The Millennium Semiconductor Survival Guide

Now in the first decade of the 21st century, Intel is not yet the former giant whose market value was surpassed by Nvidia, whose PC market was constantly being eaten away by AMD, and who achieved nothing in the mobile Internet field.

At this time, Intel is the absolute king in the PC field, and it is also a presence that cannot be ignored in the chip field. It handles everything from chip design to chip manufacturing and has acquired a large number of companies.

Intel is definitely a company that likes to acquire other companies. From 1988 to 2006, Intel acquired more than 30 companies in almost two decades.

Especially between 1999 and 2002, they spent a total of US$11 billion to acquire companies, triggering a wave of acquisitions in the communications industry.

Especially after the Nasdaq bubble burst, Intel relied on its massive cash flow. Their management believed that these communications companies were affected by the burst of the Internet bubble, and the price was simply too cheap.

Barrett is the direct driving force behind most of Intel's acquisitions. He likes acquisitions and likes to expand Intel's business territory through acquisitions.

Even if these acquisitions are reduced to nothing, the employees of these companies who are sucked into the body are reduced to wandering warriors with no work.

For Intel, this is not important, they can support so many engineers, and even some acquisitions are simply to reduce competitors.

But Intel or Barrett are really interested in new cores.

To be precise, I am interested in new core lithography machines.

As mentioned before, Intel also has a chip manufacturing segment. Their business model is similar to Samsung, from chip design to chip manufacturing to product sales directly to consumers.

Even because the products are selling so well that they cannot produce them themselves, Intel has to hand over some of its orders to TSMC, and they hold a large amount of TSMC shares.

Therefore, in the field of chip manufacturing, before the rise of Xinxin, the main players on the market were Intel, Samsung, TSMC, Texas Instruments and IBM.

As we all know, IBM is a company that attaches great importance to profits. They will gradually throw away unimportant businesses. Therefore, they gradually began to fall behind at the turn of the century. They switched from leaving the company personally to selling patent licenses to several others who were still on the track. enterprise.

Later, Texas Instruments also slowly fell behind in advanced manufacturing processes. Of course, Texas Instruments has not fallen behind yet.

So now to explore the limits of Moore's Law, there are only five players on the advanced process table, Intel, Samsung, TSMC, Texas Instruments and Xinxin.

Intel is the absolute overlord. Before the Tokyo Semiconductor Equipment Show, they had never thought that new chips could pose a threat to them.

Xinxin's lithography machine will change everything. Barrett was keenly aware at the venue that Xinxin's lithography machine will bring them huge advantages, unless Intel does not produce 40nm chips and ignores the majority of people's interest in advanced processes. users in need.

But this is impossible, because Intel has another competitor, AMD, and they cannot watch AMD use the new core's 40nm process to create chips with better performance than Intel.

This is a huge blow to Intel.

Therefore, they thought of another way, which is to acquire Xinxin, or only acquire one of the sectors, which is the lithography machine.

The world is always changing dynamically. Xinxin is an important rival of Intel. It acquires Apple, enters the computer field, launches new Apple computer chips and smartphone chips based on ARM architecture, etc. They have competitive relationships in so many fields.

This does not prevent everyone from cooperating, just like TSMC competes with Intel, but TSMC will also help Intel produce products. Competition is competition, and business is business.

After receiving a reply from Zhou Xin agreeing, Barrett boarded a flight to Pudong International Airport that evening without spending too much time in Tokyo.

Time is important. The bigger the business, the more important it is to have face-to-face meetings.

Telephones, emails, WeChat and other communication methods have made it easier for people to communicate with each other, but face-to-face chats will always have an irreplaceable status. Only through face-to-face chats can you know the other person’s true thoughts.

“I think New Core is unlikely to accept our acquisition. It’s unrealistic to sell them to us just after their lithography machine business has achieved a breakthrough.

Newman is not short of money. You know, Newman has always been famous for not being short of money. "Otellini said.

He is the CEO of Intel. This year he led the largest round of layoffs in Intel's history, laying off tens of thousands of people. Therefore, foreign companies also have the risk of graduation.

Otellini is also an alumnus of Zhou Xin and graduated from Berkeley. He likes Apple's operating system very much. The criterion for evaluating the quality of previous Windows systems is how similar it is to Mac OS. He is a pure iron fruit fan.

On the private plane, Barrett nodded: "I know that Newman is a difficult person to convince. He is very difficult to deal with and very discerning.

But one good thing about him is that as long as we can offer a suitable price, he will never be stingy about selling the company. This is true from Quora to Weibo. "

Barrett continued: "So we have to talk, even if we can't acquire it, we can still reach a cooperation."

In Barrett's mind, being able to acquire new cores is the best, and not being able to acquire shares is the second best option. Taking shares can ensure that Intel will not fall behind in the next advanced manufacturing process.

He believes that in the long run, Nikon and ASML still have advantages, and they will regain the initiative in the EUV field, but that will take time, and Intel needs technology that can be used immediately.

Barrett seemed to remember something: "Paul, the Shenhai Zhangjiang area we are going to this time is the area where Xinxin's headquarters is located. Isn't there a large number of semiconductor companies gathered there?

I read on Quora Encyclopedia that it is similar to Silicon Valley, full of semiconductor companies. Why is Intel's semiconductor factory not located here, but in Dalian? "

Otellini's full name is Paul Otellini. Under his leadership, Intel will invest US$3 billion to build a semiconductor factory in China next year, with the specific location in Dalian.

The cluster effect in the semiconductor industry is very important, and clusters can effectively reduce costs.

"Because Dalian has an outlet to the sea, it is convenient for us to transport the produced chips to where they should go."

The factory that Intel will build in Dalian is a memory chip manufacturing plant, which does not have such high process requirements. It was packaged by Intel in 2020, integrated into its memory chip business, and sold to Hynix.

Barrett asked: "Don't Shenhai also have a port?"

Otellini said: "Because the factory we plan to build in Dalian mainly produces memory chips, it requires more cooperation with Korea's semiconductor companies, while Shenhai's semiconductor industry is more about logic chips than memory chips."

This is a perfect explanation. Ever since Americen sold memory chip technology to Korean companies, Korean companies have been very strong in this area.

By now, Neon's memory chip manufacturers are almost bankrupt. The upcoming financial crisis will kick all of Neon's semiconductor manufacturing companies into the trash can of history, never to return to their original position.

Barrett didn't ask any more questions. This answer was enough. He then closed his eyes and thought about how to negotiate with Zhou Xin tomorrow.

Zhou Xin needs Intel's support. To be precise, he needs time, and Xinxin needs time.

Barrett believes that the development of EUV lithography machines will only take two or three years to complete, because Nikon has already produced prototypes in the laboratory, and ASML has also produced laboratory samples.

Zhou Xin knew that it would take five years at the earliest. Five years would be enough for Xinxin to advance the process of immersion lithography machines to 7nm.

Zhou Xin is ready to be surpassed in the field of EUV lithography machines, because whether it is Nikon or ASML, behind them is a government-led consortium of enterprises, research institutions, and universities, which took more than ten years to master the technology.

The reason why Nikon did not win is not only because behind ASML is a coalition of European and American research institutions, and the accumulation of technology is deeper than theirs, but more importantly, because after the financial crisis of 2008, all chip manufacturing companies of Neon disappeared, and the equipment they built Unable to enter the production environment for actual verification, it will always stay in the laboratory.

Xinxin has advantages in chip manufacturing, but Xinxin also has a disadvantage, that is, they have almost zero accumulation in EUV. Even with Zhou Xin's guidance, they could not develop it in six years.

So what benefits can be obtained from Intel this time is crucial.

It wasn't until 2012, when Intel invested US$4.1 billion in ASML and acquired 15% of ASML's shares, that they had more say.

After this round of investment, ASML’s equity structure changed to Intel 15%, TSMC 5%, and Samsung 3%. Intel is not yet a major shareholder in ASML, they only have a negligible stake at this time.

Intel is only a powerful partner at this point.

The relationship between TSMC and ASML should be closer.

Interestingly enough, when EUV was still in the laboratory stage, TSMC said they would never use EUV.

Why is TSMC starting to use EUV technology again? Because Apple is coming, Apple is coming with a check, and we need EUV technology to make chips with better performance.

ASML negotiated with TSMC management and reached a deal. TSMC almost overnight changed from a company that did not believe in EUV to a company that absolutely supports EUV, and became the largest customer and user of EUV lithography machines.

"Craig, I didn't expect you to come to see me. It seems that Xinxin's immersion lithography machine has performed well in Tokyo and attracted you." Zhou Xin said with a smile.

He was very familiar with the two people in front of him. Craig Barrett had met him several times when he was in Silicon Valley. He had met him at the Donkey Party’s fundraising dinner, at Microsoft’s celebration dinner, and also during business negotiations. Pass.

Paul Otellini may seem unfamiliar, but in fact he is very familiar. In Zhou Xin's previous life, Otellini retired in 2012. After retirement, he often went to Berkeley to give lectures. Zhou Xin attended and had a long talk with Otellini.

Barrett said: "Newman, is this how low-key the Chinese are?

Xinxin always gives us a huge shock inadvertently. When the industry is thinking about how to cross the 65nm process, Xinxin suddenly gives its own answer.

Xinxin has a group of outstanding engineers who have helped the semiconductor industry take a big step forward again.

It's the chip industry's luck that we're not stuck with 65nm. "

Zhou Xin said: “Indeed, this is the credit of the engineers.

We were lucky enough to discover the infiltration method and stick with it. "

Barrett was helpless: “We have been supporting ASML and hope that they can make technological progress, but even with the support of us and TSMC, ASML still lags behind Nikon in technology research and development.

Now it lags behind Xinxin and Nikon. "

After a brief exchange of greetings between the two parties, Barrett went directly to the topic: "Newman, Intel wants to acquire Xinxin's lithography machine business.

We will give you a price that is sufficient to satisfy you.

20 billion US dollars, half stock and half cash, what do you think? "

Barrett didn't want to beat around the bush and showed the price directly.

If this price is offered to Internet companies, it is too low. If it is offered to semiconductor companies, it is absolutely fine. How much does a photolithography machine sell in a year? A business with an annual revenue of less than 5 billion US dollars. , profits are pitifully low, and R\u0026D investment is staggeringly large.

(The annual revenue of lithography machines in 2006 is estimated based on ASML’s annual revenue of 5.245 billion euros in 2012)

Lithography machines are not a good business.

In other words, semiconductor equipment has never been a good business. The only advantage of this business is that after you reach the top level, the moat is deep enough and you can maintain it unless extreme events occur.

Nikon's technical advantage has been maintained for 25 years, and ASML's technical advantage has also been maintained for 20 years. Unless there is a technological dimension reduction attack or off-site forces, their advantage can be maintained for even longer.

ASML defeated Nikon by relying on off-site power and the dimensionality reduction attack of EUV technology, both.

Zhou Xin said: “Craig, you know, I’m not short of money, this little money can’t impress me.

Let's change the price, share swap, use 20 billion US dollars as the valuation of the new core lithography machine, I will take 30% of the shares, which is a value of 6 billion US dollars, in exchange for Intel shares.

I understand your intention, and Xinxin also hopes to achieve a closer cooperative relationship with Intel.

I think you also know that I will not sell Xinxin’s lithography machine business. Not before the immersion lithography machine is successfully developed, and even less likely after it is successful.

Equity swap, this is the most reasonable way. "

After Zhou Xin finished speaking, Barrett and Otellini both fell into thinking.

He didn't speak, and after about two minutes, Zhou Xin continued: "Xinxin lithography machine is doing well under Xinxin's management team and has surpassed Nikon.

Since Xinxin Tuan is good, let Xinxin Tuan continue to work, instead of being like those companies acquired by Intel. Most of them disappeared after being absorbed into Intel, and no such person was found.

Intel is too bloated, and it is not conducive for teams other than Intel's core business to enter. "

Zhou Xin would not tell the real reason why he did not want to sell the new core lithography machine. He found a good excuse.

Barrett shook his head: "If it is an equity swap, then the scarcity of Xinxin's lithography machine business will be lost.

It is impossible for us to give such a high valuation to Xinxin lithography machine. "