Late-onset Alzheimer's disease (LOAD) is framed here as progressive astrocyte-neuronal metabolic and neurovascular uncoupling initiated by astrocytic bioenergetic collapse. In genetically or environmentally predisposed brains, a self-reinforcing loop of lipid accumulation, inflammation, vascular impairment, glucose-handling defects, and mitochondrial dysfunction erodes astrocytic functional capacity. Subsequent cerebrovascular dysfunction and loss of blood-brain barrier (BBB) integrity perpetuate the neuroinflammatory response and drive amyloid-β deposition. Astrocytic failure then disrupts astrocyte-neuron metabolic and neurovascular coupling, compromising lactate shuttling, glycogen mobilization, glutamate uptake, potassium buffering, antioxidant support, lipid handling, and demand-perfusion matching. Neurons deprived of this support enter chronic energy stress with sustained AMPK activation, which enhances tau hyperphosphorylation, perturbs proteostasis, and reduces tau-GlcNAc protection, fostering pathological tau assembly. Amyloid-β deposits are enriched with heparan sulphate proteoglycans that provide a polyanionic scaffold which, together with persistent AMPK and inflammatory signaling, concentrates and misfolds tau into paired helical filaments. Tau-mediated mitochondrial injury further amplifies neuronal energy failure and feeds back via inflammatory pathways to worsen astrocytic dysfunction, closing the loop. Failure of the astrocyte-neuron lactate shuttle is identified as a key bridge between astrocytic and neuronal bioenergetic failure, where reduced lactate shuttling is proposed to impair long-term potentiation, thus accounting for the typical amnestic presentation of LOAD. Astrocytic bioenergetic load is predicted to peak in default-mode network hubs and other cortices with high resting aerobic glycolysis, reflecting reliance on astrocytic glycolysis for lactate shuttling and thereby accounting for the regional vulnerability observed in LOAD. This bioenergetic failure model integrates amyloid-β, tau, vascular, metabolic, and inflammatory findings into a single framework that accounts for genetic risk factors such asand. Falsifiable temporal sequencing predictions for LOAD and specific forms of early-onset Alzheimer's disease are generated from the model. O APOE TREM2
