On August 3, Alibaba released its next-generation flagship model Qwen3.8-Max, with a total of 2.4 trillion parameters and roughly 950 billion activated parameters. The model shows further improvements in reasoning, coding, and Agent capabilities. Meanwhile, DeepSeek announced on August 6 that it plans to raise its API service pricing overall in the near future. Hon Hai Precision Industry Co. reported record monthly revenue exceeding NT$900 billion for the first time in July, setting a new historical high. TrendForce has revised its 2026 global AI server shipment growth forecast upward to nearly 31% year-on-year. The global AI computing power industry remains in a strong cycle, warranting continued attention to potential rebound opportunities among oversold stocks in the AI computing power supply chain.
Where to focus
Alibaba's release of the Qwen3.8-Max model, with its 2.4 trillion total parameters and 950 billion activated parameters, enhances capabilities in reasoning, coding, and Agent functions. The company has also started public beta testing of its enterprise-level Agent, Qianwen Office. This signals that the competition among major models is shifting from pure model capability comparisons to accelerated deployment in Agent and commercial applications. DeepSeek announced on August 6 that it plans to raise API service prices overall, with a potentially significant increase, though the specific plan awaits formal notice. Previously, DeepSeek had introduced a peak-and-valley pricing mechanism, where API prices during peak hours are double those during off-peak hours.
Why the industry trends matter
China Securities believes that the expansion of model scale and the strengthening of Agent capabilities by leading players, combined with DeepSeek's move toward price increases, reflects that domestic large models are entering a phase of "capability improvement plus commercial monetization." At the same time, the expansion of model parameter sizes, increased frequency of Agent calls, and growing API demand are expected to further boost inference-side computing power consumption. The firm advises continued attention to investment opportunities in the domestic large model and domestic computing power supply chain. Hon Hai reported July 2026 revenue of NT$946.5 billion, up 54.2% year-on-year, marking the first time monthly revenue has exceeded NT$900 billion and setting a new record. This was mainly driven by robust growth in cloud and network products, with strong shipments of AI servers and AI cabinets. The company expects AI cabinet shipment momentum to continue into the third quarter. Additionally, TrendForce has recently raised its AI server demand forecast, revising the 2026 global AI server shipment growth rate to nearly 31% year-on-year, up from a previous estimate of about 28%. This revision is driven by increased procurement of cabinet-level AI servers such as GB/VR by North American CSPs, the ramp-up of self-developed ASICs by Google and AWS, and the expansion of domestic AI computing power deployment by Chinese cloud vendors. TrendForce expects combined capital expenditure by the nine major global CSPs to grow by approximately 90% year-on-year in 2026.
Key points to watch
China Securities believes that the strong revenue growth at Hon Hai and the upward revision of industry shipment forecasts by TrendForce corroborate each other, indicating that the current global AI computing power industry remains in a strong cycle. The firm advises sustained focus on potential rebound opportunities among oversold stocks in the AI computing power supply chain. Reuters, citing four sources, reported that the U.S. Federal Communications Commission (FCC) is drafting new rules that would ban the import of new Chinese optical transceiver modules used in AI data centers. China Securities notes that this event is still at the media reporting stage, and it remains unclear whether it will ultimately be implemented and how. Therefore, the actual impact cannot be accurately assessed. However, referencing similar measures taken by the U.S. on inverters, the firm suggests that the market should monitor this closely.
Risk factors
Changes in the international environment could affect the security and stability of the supply chain and impact the overseas expansion progress of related companies. Tariff impacts could exceed expectations. The development of the artificial intelligence industry may fall short of expectations, affecting the demand for companies in the cloud computing supply chain. Intensified market competition could lead to a rapid decline in gross margins. Exchange rate fluctuations could affect the exchange gains and losses and gross margins of export-oriented companies, including those in the ICT equipment and optical module/optical device sectors. The development of the digital economy and digital China construction may not meet expectations. The cloud computing business development of telecom operators may not meet expectations. Operator capital expenditure may fall short of expectations. Cloud vendor capital expenditure may fall short of expectations. Demand in the communication module and intelligent controller industries may not meet expectations.