Background: Autism spectrum disorder (ASD) affects approximately 1-2% of children worldwide, yet its etiology remains incompletely understood. Emerging evidence suggests that offspring of parents with autoimmune diseases show elevated autism prevalence. Notably, children of parents with psoriasis (OR 1.59), type 1 diabetes (OR 1.49-2.36), and rheumatoid arthritis (OR 1.51) demonstrate particularly strong associations.Hypothesis: I propose that autism is fundamentally an immune-metabolic disorder characterized by TNF-α-mediated mitochondrial dysfunction leading to cerebral energy deficiency. This energy deficit impairs three critical processes: (1) synaptic pruning during neurodevelopment, (2) real-time social cognition including gaze processing and emotion recognition, and (3) protein synthesis of critical synaptic scaffolding molecules. The primary mechanism involves TNF-α pathway dysregulation—through genetic inheritance from parents with autoimmune diseases such as psoriasis, type 1 diabetes, and rheumatoid arthritis, and/or through direct fetal exposure to elevated maternal TNF-α during pregnancy. I further propose that the well-documented "firstborn effect" in autism reflects maternal immune maladaptation during primigravid pregnancies. Additionally, for cases without parental autoimmune history, I propose a speculative secondary mechanism: mitonuclear immune conflict, where paternal immune genes may partially recognize maternal mitochondria as non-self, generating endogenous TNF-α. Implications: This hypothesis unifies disparate observations about autism pathophysiology and suggests that anti-inflammatory interventions targeting the TNF-α pathway may have therapeutic potential, particularly when administered early in neurodevelopment.