KERALA RISING · FROM EXPORTING PEOPLE TO EXPORTING IDEAS data products mature into their own revenue line, and the shared revenue base crosses the operating cost. From that point the institution is self-funding, and surplus is reinvested under the published policy to activate the next programs faster. The whole trajectory turns on one fact established in Chapter 5 and worth repeating: because inference is a rounding error, the crossover point is governed almost entirely by how fast the revenue ramps, not by the cost of the AI. A finance reviewer stresstesting this model should therefore pressure the revenue-ramp assumptions; corridor activation speed, partner sign-on, patron renewal, because that is where the real uncertainty lives, not in the compute bill. WORKED EXAMPLE: WHERE A FINANCE REVIEWER SHOULD PUSH HARDEST The model is most sensitive to three numbers, and none of them is the AI cost. First, corridor activation speed: how many corridors go live per period, since revenue scales with corridors while build cost does not. Second, Reward Partner sign-on and retention: the subscription base is the steadiest revenue line and the cheapest to grow once cleanliness has opened the door. Third, patron renewal: long-cycle diaspora and corporate capital is the line that most de-risks the early periods. Move those three to their pessimistic ends and the institution still works, it simply needs state 255