Climate change mitigation requires to replace preexisting carbon-intensive capital with different types of cleaner capital. Coal power and inefficient thermal engines may be phased out by gas power and efficient thermal engines or by renewable power and electric vehicles. We derive the optimal timing and costs of investment in a low- and a zero-carbon technology, under an exogenous ceiling constraint on atmospheric pollution. Producing output from the low-carbon technology requires to extract an exhaustible resource. A general finding is that investment in the expensive zero-carbon technology should always be higher than, and can optimally start before, investment in the cheaper low-carbon technology. We then provide illustrative simulations calibrated with data from the European electricity sector. The optimal investment schedule involves building some gas capacity that will be left unused before it naturally depreciates, a process known as mothballing or early scrapping. Finally, the levelized cost of electricity (LCOE) is a misleading metric to assess investment in new capacities. Optimal LCOEs vary dramatically across technologies. Ranking technologies according to their LCOE would bring too little investment in renewable power, and too much in the intermediate gas power.
