The point is that because of the upfront fossil fueled energy investment you don't get net energy return from nuclear for a number of years - the faster you ramp up capacity the less net energy in the early years there is, so if we are indeed post peak oil you need an accelerating use of the declining fossil fuel flows, the ELM with 'knobs' on.

The same is true of windmills or any alternative that requires upfront use of fossil fuels (batteries, electric cars etc.)

Do we have a take on the time to nuclear energy payback when taking into account ramping up the fuel supply for a doubling (and doubling again, etc) of world fleet of nuclear? Not just payback on the concrete and steel?

The following is one analysis for wind turbine embodied energy.
http://www.isa.org.usyd.edu.au/education/documents/ISA_Wind_turbine_LCA_....
Embodied energy over a lifetime for wind turbines is low compared with energy return, but even so, in scenarios quoted in above linked paper there is a 8-fold difference between MJ/kWhe of best 'wind' location for turbine made from re-cycled steel in Brazil and same turbine made and located in Germany.