Alzheimer's disease (AD) is characterized by metabolic dysfunctions, making cerebrospinal fluid (CSF) an ideal biological matrix for characterizing brain metabolism. Very high-field 900 MHz nuclear magnetic resonance (NMR) spectroscopy was used to profile CSF metabolites from 47 patients (n = 27 AD, n = 20 non-AD). A total of forty-eight metabolites were identified and quantified. CSF concentrations of four nonpolar amino acids were reduced in AD: l-alanine (p = 0.001), l-valine (p = 0.02), l-leucine (p = 0.04), and l-phenylalanine (p = 0.04) (two-way ANOVA, pathology and sex as factors). No significant differences were observed for the other identified metabolites. A nonpolar amino acid factor (NAAF) was computed as the sum of these four metabolites (p < 0.001). Partial least-squares discriminant analysis (PLS-DA) based on the NAAF yielded moderate group separation (accuracy = 72%, Q2 = 0.22, p = 0.006). Receiver operating characteristic (ROC) analysis demonstrated that the NAAF achieved the highest discriminatory performance over individual nonpolar amino acids (AUC = 0.83). l-Alanine correlated negatively with CSF tau markers (p-tau: r = -0.40, p = 0.005; t-tau: r = -0.42, p = 0.003); the NAAF displayed similar patterns (p-tau: r = -0.38, p = 0.009; t-tau: r = -0.40, p = 0.006). Very high-field NMR spectroscopy revealed a depletion of nonpolar amino acids in AD CSF, reflecting metabolic dysregulation in this neurodegenerative disease.