AI could transform the economy, wipe out many of its builders, and still produce one or two industrial giants. Five histories explain who profits, who fails, and what government decides.
Real demand and scarce compute today; a later correction can transfer infrastructure from heavily indebted owners to companies with cash. One or two labs or their successors can reach GM-like scale through integrated products, then face a political settlement. Reviewed quarterly.
Each case seats the people who faced the same shape of decision before, and lets them advise. Their voices are informed reconstructions, never quotations.
Ask Otto von Bismarck how he would get a fleet out of the Strait of Hormuz once it had sailed in without a defined end state.
Case No. 1 · The Hormuz Trap →Ask J. Pierpont Morgan whether the AI build-out is a bubble, and who should be holding cash when the due dates on the data-center debt arrive.
Case No. 6 · The New Gilded Age? →Ask Andy Grove whether a frontier lab should walk out of the model business the way Intel walked out of memory chips in 1985.
Case No. 3 · Memories or Microprocessors →Ask Bill Gates what Microsoft should do about AI, three decades after his 1995 memo told the company the internet was a tidal wave.
Case No. 4 · The Second Tidal Wave →Ask Franklin Roosevelt what a party does with an insurgent wing it needs in the cities and fears in the suburbs, as he did in 1936.
Case No. 2 · The Rose at the Gates →Ask Alexis de Tocqueville why an exhausted political middle hands recruits to both extremes at once.
Case No. 5 · Extremes Travel in Pairs →MAGA holds one party outright and the DSA is taking the other's future. Across 2,400 years the record shows extremes rising in pairs, and it shows the middle escaping exactly once, from power, with an assassin's help.
v1.2 · updated September 1, 2026
An incumbent whose revenue model is a derivative on human headcount, facing suppliers turned competitors who sell headcount reduction, while holding the distribution, data, identity, and compute assets the attackers need.
v1.1 · updated September 1, 2026
A capital-intensive quality race in which the core product commoditizes on a known schedule, the challengers give their version away as strategy, and the incumbents must decide each quarter which layer of the stack they are in.
v1.1 · updated September 1, 2026
An insurgent faction whose members profit from the leader's punishments, inside a coalition that needs the faction's energy in the cities and bleeds from its brand in the suburbs, facing two leaders whose own city has already fallen and whose deadline is fourteen weeks away.
v1.1 · updated September 1, 2026
A no-exit middle position on peripheral terrain, entered without a defined end state, sustained by fear of humiliation, and hardening this week into a published rate war.
v1.1 · updated September 1, 2026