Sharp Drop! Is China’s Lithography Machine Scaring Wall Street?

On July 27th Eastern Time, the three major U.S. stock indexes appeared calm on the surface.

• The Dow Jones Industrial Average rose 0.51% to 52,210.08 points;
• The S&P 500 index edged up 0.02% to 7,413.18 points;
• The Nasdaq Composite Index fell 0.18% to 24,932.08 points.

However, that night, the red-hot AI hardware sector of the past six months became the main driver of the market's decline. Leading the losses was the semiconductor sector, with the Philadelphia Semiconductor Index (SOX) falling over 5% intraday before closing down 2.23%.

Looking at the constituent stocks, both the established leaders and the new entrants all fell flat:

Lithography giant ASML (ASML) closed down 5.80% at $1,655.26, with intraday losses widening to over 8% at one point; AI chip leader Nvidia (NVDA) closed down 4.99% at $196.51, marking its largest single-day drop since June 5th, with its market cap shrinking by nearly $250 billion in a single day to $4.76 trillion, allowing Apple to reclaim the title of the world's most valuable company. AMD closed down 5.17% at $494.95.

Memory chip stocks fell even more sharply. SanDisk closed down 11.02%, SK Hynix ADR plunged 7.47% to $143.02, breaking its $149 IPO price just 12 trading days after listing; Micron Technology's intraday losses exceeded 7% before narrowing to a 2.25% close. Semiconductor equipment makers like Applied Materials, KLA Corporation, and Lam Research all fell over 3%.

Notably, the market showed a clear rotation from high-growth to value stocks: funds flowed out of crowded AI infrastructure trades and into platform companies with more stable cash flows, with Apple's stock hitting a new all-time high. In contrast, Chinese assets bucked the trend, with the China Internet Index (KWEB) rising over 2.5%, Xiaomi ADR surging 8.97%, and Tencent and Alibaba both gaining over 2%.

Why did a single report on 'Chinese lithography machines' shake ASML?

During trading on July 27th, US tech media outlet The Information published an exclusive report citing two知情人士: A Chinese company backed by the Shanghai government has begun manufacturing immersion Deep Ultraviolet (DUV) lithography machines. The first batch is expected to be delivered to SMIC, Hua Hong Semiconductor, and ChangXin Memory Technologies by 2026, targeting the 28nm process. The production capacity is planned for about 5 units in 2026, expanding to 20 units by 2027.

The report stated that most components are sourced domestically, but several key parts still rely on Japanese suppliers, and the machines lag behind ASML's comparable models in terms of throughput, overlay accuracy, and long-term stability.

Upon the news, ASML shares plunged over 8% intraday. Within half an hour, the Philadelphia Semiconductor Index and the entire US semiconductor equipment sector dropped in tandem.

A Real Question: Where Exactly Does China Stand with its Lithography Machines?

Lithography machines, once the most famous symbol of the "bottleneck" for China's semiconductor industry, occupy a uniquely critical position in the entire supply chain.

Simply put, a lithography machine is the "printer" for chip fabs. The shorter the wavelength, the finer the lines it can print.

Specifically, DUV (Deep Ultraviolet Lithography) determines whether a country can "make chips." DUV uses a 193nm wavelength and supports the vast majority of chips globally, including automotive, industrial, memory, and IoT chips. All mature process nodes down to 28nm rely on DUV. While ASML currently dominates this field, the technological barrier is relatively "climbable."

EUV (Extreme Ultraviolet Lithography), on the other hand, determines whether a country can "make the best chips." With a wavelength of just 13.5nm, EUV is indispensable for cutting-edge mobile CPUs (sub-7nm) and AI training chips. Currently, only ASML can build EUV machines, each costing over $150 million, and their export to China is strictly prohibited by the US.

If we compare the global chip supply chain to an assembly line, China originally had no place: The US controls EDA design software and core IP, Japan monopolises photoresist and high-purity chemical materials, Dutch ASML holds the critical choke point of lithography machines, while Taiwan and South Korea handle the final "integration" of design and equipment into finished chips.

Therefore, the reason this particular report caused such a shock isn't its technical prowess — 5 units targeting 28nm from a domestic Chinese DUV source is almost negligible compared to ASML's annual shipments of hundreds of machines. What truly made Wall Street nervous is the broader picture of China's semiconductor self-sufficiency that this report reflects.

At the DUV level, China may have already achieved the transition from 0 to 1 and is beginning to scale up. According to public information, Shanghai Micro Electronics Equipment's (SMEE) SSA800 28nm immersion DUV tool has been delivered to SMIC in batches, boasting a yield rate of over 90% and over 85% domestic localization. KLA's* dry DUV tool, combined with SAQP (Self-Aligned Quadruple Patterning) technology, has completed verification on SMIC's 28nm production line with a yield rate of 85%, potentially enabling processes equivalent to 5nm. Combined with previously stockpiled ASML DUV equipment (estimated by Dutch media to be "sufficient for 5 to 10 years"), China's supply chain security baseline for mature process nodes is being reinforced.

However, at the EUV level, China remains on a long journey from fundamental principles to prototypes. The highest power achieved by domestic laboratory EUV light sources is 5080W, only one-fifth of ASML's commercial standard; the mirror surface roughness achieved by the Changchun Institute of Optics, Fine Mechanics and Physics (CIOMP) is 0.12-0.2nm, still about four times worse than Zeiss's 0.05nm; the machine requires over 100,000 parts, and the completeness of the domestic supply chain is less than 20%. Industry consensus suggests that small-scale validation may only be possible after 2030. High-end mobile chips (sub-7nm) and AI training chips remain difficult to produce independently in the short term.

So, Does China's Hard Tech Really Have the Power to 'Rattle US Capital Markets'?

The answer is half and half.

"Half is real": The self-sufficiency of the mature process node supply chain has transitioned from a "lab story" to a "production line reality." China is no longer a "must-have" option for ASML in its single largest global market. This is the core sentiment shock that caused ASML to plummet 8% and dragged down the US semiconductor equipment sector.

"Half is illusory": The impact of 28nm DUV is not powerful enough to demolish the entire Philadelphia Semiconductor Index. Even if 5 DUV machines (priced around €200 million each according to ASML) completely replace imported ones, the impact on ASML's annual revenue (around €30 billion in fiscal 2025) would be negligible. Furthermore, these machines still require months or even years of production line verification for precision, yield, and stability before they can truly replace ASML's offerings.

So, this report was a spark — but it ignited a haystack already soaked in anxiety. The real source of the fire lies elsewhere.

Nvidia's $750 Billion Flywheel is the True Root of the Decline

Once we strip away the China DUV news, what remains is a brewing internal confidence crisis within the US stock market.

The more fundamental reason for the recent collective slump in chip stocks is the market's renewed concern over Nvidia's "circular financing" model, which has now spilled over into the credit market — Nvidia's credit default swap (CDS) spreads surged 14 basis points in a single day, a record high.

Nvidia has been making a series of major moves recently. Firstly, it announced a massive industrial chain cooperation with South Korea's SK Group worth over $500 billion. Secondly, reports emerged yesterday that it plans to provide OpenAI with computing power lease guarantees worth up to $250 billion, specifically supporting SoftBank's development of a 10-gigawatt super data center project in Ohio.

The total scale of these two initiatives alone exceeds $750 billion.

The market rally over the past few months essentially ignored the "elephant in the room" — Nvidia's growth model is essentially about providing financing, investment, and equity to downstream companies. These companies, in turn, use the funds to buy more Nvidia chips. Funds circulate within the industry chain; Nvidia's revenue growth is highly dependent on its customers' ability to raise capital. If any link in this chain breaks, the model becomes unsustainable.

From Goldman Sachs to 'The Big Short' investor Michael Burry, serious warnings have been issued over the past few months about this "circular financing": if the financing environment deteriorates, or if the companies heavily investing in AI ultimately fail to achieve profitability, the entire AI spending chain could face amplified losses.

Goldman Sachs analyst Chris Hussey put it more bluntly: The core reason for the S&P 500's stagnation over the past two months is market skepticism about the sustainability of AI infrastructure investment, rather than macroeconomic factors like oil prices or interest rates. A piece of evidence supporting this is that yesterday, when AI-related stocks were excluded, the S&P 500 actually rose by 0.80%.

Outlook

In fact, this US stock market correction did not start just last night. As early as mid-July, the SOX index had already corrected over 20% from its June high, entering a technical bear market. The weight of chip stocks in the S&P 500 has risen from around 8% a few years ago to over 20%, and the daily rebalancing mechanism of leveraged ETFs has further amplified the downward momentum.

However, every sell-off needs a comforting narrative. This time, it happened to be the "myth" of China's lithography machine breakthrough.

Looking back at China's semiconductor industry, self-sufficiency is a long-term process occurring in different phases and different tracks. We are closing the gap on some tracks, but we are far from the point where Wall Street would feel 'genuine fear.'

Wall Street's panic last night was more about staring into the cracks forming in its most crowded AI trade. China's hard tech, meanwhile, is quietly taking root on its own, much longer track.