万亿美元公司、创始人野心与 Token 预算：AI 周期里的资本配置新框架

The recent cluster of trillion-dollar outcomes resembles a punctuated technological equilibrium rather than a permanent annual production line.

A market can support a hundred-billion-dollar company eventually while still being unable to produce fifty to one hundred billion dollars of revenue within five years.

Outcome-based pricing allows AI applications to capture a share of service value instead of remaining limited by legacy per-seat software budgets.

Some exceptional founders are choosing defensible but narrow niches because they overestimate how naturally frontier labs will absorb every application market.

Boards should schedule a recurring, non-emotional review of whether the next six to eighteen months represent the company's best exit window.

The decisive opportunity cost of keeping a mediocre company alive is often the founder's most productive years, not merely the next financing round.

Because one year of AI time can compress several years of a normal technology cycle, strategy and financing assumptions require more frequent review.

When compute is scarce, laboratories allocate disproportionate token budgets to the small number of researchers expected to produce most of the progress.

Enterprises will move from encouraging universal experimentation to measuring return on invested tokens by person, project, and business outcome.

A safety regime that measures only downside risk and ignores expected benefit can entrench incumbents and slow socially valuable innovation.