Geothermal energy systems have the potential to significantly contribute towards the decarbonisation of heating and cooling. While robust methods exist for systems at local scale, assessing the geothermal potential at city scale remains prohibitively difficult. This constitutes a significant barrier to wider geothermal utilisation, as uncertainty can prevent investment. This paper introduces a methodology for city-scale geothermal potential quantification, building on previous work on quantifying the thermal state beneath cities. The methodology utilises statistical clustering to identify locations with similar thermal responses, producing ‘geothermal archetypes’, significantly reducing computational costs. Through the synergistic combination of numerical and analytical approaches, the potential for large-scale geothermal exploitation is quantified, considering downstream effects. This methodology is applied to two locations, Cambridge, UK, and Berkeley, USA, to perform a range of analyses, demonstrating its wide applicability. The importance of local demand is showcased, highlighting the inability of city centre areas to satisfy heating/cooling demand using solely geothermal energy. Notably, coordinated provision of heating and cooling, e.g., using networks, is shown to be a promising way to utilise geothermal energy holistically, increasing the proportion of the city-wide thermal demand that can be met from 45 % to 70 % for Cambridge and from 80 % to 100 % for Berkeley.