Author: Hanya Hu, Silicon Valley Investor + Online Novelist
Today, a piece of news about the mass production of a domestic DUV lithography machine sent many people into a frenzy.
Actually, this isn't the first time this has happened. Whenever certain stocks need hype, someone always brings up import substitution.
Today, we'll neither hype nor criticize; let's see what's really going on.
Let's start with the conclusion. The "DeepSeek moment" for domestic lithography machines has not yet arrived.
This article aims neither to hype nor criticize, but to help investors establish a fact-based understanding.
What exactly happened: A Shanghai-based state-owned enterprise has begun small-batch production of a domestic immersion DUV lithography machine. It plans to ship about 5 units this year to SMIC, Huahong Semiconductor, and ChangXin Memory Technologies (CXMT). The production target for 2027 is around 20 units. The equipment model corresponds to a 28nm process. Officially released data indicates an overall domestic content rate exceeding 85% and stable mass production yield above 90%. Almost simultaneously, SMEE (Shanghai Micro Electronics Equipment) announced full mass production of its SSA800 series, with an annual capacity target of 50 units, doubling year-on-year.
These two pieces of news are often mixed together, accompanied by headlines like "The DeepSeek Moment for Lithography Machines," making it seem as if China has already conquered the lithography machine barrier. Upon closer examination, the reality is not as optimistic as imagined.
For investors, this news is worth continuous tracking, but it is currently insufficient to directly conclude that the moat of lithography leader ASML has been breached, or to prove that related domestic equipment companies are about to see large-scale profit realization.
What Can Be Confirmed Now
On July 27th, Reuters cited a report from *The Information* stating that an unnamed Chinese state-owned enterprise had begun manufacturing domestic immersion deep ultraviolet lithography machines.
The report mentioned that the first batch of equipment is expected to be delivered to SMIC, Huahong, and CXMT in 2026, with plans to produce about 5 units this year and increase to about 20 units in 2027.
At the same time, the report clearly stated that this equipment still lags behind ASML in performance and reliability, requires further testing, and is some distance away from truly large-scale commercial production. The EUV equipment under development in China remains in the prototype stage.
These qualifiers are very important.
1. A more accurate description at present should be "began manufacturing and entered the customer introduction phase." "Already in full mass production" can easily be misinterpreted as the equipment having completed customer acceptance and being capable of long-term, stable, full-load operation in fabs, which public information has not yet demonstrated.
2. Numbers like "supports 28nm," "domestic content rate over 85%," and "mass production yield over 90%" are also circulating online. Regrettably, we have not seen a complete test report published by the equipment manufacturer, customer fabs, or authoritative agencies, nor have we seen the statistical basis for these numbers.
3. Particularly "90% yield"—does it refer to equipment factory acceptance rate, single-layer exposure pass rate, wafer yield, or full-chip yield? The differences are enormous. Lacking test conditions, product type, wafer quantity, and customer acceptance criteria, this number currently lacks sufficient analytical value.
Therefore, this article does not treat these unverified data points as established facts.
What Exactly Has Been Made This Time
Lithography machines mainly fall into two categories: DUV and EUV. This achievement is in DUV.
What's the difference between DUV and EUV? In terms laypeople can understand, it's like the difference between ordinary light and X-rays.
Of course, the DUV category also includes KrF equipment using 248nm light sources, ArF dry equipment using 193nm light sources, and the more advanced 193nm ArF immersion equipment.
Immersion lithography involves adding a layer of high-purity water between the projection lens and the wafer to improve resolution by increasing the numerical aperture. ASML's current immersion DUV equipment can be used for mature processes and, through multiple patterning techniques, also participate in advanced process production.
EUV uses extreme ultraviolet light with a 13.5nm wavelength. Since most materials absorb EUV light, EUV equipment cannot use traditional lens systems and must use multilayer mirrors in a vacuum environment for optical path transmission.
This means moving from DUV to EUV involves comprehensive changes in light source, reflective optics, vacuum systems, masks, photoresists, contamination control, precision stages, and system control, representing a massive leap in technical complexity.
Therefore, progress in domestic immersion DUV primarily fills the gap in the supply capability of advanced DUV equipment domestically. It can enhance supply chain security for mature processes and some advanced processes but does not directly represent the resolution of the EUV challenge.
What Roles Do DUV and EUV Respectively Play
In modern fabs, DUV and EUV will coexist long-term.
A single chip typically contains dozens or even more layers of patterns. The linewidth and precision requirements are not the same for different layers. The most critical and finest few layers will use EUV, while a large number of relatively less demanding layers will still use DUV.
ASML has also clearly stated that EUV is primarily responsible for the most complex and critical pattern layers in advanced chips, while a vast number of other layers continue to use various types of DUV equipment.
DUV capabilities also cannot be simply understood as "only capable of making mature chips."
Through double, quadruple, or even more complex multiple patterning, DUV can extend the process to the 7nm level. In the past, SMIC's ability to manufacture 7nm-level chips without EUV equipment relied precisely on imported immersion DUV equipment paired with complex multiple patterning techniques.
TSMC's initial N7 process primarily used DUV; it was the N7+ process entering mass production in 2019 that became TSMC's first large-scale EUV process.
The issue is, the more advanced the node, the more exposure passes, mask counts, overlay precision requirements, and process steps DUV requires, lengthening the production cycle and increasing defect probability and manufacturing costs.
The primary value of EUV is reducing complex DUV multiple patterning steps for some critical layers, thereby lowering mask count, process complexity, and defect risk. ASML's annual report shows EUV can help customers convert some complex multiple patterning into relatively simpler single patterning.
Therefore, even if domestic immersion DUV passes customer verification, it mainly addresses China's ability to independently obtain high-performance DUV equipment. It may have strategic significance for 7nm-level processes but remains completely unable to solve the economic viability issues for mainstream advanced processes like 5nm, 3nm, and 2nm.
How Significant is the '5 Units' Volume
Looking solely at 5 units, it's easy to fall into two extremes.
Some think the number is too small to be meaningful. Others believe that just making them means ASML's monopoly is over.
Both judgments are oversimplifications.
According to ASML's annual report, ASML sold 131 ArF immersion lithography machines in 2025. In Q2 2026, ASML announced its current annual immersion DUV capacity for 2026 is approximately 130 units, with plans to increase it by about 30% in 2027.
Roughly comparing by unit count, the 5 units planned for domestic production this year represent about 4% of ASML's annual immersion DUV sales.
Moreover, lithography machines cannot be simply compared on an equal unit basis.
What truly determines wafer capacity includes metrics like wafers per hour throughput, overlay accuracy, equipment availability, continuous operation time, defect rate, maintenance cycles, and compatibility with existing production lines.
Public parameters for a current ASML immersion DUV model show throughput can reach 330 wafers per hour, with machine matching overlay accuracy around 2.5nm. Comparable complete parameters for domestic equipment have not been publicly disclosed yet.
Therefore, the primary significance of these 5 units in the short term is entering real production lines at SMIC, Huahong, and CXMT to verify whether the equipment can operate stably, integrate smoothly with etching, thin-film deposition, cleaning, inspection, and metrology equipment, and secure follow-on orders from customers.
If any single key metric fails to meet mass production standards, the equipment could remain stuck in the trial and validation phase for a long time.
From this perspective, the 5 units are the starting point of industrialization, not yet forming a capacity that can large-scale replace imported equipment.
It's Inappropriate to Pile on Rumors About SMEE
There's another narrative in the market that SMEE's SSA800 series is already in full mass production, with annual capacity reaching 50 or even 120 units.
These data exist in multiple versions and also lack clear company announcements, customer acceptance reports, and fab purchase disclosures.
As of May 2025, when Reuters investigated China's semiconductor equipment industry, it still described SMEE's commercially available front-end lithography equipment capability as the 90nm level. At that time, although SiCarrier had applied for several DUV-related patents, it had not publicly displayed a complete product.
It's entirely possible for technological R&D to make progress within a year, but until new formal evidence is available, it's inappropriate to directly add the rumored SSA800 production volume to the reported 5 units, let alone conclude that "China can already produce hundreds of immersion DUVs annually."
For highly complex equipment like lithography machines, formal release, manufacturing completion, entering customer fabs, passing acceptance, and achieving stable mass production are completely different stages.
A Larger Gap Remains Regarding EUV
China's EUV R&D has made some engineering progress.
A Reuters investigation in December 2025 revealed China built a domestic EUV prototype in Shenzhen. This machine was already capable of generating extreme ultraviolet light and had entered the testing phase, but had not yet produced functional chips at that time.
The related project proposed using the domestic prototype to manufacture working chips by 2028, while individuals close to the project considered 2030 more realistic. The report also pointed out that China still faces significant difficulties in core areas like high-precision reflective optics.
This indicates domestic EUV has taken an important step, with engineering feasibility gradually being verified.
However, moving from "capable of generating EUV light" to "capable of stably manufacturing advanced chips" requires solving a series of issues: source power and stability, mirror precision, masks, photoresists, contamination control, overlay accuracy, throughput, and long-term reliability.
ASML spent nearly two decades and invested tens of billions of euros in R&D from its first working EUV prototype to achieving commercial mass production.
Domestic EUV may accelerate its catch-up in the future, but it currently remains in the prototype testing phase and should not be discussed at the same level of industrial maturity as immersion DUV, which has begun initial production.
Why the "DeepSeek Moment" Analogy Doesn't Hold
"The DeepSeek moment for lithography machines" is catchy but can mislead investors.
The impact brought by DeepSeek mainly stems from algorithm architecture, training methods, and engineering efficiency. Once a software solution is proven effective, other teams can quickly read papers, run code, tweak models, and verify results, with iteration cycles typically measured in days or weeks.
Lithography machines deal with precision manufacturing and complex physical systems.
Whether a machine can maintain overlay accuracy long-term requires validation through a large number of wafers and extensive runtime. Source stability, lens lifetime, vibration control, contamination control, stage fatigue, component consistency, and maintenance systems all require data accumulation from real production.
Occasional successful exposures on a prototype are entirely different engineering requirements from stable 24/7 operation.
Therefore, this domestic DUV progress is closer to a long-term engineering project entering the industrial validation phase. It relies on sustained investment, talent accumulation, supply chain磨合 (磨合磨合), and collaborative debugging with customers. It's unlikely to achieve an industrial reversal instantly through a single technology announcement.
The real turning point for China's lithography machines in the future might not manifest as a sensational press conference. It's more likely to appear as a continuous increase in customer acceptances, steadily improving equipment availability, expanding repeat orders year after year, and gradually decreasing import ratios for key components.
How Far Has Import Substitution Progressed Now
The domestic substitution of China's semiconductor equipment has progressed rapidly in recent years, but the gap varies greatly between different equipment categories.
Reuters' 2025 investigation showed that China's domestic substitution rate had reached about 50% for some equipment like photoresist removal and cleaning, while the overall self-sufficiency rate for equipment supporting 7nm and below processes remained below 10%. Lithography equipment is one of the most pronounced shortfalls.
On the other hand, Chinese fabs have accumulated a substantial stock of imported DUV equipment.
A 2026 report by the American Enterprise Institute estimated that Chinese customers purchased about 70% of ASML's global immersion DUV equipment in 2024, possibly close to 90 units. These imported machines form an important foundation for China's 7nm-level chip manufacturing.
It's important to note this report comes from a US policy think tank with a stance advocating for expanded export controls; its capacity estimates cannot be directly taken as neutral industry statistics.
But it at least illustrates one point: China's current mature and some advanced process capabilities still heavily depend on previously imported ASML equipment stock. Even if all 5 units of domestic equipment planned for this year pass verification, they are unlikely to replace the already established scale of imported equipment stock in the short term.
Its strategic value lies primarily in establishing a second source for the future, reducing the singular dependence of new capacity on imported equipment, and accumulating engineering experience for more advanced domestic equipment.
What This Means for Investors
First, it must be stated that this news is a positive long-term signal for the domestic semiconductor equipment industry chain.
It shows that years of investment in light sources, objective lenses, stages, control systems, metrology systems, and system integration are beginning to show signs of moving from R&D to customer sites.
However, investors cannot yet conclude, based on a single piece of news, that domestic equipment companies are about to see large-scale volume ramp-up, or that ASML's global competitive advantage is about to vanish.
ASML sold 131 immersion DUV and 48 EUV units in 2025. In 2026, the company still plans to expand both DUV and EUV capacity, indicating strong global demand from advanced logic and memory customers remains.
Domestic equipment may receive policy, capital, and customer synergy support in the Chinese market. But after entering fabs, it ultimately must withstand the test of production efficiency and manufacturing cost.
Over the next two to three years, the metrics truly worth tracking include: whether customers complete formal acceptance; how many wafers per hour the equipment can process; whether overlay accuracy can be maintained long-term; whether equipment availability meets commercial production requirements; whether defect rates are stable during continuous operation; and whether initial customers are willing to place second and third batches of orders.
It's also important to watch whether key optics, light sources, precision motion systems, and control components can gradually achieve domestic substitution, and whether after-sales service teams can perform repairs and parts replacements quickly.
For specific listed companies, it's necessary to verify exactly what components they supply, whether they have secured formal orders, how much revenue and profit the related business can contribute, and whether order growth can cover R&D and capacity expansion investments.
Final Judgment
Based on available public information, this progress is more appropriately defined as a step for domestic immersion DUV, moving from R&D and prototype validation towards initial production and customer introduction.
Planning to produce about 5 units is a limited number, but sufficient to initiate real fab line validation. As long as the equipment can pass customer acceptance and secure repeat orders, its significance will far exceed the 5 units themselves.
At this stage, it is insufficient to alter the global lithography competition landscape or to prove China has already solved the advanced process lithography equipment bottleneck.
A real industry turning point requires seeing several consecutive signals: stable operation of the first batch of equipment, customers placing follow-on orders, annual production volume increasing from single digits to dozens or even hundreds of units, key parameters approaching international mainstream levels, and import dependence for core components continuously decreasing.
Until these indicators materialize, the most reasonable attitude is to acknowledge progress, respect engineering realities, and avoid prematurely converting technical news into industrial victory and investment returns.
As for the so-called "DeepSeek moment for domestic lithography machines," it's still far from having arrived.
