The committee responsible for securing the global semiconductor supply chain had recently taken a decision that was, by its own internal metrics, a triumph of consensus. They had successfully bifurcated the process into two distinct, self-contained streams of activity. Stream A was the Control of Things, and Stream B was the Expansion of Things. The beauty of the system, as explained in a memo that was widely circulated and never read, was that the two streams operated in perfect harmony, each achieving its stated goals with a bureaucratic elegance that was a joy to behold, provided you never, ever, looked at the two streams at the same time.

Stream A, the Control of Things, was concerned with ensuring that the most advanced lithography machines, the ones that can draw circuits with beams of light so fine they’re almost embarrassed to be seen in public, did not fall into the wrong hands. This was a sensible and prudent aim, and was achieved by a sub-committee on Export Controls, which drew up lists, checked them twice, and established a rigorous, multi-layered approval process for any machine travelling to any destination that had ever looked at a silicon wafer with what might be construed as undue interest. The process was watertight. It was so effective that it had, almost as a by-product, created a new and vibrant secondary industry in interpreting the lists, which was widely seen as a net economic positive.

Stream B, the Expansion of Things, was concerned with ensuring that the world had enough semiconductors to power everything from toasters to orbital defence platforms. This was also a sensible and prudent aim, and was achieved by encouraging, with subsidies and supportive nods, the expansion of fabrication capacity in geographically diverse locations. The logic was impeccable: if you build more fabs in more places, you reduce the risk of a single point of failure. This stream was managed by a sub-committee on Capacity and Resilience, which produced charts showing wonderful upward curves and was generally felt to be doing a sterling job.

The problem, which the main committee had not yet had a meeting to consider, and for which there was no scheduled agenda item, was that Stream A and Stream B were, in fact, the same stream. Or rather, they were two parts of a single process that was now working at cross-purposes with such quiet diligence that it had achieved a kind of bureaucratic perfection.

You see, the most advanced lithography machines, the ones being so diligently controlled, are built by a company that exists in a place. And the expansion of fabrication capacity, being so enthusiastically encouraged, requires those very same machines. So, the system had evolved, without anyone deciding it should, into a beautifully balanced mechanism whereby one sub-committee was carefully limiting the availability of the key tool, while the other sub-committee was vigorously promoting projects that could not proceed without it. It was like a town council that, on public health grounds, strictly rationed the sale of bricks, while simultaneously, on economic development grounds, offering generous grants for the construction of new hospitals.

The individual engineers at the lithography company were not stupid. They were brilliant. The individual planners at the fabrication plants were not stupid. They were visionary. The individual members of both sub-committees were not stupid. They were diligent, well-meaning, and had excellent Power Point skills. Yet the collective output was a kind of sublime stasis, a state of affairs where the goal of security had been so perfectly balanced against the goal of supply that the entire system had achieved a graceful, hovering equilibrium, producing reams of excellent policy documentation and a growing, global shortage of the very thing it was designed to produce.

The system was working exactly as designed. This was the problem. It had been designed to demonstrate control and demonstrate expansion, two politically vital forms of demonstration. It had not been designed, in the end, to actually produce semiconductors in a coherent manner, because that would have required a single committee with a single, measurable objective, and such a thing was clearly an administrative fantasy. The real, practical outcome - the stress, the bottlenecks, the geopolitical friction - was merely the exhaust fume of a perfectly tuned engine running smoothly in a locked garage. Everyone could smell it. Everyone was busy writing reports about air quality. And the engine, purring away, was doing exactly what it was built to do: run.