AI Did Not Demand Centralised Power. Vested Interests Did.

Aldo Grech ,
Linked-In From Legacy to Lifeforce. December 28, 2025

here is a comforting lie spreading fast, repeated by governments, legacy energy players, defence contractors, and now increasingly by parts of the tech sector itself. It says that artificial intelligence has created an unprecedented energy emergency, that AI requires vast, centralised, always-on power, and that this necessity leaves us no choice but to return to fossil fuels and nuclear energy.

The lie works because it sounds technical, urgent, and inevitable. It borrows the language of engineering while quietly abandoning engineering logic altogether.

AI did not demand this outcome. Vested interests did……………………………………………..

The claim that AI “needs” fossil fuels or nuclear power collapses almost immediately when examined through any serious engineering lens. AI’s energy demand is real, but its characteristics are consistently misrepresented. AI workloads benefit from flexibility, modularity, geographic distribution, rapid deployment, and declining marginal costs. These map naturally onto renewables paired with storage and intelligent demand management. Solar, wind, batteries, and grid-level AI can be deployed in months, not decades. They scale incrementally. They fail locally rather than catastrophically. They reduce exposure to geopolitical shocks and price volatility.

What AI does not require is a small number of hyperscale generation points that take decades to build, depend on political stability, demand water and security, and concentrate failure risk. Those attributes belong not to AI’s needs, but to institutional preferences.

The renewed enthusiasm for fossil fuels and nuclear energy is therefore not a response to AI’s requirements. It is a response to AI’s implications.

Distributed intelligence paired with distributed energy is profoundly destabilising to legacy power structures. It undermines the logic of scarcity on which energy rents, geopolitical leverage, and industrial monopolies depend. It reduces the strategic value of chokepoints. It weakens the ability of states and corporations to weaponise supply. Most importantly, it enables communities, regions, and smaller actors to operate with greater autonomy.

That is what is being resisted.

Fossil fuels and nuclear energy share one critical feature that renewables do not. They require centralisation. Massive capital expenditure, state-scale regulation, long approval chains, security apparatuses, liability shields, and political alignment are not side effects of these energy sources. They are their operating conditions. Control is not an accidental by-product. It is the point.

Nuclear energy exposes this more starkly than any other technology. It is routinely defended as pragmatic, scientific, and unavoidable, yet it fails the most basic test of rationality: insurability. There is no private insurance market willing to underwrite nuclear risk in full. Not accidents, not waste, not abandonment, not war, not institutional failure, not the passage of centuries. Every nuclear programme on Earth exists only because the state absorbs the downside, caps liability, and transfers risk to future generations who had no voice in the decision.

This is not a marginal flaw. It is the defining characteristic of the technology.

Nuclear waste remains hazardous for tens of thousands of years. That timescale dwarfs governments, nations, languages, and civilisations. We are being asked to believe that societies struggling to meet five-year climate targets and maintain basic institutional trust will flawlessly manage radioactive stewardship across millennia. This is not engineering optimism. It is narrative denial.

The uninsurable nature of nuclear energy should end the debate in any system that claims to respect markets, responsibility, or realism. If a technology cannot exist without forcing unborn humans to absorb its catastrophic downside, it is not a solution. It is a temporal externalisation of risk disguised as progress.

Artificial intelligence does not mitigate this problem. It intensifies it. AI accelerates feedback loops, increases coupling between systems, and magnifies rare failures into systemic ones. Nuclear energy, by contrast, depends on extreme institutional continuity, conservative change, and zero-failure tolerance. Pairing AI urgency with nuclear timelines is not strategy. It is a category error designed to launder political decisions through technological rhetoric.

The same pattern now plays out visibly in transport……………………………………………………………..

The world is not racing back to fossil fuels and nuclear energy because the future demands it. It is being pulled back because decentralisation works too well, too fast, and too broadly for existing power structures to absorb.

AI did not ask for this future. Neither did consumers. Vested interests did. And that is the story we are not being told.

If decentralisation were truly the threat it is portrayed to be, its consequences would already be visible wherever it has quietly taken root. Instead, what we see is the opposite. Where energy becomes local, intelligence distributed, and dependency reduced, human systems do not fracture. They stabilise.

The gaslit fear that drives the defence of centralised fossil fuels, nuclear power, and legacy mobility is not fear of collapse. It is fear of irrelevance. It assumes that without top-down control, societies cannot function, markets cannot coordinate, and people cannot be trusted. That assumption says more about the architecture of the institutions making it than about human capability.

Decentralised systems shift the locus of resilience. When energy is produced close to where it is consumed, outages shrink in scale and duration. When storage is modular rather than monolithic, failure becomes manageable rather than catastrophic. When intelligence is embedded at the edge rather than concentrated at the centre, systems respond faster, adapt locally, and degrade gracefully instead of collapsing wholesale. These are not ideological preferences. They are properties of robust systems.

The human consequences of this shift are profound……………………………………………

Decentralising vested interests does not mean eliminating institutions or abandoning expertise. It means removing their ability to externalise risk, hoard rents, and dictate futures without consent. It means designing systems that assume change rather than deny it, and that distribute both power and responsibility accordingly.

If allowed to unfold, decentralisation does not weaken societies. It humanises them.

And that is precisely why it is being resisted. https://www.linkedin.com/pulse/ai-did-demand-centralised-power-vested-interests-aldo-grech-hsumf/