South Korea's Kospi fell 10.84% on Tuesday, July 28, its worst session since March and only the 8th time in 2026 a marketwide circuit breaker has halted trading. Samsung fell 14.4%, its worst day since October 2008. The trigger was not an earnings report or a Fed decision. It was a single report about a Chinese chipmaking tool that most investors could not have named a week ago.

The short answer

Chip stocks crashed because The Information reported that a Chinese company has begun producing domestic deep ultraviolet lithography machines, the equipment used to print circuits on chips, previously dominated by the Dutch supplier ASML. That threatens the assumption that Western and Korean chipmakers have a multi-year technological lead China cannot close, which is a large part of why AI and memory chip stocks trade at premium valuations. Samsung and SK Hynix, both heavily reliant on that assumed lead, fell 14% or more. The S&P 500 barely moved, up 0.02%, because the damage was concentrated in specific chip names, not the broad market.

What actually happened today

The Information reported that a Chinese company linked to Huawei and semiconductor equipment maker SiCarrier has begun mass production of homegrown immersion DUV lithography machines, with initial units going to major Chinese chipmakers including SMIC, Hua Hong, and ChangXin Memory Technologies this year. South Korea's Kospi index fell 732 points to close at 6,023.66, triggering the exchange's marketwide circuit breaker, an eight-minute halt that has now happened eight times in 2026. Samsung Electronics fell 14.4%, SK Hynix fell 14.7% and its US-listed shares dropped below their original offering price. In the US session, Sandisk fell as much as 14% to 16%, Micron fell more than 8%, and ASML itself fell as much as 8% on Monday when the report first circulated. Nvidia sank at the open before recovering to close roughly flat, a detail that matters for what this selloff actually was.

Why a tool, not a chip, spooked the market this much

A stock's valuation is partly a bet on how long a company's advantage lasts. Samsung, SK Hynix, and the broader memory and AI hardware supply chain have traded at premium multiples on the assumption that Chinese competitors remain years behind on the equipment needed to manufacture advanced chips at scale. DUV lithography machines are that equipment: without them, or without a domestic substitute, Chinese chipmakers depend on ASML and are exposed to export controls. A credible report that a domestic alternative now exists does not change any company's current quarter. It changes the assumption behind the multiple investors are willing to pay for that multi-year moat, which is why the reaction was sharp and immediate rather than gradual.

This is a different kind of shock than an earnings miss
An earnings miss is backward looking: it tells you what already happened last quarter. A credible competitive threat is forward looking: it asks the market to reprice years of assumed advantage in a single session, which is why Samsung's 14.4% drop dwarfs a typical post-earnings move. This connects to the same AI capex question raised in last week's Tesla and Alphabet earnings post and is the AI stock boom a bubble: the market is increasingly sensitive to anything that questions whether today's AI infrastructure spending will hold its value.

The honest counterargument: the report describes a modest step, not a finished threat

It would be dishonest to treat today's reaction as proof the competitive threat is as large as the stock moves suggest:

  • The reported production volume is small, according to CNBC's technical breakdown of the report: roughly five machines in 2026 and about 20 in 2027, a fraction of what would be needed to meaningfully displace ASML's supply to Chinese chipmakers at scale.
  • The technology targets older nodes, not the cutting edge. The machine is described as targeting 28-nanometer production in a single exposure, with 7-nanometer and 5-nanometer chips only reachable through multiple patterning techniques, and reported yields below what industry-leading equipment achieves even at those more advanced nodes.
  • This is not the first "China catches up" scare, and the last one did not end the incumbents' lead. When Huawei's Mate 60 Pro shipped in 2023 with a chip TechInsights confirmed was made at a 7-nanometer-class node by SMIC without EUV lithography, it was called a breakthrough nobody expected. That chip was still less dense, less power efficient, and lower yielding than equivalent chips from Taiwan Semiconductor. Two years later, TSMC's global foundry market share had grown, not shrunk, even as SMIC became the world's third-largest foundry by revenue.

None of that means the competitive threat is imaginary. It means today's stock moves priced in a level of certainty about a multi-year outcome that a single, still-modest production report cannot actually provide.

⚠
Notice what did not crash. The S&P 500 closed up 0.02% and Nvidia, the largest single AI chip name, closed roughly flat after an early dip. This was a sharp, concentrated repricing in Korean memory chip makers and a handful of related names, not a broad market event. That distinction matters for deciding whether this affects your own portfolio.

What a beginner should actually do

  1. Check how you actually hold semiconductor exposure. If it is through a broad S&P 500 or total market index fund, today's move barely touched your overall return, since the index itself was flat.
  2. If you hold a semiconductor sector ETF or individual chip stocks, know that you are carrying concentrated, single-theme risk. That is the tradeoff of sector and single-stock exposure discussed in single stock vs. index funds, and today is a live example of what that concentration can do in one session.
  3. Do not treat one report as a verified, finished outcome. Modest production volumes and older-node targeting are a long way from displacing an established equipment leader, a pattern the 2023 Huawei chip scare already demonstrated once.
  4. Watch confirmed shipments and yield data over the next several quarters, not today's headline. That is the actual evidence that would tell you whether this is a real structural shift or a repeat of a scare that faded.
  5. Keep contributing on schedule. See time in the market beats timing the market for why reacting to a single sector's worst day in months has historically cost more than it saved.
💡
The actionable takeaway: a single report about modest, older-node Chinese chip equipment production triggered a sharp, concentrated selloff in Korean memory makers and related names, while the broad US market barely moved. That gap is the honest signal here: this was a repricing of a specific competitive assumption, not evidence of a broad market problem.

The quick version

  • South Korea's Kospi fell 10.84% on July 28, its worst session since March, triggering the exchange's 8th marketwide circuit breaker of 2026
  • Samsung fell 14.4% (its worst day since October 2008) and SK Hynix fell 14.7% after The Information reported China has begun producing domestic DUV lithography machines, the tool previously dominated by ASML
  • In the US, Sandisk fell as much as 14% to 16% and Micron fell more than 8%, but the S&P 500 closed up just 0.02% and Nvidia closed roughly flat
  • The reported production is modest, about five machines in 2026 and 20 in 2027, targeting older 28-nanometer nodes with yields below industry-leading equipment even at more advanced nodes
  • A similar "China catches up" scare followed Huawei's 2023 Mate 60 Pro chip, and TSMC's global foundry market share grew, not shrank, over the following two years
  • This was a repricing of a specific competitive assumption behind premium chip valuations, not a broad market event, since the index barely moved
  • If you hold semiconductor exposure through a broad index fund, today's move had little effect on your overall return; concentrated sector or single-stock exposure is where the real risk showed up