We examine the characteristics that fusion-based generation technologies will need to have if they are to compete in the emerging low-carbon energy system of the mid-twenty-first century. It is likely that the majority of future electric energy demand will be provided by the lowest marginal cost energy technology—which in many regions will be stochastically varying renewable solar and wind electric generation coupled to systems that provide up to a few days of energy storage. Firm low-carbon or zero-carbon resources based on gas-fired turbines with carbon capture, advanced fission reactors, hydroelectric and perhaps engineered geothermal systems will then be used to provide the balance of load in a highly dynamic system operating in competitive markets governed by merit-order pricing mechanisms that select the lowest-cost supplies to meet demand. These firm sources will have overnight capital costs in the range of a few $/Watt, be capable of cycling down to a fraction of their maximum power output, operate profitably at low utilization fraction, and have a suitable unit size of order 100 MWₑ. If controlled fusion using either magnetic confinement or inertial confinement approaches is to have any chance of providing a material contribution to future electrical energy needs, it must demonstrate these key qualities and at the same time prove robust safety characteristics that avoid the perceived dread risk that plagues nuclear fission power, avoid the generation of long-lived radioactive waste and demonstrate highly reliable operations. This article is part of a discussion meeting issue ‘Prospects for high gain inertial fusion energy (part 1)’.