From October 20 to 22, by the banks of the Huangpu River in Shanghai, the 2026 Sustainable Global Leaders Conference will return once again. Nobel Prize and Turing Award winners will take the stage to discuss the future and trends of technology, and there will even be a live dialogue between the internet-famous professors who "rescue" people from anxiety and the era of intense competition — Macfarlane and Xiang Biao!
Limited seats, first come first served. Let us go together to the appointment with sustainability! [Click to register] Recently, the A-share computing power supply chain has been correcting in a way that makes people somewhat uncomfortable. But interestingly, on the other side, the third-quarter earnings season for US technology giants is beginning. From Meta and Google in late October, to Apple and AMD in November, and then to Nvidia at the end of November, leading global AI computing power companies will gradually release their report cards. On one side, the A-share computing power sector is fluctuating; on the other side, the overseas giants' earnings season is about to verify the industry's prosperity. The contrast in between is worth pondering: is the continued volatility in A-share computing power due to a change in the industry logic, or does it offer a window worth watching?
Breaking down the reasons for recent volatility
The recent volatility in the A-share computing power supply chain is mainly due to a combination of several factors: The first is pressure at the trading level. In the second quarter, the computing power sector rose relatively strongly, and some funds had a need to take profits, so when market sentiment weakened, concentrated selling was prone to occur. The second is policy-level disruption. Some policy rumors overseas regarding areas such as optical communications triggered concerns about the space for the supply chain to expand overseas. Although many rumors are still at an preliminary stage, the impact on sentiment exists. The third is switching at the capital level. Some market institutions are adjusting their position structures, with some funds shifting from high-beta technology tracks to defensive sectors. As a previously popular direction, computing power bore relatively large selling pressure. But these factors are more at the short-term trading and sentiment level, and the long-term logic of industry fundamentals has not undergone a fundamental change. November is an important observation window and may also become a turning point for changes in market expectations: In November, third-quarter reports will be disclosed intensively, and the market valuation baseline may shift toward 2027; after the results of the US midterm elections become clear in November, the market will realize that the Democratic Party will still advance AI development; at the same time, as Middle East conflicts ease after the midterm elections, oil prices may also fall. In addition, if leading overseas large-model manufacturers advance listings before the end of the year, related industry launch events may also have a certain positive impact on market sentiment. The above possibilities have not yet been falsified.
US earnings season is a window to verify prosperity
According to the expected disclosure times for the 2026 third-quarter reports of key US companies, the next month or so can be described as a concentrated verification period for the global AI computing power supply chain: In late October, cloud providers such as Meta, Google, Microsoft, and Amazon will disclose earnings one after another, with focus on whether data center capital expenditures and AI-related investment continue to increase. These companies are the main source of global AI computing power demand, and their capital expenditure guidance basically determines the order visibility for upstream chips, optical modules, and servers. In early November, hardware companies such as Apple, AMD, and Qualcomm will disclose earnings one after another, and progress in AI terminals and chips is worth watching. AMD's data center business growth rate and Qualcomm's AI phone chip penetration are both important indicators for observing the prosperity of the supply chain. On November 25, Nvidia will disclose earnings. As the core leader of the global AI computing power supply chain, its performance guidance and order visibility may have a relatively large impact on short-term sentiment across the entire sector. The market is more concerned about whether next quarter's revenue guidance will continue to exceed expectations, and whether supply chain delivery conditions have improved. In December, companies such as Broadcom will also disclose earnings one after another, continuously verifying the prosperity of the supply chain. The earnings performance of these overseas leaders is a very important fundamental reference for the A-share computing power supply chain. If overseas giants continue to raise capital expenditure and AI investment expectations, it shows that the resilience of global AI computing power demand is still there, and this is expected to form a positive catalyst for related A-share supply chains. On the contrary, if earnings or guidance fall short of expectations, short-term sentiment may still fluctuate, but the industry trend will not change because of single-quarter fluctuations.
The allocation logic of the A-share computing power sector
The A-share computing power supply chain covers multiple links including chip design, semiconductor equipment, optical modules, servers, and IDC. After the previous adjustment, valuation percentiles in some sub-sectors have clearly declined, and allocation value is worth watching. Among them, the PCB and computing power leasing sectors may rank relatively high in quarter-on-quarter third-quarter performance growth. In terms of PCB, the intensive price increases for ordinary products in the second quarter may be reflected in third-quarter performance, and with the shipment of Nvidia's next-generation Rubin servers, the PCB value per unit has more than doubled. The computing power leasing industry benefits from domestic large-model companies reaching the RSI threshold, and demand for high-end computing power chips used for model training is accelerating. In addition, semiconductor materials and equipment is also one of the sub-sectors with relatively high certainty among high-growth industries in A-shares. From the perspective of industry trends, the global AI computing power expansion cycle is still continuing, and domestic computing power infrastructure construction is also steadily advancing. Whether for large-model training or inference applications, demand for computing power is growing rapidly. Moreover, with the maturity of anti-distillation technology and the rapid development of RSI, open-source model companies' demand for high-end training computing power is instead increasing, which is a change the market had not fully anticipated before. From the perspective of valuation percentiles, after this round of adjustment, the valuation percentiles of many companies in the computing power supply chain have fallen back, and allocation cost-effectiveness has improved.
Several signals to track going forward
First, the performance and guidance of US technology giants' third-quarter reports, especially the growth rates of data center capital expenditures and AI-related revenue; second, the pace of implementation of domestic computing power infrastructure and the realization of orders; third, progress in the localization of semiconductor equipment and materials; fourth, changes in sector capital flows and valuation percentiles. From an operational perspective, investors with positions can pay more attention to changes in industry fundamentals and avoid frequent trading due to short-term fluctuations; investors on the sidelines can, under the premise of matching their own risk tolerance, consider building positions in batches and diversify the risk of a single track through index products covering multiple links of the computing power supply chain. Products to watch include GF Venture Board Computing Power ETF (158049), GF STAR Chip Design ETF (589210) and its feeder funds A/C: 027520/027521, GF Semiconductor Equipment ETF (560780) and its feeder funds A/C: 020639/020640, etc., covering different links of the computing power supply chain. The sales fees for GF Venture Board Computing Power ETF are as follows: When investors subscribe for fund shares, the subscription agency broker may charge a commission at a standard of no more than 0.30%. When investors redeem fund shares, the redemption agency broker may charge a commission at a standard of no more than 0.50%. On-exchange trading fees are subject to what the securities company actually charges. For details, please refer to the fund prospectus, fund contract, and other legal documents. The sales fees for GF STAR Chip Design ETF are as follows: When investors subscribe or redeem fund shares, the subscription and redemption agency broker may charge a commission at a standard of no more than 0.50%. On-exchange trading fees are subject to what the securities company actually charges. The subscription fee rate for its feeder Class A shares is: 0.30% for amounts below 5 million yuan, and 1,000 yuan per transaction for 5 million yuan and above. The redemption fee rate for individual investors in Class A shares is: 1.50% for holdings of less than 7 days, and 0 for holdings of 7 days or more. The redemption fee rate for non-individual investors in Class A shares is: 1.50% for holdings of less than 7 days, 1.00% for 7 days (inclusive) to 30 days, 0.50% for 30 days (inclusive) to 180 days, and 0 for 180 days or more. Class C shares do not charge a subscription fee. The redemption fee rate for individual investors is: 1.50% for holdings of less than 7 days, and 0 for holdings of 7 days or more. The redemption fee rate for non-individual investors is: 1.50% for holdings of less than 7 days, 1.00% for 7 days (inclusive) to 30 days, 0.50% for 30 days (inclusive) to 180 days, and 0 for 180 days or more. In addition, Class C shares charge an annual sales service fee of 0.20%. For details, please refer to the fund prospectus, fund contract, and other legal documents. The sales fees for GF Semiconductor Equipment ETF are as follows: When investors subscribe or redeem fund shares, the subscription and redemption agency broker may charge a commission at a standard of no more than 0.50%. On-exchange trading fees are subject to what the securities company actually charges. The subscription fee rate for its feeder Class A shares is: 1.20% for amounts below 1 million yuan, 0.80% for 1 million yuan (inclusive) to 5 million yuan, and 1,000 yuan per transaction for 5 million yuan and above. The redemption fee rate for Class A shares is: 1.50% for holdings of less than 7 days, and 0 for holdings of 7 days or more. Class C shares do not charge a subscription fee. Their redemption fee rate is: 1.50% for holdings of less than 7 days, and 0 for holdings of 7 days or more. In addition, Class C shares charge an annual sales service fee of 0.30%. For details, please refer to the fund prospectus, fund contract, and other legal documents. Risk disclosure: The above funds invest in the securities market. Before investing in the above funds, investors need to fully understand the product characteristics of the above funds and bear various risks arising from fund investment. The above funds are issued and managed by GF Fund Management Co., Ltd., and the distribution agency does not assume responsibility for product investment and redemption. Before investing, please carefully read the fund contracts, prospectuses, and other legal documents of the above funds, and fully understand the details and risk characteristics of the above funds. The above are equity funds, and their risks and returns are higher than those of mixed funds, bond funds, and money market funds; the specific risk rating results are subject to the rating results provided by the fund manager and sales institutions. Investors should choose products that match their risk tolerance and investment objectives. Funds carry risks, and investment must be cautious.
Comments