I looked at one of the write-ups by Sunstealer. He seems to misunderstand the work done with the calorimeter.
Jones et al never state that the samples alone can melt steel.
They put the samples in a calorimeter -- and applied energy. At a certain temperature the red chips react -- and release a huge amount of energy -- for their size. In the process they produce iron droplets.
The iron was already there as iron oxide. When the reaction runs the iron droplets are a by product.
In larger quantities however- -the nanothermite will definitely melt steel.