Interstate natural gas transmission and storage infrastructure is facilitated using regulated, private transactions. Pipeline companies obtain long-term contracts from producers and wholesale purchasers, typically local distribution companies (LDCs). Historically, the Federal Energy Regulatory Commission (FERC) accepted these counterparty contracts as sufficient justification of need. Typically the LDCs are themselves regulated firms, which sometimes possess affiliations with pipeline companies. But with contracted costs largely passed through to retail customers via regulated prices, it is unclear whether contracting parties face sufficient competition or otherwise possess an incentive to find least-cost alternatives. To aid evaluation of past and future investments, we develop a national-level optimization model that can assess the need for new interstate pipeline and storage facilities. The model takes production and demand pathways as fixed and minimizes the infrastructure and operation costs of transport and storage in order to balance supply and demand on each day in each state. Transport of gas can be achieved using pipeline transmission of dry gas, or using truck or ship transport of liquefied natural gas (LNG), and optimal placement of liquefaction and gasification facilities. The model also accounts for international imports and exports of both dry gas and LNG. Three underground dry-gas storage facilities are considered, as well as LNG storage. We compare the model's optimized plan with observed outcomes as the sector grew rapidly with hydraulic fracturing. We find that the U.S. has built 38 percent more pipeline and 27 percent more underground storage than necessary, amounting to roughly \$179 billion in excess investment. It would have been more economic to expand pipeline far less than observed and instead satisfy critical-peak demands for gas using LNG, plus necessary liquefaction and gasification facilities. Differences between optimized and observed investments vary across the interstate network, while flows between states and into and out of storage bear a close resemblance to observed outcomes.