Recently, the global capital market has undergone a dramatic "major reshuffle." Chip giants represented by Samsung Electronics and Micron Technology have plunged more than 30% from their peaks in just one month.

For a long time, "hard technology" represented by chips dominated the AI wave. However, since July this year, capital seems to have started "abandoning hardware and embracing software," flocking to "soft technology" represented by internet platforms, staging an extreme "soft-hard rotation" seesaw market.

What is the core logic behind this adjustment?

Industry experts generally believe this is not the end of the AI rally, nor is there a permanent one-way style shift. There are two main reasons behind this:

Macro and trading levels: The Fed has sent hawkish signals leading to a rebalancing of funds; At the same time, the previous sharp gains in the hard tech sector and extremely crowded trading meant that once expectations slowed, leveraged funds would flee en masse, triggering a valuation crash.

Industry level: AI trading has moved from blindly buying "shovel sellers" (hardware manufacturers) to entering the "validation phase."

The market today is no longer satisfied with grand AI narratives but requires real financial verification: who can turn AI investment into real orders and profits? Those pure concept stocks without earnings support were abandoned by capital, while the leaders who could realize profits remained strong and even rose against the trend.

|The fiercer the rise, the more fiercely the drop becomes

In July, the once highly successful hard tech stocks experienced the most intense correction of the year.

As representatives of hard technology, chip stocks bear the brunt. Domestically, the STAR Market chip index has seen a deep pullback, with more than half of the 50 constituent stocks pulling back more than 40% from their June high. Stocks like China Shipbuilding Special Gas and Baiwei Storage even halved at one point (down over 50%).

Overseas markets are equally tough. Samsung Electronics and Micron Technology plunged over 30% from their peaks within a month, while SK Hynix and SanDisk saw drawdowns of over 40%. For these trillion-dollar giants, a short-term crash means the instant evaporation of hundreds of billions of dollars in market value.

Why did it suddenly drop so hard?
This sharp adjustment is mainly caused by a collision between macro and industry factors

Macro level: The Fed has signaled a hawkish stance, and market expectations for future policies have shifted, leading to capital rebalancing assets.

Industry level: Since the second quarter of this year, the AI hard technology sector has surged so aggressively that it has attracted a large influx of leveraged capital and led to extremely crowded trading. When the market began to doubt whether "AI capital spending can actually be profitable," capital rushed out of the market, triggering a stampede downturn.

Hard tech stocks (such as GPUs, HBM, advanced packaging, etc.) have accumulated significant gains and excessively high expectations in the early stages. Once the market shows signs of "slowing price increases" or "marginal slowing in capital expenditure," funds quickly deleverage, triggering sharp valuation plunges.

Take the STAR Market chip index as an example: before its crash, it had already gone through a frenzied upward cycle: in early July this year, it reached a peak of 5660 points, more than doubling compared to a year ago; Compared to the eve of the 2024 "9·24 rally," the price has surged more than fourfold.

As the saying goes, "The higher you climb, the harder you fall." The initial peak prosperity eventually led to a dramatic correction.

| Major Capital Shift: Aggressive Dumping on "Hard Technology," Bottom-Fishing on "Soft Technology"

Just as hard tech stocks were facing fierce sell-offs and sharp adjustments, the tech sector experienced an extreme "seesaw" rally: funds began to frantically withdraw from the hardware sector, rapidly flooding into "soft tech" stocks represented by internet platforms to "fill positions."

1. Market performance: A world of hardship

From the index performance, this divergence is very obvious. From early July to July 24 this year, the STAR 50 and the SSE STAR Market Chip Index, representing hard technology, all plunged; Meanwhile, the Hang Seng Tech Index, representing soft technology, bucked the trend and rose by 3.52%. In the US stock market, the "Seven Sisters of Technology" index, which brings together AI application giants like Google and Microsoft, also outperformed chip stocks during the same period.

2. Capital flow: selling hardware and shopping online

The flow of real money best illustrates the situation:

Hard technology faces "blood drainage": Funds raised from early long positions in hard technology are rapidly withdrawing. From July 1 to 23, the top ten stocks by net margin selling in A-share markets were all semiconductors and hardware equipment. Among them, stocks like Cambricon and Demingli saw net sales exceeding 5 billion yuan.

Soft technology is being "snatched up": southbound funds are frantically "adding positions" to Hong Kong-listed internet giants. In the first 24 days of July, southbound capital net bought Hong Kong stocks by over HKD 80 billion, three times the scale of June. Internet companies like Alibaba, Tencent Holdings, and Kuaishou became the main force for capital raising, while Hong Kong, which has hard technology attributes like Huahong Hongli, was sold off.

3. Why does this switching happen?

Industry experts point out that the shift of funds from hardware to applications is essentially the market seeking a "safe haven." On one hand, the hard technology sector had previously surged too much, with extremely crowded trading and high valuations; On the other hand, internet and other soft tech stocks have experienced long-term adjustments, with valuations at low levels and stronger resilience in their underlying business models.

Under the dual effects of macro liquidity and industry cycles, funds naturally choose to withdraw from high levels and flow toward the more cost-effective "soft tech" depressions.

| The shift period of "soft-hard rotation."

Although hard technology and soft tech stocks have recently experienced a fierce "seesaw" rally, the industry generally believes the AI wave is not over. The current shift is essentially an upgrade in AI investment logic—from the first phase of "competing for infrastructure (hardware)" to the second stage of "competing for real profits (software applications)."

In the short term: Funds are seeking "cost-effectiveness" and "real profit," and this trend of "selling hard to buy soft" is likely to continue.

Hard technology is too sensitive: The valuation of the hardware sector is highly sensitive to orders, prices, and capital expenditures, and once expectations slow, valuations can easily be cut.

Soft technology is more resilient to declines: After long-term adjustments, internet platform companies currently have low valuations and offer excellent cost-effectiveness. Moreover, they are backed by real profits, cash flow, and buybacks. More importantly, as AI model costs decrease, the application layer actually benefits more easily.

Mid-term Outlook: Soft Technology also needs to "deliver its report card," but soft technology is not complacent. If giants like Microsoft, Amazon, and Google cannot prove that their AI investments and cloud businesses have generated real revenue growth, the market will also question them again. Therefore, this is not a simple "hard retreat and soft advance," but a market test who can turn AI into real money.

In the long run: hardware earns "cycle money," software earns "ecosystem money." Looking at the longer cycle, the AI market does not have a perpetual, one-way style rotation but rather exhibits a "rotation of software and hardware."

Hardware is the foundation: iterations in computing power and technological innovation periodically drive the hardware sector's market swings. But hardware is always just a supporting component, only able to gain temporary capital expenditure dividends.

Software is the highland of value: the long-term value allocation of the industry chain will ultimately tilt toward internet applications that have users, scenarios, and monetization barriers. Therefore, in the long run, software platforms cannot be outperformed by hardware over time.

Kingtech's Perspective | The era of profiting with your eyes closed is over

The 30% plunge in chip giants and the "rotation of hard and soft capital" mark that the AI market has moved from a rough phase of "speculating on concepts and expectations" to a deep zone of "evaluating performance and looking at profits." The market no longer pays for grand narratives but scrutinizes companies' true profitability through a magnifying glass. This means that the "indiscriminate rise" phase of AI investment has ended, and the era of refined and structured investment has arrived.

Focusing on "Real Orders" and "Domestic Substitution"

For the hard technology sector, short-term valuation cuts are the only way to squeeze out bubbles. During the adjustment, investors should abandon pure concept stocks and focus on companies that deliver impressive "performance cards" during the mid-term reporting season in August.

Especially amid the global computing power supply and demand game, leading semiconductor equipment, computing chips, and advanced packaging companies with core technology barriers, high order visibility, and deep benefits from domestic substitution logic have, after a deep correction, instead ushered in an excellent "golden pit" window for layout.

Focusing on the "AI Monetization Closed Loop"

Capital flowing into soft technology is essentially seeking a safe haven with high cost-performance ratios. Against the backdrop of declining AI model costs, internet platforms and application software companies find it easier to turn AI technology into actual cost-saving and efficiency gains.

It is recommended that investors focus on two types of soft technology targets: first, internet giants with massive user bases and rich application scenarios, capable of pioneering the "AI + commercialization" closed loop; Second, high-quality application-end companies with strong cash flow, continuous buyback capability, and historically low valuations.