"Plug, Baby, Plug" - Energy is AI's Wild Card

Macron gets that it's all about the energy, baby

President Macron is on a one-man mission to make France great again by positioning it as Europe's top AI contender.

His pièce de résistance? A €109 billion AI infrastructure investment—France's answer to Stargate—most of it earmarked for AI-optimized data centers.

Naturally, Macron couldn't resist taking a swipe at Trump. Instead of "Drill, baby, drill," like our "good friend," France is ready to "Plug, baby, plug!"

France runs a 75% nuclear-powered grid, unlike its European counterparts still hooked on fossil fuels and energy imports. France even exported 90 terawatt-hours of electricity last year.

This gives France a unique advantage in scaling AI infrastructure— speed.

The bulk of the finance comes from the UAE, which—like Saudi Arabia—is deploying its oil wealth to secure strategic AI infrastructure positions.

Thirty-five AI data centre sites have already been vetted.

Macron understands a fundamental truth: Energy is the wildcard in the AI story.

---

The AI Infrastructure Build-Out

We are witnessing the birth of Industrial Intelligence—a shift that requires an entirely new type of infrastructure.

We're currently in the build-out phase—constructing the industrial-scale backbone of the AI economy.

This is a period of intense competition between companies and between nations. Those who control compute and energy will determine who wins and who falls behind.

(‘Full-stack’ players who own the entire AI infrastructure pipeline will have the greatest advantage—but more on that soon.)

The Raw Truth: AI Needs Massive Amounts of Energy.

AI is not just another software innovation—it's an industrial-scale energy consumer.

That energy must be converted into computational power aka ‘compute.’

Compute is critical for two things:

• Training AI systems – No, DeepSeek R1 does not signal an upper ceiling on deep learning's appetite for compute. (More on that soon.)

• Running AI – AI is no longer just software; it' will increasingly be embodied—in robots, autonomous vehicles, and industrial automation.

As AI expands into every aspect of the physical world, compute demand will continue to surge.

If intelligence is becoming a utility, then compute is what keeps it running 24/7.

Always improving. Always on.

And if AI has no ceiling on learning and deployment, then we're looking at a future where we need near-infinite compute.

Infinite compute requires infinite energy.

AI-Energy Crunch

AI's energy consumption is scaling at a rate few policymakers fully grasp.

In 2022, data centers consumed 460 terawatt-hours (TWh) of electricity, about 1.8% of global consumption according to the International Energy Agency.

Since 2022, the generative AI explosion has sent energy demand skyrocketing. By last year, AI's electricity consumption had already surpassed entire nations like Norway and is now on par with the aviation industry.

This is just the beginning.

AI's energy consumption isn’t linear—it’s exponential.

RAND Corporation projects that by 2030:

• Global AI data centers could require 327 gigawatts (GW)—a 460% leap from 88 GW in 2022.

• That translates to 2,865 TWh per year—roughly 10% of global electricity production.

• A staggering figure, assuming high AI adoption and no major efficiency breakthroughs.

But if these projections hold even remotely true, AI’s power demand could soon rival the entire electricity consumption of the U.S. or the EU.

Some estimates go even further—suggesting AI could consume 21% of global electricity by 2030.

Sounds extreme?

Even in conservative scenarios, one thing is clear: We’re going to need a lot more energy.

---

Energy Orthodoxies Overturned

This is where it gets political.

For decades, policymakers built energy strategies on the belief that efficiency gains and renewables would gradually phase out fossil fuels.

They didn’t anticipate AI.

AI isn’t just another software upgrade—it’s a step change in energy demand, unlike anything seen before.

Unlike past digital revolutions, which optimized existing infrastructure, AI is creating an entirely new class of industrial-scale energy consumption—stacking unprecedented new layers of demand on top of the grid.

And as AI scales, we don’t just need more renewables—we need more of everything:

• More fossil fuels

• More renewables

• More nuclear

Solar and wind have their place, but they are intermittent.

AI infrastructure requires constant, uninterrupted power—which is why Google and Microsoft are securing long-term renewable energy contracts paired with battery storage.

Microsoft’s 10.5 GW renewable purchase by 2030 is one of the largest in history.

Yet, nuclear remains the only proven, zero-carbon baseload energy source that can sustain AI’s exponential growth.

No surprise that AWS, Google, Microsoft, and Oracle are already locking in nuclear power deals.

---

Europe vs. The World

While China and the U.S. are securing energy sovereignty, Europe remains a net importer of 60% of its energy—its highest level since the early 1990s.

This structural weakness leaves it vulnerable to price shocks, supply instability, and geopolitical leverage.

The contrast is stark:

• The U.S. can drill for oil and gas, thanks to its shale revolution. It also has the world's largest operational nuclear fleet, though reinvestment is needed to maintain its long-term advantage.

• China remains the world's largest energy importer but has aggressively expanded renewables, battery storage, and nuclear. It is currently building half of the world's new reactors.

• The Gulf States recognize that fossil fuels remain critical, but they are also diversifying. They are investing heavily in AI infrastructure, data centers, chip production, and compute capacity.

And Europe?

France made its move because it could.

But Germany and the UK lack the same strategic advantage, a weakness that will cost them economically and geopolitically.

As I’ve long argued, AI is not just about technology—it’s about power.

Literally. As well as figuratively.

Germany: A Self-Inflicted Energy Crisis

Germany’s Energiewende, once hailed as a bold vision for a green future, now looks like a severe miscalculation.

A country that once led the world in nuclear energy innovation shut down its last reactors in 2023, only to turn back to coal when Russian gas disappeared.

This wasn’t just a political gamble. It was a structural failure.

Renewables are intermittent, and when the wind didn’t blow and the sun didn’t shine, Germany’s grid was left exposed.

Markus Krebber, CEO of energy giant RWE, put it bluntly last year:

“During a dark doldrum—when renewables failed—Germany’s electricity market nearly collapsed. Prices skyrocketed. The entire system was pushed to its limits. The energy system should not be on a knife edge because security of supply is a precious commodity.”

That is exactly where Germany finds itself today.

It now has some of the world’s highest industrial electricity prices, putting its once-dominant manufacturing sector under immense pressure—just as China escalates its competitive advantage in advanced industries.

Berlin faces an AI-driven energy crisis—and few good options to fix it.

The UK: Complacency at Its Finest

If Germany’s energy crisis was a miscalculation, the UK’s was complacency at its worst. Britain once had two major energy advantages:

1. North Sea oil

2. Nuclear leadership

It failed to capitalize on either.

Instead of establishing a Norway-style sovereign wealth fund (which now holds over $1.5 trillion), the UK spent its oil revenues on day-to-day government budgets.

When North Sea reserves declined, there was no transition plan. Nuclear investment stalled.

By the 2010s, Britain—once an energy powerhouse—had become a net energy importer.

The consequences?

• The highest industrial electricity prices in the world

• An unstable grid

• Heavy dependence on energy imports

Even the lacklustre Keir Starmer recognizes the urgency. In a desperate bid to revive economic growth, he is pushing a "Build, baby, build" strategy to fast-track nuclear investment.

But—because this is the UK—the plan is already running into resistance.

The latest obstacle? Concerns that a new reactor might disrupt a remote Welsh-speaking community.

You. Can. Not. Make. This. Stuff. Up.

Europe can talk all it wants about AI sovereignty. But without energy sovereignty, it’s just rhetoric.

When Starmer and Germany’s next chancellor approach the UAE for AI data center investment, they will be met with a simple but critical question:

Where is your power coming from?

Macron already has his answer.

---

PS — More on the latest developments from Paris, including the EU’s newly announced €200 billion fund for AI infrastructure, JD Vance’s fighting talk on U.S. tech giants and AI opportunities, as well as fallout from the AI Summit Declaration—coming soon.

For now, Namaste.

**
PPS Live Update From Paris: (H/T Mark Nelson). The man who wants nuclear energy meeting the one who can sell it to him.