China Securities Co., Ltd.: Google's new model release, Jensen Huang's departure from Nvidia, continued recommendation of the domestic computing power industry chain.

date
07:33 16/11/2026
avatar
GMT Eight
Continue to be optimistic about the domestic computing power industry chain, with a focus on the realization of demand in segments such as GPU/ASIC, servers, high-speed switching and optical interconnect, liquid cooling, and power supplies.
China Securities Co., Ltd. released a research report stating that Alphabet Inc. Class C (GOOGL.US) has released Gemini 4 Hydrogen, improving programming and agent execution efficiency, which is expected to reduce AI application costs and drive growth in token consumption and inference computing power demand. Semiconductor Manufacturing International Corporation (688981.SH) Q3 revenue and net profit attributable to parent grew 28.1% and 87.6% year-on-year respectively, while Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR (TSM.US) October revenue grew 59.7% year-on-year, indicating continued prosperity in wafer manufacturing. Jensen Huang has resigned as CEO of NVIDIA Corporation (NVDA.US), with CFO Colette Kress serving as interim CEO; attention should be paid to the impact of the management transition on strategic execution and ecosystem investment. China Securities Co., Ltd. believes that model iteration and AI commercialization continue to support computing power demand, and improvements in domestic chip performance, software adaptation, and cluster capabilities are expected to drive order conversion. The firm remains bullish on the domestic computing power industry chain, with focus on demand realization in segments including GPU/ASIC, servers, high-speed switching and optical interconnect, liquid cooling, and power supplies. China Securities Co., Ltd.'s main views are as follows: Alphabet Inc. Class C released Gemini 4 Hydrogen, further improving programming and agent task execution efficiency. On November 12, Alphabet Inc. Class C launched Gemini 4 Hydrogen, targeting everyday programming, multimodal processing, and agent workflows, supporting a 1 million token context window and four levels of thinking intensity. According to Alphabet Inc. Class C's disclosure, the model scored 51 on the Artificial Analysis Intelligence Index at the default high thinking level, surpassing Gemini 4 Argon's 50, with the highest thinking level reaching a score of 54; Terminal-Bench 4.1 score improved from 18.2% for Gemini 3.8 Flash to 57.0%. API input and output prices are US$1.5 and US$7.5 per million tokens respectively, with uniform pricing across the entire context window. Alphabet Inc. Class C simultaneously launched Orchestrator in Antigravity, which can dynamically schedule Carbon and Hydrogen within the same session, respectively handling complex judgment and efficient execution. Official testing shows that this mode achieves task success rates close to Carbon on Terminal-Bench 4.0 while reducing quota consumption by an average of 24% and shortening completion time by 19%. Hydrogen has covered channels including the Gemini app, Alphabet Inc. Class C Search, developer API, and enterprise services. Semiconductor Manufacturing International Corporation's revenue and profitability improved, with domestic wafer manufacturing prosperity continuing. On November 12, Semiconductor Manufacturing International Corporation disclosed its Q3 report, achieving operating revenue of RMB21.983 billion and net profit attributable to parent of RMB2.847 billion in Q3, growing 28.1% and 87.6% year-on-year respectively, mainly benefiting from increased wafer sales volume, higher capacity utilization, and product mix changes. Under IFRS, the company's Q3 revenue was US$3.19 billion, up 6.1% quarter-on-quarter; gross margin was 29.9%, up 4.6 percentage points quarter-on-quarter. The company expects Q4 revenue to grow 4%6% quarter-on-quarter, with gross margin of 26%28%, and revenue growth is expected to continue. Overseas, Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's October revenue was NT$586.89 billion, up 59.7% year-on-year and 14.7% month-on-month; cumulative revenue for the first 10 months grew 43.3% year-on-year, with wafer foundry demand maintaining strong growth. On November 12, NVIDIA Corporation announced that Jensen Huang has resigned from the positions of President, Chief Executive Officer, and Director, effective immediately. Executive Vice President and Chief Financial Officer Colette Kress has been appointed interim CEO while continuing to serve as CFO; long-serving director Harvey Jones has been appointed as the company's first Chairman. The board has simultaneously initiated a formal CEO search, which will consider both internal and external candidates. The company stated that it expects continuity in daily operations, customer commitments, and product roadmap, with the management team continuing to advance accelerated computing and AI businesses. The firm believes that improvements in model capabilities and operational efficiency are expected to continue driving computing power demand, with attention to overseas management adjustments and realization of domestic computing power demand. Gemini 4 Hydrogen's upgrades in programming, agent execution, and multi-model collaboration are expected to reduce the execution cost of complex tasks and accelerate AI penetration into enterprise production processes. With the proliferation of agent applications, tool calling, multi-turn reasoning, and parallel tasks will expand token consumption scale, providing support for growth in inference computing power demand. Jensen Huang's departure may increase short-term uncertainty regarding NVIDIA Corporation's strategic execution, ecosystem investment, and customer cooperation; follow-up should track the formal CEO selection, product delivery, and capital allocation arrangements. Semiconductor Manufacturing International Corporation's operational improvement reflects continued prosperity in domestic wafer manufacturing; as domestic chip performance, software adaptation, and cluster collaboration capabilities improve, domestic AI application demand is expected to further convert into domestic computing power procurement and construction orders. We remain bullish on the domestic computing power industry chain, with focus on demand realization in supporting segments including domestic GPU/ASIC, servers, high-speed switching and optical interconnect, liquid cooling, and power supplies, while also watching for industry opportunities brought by overseas computing power investment and technology iteration.