Dopaminergic neurons in the substantia nigra pars compacta (SNc) are selectively vulnerable in Parkinson's disease, whereas closely related neurons in the ventral tegmental area (VTA) are largely preserved. Many molecular stressors have been implicated -- mitochondrial dysfunction, calcium load, α-synuclein toxicity -- but none by itself explains why two anatomically adjacent dopaminergic populations with similar molecular machinery diverge so sharply in their long-term fate. Here we analyze a minimal two-variable energetic model that captures only mitochondrial functional capacity, energetic reserve, and the combined structural and physiological loads imposed by axonal arborization and Ca(2+) handling. The model reveals a load-induced saddle-node bifurcation: above a critical combined load, the system develops **coexisting high-energy and low-energy states**, separated by a narrow separatrix. VTA-like neurons (low arborization, modest Ca(2+) load) reside in a monostable high-energy regime, whereas SNc-like neurons (extreme arborization, elevated Ca(2+) demand) fall inside a bistable window under the representative calibration used here. In this regime, modest energetic insults can drive trajectories across the separatrix into an irreversible collapsed state, providing a compact dynamical explanation for long periods of stability followed by abrupt decline. We emphasize that this is a sufficiency result within a provisional calibration, not a quantitative claim that the exact experimental placement of every SNc and VTA neuron has already been validated. Across parameter sweeps, two-parameter stability maps, and joint perturbation analyses, the bistable structure persists within the sampled nondimensional ranges. These results suggest that selective SNc vulnerability can arise from the geometry of load-dependent energy-mitochondria coupling, rather than from unique molecular defects, within this minimal framework and pending quantitative validation against bioenergetic measurements. The present model is spatially homogeneous and does not yet test whether the tipping point survives an explicit compartmental representation of the arbor; interventions shifting neurons away from this load-induced tipping regime may provide robust therapeutic leverage.