Experimental data in conjunction with mathematical models are used to explain various aspects of laser casting and laser cladding by the preplaced powder method. For increasing speed, the data can be distinguished between substrate melting, dilution-free clad bonding, bond-free casting, and poor contact breaking the melt track into spheres. Results include a thermodynamic explanation of the wide range of process parameters over which dilution-free clad deposits can be produced, as the process switches from heating of the insulating powder to additional cooling when the melt front reaches the substrate. Also, the interaction of the melt pool with the powder bed is analyzed to identify why laser castings have microscopically uneven surfaces and do not bind with the substrate. The advancement of the melt front through the powder layer is governed by heating, melting, and incorporation of each individual grain. Although most powder grains are in the small size range for the case studied, the few particles up to a factor 3 larger delay and therefore govern the front advancement due to much slower melting and surface tension driven incorporation, depending on the particle size in a nonlinear manner.