To The Moon

During the National Day holiday, the AI theme in overseas markets did not pause because of high interest rates or elevated oil prices. On October 6, the Nasdaq hit a fresh stage high, and on October 7 it pulled back as Treasury yields and oil prices spiked, yet the core pricing logic of the AI industry remained unchanged.

But on the A-share side, the first trading day after the holiday painted a completely different picture. The ChiNext Index rose intraday before reversing lower and closing down 2.13%, while the STAR 50 fell 3.66%, with CPO, optical chips, semiconductors and other technology sectors becoming the main sources of pressure.

Some observers summed up how A-shares broke several patterns today: first, the rule that a pre-holiday decline must be followed by a post-holiday rally; second, the pattern of Hong Kong stocks surging while A-shares are on a long holiday; third, the pattern of a big post-holiday gap-up followed by a pullback alongside A-shares; and fourth, the pattern that sectors which rallied strongly in US stocks during the holiday would also rally in A-shares afterward.

In the third quarter, the ChiNext Index and the STAR 50 had already fallen 27.8% and 30.7% respectively, with electronics and communications among the worst-performing industries. Today's sharp decline was directly triggered by the FCC.

On October 7, the US Federal Communications Commission announced plans to vote on October 29 on new certification rules that would bar Chinese laboratories from testing and certifying electronic products such as phones, cameras and computers destined for the US market, with the rules expected to take effect in December 2028. Since roughly 82% of electronic product testing is currently completed in China, this will raise future compliance and relocation costs for the relevant supply chains, but it does not mean China's electronics industry will lose access to the US market tomorrow.

What truly deserves renewed market expectations and valuation is where the policy direction is heading. Another thread circulating in the market is that the US may impose new supply chain requirements on next-generation 3.2T optical modules. Morgan Stanley's early-October assessment was that potential restrictions are more likely to fall on the 3.2T generation rather than the 800G and 1.6T modules already shipping in large volumes; one possible exemption path is that China-made modules could still enter the US market if roughly 65% of their BOM value comes from US companies. This proposal remains a policy expectation rather than a formally implemented rule.

The significance of this matter is not whether a particular optical module company can still ship, but that the global AI supply chain is gradually moving from "cost and efficiency competition" toward "industrial control competition." So today's CPO decline is not entirely a trade on an FCC headline, but a trade on a deeper question: in the next round of AI industry capital expenditure, who has stronger financing capacity, a more complete ecosystem, faster engineering speed and stronger policy coordination. In other words, the AI supply chain is moving from pure global division of labor toward a supply chain reconstruction shaped by geopolitical constraints. And this round of adjustment in A-share technology assets is happening precisely at this stage.

Over the past few months, the market has been debating whether US AI is a bubble and whether AI valuations should be rebalanced. AI companies also face real pressures such as overly rapid capital expenditure, rising financing costs and business models that still need validation. The recent pre-IPO valuation cut by Australian data center company Firmus is one reminder. But the other half of the facts must also be seen: AI industry demand has not disappeared because of these controversies.

Samsung's preliminary third-quarter results released on October 8 showed operating profit expected to reach 107.4 trillion Korean won, nearly a ninefold year-on-year increase, driven mainly by AI-related memory demand and rising prices. TSMC's third-quarter revenue reached 1.49 trillion New Taiwan dollars, up 50% year on year, also mainly driven by AI-related demand. Micron's latest quarterly guidance reached 61.5 billion US dollars, with long-term customer committed orders rising to 32 billion US dollars. These figures do not necessarily mean AI will grow at high rates forever, but they at least show the industry's prosperity has not reached a point that requires wholesale rejection.

What truly widens the gap between China and the US may not be a particular generation of GPUs or a particular model, but the ability to form capital. According to Stanford's "2026 AI Index," US private AI investment reached 285.9 billion US dollars in 2025, versus 12.4 billion US dollars in China, more than 23 times higher. This figure cannot be simply interpreted as total China-US AI investment differing by 23 times, because much of China's AI investment comes from government-guided funds and industrial capital and is not fully included in private investment statistics, but the difference in the concentration of capital markets, venture capital and industrial capital does deserve serious attention.

Looking at overseas markets, as of September 30, the market capitalization of the MSCI Emerging Markets ex-China Index had reached 9.84 trillion US dollars, with its top three weights being TSMC, Samsung and SK Hynix. This does not mean global funds are abandoning China, but it does mean overseas institutions can fully bypass China and build a separate emerging market allocation framework centered on technology assets in Korea, Taiwan and elsewhere. That is also why an interesting price signal appeared during the holiday.

So what truly should be reviewed today is not why A-shares did not rise along with overseas markets. Rather, it is why global funds are willing to keep pricing "AI investment" while the risk discount on A-share technology assets continues to widen. The core difference is becoming increasingly clear. The US is placing AI together with chips, data centers, energy, defense and scientific research into the same industrial investment framework. AI companies can raise funds, cloud providers continue to increase capital expenditure, and financial institutions are beginning to directly participate in AI infrastructure financing. China is also increasing investment in computing power, semiconductors, robotics and AI applications, and its supply chain is advancing rapidly. What truly needs to be caught up on is capital investment density, self-sufficiency in key links, and the speed from technological breakthroughs to large-scale commercial application.

On the contrary, this determines where the most valuable industrial opportunities will be in the coming years. In the past we often said China's manufacturing advantage comes from scale, the engineer dividend, a complete supply chain and infrastructure capacity. These advantages will not suddenly disappear. But the most cautionary aspect of technological competition is that rivals do not necessarily compete along your strengths. In the 5G era, China pushed base station and network coverage to the extreme, while the US did not simply copy that but developed satellite internet such as Starlink. China has clear advantages in power grids and new energy infrastructure, while the US is advancing nuclear power and small modular reactors to find new energy supply methods for AI. China has advantages in rare earths and some materials, so rivals look for substitute materials and new processes. AI is the same. What it may truly change is not just the software industry, but manufacturing, robotics, scientific research, energy and even the entire production system. If AI in the future can significantly improve robot manufacturing efficiency and further promote robots participating in making robots, then the cost structure and production organization of manufacturing will change. That is what truly deserves our attention in this AI industrial cycle.

Therefore, what the capital market now needs to reprice is not "whether AI has a bubble," but "who in the AI industry can keep investing and who can turn investment into productivity." In the short term, high-valuation technology assets such as CPO, optical chips and semiconductors still need to digest valuation and positioning pressure; segments such as memory, PCB and optical communications are more worth observing through orders, capacity utilization and profit realization. The market also needs to restore turnover and institutional support before the technology rally can return from sentiment-driven to industrial-trend-driven.

Longer term, China-US technology competition has gradually moved from trade friction toward competition in industrial productivity. The first stage was about supply chains and manufacturing capability. In the next stage, the core variable remains the AI race. Our advantages must continue to be consolidated; what the US has already achieved, we need to achieve; what the US has not yet solved, we also need to prepare for in advance. The key to winning in AI lies in who can keep investing and innovating through the next round of industrial iteration and ultimately turn technological advantage into productivity advantage. The AI industry race is a contest over national foundations that cannot be lost, and a contest over national destiny that must be won, and the role of the capital market is irreplaceable.

If you are still confused about when the market will stabilize; if you are puzzled about industrial investment; if you feel powerless about how to find industrial leaders and potential companies; if you feel at a loss about capital rotation and sector volatility; then please scan the code to join us, obtain more detailed supply chain analysis and investment strategies, and follow Gelonghui Research Institute to seize opportunities in the investment wave and share in the dividends of the industrial revolution.

Risk warning: Investing involves risk, and decisions must be made cautiously. The above analysis is only a reference for industry research and market strategy and does not constitute specific investment advice. To obtain the latest updates and operational strategies on targets in a timely manner, and to capture signals, identify risks and position for opportunities as quickly as possible, please scan the QR code below for more timely and detailed professional investment strategies. Disclaimer: The copyright of this article belongs to the research team of Beijing Gelonghui Investment Advisory Co., Ltd. (Duan Yuehan: A0160625050003). This report was produced based on the principles of independence, objectivity, fairness and prudence, and all information comes from public sources and has been collected, summarized and edited lawfully, reasonably and appropriately. The stock market involves risk, and caution is required when entering. None of the investment advice in this article should be used as the basis for your investment buying and selling decisions; you must make investment decisions independently and bear the risks yourself. Any unit or individual that publishes, copies, spreads, posts online or reprints any part of this content without the company's permission will be deemed to have infringed rights, and the company will pursue legal liability in accordance with the law.