companies that can solve for those inputs — quietly, efficiently, and with long-term strategy — will define the next decade.

These aren’t just stock picks. They’re pieces of a larger flywheel. A convergence of compute, energy, and national resilience.

The government’s executive order last Friday only confirmed what many of us already knew: the AI revolution isn’t about who can build the biggest model. It’s about who can keep the lights on while doing it.

The AI–Nuclear Flywheel

This isn’t just correlation. It’s a feedback loop.

AI and nuclear are locked into a flywheel dynamic — each one accelerating the other in a loop that’s becoming harder to escape, and even harder to compete with if you’re on the outside.

Here’s how it works:

1. AI needs energy — and not just any energy.

AI workloads are voracious. We’re not talking about one-time model training anymore — we’re talking about real-time inference at global scale, streaming into every consumer app, enterprise workflow, and industrial control system. That means 24/7 uptime, massive parallel processing, and latency that kills if the grid fluctuates.

The grid isn’t ready. Renewables can’t shoulder base load. Fossil fuel volatility creates risk. Only nuclear provides the scalability, stability, and sovereignty to feed AI without compromise.

2. Nuclear fills the gap — especially SMRs.

Legacy nuclear is too slow and too political. But small modular reactors (SMRs)? They’re fast to deploy, easier to permit, and perfect for co-location with data centers.

They represent a new energy form factor: clean, constant, local power. The kind AI needs. No weather variability. No carbon offsets. Just plug-and-scale.

SMRs aren’t theoretical — they’re already being deployed in Europe, Canada, and the U.S. They’re not just the future of energy — they’re the future of compute infrastructure design.

3. That energy unlocks infrastructure scale.

With sovereign, always-on power, hyperscalers and infrastructure startups can finally break out of the grid chokepoints. We’ll see next-gen campuses designed around energy sources, not cities.

Think vertically integrated compute parks: SMRs + fiber + cooling + GPUs. That’s when the real edge advantage kicks in — when AI infrastructure stops being hosted and starts being engineered from scratch.

Companies like Applied Digital are already doing this — thinking of infrastructure in terms of energy ROI, not just rack space.

4. That infrastructure feeds the next AI boom.

The more energy you have, the more compute you can unleash. And the more compute, the more powerful and pervasive AI becomes.

That includes models that run at the edge, real-time copilots for industry, AI-driven simulations for climate, medicine, defense — you name it. And all of that loops back to… more energy demand.

More demand = more nuclear deployment = more infrastructure = more AI capacity.

That’s the flywheel.

This isn’t a one-time opportunity. It’s a system. A compounding one.

Investors who only look at the front end — chatbots, apps, APIs — are playing the shallow game. The real leverage is behind the curtain. In the energy sources, materials, land, and infrastructure that turn possibility into scale.

And once this flywheel hits full speed, it’ll be nearly impossible to catch up.

Owning the Inputs

This isn’t just about clean energy or tech infrastructure. It’s about sovereignty, scale, and systems thinking — the kind that separates short-term hype from decade-long advantage.

On Friday, the executive order simply put into policy what forward-looking investors already understood: AI can’t scale without energy, and energy security will define who leads in the AI age.

That’s why I’ve spent the past six years investing in companies building this foundation, like:

• Uranium Energy Corp (UEC) — Restoring U.S. control over nuclear fuel supply.
• Energy Fuels (UUUU) — Bridging clean energy and critical minerals from the ground up.
• ASP Isotopes (ASPI) — Solving silent bottlenecks in isotope production for medicine and fuel.
• Applied Digital (APLD) — Rethinking data centers as energy-first infrastructure for AI workloads.

Each of these companies represents a different node in the flywheel — but together, they form a system. A system that feeds itself. A system that scales with demand. A system that others are only now beginning to notice.

While the market obsesses over ChatGPT plug-ins and VC-backed LLMs, the smart money is flowing to the picks, the shovels, and the power.

Because when AI becomes infrastructure, energy becomes everything.

And those who own the energy — and the land, and the rare earths, and the isotopes — don’t just survive the next cycle. They shape it.

This is just the beginning.

Financial Advice Disclaimer: This article is provided for informational purposes only and is not intended as financial advice. The content reflects the personal opinions of the author and should not be construed as specific investment advice. Investors are advised to conduct their own research or consult with a qualified financial advisor before making any investment decisions. The author and publisher are not liable for any financial losses or damages resulting from decisions made based on this article. Investments in the stock market involve risk, including the potential loss of principal.

Disclosure: The author of this article holds shares in the companies mentioned in this article. This disclosure is to inform readers of potential biases in the opinions expressed herein.

Written by James Soldinger