Policies and institutions
This is a bit tricky to answer objectively, without ideology seeping in. The authoritarian-minded will say you need an authoritarian state, and give their reasons and justifications. Democratically-minded folks will do the same.
The Indus Valley Civilisation has cities of up to 60,000 in the Copper Age, and all evidence suggests it had an egalitarian social structure: most houses are around the same size, and there are civil works like a system of roads and sewers, which must have been organised by an organisation which worked for the people's interests. A hundred years of excavating has revealed no signs of kings.
Background elements
A good level of rainfall and probably a river and/or coast.
A reliable food source, ideally something like a potato. A potato gives 10× as much calories per unit area as a grain. There is much discussion on how potatoes contributed to urbanisation. Doesn't have to be literally potatoes, could also be taro or anything like that.
A forest for growing wattles for building (plus all the other forest-resources). The Cucuteni-Trypillia culture had cities of 46,000 people in the Stone Age and housed them mostly in wattle-and-daub. Wattle-and-daub was universal around the world.
Control of the hinterlands
The city needs to be fed by the countryside. This load can be lightened a bit by fishing on the sea. If the same civilisation/country/trading bloc that controls the city controls the countryside for about 50km from their walls, that means about a hectare of farmland feeds every mouth in the city.
There will need to be reliable transport networks to get the food to the city. Roman roads would work, but river rafts and canals are easier. A horse can pull 30-50 times more on a canal's towpath than it can on a road.