Infrastructure—communications, energy, transport, waste, and water networks—is critical for economic activity and social well-being. Practitioners, politicians, and economists advocate high levels of investment in infrastructure under the rubric of 'planning for growth' (or the 'push' paradigm). This paradigm relies on complex public-private arrangements in the name of public interest. These seemingly reasonable arrangements are, however, not delivering their promise. Evidence shows that the needs of infrastructure users in rich and poor countries are not being met, many private providers of infrastructure earn rich returns, assets are rarely built in time or on budget, and there is tremendous waste in the operation of many infrastructure industries. No other sector could survive the profligacy and slack common in infrastructure. I distil the following primary propositions of the accepted wisdom, which is inspired by mainstream economics: First, infrastructure assets necessarily entail high sunk costs and large economies of scale. Consequently, assets last for very long periods of time, and they cannot be readily moved. Second, infrastructure outputs are homogeneous. Third, one network fits all users (large and small). Fourth, infrastructure users, even large ones, are likely to have weak bargaining power in procurement of infrastructure outputs. I challenge these four propositions of the conventional wisdom by putting forward alternative hypotheses. First, instead of being monolithic and costly, infrastructures can be assembled (and disassembled) as flexible modules for specific users in specific places. Drawing on option pricing theory in quantitative finance, I recast infrastructures as 'portfolios of real options'. Second, infrastructure outputs are, in fact, heterogeneous and differentiated services. Third, one infrastructure network cannot fit all users, either today or in the future. Users are remarkably heterogeneous, not only in terms of unique user preferences but also in terms of spatial location. Infrastructure networks need to evolve in tandem with user needs or risk spatial, temporal, and relational obsolescence. Finally, users, large and small, are adept at exerting strong bargaining power in procuring infrastructure both prior to and after rendering durable and immobile investments. Users also strategically deploy intermediaries, e.g. futures and Over-the-Counter (OTC) exchanges, and real estate developers, to negotiate private contracts for infrastructure services. These findings are supported by two case studies. The first case study details the process by which ThyssenKrupp, a large steel company, bargained for its infrastructure by locating to a manufacturing site in the U.S. The second case study focuses on residents of Lavasa, one of the largest property developments in India. Here, small users of infrastructure exert strong bargaining power with the aid of intermediaries—the real estate developer and the property asset manager. New departures in infrastructure provision are urgently needed at a practical level. Poor investments rendered today—particularly if costly, inflexible, and durable—will suffocate tomorrow’s possibilities. The spatial, temporal, and relational approach proposed in this dissertation begins to offer an alternative account of how tomorrow can be modularly shaped.