The latter half of 2024 marked a period of explosive growth in China's commercial deployment of large AI models, with the daily average token consumption skyrocketing nearly tenfold and the annual number of public cloud large model invocations hitting 114.2 trillion tokens. This surge has significantly increased token usage within data centers, subsequently boosting east-west traffic and communication energy consumption, positioning the network as a critical bottleneck in AI infrastructure. To alleviate this challenge, the high-end network hardware market has surged, with China's shipments of high-end Ethernet ports exceeding 6 million in 2024 and projected to maintain a robust compound annual growth rate of 45.6%. Furthermore, expenditures on network hardware associated with generative AI have witnessed a remarkable surge, escalating from 6.5 billion yuan in 2023 to an anticipated 33 billion yuan by 2028, representing a compound annual growth rate of 38.5%. Notably, China's daily average token consumption has surged from 100 billion to an astonishing 30 trillion within just one and a half years, marking an over 300-fold increase.