This paper describes strategies to lower the cost of providing Automated Teller Machine microdeposit services in rural contexts. Microdeposits represent a growing market in the developing world, but the cost of running a conventional ATM network is prohibitive due to the capital investment required to deploy networks and terminals. Our novel contributions are to use the Short Message Service (SMS) over high-penetration GSM cellular networks in conjunction with a system using location awareness to intelligently distribute available balances among machines. This allows us to provide high levels of service while simultaneously reducing risk to the financial institution and lowering per-transaction cost. Using a simulation of ATM usage patterns and distributions, our primary results under our model are: (1) transaction cost per user per year can be optimized to less than USD 0.18 given an SMS loss rate of approximately 10% while (2) customer withdrawal success rate can be maintained at approximately 98% with (3) a maxim urn of 5% of funds on deposit available in cash in ATMs at any given time. These results make wide deployment of rural ATM services by financial institutions feasible and economically viable in the near term using existing commodity technology.