AI hardware currently lies in popularizing the use of their large AI models. Especially on the training side, Scaling Law has reached a bottleneck due to the limitation of training data, but the model can still continue to learn and think on the reasoning side. The key is to deeply combine private domain user data learning to improve accuracy while reducing reasoning costs.
In addition, the iteration of smart wearable hardware often revolves around the improvement of major functions. For example, the previous requirements were low latency, low power consumption, and fast response, while the focus in the future will be on multimodal processing capabilities, such as visual and auditory discrimination perception, which naturally puts forward new requirements for the system integrated chip SOC.
02
Can it be sustained?
In fact, it is not surprising that sectors such as semiconductors, robotics, and uae phone number list consumer electronics have strengthened.
First of all, there were a lot of catalysts in the technology circle in December.
Open AI, Google, Tesla, Apple, Broadcom from abroad and ByteDance, Huawei, Tencent, BYD, etc. from China have all announced new developments in AI technology and products.
Such intensive catalysis has given new stimulation to the AI topic which has been dormant for some time.
Secondly, since midNovember, both the overall market and the technology sector have undergone a round of adjustments. Coupled with the overall loose liquidity, it is beneficial to technology stocks. From a technical perspective, these sectors also pulled back to support levels.