Artificial intelligence is reshaping the landscape of gastric cancer surgery. Deep learning-based technologies, with their powerful capabilities in multidimensional data mining and pattern recognition, have demonstrated significant value across key clinical scenarios, including preoperative accurate staging, screening of occult peritoneal metastasis, prediction of neoadjuvant therapy response, intraoperative lymph node dissection and functional preservation, and postoperative individualized adjuvant therapy decision-making. The integration of radiomics and deep learning has increased the area under the curve (AUC) for predicting occult peritoneal metastasis to over 0.92. Intraoperative AI combined with indocyanine green fluorescence navigation and three-dimensional reconstruction offers a novel approach to preserve tumor-draining lymph nodes (TDLNs) without compromising oncological radicality. In the field of adjuvant therapy, AI can efficiently identify molecular features such as microsatellite instability-high (MSI-H), Epstein-Barr virus positivity, and HER2 expression from routine histopathological sections, facilitating precise selection of patients who are most likely to benefit from immunotherapy and targeted therapy. However, clinical translation of AI still faces challenges, including a lack of high-quality datasets, insufficient model interpretability, and weak prospective validation. In the future, surgeons should engage deeply in the development and iteration of AI models as definers of clinical problems and producers of high-quality data, and promote innovation through multimodal integration, federated learning, and explainable algorithms, thereby driving gastric cancer surgery from an experience-driven discipline toward a new paradigm of data-driven precision minimally invasive surgery.