Rewrite the Technological Landscape

The reason why Jiang Shangyi and Liang Mengsong are important is not only because of their abilities and status, but also because of an important person in TSMC, that is, Cai Lixing, the current CEO of TSMC.

Cai Lixing can be regarded as the most disappointing CEO in the history of TSMC, because he did a thing that is common in large companies but really taboo, cultivating his own power.

After taking over from Zhang Zhongmou in 2005, Cai Lixing devoted himself to cultivating his own power, but the financial crisis turned TSMC into a mess, and its performance declined rapidly. Four years later, Zhang Zhongmou had to withdraw Cai Lixing himself. Let Cai Lixing change from the CEO of TSMC who manages 20,000 employees overnight to the general manager of the new business department who manages less than ten employees.

But all this really deserves Cai Lixing. Since he took office, TSMC has started a wave of management resignations. In the first year of his career, Wu Ziqian, the vice president of TSMC, announced that he was ready to resign. Hu Zhengda, the deputy general manager of marketing, resigned. As a consultant, Jiang Shangyi, the company's highly respected senior vice president of research and development, proposed to retire.

Looking at Liang Mengsong’s matter again, after Jiang Shangyi’s retirement, the company will have two new R\u0026D vice presidents. As a result, Cai Lixing not only gave one spot to Liang Mengsong’s biggest rival, Sun Yuancheng, but the other spot would rather go to Intel to poach Luo Weiren than pay for TSMC. Liang Mengsong for so many years.

Although in the matter of Liang Mengsong, TSMC has always attributed the fault to Liang Mengsong, saying that he is a traitor, that he has a flawed personality and does not fit in with the group, that he has no management ability, and that he has no vision.

But Meng Qian judged one thing only by looking at the facts in front of him. The fact is that after Cai Lixing took office, TSMC did leave a group of management for several years in a row. It is true that Zhang Zhongmou regretted trusting Cai Lixing four years later, and TSMC said that Liang Mengsong’s personality was not suitable. When the manager went to Samsung, he took a team of more than 100 people. Others included Huang Guotai, Xia Jinqiu, Zheng Junlong, Hou Yongtian, Chen Jianliang and other TSMC elite teams, and they were all old employees. Why, these elites have different personalities. Actually playing with Liang Mengsong?

Besides, Liang Mengsong has no vision. Looking at the achievements of Liang Mengsong and Sun Yuancheng in the future, who has no vision.

Moreover, as soon as Zhang Zhongmou beat Cai Lixing down, Jiang Shangyi, who retired because Cai Lixing took office, immediately returned to TSMC.

No matter how TSMC vilified Liang Mengsong in public relations, Meng Qian only believed in the facts before him.

Therefore, another important reason why Meng Qian wants to poach Liang Mengsong and Jiang Shangyi is to directly poke at this contradiction within TSMC in the past few years, which affects a group of old employees in TSMC who were suppressed by Cai Lixing. Digging as many pits as the oppressed old employees of TSMC can dig. When Zhang Zhongmou recovers and realizes Cai Lixing's pits, it may be too late.

At noon, Meng Qian first made an appointment with Liang Mengsong. After all, it was poaching, so it was not good for the two of them to shout together.

Moreover, Liang Mengsong's situation is completely different from Jiang Shangyi's situation, requiring different communication strategies.

In fact, Liang Mengsong has laid the foundation. After all, he wants to leave very much. The two vice presidents are all booked, but there is no other position for him. Why keep it? He only attended the trial once. He only said a few words, and that was how he was suppressed by TSMC.

The main reason why Sun Xiaojing failed to deal with him was that Samsung was robbing Dafeng Group, and Samsung dispatched its CEO for Liang Mengsong, so Meng Qian had to come forward.

Liang Mengsong’s annual salary at TSMC is 7 million soft sister coins. Samsung gave Liang Mengsong an annual salary of 21 million soft sister coins. This annual salary is basically the same as the annual salary of the CEO of Samsung Electronics. This shows that Samsung attaches great importance to Liang Mengsong.

And Samsung has an indefinable advantage, that is, Liang Mengsong's wife is from Korea. Whether his wife has played a role in promoting this matter, and if so, how much it has played a role in promoting, no one knows. Things between husband and wife.

"Mr. Liang, I have admired your name for a long time."

"It's my honor to meet Mr. Meng." From the performance point of view, Liang Mengsong seemed very interested in Meng Qian.

It's normal, too many people are interested in Meng Qian now.

After a few words of pleasantries, it's time to talk about business when the hot dishes are on the table, "I heard from the HR director of our company that Mr. Liang seems to have been worried."

The chairman went out to dig people himself, there was nothing to detour, not to mention that Sun Xiaojing had talked to him several times before, and everyone knew why they were sitting here today.

Liang Mengsong made an apologetic expression, "Mr. Meng, whether it is Dafeng Group or SMIC, they are in this field."

The first thing that comes up is that Samsung is ahead of the mainland in this field in technology. This is what Meng Qian wants to hear. He hopes that Liang Mengsong is a person who cares more about technology itself.

"No matter what decision Mr. Liang makes this time, he must hope to make something valuable in this field, right?"

"Of course, at my age, I have experienced everything that should be experienced. The only regret is technology."

"Sure enough, I read Mr. Liang correctly. The semiconductor manufacturing level of Dafeng Group is indeed far from the international level, especially in the manufacture of traditional CPUs. We have little experience.

In the past few years, the cooperation with SMIC, especially the joint factory, has accelerated our development to a certain extent, but as Mr. Liang said, the level of SMIC in the world still needs to be improved.

This is indeed our disadvantage, but there are some things I want Mr. Liang to take a look at.

Let’s take a look at this document first. This is the industrial chain diagram of the 7th-generation LCD screen that we jointly created with Sony. This is the industrial planning plan for the chip manufacturing factory that we are preparing for this year. Mr. Liang has a look at it. "

Liang Mengsong took the document and glanced at it to see something special, "A 7th-generation production line has attracted nearly 200 upstream and downstream companies, and the factory that has not yet started construction has attracted more than 100 upstream and downstream companies." Move in?"

"Mr. Liang, I know that TSMC definitely does not have this appeal. Let me ask Mr. Liang, does Samsung have this appeal?"

"I'm afraid not."

"Then Mr. Liang should know the reason very well."

Liang Mengsong expressed affirmation, "Only the mainland has a full industrial chain, and the chip manufacturing plant of Dafeng Group can take advantage of China's full industrial chain."

"That's right, Huaxia Future is a world factory. Anyone who wants to develop away from the world factory, unless you have the ability and energy to build and maintain a long-distance entire industrial chain, but even if you can do it, whether it is in terms of cost or In terms of management, it is impossible to surpass the advantages of our local full industrial chain in Huaxia, and as a Chinese enterprise, building a factory in Huaxia will get the benefits brought by the entire industrial chain.”

"The large and comprehensive development policy has begun to show its advantages." Liang Mengsong couldn't help but sigh, "It's time to reap the rewards of decades of hard work. In the future, China may not be an industrial power in a short time, but everyone must An unsurpassed industrial power."

Meng Qian responded with a smile, as long as Liang Mengsong can understand, he then took out another document, "I don't know if Mr. Liang paid attention to the Global Development Conference of Dafeng Group this year."

"Of course, this is a feast for the global IT industry, and Mr. Meng has brought surprises to the entire market."

"We enjoyed the future at the conference and proposed cloud intelligence, and we are really working hard to develop in this direction, but in the process we gradually realized that the development of chips is hindering our progress. Mr. Liang? I have considered a question, how long can Moore's Law last?"

Liang Mengsong cheered up sensitively, "What does Mr. Meng mean?"

"Dennard's Law ended in 2003, and Amdahl's Law has reached its limit, so how long can Moore's Law survive?

And a more critical question is, I don't know why Moore's Law can last for so many years in Mr. Liang's view. "

"Why can Moore's Law last for so many years?" Liang Mengsong actually has his own answer, but he deliberately wants to hear what Meng Qian said. "It's really hard to say, what does Mr. Meng think?"

Liang Mengsong wanted to hear from himself, and Meng Qian also said, "Because the chip industry is driven by classic 2D scaling, we have hardly seen any innovations in the mainstream architecture for so many years, and everyone has focused on the mainstream architecture. on a geometric scale.

And the more you do this, the more monopoly an architecture like X86 can achieve, so why can Moore's Law last for so many years? Not because of Mr. Moore's vision, but because Intel doesn't want to make progress, and it doesn't want to make progress. "

Meng Qian's words are very impactful to people in this industry. On the one hand, it is because Meng Qian accused Intel and denied the so-called legend of Moore's Law, and more importantly, because Meng Qian said that many people Actually thought so too

As a top talent in the industry, Liang Mengsong naturally thought so too. After hesitating for a few seconds, Liang Mengsong chose to follow Meng Qian's words, "It is the most cost-effective way to keep scaling instead of changing the mainstream structure. It’s the safest thing for a company, especially a company with a monopoly.”

"Yeah, that's why I never think that any commercial company is qualified to be great on the issue of technology. Commercial companies often only hinder the development of technology for the sake of profit. Users may see some interesting technology products and feel that technology is great. Life has changed, but Mr. Liang must be very clear that most technology companies are hindering the development of technology. This is the case with Intel, and so is Microsoft. In fact, Dafeng Group sometimes does the same.

But Dafeng Group has always had a pure research mechanism. The pure research mechanism is not only to preserve the purity of research and development, but also to constantly remind me that Dafeng Group should try its best to be a technology company that truly promotes technological development.

So Intel doesn’t want to make progress, we want to make progress, Intel doesn’t want to make progress, we want to make progress, because we have expectations for the future, because I have vision for the future. "

Meng Qian said, and handed another document to Liang Mengsong.

Liang Mengsong's heart was once again shocked by Meng Qian's words. After receiving the document, he looked at it very seriously, and his eyes became extremely sharp. After reading it for more than ten minutes, Liang Mengsong put down the document, "That is to say, the size of the chip Change must come."

Meng Qian straightened his body, "Judging from all the current research results, the traditional CPU architecture is definitely not suitable for the era of cloud intelligence.

Over the past few decades, most applications have been similar in their processing requirements. So all the workloads were handled by the CPU, and when we started to use more graphics, video and games, everyone chose the GPU.

But in the era of cloud intelligence, our workloads will look more and more different, and each workload has its own computing requirements. Therefore, in the era of AI-Cloud, we need a variety of different architectures, each of which is optimized for a specific type of workload.

In other words, the traditional architecture in the era of cloud and intelligence will become more and more weak. As stated in the document, AI workloads require three major architectural improvements, higher bandwidth memory, higher data parallelism and lower Accuracy, traditional architecture is obviously impossible.

Therefore, in the era of cloud and intelligence, we need to promote architectural innovations for various new workloads, and we can already foresee that materials such as silicon, silicon dioxide, silicon nitride, and carbon will only take more than ten years. The zoom limit will be reached.

If we want to realize the future we aspire to, we'll need entirely new architectures, and maybe even new materials.

We call it, the hardware renaissance. "

Thank you book friend 20181114173402980 for your reward! Thank you for your reward! PS: There was a typo yesterday. The top R\u0026D talent at Wanwan has a monthly salary of one million. I wrote it as an annual salary of one million. It has been revised.