KERALA RISING · FROM EXPORTING PEOPLE TO EXPORTING IDEAS for identity, verification, audit, payments, dashboards, and field operations, in seven incompatible silos, none of which the state owns and all of which end when the contract ends. Building the institution means paying for those capabilities once, owning them in perpetuity, and configuring each new program at a fraction of the cost. The vendor model is not merely more expensive; it is structurally incapable of the reuse that makes the marginal program cheap. And it leaves the state renting, in perpetuity, a capability that the AI era makes foundational, the equivalent of renting the payments rail rather than owning it. ‘This is too ambitious. Why not just do the cleanliness app?’ Because the cleanliness app, on its own, is a line item that a future government cancels, and because the expensive part, the institution, the stack, the audited corpus, the partner relationships, the Reward Partner and patron base, is built whether you run one program or seven. Having paid to build the machine, running only one program on it is the really wasteful choice. The ambition is not a cost to be minimised; it is the thing that makes the cost worth bearing. The truer framing is the reverse of the objection: it is doing only the cleanliness app that is too expensive, because it pays the full build cost for a single return. 283