Reboot 2003

Chapter 319 100: Foundry (two-in-one)

Mr. Li probably just said it casually, but the chip does seem to be a good choice.

Not even just good.

Not only is it an asset-heavy industry, but one of the most important industries hit by the current economic crisis is semiconductors. Semiconductors are production tools. They will decline when the economy is not good, and this year happens to be a surplus cycle in the market cycle. In short, the market situation It's so bad right now, it's just the right time to buy the bottom.

And today’s incident, after all, is not a preview of future trade wars.

In the past few days, I have felt the pressure of the next version in advance. As for the trade war, other problems can be solved, but the chip ban is really unavoidable.

But Chen Qin had actually thought about this issue before.

He doesn't understand semiconductors himself, and embedded ones are enough to give him a headache.

But the so-called "high position", speaking of this industry... he probably knows a little bit.

"The scope of semiconductors is quite broad. Mr. Li, what kind of chip do you want to make?"

Even Ji Xianglin is aware of this problem. Mr. Li said: "I used to advocate the need to develop self-developed computer chips at Lenovo. Later, after leaving Lenovo, I participated in several self-researched chip projects. Although I did well, at present The progress is not very good, but now, I think high-end semiconductors are better. This is an opportunity for industrial upgrading for our country, and it is also an additional choice for domestic companies..."

He couldn't stop talking, but he quickly added: "Of course I'm just a suggestion, and Xiao Chen and I have already started working on this thing before."

Ji Xianglin was a little surprised: "Already bought it? Is it the investment start-up team?"

"No." Mr. Li said, "We made an open source instruction set..."

Ji Xianglin has never heard of this: "I don't quite understand this. What is the instruction set?"

"The instruction set is almost an upper-layer code on the chip. Whether it is an operating system or software, it needs an instruction set to complete its functions. For example, Intel's chip instruction set is X86, but their instruction set is attached to the chip. It is regarded as the connecting component between the chip and the system..."

As Mr. Li explained, Ji Xianglin seemed to have flipped through the manual at hand: "Oh, yes, the instruction set of our Kaitian mobile phone turns out to be ARM, right?"

At this time, Chen Qin said: "Yes, ARM is currently a ToB semiconductor company that provides low-power instruction sets and supporting technical solutions. We are now the leader of the Kaitian ecosystem and cannot be completely controlled by others, so we recently joined forces Mozilla, provides an open source instruction set called Risc-V."

"What a great idea. Open source is really easy to use." Ji Xianglin also understood. "That means those who make Kaitian mobile phones in the future can directly use our open source instruction set, so they don't have to pay ARM?"

Mr. Li said: "That's almost the case, but open source itself is not a panacea. For a mature market, it is usually 'second-rate companies doing open source'. It is currently only a free alternative, and it will take time to develop."

Chen Qin also said: "You can think of it as the game engine we are making. ARM's strength lies in their technical services supporting the instruction set. This business model is their first. The core is that they can allow chip companies to reduce costs. , and finally use the revenue to pay for technical services and licensing fees.”

Ji Xianglin pondered: "It sounds good. How about we consider this ARM company? Although there are alternatives, if we acquire it and join forces..."

Mr. Li pondered, but Chen Qin said: "No need, Risc-V will eliminate them sooner or later, our business model is more advanced."

Ji Xianglin was curious: "Isn't this an open source architecture?"

Chen Qin explained: "The technical solution provided by ARM essentially helps companies skip the workload of underlying instruction set design, but on the one hand, their R&D capabilities cannot cope with different customer needs. After all, not every customer is Those who make mobile phones or computers, on the one hand, their ecology is actually very closed, customers can only pay to accept it, but they cannot participate in development."

"And our Risc-V can be used by more people. Maybe some customers are willing to develop a technical solution by themselves. After success, they can sell this technical solution to others. Over time, it will definitely be more efficient than ARM's solution." Rich."

"At the same time, more developers can gain fame and income by developing the tools of our chipsets. Now Mozilla, which we are cooperating with, has started this kind of business, and even ARM can also invest in the development of our instruction sets in the future, so our Risc-V will definitely become one of the most mainstream instruction sets in the future."

Ji Xianglin was a little confused after hearing this: "I thought this was a technical problem..."

"This is a business issue." Chen Qin said, "It is true that most of the problems in the semiconductor field are business issues, so ARM should forget it."

Chen Qin talked for a long time and dismissed ARM, one of the two mainstream architectures in the future, as useless.

I didn’t mention the main reason. After all, ARM technology is really difficult, and relatively speaking, there is really no money to be made.

But these things just now were not his nonsense and deception - as a person from the future and in the technology industry, he has naturally heard of chip issues to some extent.

So last month, in addition to being idle, he also thought of a related solution. In fact, his first target was ARM, but ARM is still too small. This small means not only its small scale, but its small share. After all, mobile phones are a big piece of cake, and competition for its share has been fierce since the 90s. The current pattern has almost been delineated.

The ARM segment is really small - Qualcomm and Nokia have finished the war of the gods and have withdrawn from the market and become a net patent holder. They are the largest players on this big cake table, and ARM is under its camp. One of the minions.

Specifically, the Qualcomm tax on future mobile phones will be 5%. It will also be the absolute overlord of baseband chips and SOC chips, and the price will be calculated separately.

The ARM tax is calculated based on a fixed fee. A CPU costs a little bit. If you have to say that the base is large, it may indeed be a lot. However, if you have a conscience and say that it provides so much technical support, just take this share, compared with those patents. The gangster doesn't know where his conscience is.

So from an investment perspective, this company really doesn’t have much to offer – not only is its revenue not that good, but by 2023, ARM’s annual revenue will be just that much... According to the standards of rebirthers, ARM’s profits are less than Genshin Impact is higher, but he is almost on the same level as Li Jiaqi.

And I don’t have much say. I’m just a stinky part-time worker, which is really embarrassing.

So if you want to ensure controllability, there is really no need to make one yourself, so Chen Qin pulled out Risc-V - it is also a low-power, micro-kernel, and embedded-core open source chip architecture, so it is called this This is because there were four big brothers before, but since the war between microkernels and macrokernels in 90, this branch has been in a state of decline. According to history, it took about two years for it to appear, and after it appeared, it could not beat the first starters. Advantages of ARM.

But things are different now. Embedded systems are the absolute downstream of chips. Chen Qin told Mr. Li this before when he was working on the Kaitian system. Now that we have a fulcrum and move upstream, the first step we encounter is the instruction set. , making Risc-V is almost a way for other open source alternatives. Even if there is no forced binding, it should be able to have an alternative choice in the mobile phone instruction set, and it can also provide a fulcrum for continuing to move upstream.

"Well, since the architecture instruction set is not good, how about investing in something else?" Ji Xianglin also realized the complexity of this industry. He seemed to be searching for Bloomberg's industry reports while talking, "I read that the chip industry is currently mainly There are three models, one is IDM, which is a vertically integrated comprehensive enterprise, one is a foundry like TSMC, and the other is fabless... Which one is better?"

This time Chen Qin said directly: "If you want to ensure industrial safety, you can only choose a foundry."

"Isn't IDM okay?"

"It's not very realistic." Mr. Li said this time, "Of course I think design companies are also very good. After all, we have SMIC. Now we all know that only chips and no ecology are not enough. We now have developers." TianOS and Risc-V, then continue to move upstream, the next stop should be the design company..."

Chen Qin said: "Fabless, which specializes in design, can also do it, but I don't have a very good goal. On the contrary, the foundry has a better choice..."    Li Lao was curious: "As for TSMC, they have cooperated with the Baodao Provincial Government. The relationship is very close, I'm afraid you can't buy it..."

"It's not TSMC." Chen Qin said, "AMD is currently preparing to sell their wafer fab and its supporting production technology. We can find a way to buy it and then operate it as a foundry to produce mobile phone chips in the future. There will definitely be a lot of foundry demand, and this fab can focus on this type of foundry and replace TSMC's position in the industry."

Ji Xianglin was curious: "How is its technology? Compared with TSMC? Why does AMD want to sell it?"

"The technology is a little worse than TSMC, but not by much." Chen Qin said, "It was originally AMD's wafer factory. AMD had to spend money to develop its own chips and chip manufacturing processes. The cost was too high. , in order to maintain its leading position, AMD can only sell it."

"Then if we buy it, can we surpass TSMC?"

"We can find a way. Now that we have Kaitian and Risc-V in our hands, design manufacturers who want to enter related fields can become our potential customers. The fab we bought should have a chance to surpass TSMC."

"Yes, there are many areas of overtaking in corners in our country. Maybe we can surpass TSMC in the future..."

Ji Xianglin seems to understand, although there are still some things that he doesn't understand, such as why IDM can't work, and how can he surpass TSMC by buying a factory that has been eliminated by AMD.

Mr. Li was hesitant to speak. He probably really wanted to have a more powerful domestic Fabless company.

But he didn't say anything in the end. He just asked if he needed anything, and revealed that their bureau is also discussing how to solve the semiconductor problem. If Chen Qin is willing to step forward, he can get some resources from the country...

So the matter was settled - Ji Xianglin later volunteered and said that he hadn't returned to China yet and could go to AMD to ask about the situation. Since acquiring the GPU design company ATI, AMD has been having a really hard time in recent years. It is almost heading towards the abyss of being an anti-monopoly pendant for both parties in the future. Selling the wafer fab is now a last resort. Recently, it is indeed seeking to take over his family's wafer fab.

Chen Qin did not reject Mr. Li's proposal this time. He said that when the time comes, the domestic wafer fabs will need to participate together. It may not be a good idea to grab orders from SMIC, but since they have already bought it, there is no reason not to do it; besides that, It is probably necessary to open a factory in China. In the final analysis, the capital structure of a company is just a layer of flower shirts. It depends on whose family it belongs to and where its assets are allocated. This aspect obviously also requires some efforts from the Chinese government. force.

After this discussion, the tone for the follow-up actions of this incident was officially set.

In short... I need to buy a chip factory.

Then I want to buy AMD's wafer fab. After buying this, I have no other plans.

Does it feel a bit sudden?

"Mr. Chen, when we buy a wafer fab, neither the business nor the technology at hand have anything to do with it. It doesn't feel very integrated." This is what someone in the company said.

"Old Chen, are you sure you want to buy this? Although their asking price is not high, the current purchase price is only three billion US dollars, but their financial report is very bad, and there is an oil man from the Middle East bidding with us, so I asked When I asked them if they really wanted to buy it, they raised the offer to 40 billion on the spot..." Ji Xianglin, who had been optimistic before, soon became doubtful.

Then... even the girl sitting next to me felt unreliable:

"So after you talked so much that day, did you finally want to buy a chip company?"

In the self-study class at school, A-Yue didn't know why, but she seemed to have a lot of free time today, so she came over to look at the information Chen Qin was browsing on his mobile phone, and asked after looking at it for a while.

"How did you know?"

"Because you've been reading this kind of information every day lately."

"That's right." Chen Qin put down the phone, "Recently, I... and the people from Galaxy Software are still planning to invest here. Isn't it boring?"

A Yue quickly shook her head: "No..."

"Hey, I know that although cutting-edge technology is good, you still have to consider the daily necessities of life. Chips are the infrastructure of computer technology. It won't work if you don't do it."

A Yue said: "I don't think the chip is bad. For example, your mobile phone needs a chip."

"We can't make mobile phone chips." Chen Qin said, "All complex industries will be infinitely subdivided into a large number of branch industries. The main technology for mobile phone chips comes from design companies, followed by communication companies. This wafer factory is just I am a OEM for others, but the OEM technology is not the best..."

A Yue was curious: "That's good too. Maybe after this step, we can continue to do more. Don't you often say that you want to expand upstream and downstream..."

"That thing can't happen elsewhere," Chen Qin said, "This is the most boring business. As long as we take this step, no matter how the industry develops in the future, we can only do OEM work for others. If we make a mistake, …”

"What will happen?"

"It's not going to be a big deal." Chen Qin said, "Probably just like a salted fish, I can only go around doing OEM orders for AMD, burning a lot of money, and then I can only be the second in the world..."

A Yue didn't know why he was so sure, but Chen Qin was very sure about it. After all, he wanted to invest now, and he had chosen this AMD fab, which was actually quite famous in the previous dimension.

It will be GlobalFoundries in the future, also known as GLOBALFOUNDRIES, the king of unfinished products.

It was separated from AMD and acquired by oil guys who were interested in investing in high-tech industries.

Although it will be the world's second-largest foundry in the future, from this perspective, this wafer fab probably does seem to be a good target for picking up leaks.

But semiconductors are still not that simple -

A Yue was curious: "Why is the second largest company in the world just dawdling? And why will it become a salty fish if it expands upstream and downstream?"

She finally asked the right question, but Chen Qin did not answer immediately: "This question is not simple. First of all, do you know Moore's Law?"

Of course Ah Yue has heard and seen it: "Well, it's the law that the number of transistors on a circuit board will continue to increase every 18 months, right?"

"Yes, for the semiconductor field, Moore's Law is the answer to all basic questions. But as I told you before, it means that the performance of mobile phones will continue to increase. This is Moore's Law in the eyes of developers and users."

A Yue was confused: "Then what other perspective is there?"

“Of course it’s from the perspective of the company and investors, you can think about it – if you are a chip company and you don’t double the number of transistors on your chip within 18 months, what will happen then?

(End of this chapter)