High temperature (>900 • C) industrial waste heat recovery remains a key challenge for thermoelectric materials. The unique combination of high temperature, low heat-flux, and large surface area of waste heat generation as analyzed herein shows that active materials cost is the main metric inhibiting application. Molten compounds with semiconducting properties are therefore proposed as a cost-effective addition to solid-state materials for these conditions. A review of prior experimental results is presented, after which we demonstrate the performance of a laboratory-scale device based on molten SnS. The results allow reporting, for the first time, the Figure of Merit (Z T ) and the conversion efficiency of the candidate materials. In addition, the Seebeck coefficient of molten SnS is reported. The results confirm the opportunity offered by molten thermoelectric compounds and allow discussion of the remaining materials and engineering challenges that need to be tackled in order to envision the future deployment of thermoelectric devices based on molten semiconductors. Industrial waste heat represents an important source of energy (around 5% of the total US annual energy production, i.e. 1,000,000 GWh), a significant fraction of it being generated at high temperature (around 20% of the industrial waste heat is at T > 650• C).1 Because of such high exergy, there is an interest in directly recovering this energy in the form of electricity, potentially via thermoelectric conversion.Though exceptional progress on solid-state materials for thermoelectrics has been accomplished, including extending their maximum range of operating temperature, there are to date no material systems and devices that have been implemented in the factories which generate this high quality waste heat.We herein first analyze the cost and materials properties required to potentially enable such deployment. The analysis reveals that for large-scale (i.e. large surface of heat dissipation), low heat-flux and high temperature applications such as primary steel or glass production, the cost of the thermoelectric materials is the primary driver for materials choice. In addition, the heat-generating locations in the existing processes call for materials compatible with very high temperature. Finally, the chosen materials system will have to exhibit mechanical stability when subject to a large temperature difference at high temperature.Additionally, a survey of the prior art and available data show that molten semiconductors are likely to fulfill the majority of the proposed criteria, suggesting the opportunity to complement solid-state materials to extend thermoelectricity to high-temperature industrial waste heat recovery.Finally, the practical opportunity offered by molten thermoelectrics is demonstrated, and both experimental materials property data as well as performance of a laboratory-scale device are reported. The results enable the discussion of the remaining materials science and engineering challenges to allow further developments and fut...