Friday's close capped a week that was anything but calm. External markets were turbulent, and A-shares moved sluggishly. But the quieter the market, the more one must watch for undercurrents beneath the surface. This week, in four sectors, capital is voting with real money. Today, let's discuss what actually happened in these sectors and what logic lies behind them. First, the conclusion: solid-state battery policy implementation, a full-scale AI application breakout, confirmation of the PCB price increase cycle, and continuous central bank gold purchases — these four sectors are forming the strongest cross-market resonance this week.

Solid-state batteries: the state has given a timetable, but which companies actually have real products?

The most aggressive policy this week was the joint issuance of the "15th Five-Year Plan for the Development of the New Battery Industry" by the Ministry of Industry and Information Technology and seven other departments. This is the first national-level special plan in the battery sector. The plan is clear: by 2030, all-solid-state batteries will initially achieve large-scale application. Long-life lithium batteries must reach a cycle life of 15,000 times, and leading companies' product defect rates must reach PPB level. Once the news broke, the A-share battery sector exploded. Shidai Wanheng hit five consecutive limit-ups, with sealed orders near 250,000 lots, surging over 60% in the past five trading days. Zizhu Gaoke hit four consecutive limit-ups, Lingpai Technology hit a 20cm limit-up for two consecutive boards, and Liwang Shares rose 61.3% in a week. But there is an awkward problem. Shidai Wanheng itself announced: the company's products do not involve solid-state batteries, and its main production lines and fundamentals have not changed. Zizhu Gaoke also said that its solid-state battery-related business is still in the early exploration and R&D stage. In plain terms, this is concept hype. So what are the companies actually doing solid-state batteries doing? BYD said it will launch its first mass-production model equipped with solid-state battery technology next year. Chery said it plans to conduct on-vehicle validation of all-solid-state batteries in 2027. CATL said all-solid-state batteries are expected to achieve small-batch production in 2027. Overseas players are also catching up. Panasonic said it will start sampling all-solid-state batteries in the fourth quarter of this year, and Samsung SDI said its 2027 second-half mass production target remains unchanged. So the question is: is 2027 truly an inflection point, or another "mass production next year" pie in the sky? Industry insiders judge that 2027 will be a key inflection point for demonstration vehicle installations, while true large-scale industrialization will wait until around 2030. In between, there are still several hurdles to overcome: low yield rates, high costs, and insufficient solid-state interface stability. Also note that few companies in the sector are truly doing solid-state work; most are riding the hype. Before chasing highs, first figure out whether the company you are buying actually has real substance.

AI applications: from "selling computing power" to "selling movies" — can this story hold up?

This week, the AI application side became extremely hot. On the A-share side, AI film and television concepts saw a wave of limit-ups. Chinese Online and Mango Super Media rose 20% to limit-up, while Jishi Media, Bona Film Group, and a host of other stocks followed. The catalyst is very specific: Bona Film Group announced that China's first AI hyper-realistic theatrical film, "Sanxingdui: Future Past," is scheduled for release on October 23. The film runs 100 minutes, and all characters are original digital images, without digital replication of real actors. Previously, the fully AI-generated prime-time long drama "Post-Journey to the West" aired on Hunan TV's golden time slot, with a peak ratings of 0.3384% on its premiere night, taking first place in its time slot among provincial satellite TVs. Now look at the Hong Kong stock side. Zhipu surged over 8%, with total market value breaking HK$350 billion. The reason: Amazon Web Services' Amazon Bedrock officially integrated Zhipu's GLM-5.3 large model, with the two parties sharing revenue based on call volume. Goldman Sachs directly upgraded its rating from "neutral" to "buy," with a target price of HK$1,560. Zhipu now supports over 40 mainstream chips, achieving large-scale inference on 100,000-level domestic chips, with unit token inference cost down 80% from the start of the year. What is interesting here? Over the past two years, large model companies have been searching for ways to make money. Sell computing power? Price wars have been bloody. Do ToB customization? Profit margins are getting thinner. Zhipu chose a light path: put the model on the world's largest cloud service platform, let enterprises call it via API, and share revenue by usage. Essentially, this is a replica of the SaaS model in the large model field — no need to build a huge sales team, relying on AWS's global customer network and billing system to achieve scaled distribution. But the problem is also sharp: with revenue sharing by call volume, how large can call volume actually be? The specific revenue-sharing ratio is not public, and it is hard for outsiders to know. But at least the market sees a revenue mechanism that is "quantifiable, trackable, and predictable." That is better than a pie in the sky. The AI narrative is shifting from "selling computing power" to "selling content and applications." A-share AI film and television is the imagination on the content side, while Hong Kong-listed Zhipu is the implementation on the model side. Together, they form the complete story of "AI application monetization."

PCB price increases: an unremarkable industry is undergoing a super cycle

PCB, printed circuit boards. Sounds boring, right? But this week, in both A-shares and Hong Kong stocks, this sector rose. Hong Kong-listed Kingboard Laminates rose nearly 12% in a single day, and A+H dual-listed Shengyi Technology entered the top ten most-traded stocks on the Shanghai Stock Connect. The logic is actually not complicated: the surge in AI computing demand is driving PCB prices and volumes higher together. Kingboard Laminates has issued multiple rounds of price increase notices this year, and based on compound calculation by major thicknesses, the cumulative increase has exceeded 100% year-to-date. Since the first increase in March, after each round of price increase notices from this world's largest copper-clad laminate manufacturer, domestic manufacturers such as Shengyi Technology and Jin'an Guoji have mostly chosen to follow suit. A research report from Bank of America Securities gave an intuitive figure: assuming a 100% CCL price increase, Kingboard Laminates' earnings per share could rise by more than 200% compared with the 2026 level. Even with only a 20% increase, it would bring a 67% earnings boost. Moreover, the shortage-driven price increase trend may continue. Many AI-PCB companies have strong orders, full production and sales, with lead times extended to 2 to 3 months, and some companies even implementing purchase restrictions. Overseas copper-clad laminate manufacturers are slow to expand capacity, and the supply-demand gap is difficult to fill in the short term. More noteworthy is the action of industrial capital. Optical module leader Cambridge Technology announced this week that it plans to place new H shares and issue convertible bonds, raising a total net proceeds of over HK$6.5 billion. About 55% will be used to expand optical module capacity, and about 30% will be used for strategic investment in upstream companies. Cambridge Technology said a very honest thing in its announcement: customer demand is expected to explode in 2027 and beyond, and the growth rate of orders in hand has already exceeded the growth rate of capacity, forcing it to accelerate expansion. Isn't this a "supply-side signal"? A-shares provide the expected pricing for price increase transmission, H-shares provide real-world verification of leading companies' earnings delivery, and Cambridge Technology's H-share financing represents a real bet by industrial capital on capacity expansion. When these three signals overlap, the credibility of the PCB price increase cycle is much higher.

Gold: central bank purchases for 23 consecutive months — don't ignore this signal

Gold was also a line of cross-market resonance this week. Data released on October 7 showed that China's gold reserves at the end of September were 77.47 million ounces, an increase of 740,000 ounces month-on-month, or about 23.02 tons. The central bank has increased its gold holdings for the 23rd consecutive month. Moreover, the single-month increase in September reached the highest level since the start of this gold purchase cycle. Hong Kong-listed gold stocks rose collectively. Zijin Gold International rose 5.29%, Chifeng Gold rose 5.08%, and Zijin Mining rose over 4%. On the A-share side, Zijin Mining entered the top ten most-traded stocks on the Shanghai Stock Connect, and the precious metals sector also ranked high in industry gains. But here is a question worth thinking about: the central bank is buying, but gold prices are falling. In September, international gold prices fell 8.5% cumulatively, the U.S. 10-year Treasury yield rose about 53 basis points to 5.3%, and the dollar index rose about 2%. With rates and the dollar strengthening simultaneously, gold prices came under pressure. What does this show? Central bank gold purchases are long-term allocation behavior, not the same thing as short-term gold price movements. Official reserve management places more emphasis on safety and diversification, not primarily on investment returns over a few weeks or months. So even if the central bank is buying, short-term selling in financial markets may still dominate prices. So how should ordinary investors view this? The fact that the central bank has increased holdings for 23 consecutive months is itself more informative than short-term gold price fluctuations. It conveys the official long-term judgment on reserve asset diversification. It is not common in history for the central bank to increase gold holdings uninterrupted for 23 consecutive months. It hints not just at gold price movements, but at deeper changes taking place in the global monetary system.

What to watch next week?

Having covered the four major sectors this week, let's talk about what to pay attention to next week. The third-quarter earnings disclosure window is opening. The pricing focus will shift from policy expectations back to earnings delivery. Stocks that rose in recent weeks on policy catalysts may face tests if they lack earnings support. Specifically for several sectors: for solid-state batteries, the policy benefit has already landed, but companies with real industrialization progress and those purely riding the concept will accelerate their divergence. The key is who can show actual progress on pilot lines in their third-quarter reports. For AI applications, Zhipu's commercialization model still needs more data verification. Whether AI film and television can move from "concept" to "box office" — the performance after "Sanxingdui" is released on October 23 is an important observation point. For the PCB sector, the certainty of the price increase cycle is relatively strong, but the stock price has already risen. Next, watch whether fourth-quarter orders can continue to exceed expectations. For the gold sector, watch for changes in the Federal Reserve's policy path. If a turning point appears in U.S. Treasury yields, gold prices may have room for recovery. One final reminder: among the stocks that rose the most this week, some companies have already issued announcements saying they are "not involved in related businesses." Chasing hot topics is fine, but don't mistake concept riders for true leaders. What truly survives cycles is always companies with products, orders, and cash flow.