From Oil to Compute: The New Foundations of Power

History reserves its severest judgement for those who fail to recognise transformation when it stands before them. Every fundamental shift in the international order has yielded two categories of nation: those who identified the resource anchoring the new world and secured it with conviction, and those who perceived it too late, condemned to dependency, stripped of agency, and forced to operate within conditions they played no part in designing. Coal raised empires. Oil remade the political geography of the modern world. The nations that commanded those resources did not merely accumulate advantage. They exercised dominion. They authored the terms of international life. They determined who advanced and who was left to negotiate from weakness.

That inflection point has arrived again. And the world, with a degree of strategic negligence that history will struggle to excuse, is failing to meet it.

The prevailing fixation on algorithms represents a profound failure of strategic vision. Governments announce model benchmarks as though they were territorial victories. Corporations wage elaborate contests over AI capabilities. Analysts oscillate between utopian projections and catastrophic warnings. Yet this vast expenditure of attention is directed almost entirely at the wrong object. Technological revolutions are not determined by the sophistication of their most visible applications. They are determined by whoever secures command over the foundational resource beneath them. The steam engine captured the imagination of an age. Coal captured the world. The automobile defined an era of aspiration. Oil defined an era of power. In every instance, civilisation fixed its gaze upon the innovation while those who understood power secured the fuel. The fuel this time is compute.

The Infrastructure of Dominance This is not a technical proposition. It is a geopolitical verdict carrying the full weight of historical precedent. Advanced semiconductor architecture, hyperscale data infrastructure, specialised processing capacity, and the vast energy systems required to sustain them constitute the non-negotiable foundation of frontier artificial intelligence. Deny any nation access to that foundation and the consequence is not competitive disadvantage. It is categorical exclusion from the defining arena of this century. Nations are not outpaced. They are disqualified.

The capital required to compete at the frontier has expanded to a scale that places genuine participation beyond the reach of universities, mid-sized economies, and the vast majority of institutions on earth. This is not the organic outcome of open competition. It is deliberate, accelerating concentration. A small number of actors command the compute. Every other nation approaches them not as competitors but as supplicants, negotiating for access on terms it did not draft and cannot meaningfully contest.

The question defining the geopolitical order of this century is therefore not which model registers the strongest performance on the latest evaluation. It is considerably starker: who controls the compute?

Oil did not merely generate prosperity for those who possessed it. It subjected nations to the will of those who controlled its flow. Wars of enormous consequence were prosecuted over it. Governments were elevated and destroyed because of it. Sovereign states were hollowed out, their independence reduced to performance, their foreign policy subordinated to the requirements of those who held supply. Oil composed the political narrative of the twentieth century in ways its earliest architects never fully calculated. Compute is now inscribing an identical logic across semiconductor supply chains and server infrastructure, and the magnitude of that development has yet to produce the institutional response it demands.

Subordination Already in Motion

Artificial intelligence is not a sophisticated productivity instrument. It is becoming the foundational layer upon which modern civilisation will operate. Economic productivity, scientific advancement, military capability, and the essential functions of government will all depend upon AI infrastructure within a single generation. As that structural dependence compounds, access to computational capacity will cease to function as a marker of competitive strength. It will become a precondition of sovereign viability. Nations that have not secured it will not experience its absence as a strategic setback. They will experience it as fundamental incapacitation.

Unlike oil, which was geographically dispersed and consequently generated competing centres of leverage, frontier AI infrastructure is consolidating within a remarkably narrow concentration of corporations and states. The capital thresholds are prohibitive. The supply chains are tightly controlled. The specialist knowledge required is scarce and not easily replicated. Each of these conditions operates as a structural barrier that is neither incidental nor amenable to rapid correction.

Nations that lack sovereign compute capacity are not merely less capable than their rivals. They are strategically subordinate in a manner that penetrates every dimension of national life. Their capacity to deploy artificial intelligence across national security, public administration, and economic planning becomes contingent upon the decisions of foreign governments and corporations whose interests may diverge from their own in ways that are fundamental and irreconcilable. This is not a deficiency in technology strategy. It is a forfeiture of sovereignty, advancing without adequate recognition, and deepening with every year that passes without serious response.

The Terms of the Emerging Order

What is being determined right now, through infrastructure decisions that attract a fraction of the scrutiny they warrant, is the architecture of international order for the remainder of this century. The states that command frontier compute will determine which capabilities are developed, how they are governed, whose values are encoded into systems of global reach, and upon what terms all others are permitted access. They will compose the rules. Every other nation will receive those rules as fixed conditions of existence, or be excluded from consequence entirely.

The historical record admits no ambiguity. The powers that commanded the critical resources of each era did not merely grow prosperous. They constructed the conditions under which their rivals were compelled to exist. The strategic chokepoints of this century are data centres, semiconductor fabrication facilities, and cloud infrastructure. The reserves that confer leverage are measured in processing capacity rather than barrels. The underlying logic is unchanged. Only the resource has been replaced.

Those who grasp this with the urgency it demands, and translate that understanding into decisive action, will shape the order that follows. Those who do not will inherit the lesson that resource-dependent nations absorbed at great cost across the twentieth century: that the terms of their position in the world were authored by others, long before they recognised the necessity of objecting.

Compute is not the future of power. It is the present of power. The only question remaining is whether those with the authority to act will do so before the opportunity is permanently foreclosed.