# ANALYSIS: Chip stocks crater, and the AI spending story faces its bill

> South Korea's Kospi closed down nearly 11% on a Tuesday in late July 2026, its eighth circuit breaker of the year , according to data reported by Fortune on July 28. This is not an isolated Asian market accident.

- Author: Maxime Marquette
- Publisher: MadMax (https://mad-max.co)
- Section: Analysis
- Published: 2026-07-29
- Canonical URL: https://mad-max.co/en/article/analysis-chip-stocks-crater-and-the-ai-spending-story-faces-its-bill

## Key takeaways

- The Kospi fell nearly 11% in a late-July 2026 session, its eighth circuit breaker of the year, and the SOX fell as much as 6% in a fourth straight losing session.
- The Nasdaq-100 sat 9.7% below its record high, close to a technical correction.
- The trigger was a report by The Information that Shanghai Yuliangsheng mass-produces DUV lithography machines; the company did not publicly confirm it.
- Wedbush analyst Matt Bryson noted China has had DUV access for years and that the binding constraint is EUV, which China still lacks.
- DA Davidson's Gil Luria called the panic "indiscriminate," and Wells Fargo's Ohsung Kwon noted chip-hyperscaler correlation fell to -31%.

South Korea's **Kospi** closed down **nearly 11%** on a Tuesday in late July 2026, its **eighth circuit breaker of the year**, according to data reported by **Fortune** on July 28. This is not an isolated Asian market accident. The same day, the American semiconductor index **SOX** fell as much as **6%**, its **fourth straight losing session** — the longest losing streak of the year for that index. *A market does not panic eight times by accident. It panics when nobody trusts its own numbers anymore.*

The **Nasdaq-100** flirted with a technical correction, sitting **9.7% below its record high**. This shock strikes at the heart of the story that has carried markets for three years: the promise that artificial intelligence justifies unprecedented capital spending. The **immediate backdrop** includes an unverified **Chinese claim** in lithography, a surge for memory maker **CXMT**, and **DRAM contracts** trading **20% to 30% more expensive** this month, according to Fortune.

This **analysis** is built exclusively on market data published July 27 and 28, 2026, on statements attributed to financial analysts, and on companies' own communications. No **capex figure**, no quote, no **projection** is presented here as a **certainty** beyond what the source actually supports.

## The Kospi and the SOX, two indexes telling the same nervous story

### An eighth circuit breaker that is anything but trivial

A **circuit breaker** is not an ordinary decline: it is an emergency mechanism that automatically halts trading when a single-session drop crosses a critical threshold. That the Kospi triggered this mechanism for the **eighth time in 2026**, according to Fortune, signals a **volatility** that goes well beyond the usual noise of **tech markets**. South Korea hosts two of the world's largest memory chipmakers, which is why its benchmark index absorbs the first shocks from any tremor in the semiconductor sector.

It is no coincidence that this drop lands at the **exact moment** when reports of a **Chinese technological breakthrough** in lithography are circulating. **Markets do not always distinguish a rumor from a confirmation**. They react to perceived probability, not to established proof. Eight **circuit breakers** in seven months describe an industry living on its nerves.

### The SOX strings together an unusual run

The **SOX** index, which tracks the leading American semiconductor stocks, fell as much as **6%** during its **fourth consecutive losing session**, according to Fortune — the longest run of the year for this index. Four red sessions in a row do not amount to a structural collapse, but they mark a clear break from the near-uninterrupted upward path this sector has followed since the rise of large AI models.

The broader **Nasdaq-100** followed the same trend, edging toward a **technical correction** at **9.7% off its record high**. A technical correction is conventionally defined as a 10% drop from a recent peak. The index had not crossed that line as of the reported data, but the gap narrowed to under a single percentage point. **The safety margin evaporated** in a matter of sessions. *One percentage point still separates the Nasdaq-100 from an official correction. One percentage point, in markets, protects nothing.*

## The trigger: an unconfirmed Chinese claim

### The Information lights the fuse

According to the outlet **The Information**, the Chinese company **Shanghai Yuliangsheng** is now said to mass-produce deep ultraviolet immersion lithography machines — a technology **ASML** has mastered for roughly a decade. Several major American semiconductor stocks fell on Wall Street as early as Monday, July 27, in the direct wake of that report. **Contacted by AFP, the Chinese company did not publicly confirm the details of the report**, according to the **dispatch** relayed by Fortune and NAMPA.

This absence of direct confirmation is a central element, not a footnote. *A market that loses eleven percent on a rumor the company itself has not confirmed is not measuring a fact; it is measuring the fear of a fact.* The **available fact dossier** explicitly labels this information a **press report**, not an independently verified **technical fact**.

### CXMT soars while Wall Street shakes

The same day, Chinese memory maker **CXMT** made its Shanghai stock market debut with a **466% surge**, after raising **$8.6 billion**, according to Caixin Global. This double sequence — a lithography breakthrough report and a spectacular IPO from a Chinese memory competitor — was enough to shake entire indexes, independent of any independent technical verification of Yuliangsheng's actual capabilities.

What happened must be named plainly: a **temporal correlation** between two Chinese events on the same weekend was read by markets as a **structural threat** to Western dominance in advanced semiconductors. No available data confirms that these two events are technically linked beyond their simultaneity.

## What analysts actually say about the technological constraint

### The crucial DUV-versus-EUV distinction

**Matt Bryson**, semiconductor analyst at **Wedbush**, offered a decisive nuance that the market panic largely ignored: "China has actually had access to DUV lithography for years," he said, adding that "the actual binding constraint is EUV — a more advanced technology that China still lacks." In other words, China has had **access to DUV lithography** for years — the very technology Yuliangsheng now claims to mass-produce. **The real technological barrier remains EUV**, a technology China still does not master.

*A market that confuses DUV with EUV has not made a subtle technical error; it has made a ten-year industrial-lead error.* ASML remains the only company in the world producing EUV machines, and Western export controls keep this technology out of Chinese reach, according to the available fact dossier.

### What DUV can and cannot do

Advanced memory chip manufacturing remains possible with DUV lithography alone, but it requires **more etching passes**, which mechanically raises production costs. This **cost-and-yield gap** still separates established manufacturers — **Micron, Samsung and SK Hynix** — from a new entrant like CXMT, even after a spectacular **stock market debut**.

A successful IPO does not replace a proven technology roadmap. **The industrial gap** separating China from Western EUV standards did not close on July 27, 2026, regardless of the scale of that day's market reaction.

## The diagnosis of the panic: indiscriminate, according to an analyst

### "The panic appears to be indiscriminate"

**Gil Luria**, technology analyst at **DA Davidson**, summed up the situation in blunt terms: "Right now there's a lot of panic around the AI investment," he said, before adding that "the panic appears to be indiscriminate." This **phrasing** deserves to be taken seriously: a professional **financial analyst** himself describes the market's reaction as **non-discriminating**, meaning unable to separate companies genuinely exposed to **Chinese risk** from those that are not.

*A panic that hits without distinguishing real risk from imagined risk is no longer a financial assessment. It is a collective reflex.* Luria's language describes a correction driven by market emotion, not a methodical reassessment of each company's fundamentals.

### A correlation that collapses between chips and hyperscalers

**Ohsung Kwon** of **Wells Fargo** wrote that the correlation between semiconductor stocks and major cloud hyperscalers has fallen to **-31% this month**, calling the coming week "make or break." A **negative correlation** between two previously aligned sectors indicates that **investors** no longer treat the **AI ecosystem** as a single block, and are starting to separate presumed winners from presumed losers.

This **correlation breakdown** may be the single strongest signal in this entire sequence, more so than the **drop percentages** themselves. **Markets are starting to sort**, even if that sorting remains, for now, largely driven by fear rather than fine-grained analysis of each balance sheet.

## The earnings wall arrives at the worst possible moment

### Microsoft, Meta, Apple and Amazon under the spotlight

This market shock lands just days ahead of a crucial earnings week: **Microsoft and Meta** report Wednesday, July 29, **Apple and Amazon** report Thursday, July 30, 2026 — the same week as a **Federal Reserve** rate decision. The **calendar** could not be more crowded for an **already jittery sector**.

Each of these four companies ranks among the world's largest investors in AI infrastructure. **A disappointing result from even one of them** would, in this climate, be enough to further amplify the correction already underway on semiconductor indexes.

### Alphabet already set the tone with its capex

**Alphabet** raised its capital spending to **as much as $205 billion** for this year, up from **$91 billion in 2025** — more than double in a single year. According to **Moody's**, the six largest hyperscalers will spend a combined **roughly $785 billion in 2026**, and close to **$1 trillion in 2027**.

*Doubling capex in one year is not a budget tweak. It is a bet on growth that must now materialize, or else retroactively justify every fear the market has voiced.* These figures come from company communications and Moody's estimates, not from a single independent audit, and should be read as such.

## DRAM memory, a market already stretched before the panic

### Contracts climbing before the stock market crisis even hit

Independent of the Chinese lithography claim, **third-quarter DRAM memory contracts** were already trading **20% to 30% more expensive** this month, according to Fortune. **Google and Meta** signed contracts fixing price and volume over **five years**, a long-term contractual commitment that reflects an anticipation of durable scarcity rather than a temporary demand spike.

Analysts cited by Fortune do not expect meaningful additional supply before **2028**. **This supply ceiling** predates the **stock market panic** of July 27-28; it is one of the **structural factors** underneath it, distinct from the specific question of Chinese lithography.

### A supply crunch that goes beyond the geopolitical debate

This DRAM memory shortage touches the entire industry, including American and South Korean companies with no direct exposure to any potential Chinese technological catch-up. **The conflation of these two dynamics** — a structural memory shortage on one side, a rumor of Chinese technological catch-up on the other — likely amplified the scale of the July 27 reaction.

Separating these two threads is not an academic exercise. It is the condition for understanding why a South Korean index could lose eleven percent even as the EUV constraint, according to Wedbush, remains entirely intact. *The memory shortage did not come from Beijing. It came from five years of demand that outran every forecast.*

## What the Chinese production timeline actually says

### Mass production versus prototype, a nuance that changes everything

The Information's report describes "mass" production of DUV machines by Yuliangsheng — a term that, if confirmed, would indicate a stage of industrial maturity far beyond a simple lab prototype. But the **company itself** has not publicly confirmed these details, according to the dispatch relayed by Fortune. **The word "mass" remains, for now, the journalist's word**, not the manufacturer's.

*Between a prototype and a real production line lies the distance separating a promise from a deliverable product. Markets, for their part, reacted as if that distance had already been crossed.*

### The absence of independent confirmation weighs on the entire analysis

None of the available sources report any **independent test**, any **named client**, or any **verifiable production volume** for Yuliangsheng's machines. **This absence of third-party verification** does not mean the report is false; it means no rigorous journalist should present it as an established fact in the strict sense at this stage.

The **methodological caution** applied here consists precisely of separating what is confirmed — the drop in indexes, the capex figures, the analysts' quotes — from what remains, as of July 28, 2026, an **unconfirmed press allegation** from its direct subject.

## The impact on American, Korean and Taiwanese manufacturers

### Micron, Samsung, SK Hynix: the trio under watch

The three long-established memory manufacturers — **Micron** in the United States, **Samsung** and **SK Hynix** in South Korea — occupy a particular position in this sequence: they benefit directly from the **rise in DRAM prices** noted above, while simultaneously facing **downward pressure** tied to fear of an **emerging Chinese competitor**. **These two opposing forces** complicate any simple reading of the July 27-28 market move for these three companies specifically.

Nothing in the available data establishes to what extent the Kospi's decline directly affected Samsung and SK Hynix relative to other market factors. **The cause-and-effect link** deserves to be stated with this explicit caveat.

### A Taiwanese industry structurally apart

No source in the available dossier mentions any **specific movement** affecting the **major Taiwanese foundries** in this precise July 27-28 sequence. **This absence of mention** should not be read as proof of immunity: it simply reflects the limits of the sources consulted for this analysis, which focus on South Korean and American indexes.

What the available data does confirm, however, is that the EUV constraint — from which Taiwan and its partners benefit heavily through ASML — remains, according to Wedbush, the real technological chokepoint separating Western industry from any emerging Chinese competitor. *The EUV lock still holds. Fear, meanwhile, has already walked through the door.*

## The role of speculation in amplifying the move

### A reaction faster than verification

The **lag** between the publication of The Information's report and the **drop in American indexes**, as early as Monday, July 27, illustrates a dynamic now familiar to tech markets: **information travels faster than its verification**. Trading algorithms and human panic reactions do not always distinguish, in the moment, a press rumor from a confirmed official statement.

This gap between diffusion and verification is structural across the entire publicly traded technology sector, not unique to this episode. **It partly explains** why a single report, unconfirmed by its subject, could trigger an eighth circuit breaker in seven months on a major index.

### What next week will really tell

The **quarterly results** from Microsoft, Meta, Apple and Amazon, combined with the **Fed's rate decision** the same week, will be the **real test** of this panic. **If the capital spending on artificial intelligence** announced by these companies holds up and their AI-related revenue grows accordingly, part of the July 27-28 correction could reverse quickly.

*Markets got scared before they knew. This week, they will know.* Nothing in the data available as of July 28 allows a prediction of how those releases will land.

## What this panic reveals about the fragility of the AI narrative

### A story carried by numbers that must now deliver

The capex commitments announced by hyperscalers — up to a combined **$1 trillion** in 2027 according to Moody's — rest on an **implicit assumption**: that **demand** for AI services will keep growing at the pace needed to justify those sums. **This assumption was not invalidated** by the events of July 27-28, but it also received no additional confirmation.

The panic observed that week is therefore not proof that the AI investment story is collapsing. It is proof that this story remains **fragile enough** for an unconfirmed rumor out of China to shake entire indexes within a few sessions.

### A reading lesson for the weeks ahead

The available dossier draws a **clear distinction** between what is measured — indexes, DRAM contracts, reported capex figures — and what belongs to **independently unverified professional opinion** — analysts' quotes, market projections. **Conflating these two categories** is precisely what amplified the July 27-28 panic.

Any serious reading of this sequence must preserve that distinction, week after week, as new figures — quarterly, macroeconomic, or technological — feed or contradict the current narrative of massive AI investment. *A measured figure and a cited opinion do not carry the same weight. Markets, that week, weighed them the same.*

## The precedent of past semiconductor cycles

### A sector used to roller coasters

The semiconductor industry is no stranger to a stock market panic followed by a rapid rebound. Previous cycles, driven first by smartphones and then by cloud computing, each saw severe corrections before rebounding once production capacity adjusted to real demand. **This historical precedent** guarantees nothing for the current AI cycle, but it invites a distinction between a cyclical correction and a structural collapse.

The **difference** this time lies in the scale of the **sums committed**: never before has a sector seen individual companies pledge more than **$200 billion** in annual spending on a single investment line. **This unprecedented scale** changes the nature of the risk: a 10% correction on such an amount represents tens of billions of dollars in revalued worth within only a few sessions.

### What truly sets this cycle apart

The current cycle also stands apart in the speed at which information circulates: an unconfirmed press report, published on a Sunday or a Monday, can now send an entire national index tumbling as soon as markets open the next day. **This speed** leaves little time for factual verification before capital moves massively.

*Previous cycles had months to digest bad news. This one has only hours.* It is this compression of reaction time, more than the news itself, that partly explains the scale of the July 27-28 move.

## The voices urging caution on both sides

### Neither euphoria nor capitulation, some observers say

Beyond the Gil Luria and Ohsung Kwon quotes already noted, the overall tone of available financial analysis for this period avoids both opposite extremes: neither the euphoria that has dominated tech markets in recent years, nor the total capitulation some headlines might suggest. **This absence of a clear-cut consensus** honestly reflects the real uncertainty surrounding the situation as of July 28, 2026.

The same analysts flagging **indiscriminate panic** at DA Davidson do not go so far as to claim current sector valuations are justified. They describe a market that has temporarily lost its ability to distinguish real risk from media noise — which amounts to neither confirmed undervaluation nor confirmed overvaluation.

### The notable silence from the Chinese companies involved

Neither **Shanghai Yuliangsheng** nor any official Chinese spokesperson cited in the available sources has publicly confirmed or denied, in any detail, the real scale of the production capacity claimed by The Information. **This silence** leaves the market in a state of prolonged uncertainty, fertile ground for any further rumor.

*Silence is not an admission. But a prolonged silence, facing a rumor that has already toppled entire indexes, ends up carrying as much weight as a statement.* As long as this silence persists, the **nervousness observed** on July 27 and 28 will remain an **open option** for markets.

## What Asian markets say about supply chains

### Japan and Taiwan watched closely

Beyond South Korea, Asian stock markets tied to the semiconductor supply chain broadly followed the nervousness of the July 27 and 28 sessions, even though the data available for this analysis focuses specifically on the Kospi and American indexes. **This source limitation** should be acknowledged explicitly rather than filled in by **extrapolation**.

What remains verifiable, however, is that the global supply chain for advanced chips stays concentrated among a very small number of players — **ASML** for EUV lithography, **TSMC** for leading-edge etching — a concentration that makes the entire sector structurally sensitive to the slightest rumor touching any one of these links.

### A dependency that explains the scale of the reactions

This extreme concentration of the value chain means that a mere rumor about a single supplier, even unconfirmed, can shake entire indexes worldwide. **This is not a market communication flaw**; it is the exact reflection of an industry where a handful of factories, in a handful of countries, manufacture most of the components running the global AI economy.

*When an entire global industry depends on a handful of factories, the slightest rumor weighs like a fact. That is the vulnerability the July 27 panic laid bare.*

What this sequence establishes with certainty, as of July 28, 2026: an **eighth circuit breaker** on the Kospi, a **fourth straight losing session** for the SOX, a Nasdaq-100 within a single point of a technical correction, and a Chinese company whose technological claim remains unconfirmed by itself. What this sequence does not allow us to state: that China has closed its gap in advanced lithography, or that the narrative of massive AI investment is collapsing.

The week of July 29, with results from **Microsoft, Meta, Apple and Amazon** and the **Fed's decision**, will show whether this panic was a **healthy correction** or an **overreaction** the fundamentals will soon fix. **The market has not yet decided**, whatever the red percentages displayed that week suggest. *An eleven-percent single-session drop proves nothing about Chinese technology. It proves only that confidence, in this market, hangs by a thread that a single rumor is enough to pull taut.*

**By Maxime Marquette, columnist**

## Columnist's transparency note

### Editorial stance

I am not a journalist but a **columnist and analyst**. My expertise lies in observing and analysing the geopolitical, economic and strategic dynamics that shape our world. My work is to take political strategies apart, to understand global economic movements, to put the decisions of international actors in context and to offer analytical perspectives on the changes that are redefining our societies.

I do not claim the cold objectivity of traditional journalism, which stops at reporting facts. **I claim analytical clarity**, rigorous interpretation and a deep understanding of the complex issues that concern us all. My role is to give meaning to the facts, to place them in their historical and strategic context, and to offer a critical reading of events.

### Method and sources

This text keeps the basic distinction between **verified facts and interpretive analysis**. The rule never changes: a factual claim is published only when a verifiable source supports it, and the sources this article actually used are listed under **Sources**, never here.

**Primary source families** used by the publication, when the story has them: official statements from governments and international institutions, public statements by political leaders, reports from intergovernmental organisations, dispatches from recognised international news agencies.

**Secondary source families**: specialised publications, internationally recognised news media, analyses from established research institutions, reports from industry organisations.

When an article cites statistical, economic or geopolitical data, it comes from the institutions that produce it (intergovernmental organisations, central banks, national statistics offices), and the exact institution is named under **Sources**.

### Nature of the analysis

The analyses, interpretations and perspectives in the analytical sections of this article are a **critical, contextual synthesis** based on the information available, the trends observed and the expert comments cited in the sources consulted.

My role is to interpret these facts, to set them in the context of today's geopolitical and economic dynamics, and to give them a coherent meaning within the larger story of the changes shaping our time. These analyses reflect expertise built through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors.

This article describes a documented state of affairs on its publication date, not a prediction: later developments may shift its outlook. No update is promised in advance; when an article is corrected or completed, the change is dated in the text.

## Sources

### Primary sources

- [Fortune — Why stocks are falling: chip and semiconductor panic — July 28, 2026](https://fortune.com/2026/07/28/why-are-stocks-down-chips-panic-semiconductors/)

- [Chosun Biz — Coverage of the Kospi circuit breaker and semiconductor panic — July 28, 2026](https://biz.chosun.com/en/en-finance/2026/07/28/ZUP6LYZYRFGODFTVUFF45ZMKIA/)

- [Caixin Global — CXMT surges over 470% in debut to become China's most valuable stock — July 27, 2026](https://www.caixinglobal.com/2026-07-27/chipmaker-cxmt-surges-over-470-in-debut-to-become-chinas-most-valuable-stock-102468384.html)

### Secondary sources

- NAMPA — Dispatch on the Chinese DUV lithography claim — July 28, 2026

- [UPI — Circuit breaker on the Kospi at the Korea Exchange — July 28, 2026](https://www.upi.com/Top_News/World-News/2026/07/28/kospi-circuit-breaker-korea-exchange/7341785278461/)

- [Lycoris Technologies — The peak-cycle test for semiconductor stocks, July 2026 — July 28, 2026](https://lycoristechnologies.com/blog/chip-stocks-peak-cycle-test-july-2026/)

---

Source: Maxime Marquette, MadMax. When citing, name the author and link https://mad-max.co/en/article/analysis-chip-stocks-crater-and-the-ai-spending-story-faces-its-bill