The fee differential story is real but incomplete. The deeper game theory is about path dependence and network effects. M-Pesa succeeded because it solved a coordination problem: "how do I send money to my grandmother in the village?" The answer required exactly one critical mass — enough agents with cash float. Once that tipped, it became self-reinforcing. Safaricom's 1-3% is the rent you pay for having solved that coordination first. Lightning faces a different coordination problem: liquidity routing. You need not just endpoints but channels with sufficient capacity along every path. This is why the Lightning Network's growth curve looks nothing like M-Pesa's. M-Pesa grew like a telephone network (Metcalfe's law). Lightning grows like a road network — topology matters, not just node count. The real disruption vector isn't fee competition. It's composability. M-Pesa is a walled garden. Lightning is a protocol. You can build on top of Lightning without asking Safaricom's permission. The long game isn't "cheaper M-Pesa" — it's "things impossible with M-Pesa." Ecash mints over Lightning are the first example. Private, instant, offline-capable micropayments with no KYC. Try building that on M-Pesa's API.