WE Think: Is This Selloff in Tech Hardware a Bull-Market Correction or a Bubble Burst?

In July 2026, global technology stocks reversed sharply, erasing a significant portion of their first-half gains. The STAR 50, ChiNext, and CSI 300 fell 25.9%, 23.0%, and 7.9% respectively for the month—marking the STAR 50’s worst monthly performance on record, ChiNext’s largest drawdown in a decade, and the CSI 300’s steepest decline in five years. South Korea’s benchmark dropped 22.19% for the month (peak-to-trough: 39.61%), with Samsung and SK Hynix plunging as much as 44.85% and 55.74%, respectively.

Korean retail investors, who had captured substantial gains in the first-half rally, came under intense pressure in July, with margin calls triggering widespread forced liquidations across leveraged accounts. Domestically, mutual funds heavily concentrated in China’s tech hardware delivered record first-half performance, only to suffer unprecedented NAV drawdowns in July as the STAR Market and ChiNext corrected sharply.The speed of this reversal has been staggering—even by my standards as a market veteran. Some of the first-half winners barely had time to book their celebration dinners before the drawdown erased their gains.

Amid this swift reversal, the key question on investors' minds is twofold: What drove tech’s exceptional Q2 performance? And is the July selloff a healthy bull-market correction—or the unwinding of a speculative bubble? In this edition of WE Think, we address both.

What Drove the Q2 Tech Rally?

When we first reviewed the Q2 fund allocation data, the numbers were startling—even to us.According to Q2 2026 mutual fund disclosures, active equity funds allocated 61.34% of their portfolios to Electronics, Telecommunications, and Computer sectors. Including closely linked AI hardware supply-chain industries—Machinery, Power Equipment, and Building Materials—the aggregate allocation reached 76.61%, up 21.7 percentage points from 54.91% at Q1-end. Beyond mutual funds, several trillion RMB in quantitative private funds also participated heavily. Based on industry gatherings with peers in June, sentiment toward tech was overwhelmingly bullish—suggesting that most discretionary hedge funds and sophisticated retail investors carried tech hardware exposure at least on par with, if not exceeding, that of public funds.

(Fund sector allocation data—see original table.)

These figures reveal the underlying dynamic of Q2: traditional sectors such as Consumer, Healthcare, and Non-Ferrous Metals accelerated to the downside, while tech surged—driven by a powerful pro-cyclical feedback loop as onshore capital rotated en masse into the AI hardware theme. Food & Beverage, a long-standing institutional favorite with an average ~11% allocation since 2021, collapsed to just 1.51% by Q2-end. One of the most crowded trades of the quarter was to sell low-visibility sectors like Consumer and rotate into AI hardware names. Yet in July, the two blocks reversed course completely—prompting a question we cannot ignore: Is the market being irrational, or are we missing something in our own framework?

( As we noted in last month’s commentary: investors must distinguish between Howard Marks’ first-level and second-level thinking. First-level thinking focuses on visible catalysts—strong sector momentum, high growth, and robust earnings. Second-level thinking asks the more critical question: At current prices, is the market over-discounting, fairly discounting, or under-discounting these positives?)

How Should We Interpret the AI Hardware Selloff—Correction or Burst?

We recognize that our view diverges from a still-prevailing bullish consensus on AI hardware. Nonetheless, we believe candor with our clients is paramount—and divergence of opinion is both natural and healthy in any vibrant market.

This month's letter was written on a train journey with limited time, hence its brevity on data and details. We will discuss the full analytical framework in an upcoming online dialogue.

We view the July selloff as a classic speculative bubble unwinding—distinct from a typical bull-market correction.

Driven by Q2 momentum and the capital-siphoning dynamic, a number of cyclical tech hardware companies—many with limited competitive moats—were bid up to levels that severely discounted future cash flows, forming what we see as a deep valuation bubble.The degree of excess rivals—and in some respects exceeds—the 2015 ChiNext mania. As a result, we began to proactively reduce hardware exposure in June, well ahead of the subsequent drawdown. Our experience navigating the 2015 ChiNext mania proved invaluable. This time, we were not caught off guard by Mr. Market's deception.

We also reject the notion that “bubbles only correct by half.” In our view, the topping process in tech hardware carries greater downside risk than the 2021 consumer staples bubble peak—a period that inflicted deep and prolonged losses on trend-following investors.

Near-Term Investment Strategy 

The magnitude of the tech correction, combined with the Q2 fund disclosure data, should prompt institutional investors to reassess both the current market landscape and the essence of sound investing: What is really happening in the market, and have we lost sight of first principles?

As we wrote in last month’s letter, the market is returning to rationality. The focus must shift from “What are this quarter’s and this year’s earnings and valuations?” to a more fundamental question: Which earnings are sustainable, and which are merely cyclical windfalls at the peak of industry profitability? A time-honored lesson: The most costly investment mistake is buying cyclical growth stocks at the peak of industry profitability and holding them through the downcycle.We have learned this lesson the hard way and will not repeat it.

The July index correction and the magnitude of the decline in bubble sectors represent a necessary structural mean reversion. Had this correction been delayed, the risk of continued capital siphoning by the bubble would have been far greater—the earlier the adjustment, the lower the systemic cost. Once this process runs its course, we expect A-shares to resume a structural slow-bull trajectory, with reasonably valued, high-quality companies emerging as the primary beneficiaries.

The bursting of the tech hardware bubble does not negate opportunities across the broader technology universe. Technology companies with genuine competitive advantages, technical barriers, and strong business models will, after valuation repair, continue to deliver solid performance in a slow-bull environment. High-quality companies with reasonable valuations and superior business models stand to benefit.

From a strategy standpoint, our positioning remains largely unchanged from last month: we continue to hold high-quality assets with patience.

Wu Weizhi

2 August 2026

Written on the train from Hengyang West back to Shenzhen

本期《偉志思考》簡體中文版鏈接：

伟志思考：这一轮科技硬件公司的下跌，是牛市中的调整还是泡沫破灭？