Lithium metal electrode (LME) poises as the best candidate for truly high-energy designs for rechargeable lithium batteries (RLB) when coupled with high-voltage, high-capacity cathode materials. To enable LME function well in such designs, very dense lithium (Li) deposition without dendrite growth is essential. To achieve this goal, acquiring fundamental understanding of Li nucleation and growth during deposition and the subsequent pathways toward stability during cycling at very negative, reducing potentials; is vital. It is well known that Li reacts with electrolyte to form solid electrolyte interphases (SEI). Putting the complicated SEI issues aside, understanding Li nucleation fundamentals is still very critical to ultimate Li deposition. Currently Li electrodeposition is limited by non-uniform behavior that leads to volume expansion due to pore generation, poor Li utilization (e.g. forming inactive or “dead” Li), and accelerated SEI formation and electrolyte decomposition due to increased interfacial area, just to name a few. The key question is, “Are these undesirable consequences come from the same origin or not?” Here, multi-scale observations by cryogenic-transmission electron microscopy (cryo-TEM) and advanced reactive-molecular dynamics (r-MD) simulations were used to understand the kinetic progression of Li nucleation. The cryo-TEM imaging revealed that amorphous metal deposits can be obtained at room temperature with very slow deposition rates, traditionally only possible with very high quenching rates (e.g. >106 K s-1) on Li nucleation and reveals the transition from amorphous disordered states to crystalline ordered ones as a function of current density and deposition time. The r-MD simulations provide additional understanding of the driving forces and kinetic pathways for the nucleation and the associated amorphous-to-crystalline (disorder-order or second-order) phase transition. The incubation time for the phase transition varies with the Li canonical ensemble size, mass and energy transfer rate between an ensemble and the neighboring ones, and as a function of current density (i.e. Li deposition rate). It is important to note that this is the first time the r-MD was used to simulate kinetic pathway in a discrete manner with canonical ensembles to assess the impacts from the kinetic regime, which is very different from the conventional temporal or spatial averaging methods according to the statistical thermodynamics. Although the r-MD results were developed independently from cryo-TEM experiments, i.e. without any empirical correlation; there is a high degree of agreement with experimental observations.