Why Did Chip Stocks Drop First When AI Giants Called for a Slower Pace?
—a Market Mood Rollercoaster Triggered by an “Oral Initiative”
Hello everyone, I’m your financial journalist.
On September 14th, if you were watching the stock market, it might have felt like the sky was falling: storage chip giants from South Korea and Japan (Samsung, SK Hynix, Kioxia) tumbled, while leading companies in China’s optical modules and storage chips (such as Zhongji Xuchuang, NeoPhotonics) also experienced significant declines. Hong Kong’s large-scale AI companies suffered even more severe losses. In just one day, the market value of these companies evaporated by over 130 billion yuan.
Many individual investors might have panicked: “Is the AI industry doomed? Will the chip stocks I bought become worthless?”
Don’t worry. As a journalist who has observed technology cycles for years, I want to tell you that this sharp drop was more of an emotional outburst than a fundamental reversal. Let’s break down what happened and understand the logic behind it.
---
The Trigger: Three Industry Leaders Saying the Same Thing Rarely
The direct cause of the market panic wasn’t poor company performance or canceled orders, but rather three top AI leaders suddenly reaching a rare consensus over the weekend: we need to slow down.
1. Dario Amodei, CEO of Anthropic: He published a 150-page article and report stating that current AI models are evolving too quickly, especially their self-improving capabilities, which pose uncontrollable risks. He proposed a three-step plan to slow down and called for a safety assessment in the industry.
2. Sam Altman, CEO of OpenAI (ChatGPT): He explicitly supported this, saying OpenAI would not go public (IPO) before 2026. This signals that they are not in a rush to raise funds for expansion but want to stabilize first.
3. Elon Musk, CEO of SpaceX: He liked the post on social media, saying, “Dario is right.”
Why did this cause such a stir?
Over the past two years, the AI industry’s mantra has been “speed equals success.” Companies invested heavily in chips and data centers, believing that the one that developed the strongest model would win. Now, the most aggressive companies are saying they need to slow down for safety reasons. The market immediately feared the worst: if AI companies stop expanding aggressively, who will buy all those expensive chips, optical modules, and servers?
Fearing the worst, investors sold their assets, leading to the global AI hardware stock market crash on September 14th.
---
A Common Misunderstanding: Confusing a “Safety Initiative” with Budget Cuts
This is the most easily misunderstood aspect of the event and the root of many investors’ panic.
The current situation is this: There are only oral initiatives; no actual actions have been taken.
- No order cancellations: None of the major cloud providers (Microsoft, Amazon, Google) or AI companies have announced cuts to capital spending (Capex), nor have they postponed data center construction plans.
- No demand decline: Chip manufacturers still have full orders, with no signs of returns or cancellations due to the supposed slowdown in AI.
To put it simply:
It’s like drivers on a racetrack suddenly saying, “Be careful in the upcoming turn; we need to slow down to avoid accidents.” Tire and engine sellers, thinking the drivers are quitting, quickly sold their stocks. But the drivers just meant they wanted to drive more carefully, not stop. As long as the race continues, the demand for tires and engines remains; it’s just that the pace won’t be as frenzied.
So, the drop was more of an emotional reaction, a concentrated release of fears about a peak in AI demand, rather than a fundamental re-evaluation based on deteriorating supply-chain orders.
---
The Chinese Perspective: Our “Computing Power Anxiety” Hasn’t Gone Away
Will the overseas giants’ call for a slowdown also slow down China’s AI industry? From the domestic perspective, the answer is no; in fact, China’s demand is even more robust. Here are three key reasons:
1. Domestic substitution is a necessity, unaffected by overseas trends: Amidst Sino-US technological competition, developing independent computing power (domestic GPUs, CPUs, storage, interconnect chips) is a strategic task. Even if overseas model development slows, Chinese manufacturers’ demand for AI servers, advanced storage, and high-speed interconnects remains strong. This is a matter of survival, not business choice.
2. Shift from training to inference: While the focus is on advanced model self-improvement, the actual application of AI (inference) is accelerating. The demand for servers to handle daily user queries, AI agents, and enterprise deployments is growing. With domestic large models like DeepSeek and decreasing API prices, developer ecosystems are expanding, leading to explosive demand for computing power, storage, and cooling solutions.
3. Technological evolution continues: At the Shenzhen Optoelectronics Expo, there was significant interest in new technologies like NPO (near-packaged optics). The architecture of intelligent computing centers is evolving, and the demand for high-speed interconnects is driving orders in optical modules, chips, and switches. These trends are not reversed by overseas calls for a slowdown.
---
Why Such a Sharp Drop? Because of “Overvalued Stocks” and Market Sensitivity
Since the fundamentals are sound, why did stock prices fall so sharply? There are two market vulnerabilities:
1. Excessive gains earlier on: The optical module sector (Zhongji Xuchuang, NeoPhotonics) has seen huge gains in the past two years, accumulating many profitable positions. The storage chip sector also experienced a price surge. Any market disturbance (even an oral slowdown) made these high-priced stocks highly sensitive. Investors wondered, “If AI is slowing down, should I sell?” This led to extreme price fluctuations.
2. The cyclical nature of storage chips: Storage chips (DRAM, NAND) are highly cyclical; prices rise and fall. After a price increase, there were concerns about a peak. The combined impact of these factors put pressure on stock prices.
In short, the market didn’t think AI was becoming useless; rather, it felt that current prices had already anticipated future growth. Any slight disturbance prompted investors to sell, waiting for clearer signals.
---
What to Look Forward to Next?
For investors, the most important thing is the real financial performance of companies, not just their statements.
1. Third-quarter earnings reports: In the coming months, major hardware manufacturers (NVIDIA, Micron overseas, and Zhongji Xuchuang, ChangXin Technology in China) will release their third-quarter reports.
- What to look for: Revenue growth, stable gross margins, and healthy cash flows.
- If reports show: Despite the slowdown claims, full orders and growing profits, the current drop might be an opportunity to buy. If orders decline and inventory builds up, it indicates a fundamental shift, and caution is needed.
- Focus on inference computing power: Pay attention to the demand for inference chips, edge computing, and AI servers. As AI applications move from chat to practical tasks, the demand for inference power is likely to be more sustainable and stable.
- Progress in localization: Watch the capital spending plans of domestic cloud providers (Alibaba, Tencent, ByteDance, Baidu) and telecom operators. If they continue to invest heavily in AI infrastructure, the long-term logic of the domestic AI hardware chain remains valid.
---
In Summary
The September 14th drop in AI hardware stocks was an emotional sell-off triggered by a safety initiative.
- Short-term: Market sentiment needs to settle, and high-priced sectors will be volatile. Avoid reckless buying or panic selling.
- Medium-term: Focus on third-quarter earnings to verify if inference demand is indeed increasing.
- Long-term: China’s AI industry’s autonomy and the growth of inference power are unproven. AI is not dying; it’s just adjusting.
Advice for investors:
If you hold related stocks and your position is not heavy, wait for earnings reports to confirm the situation. If you’re short, don’t rush into the market during panic. Wait for market sentiment to stabilize and fundamental data (orders, revenue) to confirm growth trends before making decisions. Remember, in the capital market, emotions are short-term noise; performance is the long-term signal.