According to the CITIC Construction Investment Research Report, currently computing power is mainly deployed on the cloud side. The end side can not only become the gateway to the cloud side, but also provide the differentiated value of privacy protection, immediate response, low cost, and active response, so it is expected to become the next wave of AI. From a scenario perspective, end-side AI can not only be applied to large consumer terminals such as mobile phones, PCs, and glasses. The value of end-side AI also lies in the penetration of multiple fragmented scenarios. This is reflected in the fact that end-side AI will enable the upgrading of traditional mature products, and at the same time, end-side AI will also drive the birth of new terminals. Finally, in terms of implementation, the upgrades and iterations of mature products are implemented faster. Currently, we can see that some products are starting to be rolled out. The new AI terminals require a longer development cycle, and require deep integration of models, scenarios, and computing power to build pain-level system capabilities around user core points Side AI from” The last kilometer from “general capability” to “industry value” still requires mutual collaboration between model manufacturers and terminals, and will gradually be implemented in the future. The gradual penetration of these fragmented scenarios is one of the core reasons why the performance of leading domestic companies can continue to grow rapidly.

According to the CITIC Construction Investment Research Report, currently computing power is mainly deployed on the cloud side. The end side can not only become the gateway to the cloud side, but also provide the differentiated value of privacy protection, immediate response, low cost, and active response, so it is expected to become the next wave of AI. From a scenario perspective, end-side AI can not only be applied to large consumer terminals such as mobile phones, PCs, and glasses. The value of end-side AI also lies in the penetration of multiple fragmented scenarios. This is reflected in the fact that end-side AI will enable the upgrading of traditional mature products, and at the same time, end-side AI will also drive the birth of new terminals. Finally, in terms of implementation, the upgrades and iterations of mature products are implemented faster. Currently, we can see that some products are starting to be rolled out. The new AI terminals require a longer development cycle, and require deep integration of models, scenarios, and computing power to build pain-level system capabilities around user core points Side AI from” The last kilometer from “general capability” to “industry value” still requires mutual collaboration between model manufacturers and terminals, and will gradually be implemented in the future. The gradual penetration of these fragmented scenarios is one of the core reasons why the performance of leading domestic companies can continue to grow rapidly.