The Compute Map Redrawn: Scenarios and Second-Order Effects

Having assembled the pieces, minerals, diplomacy, screening and finance, it is worth stepping back to ask where the redrawn map leads. The thesis of this course is that export controls reshape the global compute map, and the honest way to test a thesis is to trace its consequences forward and to name the ways it could be wrong. Consider three broad trajectories and the second-order effects that shadow each.

The first trajectory is successful containment. In this scenario the allied bloc holds its coalition together, the controls bite, the screening perimeter seals, and the frontier of AI capability remains concentrated in the United States and its aligned partners. China, cut off from the most advanced chips and tools, falls a meaningful generation behind at the frontier even as it dominates the mineral inputs and builds vast capacity at the trailing edge. The compute map in this world is bifurcated but asymmetric: one bloc leads in capability, the other leads in the physical supply chain, and each holds leverage over the other. The stability of this arrangement depends on neither side finding it intolerable, which is a demanding condition.

The second trajectory is leakage and adaptation. Here the controls slow China but do not stop it. Determined investment, indigenous innovation, the recruitment of talent, the exploitation of coalition seams and the sheer size of China's domestic market allow it to build a parallel frontier, perhaps a step behind but self-sufficient and immune to further pressure. In this world the controls prove to have been a spur rather than a cage, accelerating Chinese self-reliance and forfeiting the revenue and influence that engagement once bought. The minerals leverage remains, but now it is wielded by a rival that no longer depends on Western technology, a far more dangerous configuration. The lesson of this trajectory is that controls buy time, and the value of that time depends entirely on what the controlling bloc does with it.

The third trajectory is fragmentation and cost. In this scenario the effort to sort technology and capital by geopolitics succeeds so thoroughly that it fractures the efficiencies that made the technology cheap in the first place. Duplicated supply chains, stockpiled minerals, screened-out capital and politically constrained markets raise the cost of compute for everyone, including the allied bloc. Third countries, forced to choose sides, resent the choice and hedge where they can. The global capacity to build AI grows more slowly and more expensively than it would in an open world, and the benefits of the technology arrive later and less evenly. This is the price of security, and whether it is worth paying is a genuine question rather than a settled one.

The real future is likely to blend all three, varying by layer of the stack and by region. What matters for the analyst is to hold the scenarios simultaneously and to watch the indicators that tip the balance among them. The cohesion of the allied coalition is one such indicator; when the Netherlands, Japan and South Korea hold the line, containment strengthens, and when they defect, leakage grows. The pace of Chinese indigenous progress in lithography, chip design and mineral-free magnet substitutes is another. The stability of the conditional bargains with Gulf and other partner states is a third. And the political durability of the whole apparatus across changes of government in the leading democracies is perhaps the most important of all, because a control regime that reverses every four years cannot sustain the credibility it needs to work.

Faraday's read is that the redrawn compute map is now a permanent feature of the strategic landscape, not a passing episode. The assumption that governed the previous era, that technology and capital flow to their most efficient use regardless of borders, has been replaced by an assumption that they flow within geopolitically defined channels policed by controls, diplomacy and screening. This does not mean globalisation is dead, but it means that in the specific domain of advanced compute, the logic of security has overridden the logic of pure efficiency, and it is unlikely to reverse while the underlying rivalry endures. The map has been redrawn, and the pencil is still moving.

The closing discipline for the expert is to resist both complacency and fatalism. Complacency assumes the controls will simply work and the map will settle in the controlling bloc's favour; the leakage scenario warns against that. Fatalism assumes the effort is futile and the costs pointless; the containment scenario warns against that. The mature posture holds the tension: controls are a real and consequential instrument that genuinely reshapes the compute map, but their success is contingent, their costs are real, and their outcome depends on choices not yet made. Whoever understands the map as it is being drawn, rather than as it was or as they wish it to be, will read the coming decade of artificial intelligence more clearly than those who mistake the frontier of models for the whole of the contest. The minerals, the megawatts and the passport on the capital are where the future is being decided.