Reborn in 2010, the AI ​​in my mind is too powerful.

Chen Yiran has always been worried about the future bottleneck of lithography machines.

As someone who spends a lot of time online, he knows quite a bit about lithography machines. It's not the worst time yet; at least we can still buy DUV lithography machines. In ten years, not only will DUV lithography machines be unavailable, but some of the core components will also be impossible to buy.

The first to be affected are the objective lens and projection lens, all of which are manufactured by Zeiss. These are the soul of the lithography machine. Until 2025, China has not caught up with Zeiss's level and can only produce some low-end lines.

What's disheartening is that catching up requires ultra-precision equipment such as ion beam polishing machines, which themselves need to be imported from Japan and Germany.

As for light source systems, although they are usable after much effort, their stability is far from perfect, and EUV is nowhere near being realized.

In the field of precision machinery, dual-stage technology is ASML’s signature technology. While one stage is exposing, the other stage is performing alignment and measurement, alternating between the two stages to double the production capacity.

Although China seems to have caught up, it still relies almost entirely on Heidenhain of Germany for grating rulers.

In 2025, although the slogan of domestic production rate of more than 7% has been proclaimed, the remaining 3% has been relying on inventory of parts, but these parts will eventually run out.

Therefore, at the management meeting, Chen Yiran directly determined the company's future development direction: don't worry about economic issues, just focus on technological advancement.

Chen Yiran has plenty of money. He directly stated that in addition to the $900 million on the books and the $800 million installment payment for Kaiden over the next three years.

All the money that WanGuoYuan earns in the Chinese EDA market, as well as the money earned from its 6-inch production line, will be invested in equipment and technology.

Furthermore, if the research and development needs require excessive funds, he can borrow money as a shareholder.

Although the meeting was attended by seasoned professionals who were past the age of casual escapism, Chen Yiran had cultivated a certain charisma over the past year. Several of the industry leaders seemed to still have a certain gravitas.

After the meeting ended that day, Qiu Ciyun even sent him a detailed summary of the entire lithography machine procurement process that evening.

Buying an immersion DUV lithography machine is not a simple matter of signing a contract, making payment, and waiting for delivery. From sending an inquiry letter to the final delivery of the equipment into the cleanroom, it normally takes one and a half to two years, with three major hurdles in between.

The first hurdle is at ASML's headquarters in the Netherlands, the second is with the Dutch government, and the third is with the Chinese government.

Of course, the first and third routes are the ones that all trade will take.

However, as a high-end project, this second hurdle is quite troublesome.

Because this involves an agreement.

Wassenaar Arrangement.

The Wassenaar Arrangement, officially known as the Export Controls Agreement on Conventional Arms and Dual-Use Goods and Technologies, was established in Vienna, Austria in 1996 and consists of forty member states.

At the heart of the agreement is a control list, on which sensitive items and technologies must be approved on a case-by-case basis by the governments of each member country before they can be exported. Semiconductor manufacturing equipment, especially immersion DUV lithography machines, is on this list.

ASML cannot decide who to sell to on its own. For every NXT lithography machine exported to China, ASML must first submit an export license application to the Economic Policy Bureau under the Dutch Ministry of Foreign Affairs.

The Dutch government reviews whether the technical parameters of the equipment exceed the limits of the Wassenaar Arrangement, while also taking into account the opinions of other member states.

Of course, to put it bluntly, this is the opinion of the United States.

Then, a decision is made on whether to issue or reject the certificate; the normal processing time is four to six months.

At that time in 2011, immersion DUVs were in a gray area within the review system. The review process was slow, but if the buyer had a clean background, a clear purpose, and promised not to resell or use them for military purposes, they were generally not rejected outright.

The Netherlands does not include it in the Strategic Materials Regulation, nor does the United States include it in the Export Administration Regulations.

SMIC made large-scale purchases of ASML lithography machines in the past few years, and none of them were rejected.

This is a short window of opportunity; once it's gone, it's gone forever.

However, at the meeting, Chiu Tzu-yun, Chi Ming-hua, and Chiang Shang-yi did not initially support purchasing the latest model of lithography machine.

The latest lithography machines are the NXT1950i and 1960i.

The 1950i was the main model in early 2008.

The 1960i is ASML's latest immersion DUV model, which will be delivered next year. It boasts a resolution of 38 nanometers and an overlay precision that can be controlled within a few nanometers.

In terms of parameters, this machine is the most suitable for 28nm and subsequent 14nm process advancements.

However, the 28nm process can also run smoothly using the previous generation 1900i or even the earlier 1700i, after a few more months of process refinement.

Several industry experts even believe that the 1900i will be better to use, due to its manufacturing know-how.

There isn't a fixed Chinese translation for "technology know-how," but simply put, it means technical expertise or skill.

Many people believe that buying the most advanced lithography machine means they can make the most advanced chips, but it's not that simple.

The lithography machine is just one link in the entire chip manufacturing process chain. A 28-nanometer chip goes through about 500 to 800 processes from wafer fabrication to packaging and delivery.

Photolithography projects circuit patterns onto a silicon wafer, etching etches the patterns into the material, thin film deposition coats various material layers, ion implantation alters conductivity, and chemical mechanical polishing smooths the surface. Between each process, there are several rounds of cleaning and measurement.

The lithography machine is only responsible for the photographing step in each cycle. After the photographing is completed, it decides how to etch and how much to etch.

The gas mix ratio, the RF power setting, the chamber pressure, the temperature control, and the etching time all require subsequent process testing.

A mature etching formula may contain dozens of parameters, which cannot be found in textbooks.

Each contract manufacturer has different etching equipment, different internal cavity conditions, and different target combinations, so all parameters have to be figured out by themselves.

A deviation of even 5% in the ion implantation energy window can cause the final threshold voltage to deviate. When to stop chemical mechanical polishing (CMP) cannot be determined by a single sensor reading; it requires experience accumulated by process engineers based on data from thousands of silicon wafers.

The most troublesome thing is the mutual influence between the various processes. If the etching in the previous process is two nanometers too much, the test results of this step may be completely qualified.

However, after several dozen steps, a deviation of two nanometers will cause the overall distribution of ion implantation to shift.

If the threshold voltage is not up to standard, the chip cannot reach the target frequency, but no one can trace back to the specific step where the problem occurred.

This causal chain, spanning dozens of processes, cannot be predicted by any theoretical model; it can only be discovered through trial and error.

TSMC has accumulated 24 years of experience since it was established in 1987.

Each parameter is the result of tens of thousands of test wafers and decades of discarded wafers; it cannot be bypassed by simply buying the latest lithography machine.

This explains why, although both SMIC and TSMC possess the same immersion lithography machines as ASML, TSMC can achieve overlay precision down to a few nanometers, while SMIC cannot at the same level. The difference lies in the different process know-how applied to the same machine.

Qiu Ciyun, Ji Minghua and Jiang Shangyi's suggestions were certainly valid, but Chen Yiran still made the final decision to ask first if it was possible to buy the latest model.

The Wassenaar Arrangement currently only restricts and reviews, not prohibits.

1960i, the review process is slow but it is not rejected outright.

In a few years, who knows if you'll even be able to buy it again?

His year had a huge impact on the tech world. Due to the butterfly effect, the subsequent course of history is unreliable. Who knows when an embargo might suddenly be imposed?

If you buy 1900i or 1860i now, and by the time WanGuoYuan's technology reaches the point where it needs the most advanced immersion lithography to make 14 nanometers, the Netherlands won't sell to them anymore. Wouldn't that be like having money to burn?

The 1960i is something to strive for, but at the very least, you should buy the 1950i.

The old men seemed to understand Chen Yiran's meaning, and finally agreed, promising to do their best to overcome the technical challenges.

Qiu Ciyun was the most interesting. After sending the file to Chen Yiran that night, he called again and started sharing his experience.

"Mr. Chen, it's time for you to start acting. If you want our company to do better, you need to leverage your current American company, FlowEDA's name, and build up your image there!"