WE Think: Why Does Tech Keep Falling — and How Do We Prevent Another "Crazy Bull" Tragedy?

Investors were left perplexed by the performance of global technology shares in October. The ChiNext Index fell 23.02%, the STAR Composite 21.7%, the Wind All A Index 12.44%, and the CSI 300 8.54%; South Korea's KOSPI dropped 16.19%. Even US equities cracked: the Nasdaq fell 5.3%, while the Philadelphia Semiconductor Index slumped 15.36%.

Philadelphia Semiconductor, KOSPI, STAR and ChiNext have the highest correlation with this AI build-out, and they were October’s weakest. Meanwhile, A-share banks (+5.33%), Food & Beverage (+5.15%), Oil & Petrochemicals (+2.85%) and Coal (+1.00%) rose against the trend on CITIC Level-1 indices.

Why do AI supply-chain names with decent earnings and orders keep falling since July? Why did Google and Intel set record single-day declines right after reporting? Why are US tech giants all refinancing? And is the regulator powerless against a runaway bull?

This edition of WE Think examines each question in turn.

1. Why Do Tech Stocks Fall on Good Earnings?

The ChiNext and STAR markets traced a textbook "A-shaped" trajectory. April to June surged; July, September and October fell like a waterfall, giving back the first half's 35% gain. By end-October the ChiNext Index was down 24.65% year-to-date, a 61.4% high-to-low amplitude. Ten months, one perfect "A."

Why did hardware tech names keep falling in October despite solid results? This was not a single-market event but a cross-market resonance spanning A-shares, Hong Kong, the US and Korea. The steepest declines came from the hottest links in the chain: optical modules, PCBs, memory chips, and compute chips. Four shocks explain "good earnings, falling prices":

· Narrative Shock: doubts over AI ROI broke the "infinite capex" assumption. The trigger was the October 9 controversy over OpenAI's annualized revenue. Then Anthropic reported Q3 revenue of $16.8 billion, with sequential growth collapsing from 147% to 45%. CoreWeave's equity and debt sold off hard; Google's free cash flow turned negative. The market began to ask whether AI's vast spending can earn adequate returns.

· Valuation Shock: US 10-Year Treasury yields breaking above 5% compressed the denominator, hitting high-multiple growth stocks hardest.

· Positioning Shock: a stampede out of an overcrowded trade. Tech hardware surged in the first half. Institutional overweighting and margin financing hit record levels; positioning became crowded and market microstructure deteriorated—the main driver of the Q3 sell-off.

· Expectations Shock: results were solid, but expectations were already priced in—good news failed to lift prices.

In short, earnings are the rear-view mirror; the market looks through the telescope at future cash flows. October re-priced the core narrative of "unlimited AI capex." Good results only confirm the past; prices discount whether growth persists, returns are delivered and cash burn peaks. When Google reported negative free cash flow for two consecutive quarters, Anthropic’s growth collapsed, CoreWeave’s CDS implied a 63% default probability and Treasury yields broke 5%—the view through the telescope had changed. AI infrastructure became a risk asset: capital-heavy, uncertain in return, accelerating in burn. No Q3 beat could reverse that repricing. With positioning extremely crowded in A-shares and Korea, any crack in the narrative set off a liquidity stampede, which redemptions then amplified.

I largely agree with the AI summary above. But AI's strength is being "too comprehensive," which can bury the key point. The most critical factor I would add: an overcrowded positioning structure that triggered a panic stampede—that is what turned a comedy into a tragedy. How to avoid the next one is the subject of part three.

2. Google and Intel Plunge Post-Earnings: Is the Faith in a "Long Bull Without Bears" Breaking?

In late October, heavyweight US tech names reported, and the picture changed. Investors opened their earnings "blind boxes" as usual. Before, they held surprises—red envelopes of varying size. This time, none. SpaceX was the exception: it beat on results and guidance and gapped up 7.79% to $160.55. Over seven sessions it drifted down to $134.72 at month-end—still 8.84% below its pre-earnings close of $147.79 on October 20.

· For most tech giants, earnings releases turned into horror shows—large bills instead of red envelopes. Google reported Q3 on October 28. On October 29 the stock plunged 12.64%—its largest single-day drop since listing, wiping out nearly $500 billion in market value overnight. That is a Tencent gone: Tencent closed October at $444.7 billion. Sell-side analysts and the media blamed a "triple miss" plus a wobble in the narrative. Ad growth slowed as AI eroded monetization. A fair-value loss on Q3 holdings, SpaceX and Anthropic included, dragged down GAAP profit. Free cash flow turned negative for a second straight quarter, with no end in sight to the AI burn. Microsoft, reporting the same day, was the contrast: cloud revenue grew 48%, it generated +$18.6 billion in free cash flow, and it cut capex guidance from $175 billion to $170 billion. The market rewarded capital discipline and positive cash generation.

· Intel’s post-earnings collapse opened my eyes—and I am an old hand in these markets. The business was unchanged overnight, yet its value swung nearly 30% simply because a quarterly report came out. ServiceNow crashed 26.95% on October 29, also a record single-day drop.

When rock-solid giants crash like this overnight, it should make us think. If you hold such names, how do you hedge?

Why has overnight repricing become so violent? I have spent years tracking these patterns in both markets. After an 18-year US bull market, most active managers have long underperformed passive benchmarks, so investors abandoned them. From 2016 to 2025, passive index funds attracted $6.42 trillion in net inflows, while active funds suffered $2.40 trillion in net outflows. Index funds are now 64.8% of total US domestic equity fund assets.

Owning Nasdaq or S&P 500 ETFs has become an article of faith for American investors. But without active capital, the investment ecology breaks down: no one sells early to hedge before bad news, and no contrarian steps in to buy the crash. A-share quant index-enhanced funds show the same pattern—slow climbs, fast dives. When passive and quant flows dominate and active managers are marginalized, post-hoc stampedes become inevitable.

Back to today's US market: this bull has run nearly two decades—unprecedented in US and global equity history. I once said in a media interview: "Every article of faith in capital markets exists to be broken." Now: is the long-held faith in a permanent US 'long bull without bears' reaching its breaking point?

3. How Do We Avoid the Next "Crazy Bull" Tragedy? Should Regulators and Investors Both Review and Upgrade?

I lived through the 2015 growth-theme bubble, and have just watched the Q2–Q3 hardware bubble repeat it. Because of 2015—the euphoria in late June, the irrational concentration of mutual fund holdings at end-Q2—we knew the tragedy was unavoidable. In late July, my WE Think essay argued this was not a bull-market correction but a bubble bursting. It drew controversy and real pressure; on Xueqiu it received more criticism than anything I had written. I understand the critics: separating a wish from rational analysis is not easy.

Why does institutional industry concentration reach such high levels, and how can the industry prevent a repeat?

First, the industry must curb the disorderly spread of a speculative "gambling culture" in mutual fund management.

Mobile internet and digital financial media have made online distribution the main channel for new money. Only the year's top-ranked funds get media coverage and shelf space; middle-ranked funds get neither. A small mutual fund then grew fast with sector- and style-focused funds, topped the rankings and scaled up. Peers copied it, learning that overnight fame required extreme concentration in industries and holdings.

Regulation has long capped a fund's holding in any single stock at 10% of assets, and its stake in any single company at 10% of that company's shares—the "double 10%" rule. That has kept China's mutual fund industry free of single-stock liquidity problems, but it sets no limit on a fund's exposure to one industry. In Q2, star tech funds were concentrated almost entirely in optical modules or semiconductors. Putting 80% to 100% of assets into one hot sub-sector—with the powerful lure of creating a star—became a tempting template. If you can do it, why can't I? That is the logic behind Q2's spectacle of funds holding over 60% of assets in just two industries. What began as a celebration of prosperity ended as the wreckage of a bubble. Without reform, the next abnormal overweight is probably unavoidable. Is it time to cap single-industry allocation for non-sector, non-style funds?

Second, individual investors themselves need to reflect.

Asset allocation also needs balance: no one should put all their eggs in one basket. In 2015 and again this cycle, many high-net-worth investors chased hot tech stocks midway up the rally. Few exited at the top; most rode the roller-coaster, watching paper gains turn into losses. Traders who truly understand trend investing could still take profits with discipline. Most, though, mixed trend with fundamentals—and those who tracked fundamentals closest found retreat hardest.

Investing is like surgery: without real expertise you may get it right once or twice by luck, never over the long run.

How to avoid the stampede? Return to owning quality long-term assets. Bought at reasonable valuations, they have little to do with such a stampede. As sentiment settles, the muddy water clears, and they again deliver sustained returns.

4. Investment Strategy

The tech hardware sell-off has hit sentiment hard; the situation resembles late 2015. Having been through such episodes, solving today's problem is not so hard.

Our strategy returns to "each flower represents a world of its own." Industries differ: some are mid-bear, others turning from bear's end to bull's beginning. We strictly avoid sectors that remain valuation-expensive even after their bubbles burst. There may be rebounds, but those are chances to escape—as August gave tech investors; making rebound money in bear sectors is too hard. Better to pick easier questions.

For industries and companies that have fallen long and are in late bear, we see many positive signals—spring is arriving. We will embrace them and invest more.

Wu Weizhi

October 31, 2026, Shenzhen

本期《偉志思考》簡體中文版鏈接：

偉志思考：為何科技板塊跌跌不休？如何才能避免瘋牛悲劇的重演？