Warburg phenomenon refers to the development of unique metabolic patterns during the growth of tumor cells. This study stratified gastric cancer into prognostic metabolic subgroups according to changes in gene expressions related to glycolysis and cholesterol synthesis. The RNA-seq expression data, single nucleotide variants (SNV), short insertions and deletions (InDel) mutation data, copy number variation (CNV) data and clinical follow-up information data of gastric cancer tissues were downloaded from The Cancer Genome Atlas (TCGA) database. ConsensusClusterPlus was used to stratify the metabolic subtypes of gastric cancer. Four metabolic subtypes (Cholesterogenic, Glycolytic, Mixed and Quiescent) of gastric cancer were identified, and patients with cholesterogenic tumors had the longest disease-specific survival (DSS). Genome-wide analysis showed that aberrant amplification of TP53 and MYC in gastric cancer was associated with abnormal cholesterol anabolic metabolism. The mRNA levels of mitochondrial pyruvate carriers 1 and 2 (MPC1/2) differed among the four subtypes. Tumors in the glycolytic group showed a higher PDCD1. A genomic signature based on tumor metabolism of different cancer types was established. This study showed that genes related to glucose and lipid metabolism play an important role in gastric cancer and facilitate a personalized treatment of gastric cancer.