Although the chip factory is not a big problem, Wang Yi can save billions because the Imperial City provides it.

But the production lines, photolithography machines and other equipment inside have to be handled by Wang Yi himself.

These are the most expensive ones, and a conservative estimate is that the subsequent investment will start at 15 billion.

For example, in 2010, ASML launched the EUV lithography machine TWINSCAN NXE:3100.

In the previous life, ASML launched the EUV lithography machine NXE:3300 in 2013.

However, Wang Yi will not consider these advanced EUV lithography machines because they are unrealistic and impossible to sell to mainland companies.

In the previous life, SMIC ordered an EUV lithography machine and spent 200 million US dollars. All the money was paid. In the end, because of the empire's obstruction, ASML also delayed delivery.

On the contrary, what mainland companies can buy are DUV lithography machines that are many generations behind.

For example, the NXT 1980Di released in 2016 is one of the most advanced lithography machines purchased by mainland companies in the previous life, and it is also the main model of SMIC.

The more advanced DUV lithography machine NXT:2000i is not available, let alone NXT:2050i.

NXT 1980Di was originally used to manufacture the 300nm-38nm process, and can reach the 19nm process after double exposure.

SMIC, Huali and other companies all use NXT 1980Di double exposure to mass-produce 28-19nm chips.

Similarly, if you want to use NXT:1980Di to manufacture 14nm and 7nm chips, you have to rely on multiple exposure technology. The technical requirements for manufacturers are very high, and the cost of mass production of chips is also very high.

Initially, TSMC used NXT:1980Di to manufacture 7nm chips, but the results were unsatisfactory, with low yield and high cost. Finally, they replaced it with EUV lithography machines to achieve perfect 7nm.

However, SMIC cannot buy an EUV lithography machine, so it can only use NXT 1980Di to perform triple exposure, quadruple exposure, or even quintuple exposure.

Fortunately, Mr. Liang has outstanding abilities and is very knowledgeable in the field of multiple exposures. He just launched the N+1 and N+2 processes and used a backward lithography machine like NXT 1980Di to achieve 14nm, 10nm, and even 7nm.

Liang Lao's double exposure technology has not only increased SMIC's 28nm yield rate from 50% to more than 90%, but the cost will not increase too much.

This enables SMIC's 28nm process to be mass-produced on a large scale and can meet the needs of most chips.

14nm requires triple exposure. With Liang Lao's technology, the yield rate and cost can be guaranteed.

But the 7nm process requires the use of NXT 1980Di quadruple exposure. In terms of technology, Mr. Liang can solve it, but the yield rate and cost are hard to say.

Make bricks without straw.

It is really difficult to use NXT 1980Di as EUV3300.

However, this is also a last resort before the breakthrough of domestic lithography machines.

Of course, the 2011 NXT 1980Di has not been released yet.

The lithography machine models that Wang Yi can buy at the moment are the NXT:1950i, which was launched in 2009, and the NXT:1960Bi, which was just launched this year.

In fact, 1950i/1960Bi/1970ci/1980Di, these four lithography machines all have the same light source system. A single exposure is above 38nm, and multiple exposures can produce 28nm, 16nm, 14nm, 12nm, and even 7nm.

The subsequent NXT:1960Bi, 1970Ci, and 1980Di were only partially upgraded on the NXT:1950i to improve the yield rate and accuracy. In essence, they are not much different.

For example, Samsung TSMC's 28nm, 22nm, 20nm, 16nm, 14nm, and 10nm in the next few years, many of them will use NXT:1950i!

After all, when Samsung and TSMC mass-produced 10nm chips in 2016, the NXT:1980Di lithography machine began mass production!

Before 7nm, Samsung TSMC used old models such as 1950i, 1960Bi, and 1970Ci.

As long as 7nm is not mass-produced, there is no pressure.

Similarly, if Wang Yi's 12-inch wafer fab buys NXT:1950i and 1960Bi, it can also use 22nm, 20nm, 16nm, 14nm, 10nm, and even 7nm from next year's 28nm.

But the 1960Bi just released this year is not realistic, and ASML probably won’t sell it.

However, if the old 1950i is put into operation, there is still hope.

At present, the domestic blockade restrictions in Europe and the United States are not that strict.

If you take advantage of the opportunity and order more batches, you will have peace of mind in the future.

Even if you buy it at a premium, it won’t be a problem!

Wang Yi remembers that EUV lithography machines cost US$200 million in the previous life and US$72 million in 1980.

The current 1950i is probably cheaper.

For this reason, Wang Yi took out his mobile phone and started checking the data.

Sure enough, a 1950i lithography machine costs 40 million euros, 54.7 million US dollars, or 500 million yuan.

If you enter ten units, it will be 3.5 billion.

20 units is 7 billion!

Why did you invest so much at once? It’s not because you’re afraid of getting hammered if you get bigger!

Of course, take advantage of the fact that the limelight is not strong at the moment, so hurry up and make more!

This will be enough for subsequent production capacity expansion, and 1980di will no longer be banned for use until 2018.

Of course, you can’t just suppress Asmai, you also have to look at the little days.

Originally, the three giants of photolithography machines were ASML, Nikon, and Canon.

Now Canon has given up on high-end, staying at 55nm, and can't do anything higher.

ASML not only leads the field of DUV lithography machines, but also begins to develop EUV lithography machines and enter processes below 7nm.

And Nikon is still struggling. There is no hope in the high-end field, but there is still no problem in the mid- to low-end field.

For example, Nikon's NSR-S620D and ASML's 1950i are not much different in performance. They can also do 10-28nm or even 7nm after multiple exposures.

And the price is cheaper, only $30 million.

Intel's current orders are mainly for Nikon NSR-S620D, with a small amount of ASML 1950i.

Similarly, Wang Yi can also bet on both sides.

If ASML doesn't ship much and things don't go well, Nikon can definitely use it.

It was not until below 7nm that Nikon completely fell behind.

At 28-10nm, Nikon can still compete.

Wang Yi had some concerns. When the chip factory situation was settled, he would have to find time to go to Nikon and ASML to discuss the photolithography machine order.

This kind of order is not available as soon as the order is placed. On the contrary, it is placed one year in advance.

Wang Yi orders now, it will be good to have the goods next year.

By then, the chip factory will be almost completed, and the equipment will be installed directly, and then installed and debugged. If everything goes well, chip production will be possible the year after next.

Of course, I’m afraid it won’t go well!

After all, if you want to buy an ASML lithography machine now, ASML will send an inspection team in advance to inspect it.

Shipments will be made only after a comprehensive assessment of the scale, production capacity, and financial status of the semiconductor company.

Ordinary small businesses and new enterprises, ASML doesn't even bother to pay attention to them.

After all, the current photolithography machine industry is an absolute seller's market, and demand simply exceeds supply.

But this kind of inspection team is easy to operate.

For example, Seoul Hongchip in the previous life, even though it was just established, even though it was a 100 billion scam, it still got the 1980Di lithography machine in 2018.

As for the reason, it was because Mr. Jiang was in charge.

Mr. Jiang has worked at TSMC for so many years and has received hundreds of ASML lithography machines. All ASML senior executives know him.

Just for Mr. Jiang’s sake, all the so-called inspections will be passed smoothly and the photolithography machines will be delivered.

Although this lithography machine was immediately mortgaged to the bank by Hongxin and obtained a loan of 800 million yuan, Mr. Jiang was also deceived and almost lost his reputation...

I can only say that Hongxin’s capital structure is too high-end.

Similarly, if Xingyi Semiconductor can order photolithography machines directly, that would be the best.

But if you can't place an order, you still have to find a big boss to take charge, so there's no pressure.

What Wang Yi values ​​most is Mr. Liang.

But unfortunately, Mr. Liang has just joined Samsung Semiconductor and has been sued by TSMC again.

This lawsuit lasted until 2015 before the result was reached. In his previous life, Mr. Liang entered SMIC in 2017.

Wang Yi wanted to recruit Mr. Liang, but it was also very troublesome.

Putting aside whether Samsung is willing to let him go, how TSMC will deal with it is a big problem!

Unless Xingyi Semiconductor, TSMC, and Liang Lao reach a cooperation, Liang Lao leaves Samsung Semiconductor to join Xingyi Semiconductor, TSMC withdraws the lawsuit, and Wang Yi gives TSMC a large order and gives up some of the benefits.

In this way, TSMC benefits and also blocks the rise of its arch-rival Samsung Semiconductor.

Xingyi Semiconductor's introduction of Mr. Liang can also lead to great development.

As for the threat of Xingyi Semiconductor to TSMC? This is not a problem.

Compared with behemoths like Samsung, Xingyi Semiconductor is not worth mentioning, and TSMC doesn't take it seriously at all.

What's more, Europe and the United States have restricted EUV lithography machines, and TSMC doesn't think Xingyi Semiconductor can grow bigger.

At most, low-end chips are produced by ourselves, and orders for high-end chips are still obtained by TSMC. Having one more big customer can be regarded as the basis for cooperation.
To be continued...