At present, the Chinese AI chip market is experiencing two pivotal shifts. Firstly, NVIDIA's market share is gradually being whittled away by domestic chipmakers. Secondly, the shipment volume of non-GPU accelerators has, for the first time, eclipsed that of GPUs, with the market share gap between NVIDIA and Huawei's Ascend series narrowing to a mere 6 percentage points. Behind these transformations, major companies with comprehensive AI strategies, such as Alibaba and Baidu, are taking decisive steps. They are not only propelling the research, development, mass production, and commercialization of their self-developed AI chips but also harnessing their proprietary models, cloud services, and high-traffic applications to refine and validate their products. Their goal is to capitalize on the burgeoning market opportunities arising from the swift growth in inference demand. When it comes to monetization strategies, domestic AI giants with full-stack capabilities mainly adopt two approaches: selling hardware and spinning off chip businesses for independent financing. The viability of this model has been demonstrated by Amazon's similar endeavors. Currently, the AI chip industry is characterized by intricate relationships among various players, with major companies simultaneously assuming multiple roles as buyers, shareholders, sellers, and competitors. Furthermore, production capacity shortages are driving the industry's competitive focus from performance benchmarking to capacity acquisition, with the future landscape of capacity allocation still up in the air.