The Huaxin Securities Research Report pointed out that Mu Xi Co., Ltd.'s GPU shipments boosted performance and accelerated the implementation of domestic computing power ecosystem cards. Revenue volume resonated with cost optimization, and the Q2 single quarter deduction was corrected without return to parent. The demand for AI computing power extends to the entire process, and multi-card interconnection and software ecosystem have become competitive factors. According to statistics and forecast data from Toubao Research Institute, from 2021 to 2030, the total scale of AI accelerators in global data centers will grow from US$9.17 billion to US$730 billion. Of these, GPUs will dominate the market with ecological and performance advantages, and the market size will jump from US$6.67 billion to US$600 billion. At the level of demand structure, computing power requirements are shifting from “heavy pre-training” to “full-process balance”. At the hardware level, the impact of multi-card interconnection capabilities on the effective computing power of clusters continues to expand. In terms of ecology and customer implementation, the company continues to implement the “1+6+X” industrial ecosystem strategy. Among them, “1” is the digital computing power base, “6” is the six key industries of finance, healthcare, energy, scientific research, transportation, and entertainment empowered by AI applications, and “X” is cutting-edge application directions such as embodied intelligence, low-altitude economy, and space computing power. Considering the high demand for domestic AI chips, the company's full-stack self-developed GPUs and the MXMACA software ecosystem with good compatibility was covered for the first time, giving it a “buy” investment rating.

The Huaxin Securities Research Report pointed out that Mu Xi Co., Ltd.'s GPU shipments boosted performance and accelerated the implementation of domestic computing power ecosystem cards. Revenue volume resonated with cost optimization, and the Q2 single quarter deduction was corrected without return to parent. The demand for AI computing power extends to the entire process, and multi-card interconnection and software ecosystem have become competitive factors. According to statistics and forecast data from Toubao Research Institute, from 2021 to 2030, the total scale of AI accelerators in global data centers will grow from US$9.17 billion to US$730 billion. Of these, GPUs will dominate the market with ecological and performance advantages, and the market size will jump from US$6.67 billion to US$600 billion. At the level of demand structure, computing power requirements are shifting from “heavy pre-training” to “full-process balance”. At the hardware level, the impact of multi-card interconnection capabilities on the effective computing power of clusters continues to expand. In terms of ecology and customer implementation, the company continues to implement the “1+6+X” industrial ecosystem strategy. Among them, “1” is the digital computing power base, “6” is the six key industries of finance, healthcare, energy, scientific research, transportation, and entertainment empowered by AI applications, and “X” is cutting-edge application directions such as embodied intelligence, low-altitude economy, and space computing power. Considering the high demand for domestic AI chips, the company's full-stack self-developed GPUs and the MXMACA software ecosystem with good compatibility was covered for the first time, giving it a “buy” investment rating.