The current approaches for detecting most colorectal polyps and early neoplasms lack sufficient sensitivity and specificity, potentially hindering treatment and ultimately reducing survival rates. Here, we performed a metagenomic analysis to identify microbiome markers in stool samples from patients with early-stage colorectal cancer (CRC). We compared the composition of gut microbiota between patients with CRC and healthy individuals, specifically focusing on patients with early-stage CRC, defined as those without core mutations (KRAS, BRAF) for CRC diagnosis, stable microsatellite instability, and distant metastasis. The aim of our study is to identify potential biomarkers from gut microbiota at different cancer stages in colorectal cancer (CRC) patients through 16S rRNA amplicon sequencing, thereby proposing a novel non-invasive method for the early diagnosis of CRC. Specific microbes were detected from groups divided based on the TNM criteria, with one group classified by tumour size only (named the T group) and another group with lymph node metastasis (named the TN group). Aerobic bacteria, such as Delftia, Stenotrophomonas, Sphingobacterium, Rhodococcus, Devosia, Ensifer, and Psychrobacter were predominantly detected in patients with CRC without lymph node metastasis. The diagnostic prediction was evaluated using the CatBoost algorithm; these microbes presented high diagnostic accuracy with a receiver operating characteristics-area under curve of 0.8, which was validated using qPCR. In conclusion, this study identified specific aerobic microbial groups as non-invasive biomarkers for early diagnosis in patients with CRC without genetic or environmental factors.