Chip stocks have "cooled off," but the AI cycle may not be dead yet

In the past few weeks, the semiconductor sector has experienced a breathtaking rapid pullback.

The DRAM ETF has retreated approximately 25% from its June 22 peak, the semiconductor ETF (SMH) lost 12% in two weeks, and the Philadelphia Semiconductor Index faced consecutive declines. Micron delivered a "blowout" earnings report—revenue of $41.4 billion, with forward guidance pointing towards $50 billion—but the stock price reversed downwards under the logic of "buy the rumor, sell the news."

And Meta, the most aggressive AI infrastructure buyer over the past two years, officially announced last week that it would rent out its spare computing power to external customers, which became the catalyst for the pullback in the semiconductor sector.

Even the most fervent "chip hoarders" have started selling their idle inventory.

This signal quickly triggered a chain reaction in the market. Subsequently, Samsung's earnings report further cemented investor concerns: the profit surge in the memory chip industry essentially means the entire sector is riding the same super-cycle wave, rather than any single company monopolizing an irreplicable moat. Many investors are already starting to worry if the AI narrative has reached its end.

1. This May Be a "Rotation," Not an "End"

Amidst the widespread bearish sentiment on chips, one of Wall Street's most influential strategists offered a completely different assessment.

Michael Wilson, Chief US Equity Strategist at Morgan Stanley, directly stated in the latest weekly report: Underweight semiconductors, rotate into hyperscale cloud computing vendors.

The weight of this statement lies in the fact that it's not saying "AI is over," but rather "the direction of profit distribution has changed."

Wilson's analytical logic is this: Over the past two years, the most profitable link in the AI value chain has been the "pickaxe sellers" – NVIDIA, Micron, SK Hynix. They provide the foundational computing and storage infrastructure for the AI revolution, enjoying pricing power and ultra-high margins driven by supply shortages. However, as cloud vendors' capital expenditure growth approaches an inflection point, the demand growth rate for these "pickaxes" is shifting from an explosive phase to a stable one.

Moving forward, the profit center of the value chain is shifting from the hands of the "pickaxe sellers" to those "mining with the pickaxes."

The cloud vendors – Microsoft, Google, Amazon, Alibaba – they are the ultimate integrators and commercial outlets for AI capabilities. They build computing power using chips, package that power into cloud services, AI assistants, and enterprise solutions, and sell them to billions of users and companies globally. As the price increase cycle for upstream chips begins to slow, the cost pressure on cloud vendors will actually ease, while their revenue from AI services continues to climb.

This marks the peak of the *rate of increase*, not the peak of the entire capital expenditure cycle. In other words, cloud vendors won't stop investing in AI, but the phase of the fastest investment growth has likely passed. Consequently, the valuation of chip stocks needs to transition from "explosive growth" pricing to "mature growth" pricing – and this transition process is often accompanied by significant capital migration and stock price volatility.

2. Memory Drops First, Potentially the Start of a Capital Rotation

Let's see how the data validates this logic.

Since Meta announced its computing power rental plan, memory chips have become the epicenter of the pullback storm – because memory demand is most directly dependent on the capital expenditure intentions of cloud vendors. When the biggest buyer starts signaling "we've bought enough," cracks naturally appear in the growth narrative of memory suppliers.

Micron's case is the most typical. The company reported Q3 revenue of $41.4 billion, with management's guidance reaching as high as $50 billion – a proverbial "trump card" in any normal environment. Yet the market's response was a falling stock price. When "beating expectations" becomes the norm, what truly determines the stock price is no longer the numbers themselves, but "can we continue to exceed expectations?" This "sell the news" pattern is, in itself, a characteristic of capital pricing in a cyclical peak.

But if the AI cycle were truly over, what should we see? We should see a complete collapse of all AI-related assets – cloud vendors, AI application companies, Chinese tech stocks – none spared.

However, the market paints a different picture: Capital flows *out* of chips, but *into* another thread of the AI narrative.

3.A：Alibaba Surges 11%: The Flare Signal of a New Rotation

One of the most convincing pieces of evidence occurred during US trading hours yesterday.

Alibaba's US-listed shares surged 11% in a single day. This happened while the semiconductor sector was experiencing a brutal selloff. If the AI narrative had truly broken, Alibaba, as a downstream company in the AI value chain, should logically have fallen in tandem. But the opposite happened – capital retreated from chips and flowed towards Chinese cloud computing and AI platform companies represented by Alibaba.

Two layers of logic are resonating behind this.

The first is the rotation itself. As Wilson stated, the profit center of the value chain is shifting from hardware infrastructure to software platforms and cloud services. Alibaba, as China's largest cloud vendor and a major AI large language model developer, is positioned to benefit from this shift in profit distribution.

The second layer carries a geopolitical tint. Market rumors (unconfirmed by official sources as of press time) suggest that the Chinese government plans to impose stricter restrictions on the export and overseas access of cutting-edge AI models. Such a policy signal implies that the AI competition between the US and China is further heating up – each country attempting to build its own self-controlled AI tech stack. Against this backdrop, the strategic value of local platform companies with a complete AI ecosystem (cloud computing, large models, application scenarios) is being reassessed by the capital market.

Notably, US-based cloud vendors – Microsoft, Google, Amazon – haven't yet shown the drastic positive reaction seen with Alibaba in their recent stock performance. This might be because their valuations were already relatively high, or because the market is still digesting the dampening effect of "peak capital expenditure growth" on short-term profit statements. However, based on the rotation logic, if capital is indeed moving from chips to cloud platforms, a valuation recovery for these giants is likely a matter of time.

4. Final Thoughts

The pullback in semiconductor stocks is painful, but the AI story is far from over.

It's just that the protagonists of the next chapter may no longer be those selling GPUs and HBM, but those using GPUs and HBM to build the next generation of the internet. Capital is flowing from the hands of "pickaxe sellers" to those "mining with the pickaxes" – this isn't AI's funeral; it's a redistribution of profits within the AI value chain.

And on the BIT (formerly Matrixport) platform, you can not only trade semiconductor stocks like Micron and NVIDIA but also directly buy real US stocks of cloud platform companies like Alibaba, Microsoft, Google, and Amazon. Covering over 10,000 US stocks and ETFs, BIT allows you to quickly complete portfolio adjustments whenever a rotation occurs.

[Disclaimer] This article is market commentary and analysis content. It represents the author's analysis of public information and personal views only, does not constitute investment advice, and does not constitute a recommendation or offer for the purchase or sale of any securities. The views of third-party institutions cited herein (including Morgan Stanley and its analyst Michael Wilson) represent only the stance of the individuals or their institutions, not the views or judgments of BIT. Market trends and individual stock performance are affected by multiple factors. The attribution analysis in this article is only one possible interpretation, is not guaranteed for accuracy, and does not guarantee that the market will follow this logic in the future. Investing involves the risk of loss of principal. Please make decisions with caution.