From Petrostates to Electrostates
The 1970s redux: AI drives the next great energy dislocation
🇺🇸 I've just returned from three weeks in California, where conversations at the crossroads of advanced manufacturing and energy left me with more questions than answers. From Silicon Valley's obsession with artificial intelligence and how to power it, to discussions about reshoring critical supply chains, a pattern emerged: we're witnessing an energy transition that will reshape global power as fundamentally as oil did in the 20th century.
This essay launches a series exploring how electricity is becoming the new foundation of geopolitical strength. Future pieces will examine the State Grid Corporation of China's global infrastructure strategy and trace the remarkable history of France's civil nuclear programme, among other deep dives into the institutions and technologies driving this transformation.
The stakes could hardly be higher. Just as Britain's coal advantage and America's oil abundance defined previous eras, the countries that master electricity generation, storage, and application will shape the next century. China grasps this reality and builds accordingly. The West remains caught between old assumptions and new possibilities.
What follows traces energy transitions through history to illuminate today's moment—why artificial intelligence is really an electricity story, how synthetic fuels might let some nations leapfrog traditional energy powers, and why the age of digging for power is giving way to manufacturing it. The transition from petrostates to electrostates has begun, and the early movers will inherit outsized influence in a world powered by electrons rather than molecules. Read along 👇
1/ Power used to come from the ground
Most of my recent dives into geopolitics have come through the work of Peter Zeihan, whose analysis I’ve followed closely since 2020.
His book The End of the World Is Just the Beginning, published in 2022, opens with a sweeping introduction that has stayed with me. In it, Peter traces the arc of human history through the lens of energy—how each major shift in energy use has shaped the rise and fall of power across regions and empires. This idea—that access to and control of energy underpins geopolitical strength—feels especially relevant now.
As told in this landmark introduction, each major energy shift reshaped how geography influenced power and development:
Early civilisations developed around rivers that provided year-round farming, transport, and water-powered mills. The Nile, Tigris-Euphrates, and Yellow River valleys supported complex societies because water infrastructure enabled food production, goods movement, and mechanical energy. Mountains offered defence; plains allowed expansion. Geography determined which societies could harness energy for growth.
Wind power in medieval Europe marked the first break from strict geographic determinism. After the seventh century, settlements could thrive away from rivers as windmills enabled grain processing and water pumping across open plains. This broader access to energy fragmented political power. Multiple centres could now develop independently, rather than clustering around scarce water resources. Wind power, the technological revolution of the day, helped shape the Europe we recognise today: politically divided, often at war, with fiercely competing power centres scattered across the landscape.
Then coal and oil restored geographic advantage with force. British dominance from 1750 to 1900 came directly from Welsh and Northern English coal seams that powered the first factories. In turn, America built its 20th-century strength on oil—first from Pennsylvania and Texas wells, then through military control of Middle Eastern reserves, notably via Saudi Arabia. After 1945, global order rested on three energy-related pillars: the dollar as the reserve currency for oil, US Navy control of tanker routes, and American influence over petroleum distribution networks—still traded in dollars.
Oil, like coal before it, remained tied to location. The coal mines of Wales, the oil fields of Texas, the gas deposits of Siberia—accessing energy meant controlling territory, or in the US case, wielding enough military power to compel producers to fall in line. The fact that the US became the leading nation in the age of oil, automobiles, and mass production had everything to do with its privileged and sustained access to oil reserves, both domestic and foreign.
Today, that geographic link between energy and power—central since civilisation began—faces its deepest disruption. Solar panels, wind turbines, and batteries are maturing at the same time, allowing energy generation anywhere with basic industrial capacity. Small modular nuclear reactors, an almost mature technology, promise factory-built units that ship like heavy machinery. The constraint now shifts from controlling land to building and deploying technology.
As I’ve argued in my late-cycle investment theory, today we're generally seeing dynamics much like the 1970s. That decade saw oil markets fragment when producer nations seized control of reserves from Western oil majors. Libya nationalised in 1970, followed by Iraq, Saudi Arabia, and others. By 1973, these newly empowered producers could restrict supply through OPEC, quadrupling prices. The smooth, integrated system that had delivered cheap energy for decades shattered overnight into volatile spot markets needing new intermediaries.
Today's fragmentation follows a different path but causes similar market disruption. Instead of producers limiting supply, consuming nations are building domestic generation to escape import dependence—all while energy demand rises fast, notably due to AI. China's fossil fuel imports peaked in 2019 as renewables scaled. Europe aims for energy independence through vast wind and solar projects. Like in the 1970s, today’s transition creates profitable complexity for those able to navigate fragmented markets, volatile pricing, and shifting trade flows.
2/ AI is redrawing the industrial map through soaring, concentrated energy demand
Training GPT-4 used around 50 gigawatt-hours of continuous electricity (twice the annual electricity use of a small city with 20,000 people). The next generation of models will need 10 to 100 times more computing power—and thus more electricity. Microsoft’s acquisition of the 837-megawatt Three Mile Island nuclear plant for its data centres shows how access to energy now shapes competitive advantage in AI. This marks a major shift in power requirements.
Unlike earlier industries that clustered near ports or large populations, AI infrastructure goes where electricity is abundant. Iceland’s aluminium smelters offered a precedent, existing only because geothermal power offset the disadvantages of a remote location. AI takes this further: data centres need near-perfect uptime and power quality, so reliability now matters as much as price. Quebec’s hydro-nuclear grid, Nevada’s mix of solar and geothermal, and other regions with similar conditions are gaining strategic value beyond their traditional economic weight.
This shift has compounding effects. Data centres will go where power is both cheap and reliable. That in turn draws new investment into local generation and grid infrastructure, helping those areas keep up with rising demand. Once power is both abundant and scalable, other electricity-intensive industries follow. The result is a feedback loop: energy abundance attracts energy users, which in turn justify and fund even more energy generation. Over time, this clustering effect can transform regional economies.
The impact is comparable to the oil shocks of the 1970s, though the mechanism is reversed. Then, a supply shock in oil triggered inflation, stalled growth, and exposed energy dependence as a systemic risk. Today, AI drives a demand shock in electricity, pushing grids to their limits and repricing energy from the demand side. In both cases, energy costs rise without a corresponding rise in employment. Fuel and electricity underpin all production, but neither shock is labour-intensive—resulting in inflation without job growth, the hallmark of 1970s-style stagflation.
Both episodes force long-term shifts in energy strategy. The oil shocks led to conservation policies, advances in energy efficiency (with the Japanese car industry leading the way), alternative fuels (notably nuclear), and new geopolitical alignments. Today’s electricity crunch is accelerating grid investment, electrification, and the search for stable, scalable generation. Countries with reliable electricity systems will gain: data centres will follow cheap power, drawing in further industrial activity and reinforcing the cycle. Others will face rising costs, constrained growth, and declining competitiveness in both energy and manufacturing.
The clustering effects of electricity abundance go well beyond AI. Carbon removal technologies only become viable near ultra-cheap electricity. Green hydrogen production for ammonia and steel demands huge, steady energy inputs. High-temperature industrial processes may depend on nuclear heat. These industries will also concentrate in regions with the right mix of generation, geography, and reliability. Traditional advantages—skilled labour, transport links, regulatory stability—now matter less than the ability to deliver large volumes of electricity, continuously and cheaply. We hear a lot about the AI race—in fact, it may turn out to be an electrification race.