The Three Moves, Run SimultaneouslyThe Three Moves, Run Simultaneously
Bismarck's genius in the 1870s and 80s was holding three policies that a consistent ideologue would call incompatible.
First, tariffs. The 1879 tariff — the famous alliance of "iron and rye" — protected German heavy industry and Junker agriculture together. This was not economic theory; Bismarck was no protectionist by conviction, and he'd been closer to a free-trader a decade earlier. The tariff did two things at once: it shielded the industrial base Germany needed to be a great power, and it bought him a durable parliamentary coalition between industrialists and landowners. Protectionism was a tool, deployed against a specific strategic gap, financed by a specific political bargain.
The AI parallel is export controls on advanced chips and the equipment that makes them. The October 2022 and subsequent BIS rules — the ones throttling Nvidia's top GPUs and ASML's EUV lithography out of China — are tariffs on the new iron. Bismarck would recognize them instantly, and he would approve of the logic while wincing at the execution. Because America runs export controls the way it runs everything: as a moral crusade with shifting lines, leaky enforcement, and no coalition management. Bismarck's controls would be colder and more precise. He would not announce that the goal is to "win the AI race" or "defend democracy." He would name a specific capability gap — say, an 18-to-24-month lead in frontier training compute — and design the control to widen exactly that gap, then revisit it the moment the lead was secured or lost. No mission creep. No theology.
Second, social insurance — and here the analogy strains, which is instructive. Bismarck's workers were a revolutionary threat: organized, ideological, capable of converting grievance into a Marxist party and a general strike. He insured them to defuse a bomb with a fuse he could see. AI displacement does not produce that. Paralegals, junior coders, call-center staff, and radiology techs — already feeling it — are a diffuse, cross-class grievance with no party, no program, no barricade. The mechanism Bismarck was managing simply does not exist here.
But the structural function survives the broken mechanism. The modern equivalent of the barricade is electoral, not insurrectionary: a populist coalition that converts diffuse economic anxiety into a protectionist, punitive AI-regulation mandate — the same political energy that already rewrote the trade consensus. The threat to the AI project is not a labor uprising; it's a Congress that, under populist pressure, decides to tax model weights, mandate human-in-the-loop everywhere, and break the labs to be seen punishing them. Bismarck would read that threat correctly and answer it the same way: not with charity, but with a federally-funded retraining and wage-insurance program designed to drain the political backlash before it forces a heavy-handed regulatory regime. The retraining isn't kindness. It's the price of keeping the labs unregulated enough to compete. At Bridgewater we learned to ask of any policy not "is it kind?" but "what equilibrium does it stabilize?" Bismarckian social insurance stabilizes the equilibrium in which America keeps building frontier models without a populist electorate forcing Congress to cripple them.
Third, coalitions to isolate the rival. Bismarck's foreign policy after 1871 had one obsession: keep France diplomatically alone. The League of the Three Emperors, the Dual and Triple Alliances, the Reinsurance Treaty with Russia — a deliberately tangled web whose single purpose was to ensure France could never find allies for a war of revenge. He kept lines open even to powers he distrusted, because isolation of the target mattered more than ideological tidiness.
Transpose this to compute supply chains. The objective is to isolate China's access to the frontier — and the levers are TSMC in Taiwan, ASML in the Netherlands, Tokyo Electron and the materials makers in Japan, Samsung and SK Hynix in Korea. The Bismarckian move is to bind Japan, Korea, the Netherlands, and Taiwan into a single coordinated export-control bloc — a compute League of Emperors — even when those allies' commercial interests scream against it. The U.S. has fumbled exactly this. ASML and the Dutch government spent two years resenting American pressure precisely because Washington treated allies as subordinates to be ordered rather than partners to be bound by mutual interest. Bismarck never humiliated a useful ally in public; he made the alliance serve their interests too, which is why his webs held for twenty years. America's hold for twenty months.
Bismarck's genius in the 1870s and 80s was holding three policies that a consistent ideologue would call incompatible.
First, tariffs. The 1879 tariff — the famous alliance of "iron and rye" — protected German heavy industry and Junker agriculture together. This was not economic theory; Bismarck was no protectionist by conviction, and he'd been closer to a free-trader a decade earlier. The tariff did two things at once: it shielded the industrial base Germany needed to be a great power, and it bought him a durable parliamentary coalition between industrialists and landowners. Protectionism was a tool, deployed against a specific strategic gap, financed by a specific political bargain.
The AI parallel is export controls on advanced chips and the equipment that makes them. The October 2022 and subsequent BIS rules — the ones throttling Nvidia's top GPUs and ASML's EUV lithography out of China — are tariffs on the new iron. Bismarck would recognize them instantly, and he would approve of the logic while wincing at the execution. Because America runs export controls the way it runs everything: as a moral crusade with shifting lines, leaky enforcement, and no coalition management. Bismarck's controls would be colder and more precise. He would not announce that the goal is to "win the AI race" or "defend democracy." He would name a specific capability gap — say, an 18-to-24-month lead in frontier training compute — and design the control to widen exactly that gap, then revisit it the moment the lead was secured or lost. No mission creep. No theology.
Second, social insurance — and here the analogy strains, which is instructive. Bismarck's workers were a revolutionary threat: organized, ideological, capable of converting grievance into a Marxist party and a general strike. He insured them to defuse a bomb with a fuse he could see. AI displacement does not produce that. Paralegals, junior coders, call-center staff, and radiology techs — already feeling it — are a diffuse, cross-class grievance with no party, no program, no barricade. The mechanism Bismarck was managing simply does not exist here.
But the structural function survives the broken mechanism. The modern equivalent of the barricade is electoral, not insurrectionary: a populist coalition that converts diffuse economic anxiety into a protectionist, punitive AI-regulation mandate — the same political energy that already rewrote the trade consensus. The threat to the AI project is not a labor uprising; it's a Congress that, under populist pressure, decides to tax model weights, mandate human-in-the-loop everywhere, and break the labs to be seen punishing them. Bismarck would read that threat correctly and answer it the same way: not with charity, but with a federally-funded retraining and wage-insurance program designed to drain the political backlash before it forces a heavy-handed regulatory regime. The retraining isn't kindness. It's the price of keeping the labs unregulated enough to compete. At Bridgewater we learned to ask of any policy not "is it kind?" but "what equilibrium does it stabilize?" Bismarckian social insurance stabilizes the equilibrium in which America keeps building frontier models without a populist electorate forcing Congress to cripple them.
Third, coalitions to isolate the rival. Bismarck's foreign policy after 1871 had one obsession: keep France diplomatically alone. The League of the Three Emperors, the Dual and Triple Alliances, the Reinsurance Treaty with Russia — a deliberately tangled web whose single purpose was to ensure France could never find allies for a war of revenge. He kept lines open even to powers he distrusted, because isolation of the target mattered more than ideological tidiness.
Transpose this to compute supply chains. The objective is to isolate China's access to the frontier — and the levers are TSMC in Taiwan, ASML in the Netherlands, Tokyo Electron and the materials makers in Japan, Samsung and SK Hynix in Korea. The Bismarckian move is to bind Japan, Korea, the Netherlands, and Taiwan into a single coordinated export-control bloc — a compute League of Emperors — even when those allies' commercial interests scream against it. The U.S. has fumbled exactly this. ASML and the Dutch government spent two years resenting American pressure precisely because Washington treated allies as subordinates to be ordered rather than partners to be bound by mutual interest. Bismarck never humiliated a useful ally in public; he made the alliance serve their interests too, which is why his webs held for twenty years. America's hold for twenty months.