资管舵

Jingji Xuchuang (300308.SZ) announced its 2026 semi-annual dividend: based on the total share capital of 1.11 billion shares after excluding repurchases, it will pay 12 yuan per 10 shares (inclusive of tax), totaling 1.331 billion yuan. The record date is October 9, and the ex-dividend/ex-rights date is October 12.
Third, spell it out clearly.
First, this is the payment time point. This is the semi-annual cash dividend, not the annual report. Earning more than 1.3 billion yuan in cash in just half a year shows that the cash flow can hold up, and the first half looked solid. For a heavy-asset hardware company to be able to distribute this amount, that in itself is a performance report.
Second, don’t mix up the two dates. You must still hold the shares at the close on October 9 to receive the dividend; on October 12 the shares will go ex-dividend and ex-rights, and the stock price will adjust downward accordingly. The “amount you receive on paper” and the “stock price drop” basically offset each other, so don’t treat the pre-ex-dividend price as free money you get for nothing.

In the U.S. stock market’s pre-market trading, the storage sector is collectively weakening. As of the time of writing, Micron Technology is down 2.40%, SK Hynix is down 4.12%, SanDisk is down 3.68%, Western Digital is down 2.80%, and Seagate Technology is down 2.44%.
These five aren’t randomly grouped—Micron and SK Hynix make memory chips, while SanDisk, Western Digital, and Seagate make hard drives and flash memory modules. They’re all on the storage chain. When the whole sector moves down together, it suggests funds are withdrawing from the “storage” line, not that any single company has gone off track.
Pre-market volatility does not equal a guaranteed outcome at the close. Pre-market liquidity is thin, and just a few orders can push the price down. After the market opens, these moves are often erased. This data is only a snapshot at a certain moment—use it to gauge direction, but don’t treat it as a conclusion for the whole day.

On the day trade-favorable news landed, A-shares actually fell in the opposite direction: the ChiNext index dropped more than 4% within half a day, more than 4,800 stocks were in the green, and high-flying popular names saw widespread limit-downs.
The hardest hit were the sectors that had risen the most earlier and were most tightly held by institutions—CPO, 5.5G, PCB, electronic chemicals, and precious metals all led the decline together, a typical pattern of “many-to-one” selling (panic liquidation).
Shares in Asia-Pacific Technology first fell, with the Korean market triggering circuit breakers multiple times within the year and heavy sell-offs hitting leading weights. Risk-averse sentiment followed foreign capital pressure and flowed into A-share tech stocks. U.S. Treasury yields remained at historic highs, the interest-rate differential between China and the U.S. stayed inverted, and the RMB faced pressure—foreign capital outflows remained a lurking concern. Ahead of the holiday, investors cashed out and saw frequent selling, and supply-side pressure became clearly evident. Hot sectors such as optical modules also carry expectations of overseas trade restrictions.

The AI world has produced yet another self-made female billionaire. At age 39, Daniela Amodei, with a net worth of $15.5 billion, ranks on Forbes’ US 400 list as one of the youngest self-made women billionaires. She is a co-founder and the President of Anthropic, and the sister of CEO Dario Amodei. A liberal arts graduate with a background in English literature, she now runs an AI giant valued at $965 billion.
Anthropic plans an IPO in November with an expected valuation of $2 trillion. If it happens, it will surpass SpaceX’s June record and become the largest IPO in history. The revenue curve looks more like science fiction: in January 2025, annualized revenue was only $1 billion, but by July 2026 it had already exceeded $65 billion—up 47x in under 18 months. In the private secondary market, valuations have been pushed to $1.5 trillion, with "almost nobody willing to sell."

Treasury yields have surged to near the highest levels in nearly two decades, yet the U.S. stock market is still holding steady—this divergence makes many people uneasy.
In history, whenever yields have spiked, the U.S. stock market outcomes have varied: 2022 was the worst—the S&P briefly dipped into a technical bear market, with global stock market value losing about $18 trillion over a year. In contrast, 2016, 1999, and 1994 all rose—bonds and stocks moving higher together, or stocks falling first then recovering. 2006, meanwhile, saw a gradual slide amid volatility. The difference isn’t the yield surge itself, but whether the economy at the time was improving or running into problems.
Right now it feels a bit like the tech frenzy of 1999—AI infrastructure is supporting profit expectations, and the stock market isn’t panicking. But in the bond market, the “MOVE panic index” surged about 29% last week, the biggest one-week jump since April 2025.

Spot gold and silver both weakened today. As of the time of writing, spot gold is down 2.22% on the day, losing the $4190/ounce level; spot silver’s decline has widened to 4%, trading at $61.7/ounce.
Gold and silver usually move in the same direction, but their timing differs: silver has more volatility—when it rises it’s more aggressive, and when it falls it can drop faster too. Today’s 4% decline is a good example. When both fall together, it suggests this isn’t just a problem with one specific commodity. It’s more like capital is withdrawing from precious metals overall—possibly profit-taking after a strong prior run, or related to short-term shifts in the U.S. dollar and real interest rates.
Looking at day-to-day gains and losses alone is of little significance. This kind of broad-based pullback is often more worth watching than the volatility of a single product. It’s not telling you to chase or run; it’s simply a reminder: even in an uptrend, there are pullbacks. Don’t load the position too heavily—leaving some room is more practical than trying to guess the top or bottom.

Meta’s AI agent Muse has taken off. It only launched in early September, and within two weeks it shot to the top of the App Store charts—so much so that even JPMorgan Chase’s evaluation ranked it first. It’s not like a chat bot that only answers passively: it can call on your behalf to negotiate with insurance companies, cancel streaming subscriptions you don’t need, and search the web for discounts—it actually gets things done.
OpenAI can’t sit still and is rushing to launch its always-on assistant “o” to take on Meta Muse. The product is expected to be unveiled at the September 29 developers’ conference. It will be driven by a GPT-6 Aeon variant, focused on handling long-running tasks, and will come with a donut-shaped piece of hardware. A new Pro Max subscription will cost $500 per month and will run on Cerebras’ wafer-scale engine. On Meta’s side, each Muse agent runs on a dedicated secure computing device, making safety a selling point too.