Every generation has fought its own fuel wars. Our ancestors cleared forests for firewood, while the industrial revolution demanded mountains of coal. Oil reshaped the twentieth century, driving mobility, global trade, and conflict. Now we are entering a new chapter. Artificial intelligence is no longer confined to laboratories. It has become a planetary system with an appetite for electricity that rivals small nations. The future will not be measured in barrels of oil or cubic meters of gas but in terawatts of power. These are the first steps into what I call the energy wars of the algorithmic age.
Artificial intelligence does not simply operate on abstract code. It is machinery that consumes power and transforms it into decisions, predictions, and language. Recent studies suggest that data centres worldwide consume somewhere between 240 and 340 terawatt hours of electricity each year, with some estimates placing the figure closer to 400 TWh depending on what is included in the calculation. When I look at these numbers, I cannot treat them as a line in a sustainability report. This is a new foundation of power in the truest sense of the word. Energy is no longer just the fuel of nations. It is the oxygen of intelligence. Whoever controls electricity controls intelligence, and whoever controls intelligence will shape the future.
Artificial Intelligence and the Rising Demand for Power
Artificial intelligence grows by consuming computation, and computation is electricity organized into logic. Each advance in model size or training data requires exponential increases in energy. Training a frontier model is not the work of a few computers. It requires gigawatt hours of electricity, equal to the needs of entire towns. Once these systems are trained, they do not sit idle. They generate answers for millions of users, turning inference into a constant draw on the grid. The very success of artificial intelligence ensures that its hunger will never fade.
I find it frustrating when public debate reduces this to the carbon cost of a single training run. That argument misses the scale of what is happening. The real issue is whether we are willing to dedicate national-scale resources to machines that think, while homes and industries share the same fragile grid. Algorithms do not eat food, but they are the hungriest machines we have ever built. Artificial intelligence has become a sovereign consumer of power. It is no longer a luxury application. It is a competitor for energy alongside every other priority in society.
AI Data Centers and the Risk of Geopolitical Energy Wars
Energy scarcity has always created conflict, and AI adds a new layer to the struggle. Data centers cluster where electricity is abundant and cheap. Oregon’s hydroelectric facilities, Iceland’s geothermal wells, and the vast solar arrays of the Gulf states have become magnets for AI infrastructure. These locations are not simply attractive, they are strategic. They are the new oil fields of the twenty-first century. Nations that control these flows of energy will also control the flows of intelligence.
I believe governments will soon treat data centers as strategic assets rather than ordinary commercial real estate. In practice, some will be guarded as tightly as refineries once were, while others will function as bargaining chips in international negotiations. At the same time, these facilities are likely to emerge as high-value targets for cyber sabotage. None of this drifts into speculation. The logic of geopolitics makes the outcome unavoidable. When electricity and intelligence become inseparable, whoever controls the power to compute controls the power to decide.
National Security in the Age of Algorithmic Energy
National security now depends as much on kilowatts as it does on soldiers. Governments are already rewriting permitting laws to accelerate grid expansions that support AI. Militaries are planning for cyberattacks that could cripple electricity flows to critical data centers. Smaller nations with abundant renewable energy are being courted with offers that resemble the oil politics of the last century.
I believe we are underestimating how fast this transformation is taking shape. Within a decade, treaties will be signed not over crude oil but over guaranteed energy allocations for artificial intelligence. Transmission lines and substations will become as contested as pipelines once were. Without a pivot toward abundant energy sources, scarcity politics will dominate the algorithmic age as surely as they dominated the carbon age.
Fusion Energy and the Future of Artificial Intelligence
The bleak scenario is easy to imagine, but I do not think it is inevitable. Fusion offers the clearest path out of this trap. For decades, critics joked that fusion was always thirty years away. That line no longer holds. The U.S. National Ignition Facility has achieved repeated ignition events, producing more energy than consumed by its lasers. At the same time, private ventures such as Commonwealth Fusion Systems are building compact reactors that aim to achieve net-positive energy before the decade ends.
Fusion is not simply another energy option. It is the only credible technology that can deliver virtually limitless, carbon-free power at the scale required to sustain artificial intelligence. It avoids the dangers of meltdown, produces no long-lived radioactive waste, and relies on hydrogen isotopes that are both abundant and safe. To feed tomorrow’s algorithms, we may need to steal fire from the stars. If humanity succeeds in mastering fusion, the energy wars of the algorithmic age may never ignite.
The Current Landscape of Fusion Projects
To understand how real this is, we must look at the projects underway. ITER in France remains the largest multinational science project on Earth, aiming to demonstrate sustained plasma confinement. The United Kingdom is developing its STEP demonstration plant, with the goal of operation by 2040. These efforts show global commitment but also highlight the bureaucracy that slows progress.
The most exciting progress, in my view, comes from smaller ventures. Commonwealth Fusion Systems, Tokamak Energy, and several others are pushing aggressively toward reactors that can prove commercial viability within years. We have already crossed scientific thresholds many thought unreachable. The challenge now is scaling the technology and funding it at the level it deserves. If we continue treating fusion as a research experiment, it will take too long. If we treat it as urgent infrastructure, it may arrive in time to reshape the trajectory of AI.
Geothermal Energy and the Untapped Baseline of Power
Fusion dominates the headlines, but geothermal deserves far more attention. Enhanced geothermal systems can provide stable, dispatchable energy almost anywhere, not just in volcanic regions. Unlike solar or wind, geothermal does not depend on weather or time of day. Unlike fossil fuels, it does not poison the atmosphere. It is proven technology with political neglect as its main barrier.
From my perspective, the solution is obvious. Every subsidy spent on extending the life of coal plants should instead fund geothermal expansion. This is not glamorous science fiction. It is practical, immediate, and reliable. If we want to secure the future of AI while protecting the climate, geothermal must become part of the backbone of national energy policy.
Why Deinvestment in Carbon Fuels Cannot Wait
The temptation to rely on fossil fuels during moments of scarcity is strong. Leaders see short-term stability and voters see affordable power. Yet every dollar used to extend fossil fuels is a dollar stolen from the future. Artificial intelligence running on coal and oil produces a double disaster. It accelerates climate collapse while deepening the competition for dwindling resources.
I argue that we must deinvest from carbon fuels with urgency. This is not just about morality. It is about survival. If AI becomes locked into fossil grids, energy wars are unavoidable. Scarcity and pollution will reinforce one another. Feeding algorithms with carbon is like teaching the future to choke on the past. The only responsible strategy is to cut carbon out of the equation and redirect resources into geothermal, fusion, and other frontier technologies.
AI as Both Problem and Solution in the Energy Crisis
Ironically, artificial intelligence is not only a consumer of energy. It is also part of the solution. Machine learning is already being used to simulate fusion plasmas, balance electricity grids, and integrate renewables more effectively. The same algorithms that strain our resources can help us stretch them further.
This is why I do not support calls to slow AI in order to save energy. That idea misunderstands the geopolitical reality. Nations will not abandon the race for intelligence supremacy. Instead, we must demand that AI be applied to accelerate the transition to abundance. Let the very tools that hunger for electricity become the architects of efficiency. That is the only path that balances ambition with responsibility.
The Road Ahead for AI Energy and Fusion Futures
Two futures are emerging. In one, AI outpaces renewable investment, fossil fuels surge again, and nations slide into scarcity politics. Power grids fail under strain, cyberattacks target rivals, and militaries entrench themselves around data centers. The energy wars of the algorithmic age arrive in full force.
In the other future, we act decisively. Geothermal scales rapidly. Fusion achieves net-positive energy within a decade. Solar and wind keep expanding while AI itself optimizes energy systems. Compute becomes abundant, and the age of intelligence flourishes without scarcity wars. Which fork we take depends on decisions being made right now.
Choosing Abundance Over Scarcity
The energy wars of the algorithmic age are not destiny. They are a warning. The future will not be written only by silicon chips. It will be written by the choices we make about power.
Slowing artificial intelligence is not a realistic option. The appetite for intelligence is too strong, and the competition too fierce. The only path that respects both progress and survival is to scale energy systems to match. That means investing in geothermal, accelerating fusion, and refusing to subsidize fossil fuels that drag us backward.
We can ration and fight, or we can build and thrive. The future will not be decided by silicon. It will be decided by how we power it. Energy is the oxygen of intelligence, and I refuse to accept a future where we choke ourselves.
