Once Xingyi Semiconductor starts the "micro-profit foundry" model, it will rush for orders like crazy.
By then, the entire chip foundry industry will enter an era of "zero profit" or even "negative profit"!
It depends on who dies first!
There is no doubt that Intel, the son of Emperor M, has a large scale and the highest cost, so it will definitely have "negative profits" and may become the first of the Big Three to die!
Samsung Semiconductor has high costs, but with the support of its own mobile phones, tablets, TVs and other businesses, it may be able to survive, but it will still be half disabled.
TSMC is also not optimistic. Once it loses a large number of orders and profits, it is very likely that it will die or be disabled. If there is another earthquake, it will be completely over.
Xingyi Semiconductor, which has the lowest cost, is more optimistic.
Friends have zero profit or negative profit, but Xingyi Semiconductor can still maintain a small profit, which is enough.
This is the advantage of low cost!
As for GlobalFoundries, STMicroelectronics and the like, they can’t roll out high-end processes, but they can’t roll out mid- to low-end processes. They all have negative profits, and they will probably die faster.
To put it bluntly, Wang Yizhen flipped the table and killed Intel and TSMC in "zero-profit" foundry. He sent away half of his competitors, and then slowly raised the price to continue making money.
This is how domestic Internet giants played in previous lives.
First burn money to grab the market and grab customers, then kill your friends and form an oligopoly, and then raise the price together to make a lot of money.
If the other party is in a hurry, Wang Yi doesn't mind applying this method to the semiconductor field and wafer foundry.
The strategy of killing one thousand enemies and losing eight hundred to yourself is not a good strategy, but it is useful and invincible!
But when the time comes, the U.S. imperialists may have to give up...
Unless they want to watch Intel die in the foundry price war.
The prerequisite for all this is to build more factories, buy more photolithography machines and other equipment, and realize self-research of photoresist!
In this regard, Wang Yi made a decision:
"There are no problems with the second phase of the expansion of the Didu wafer fab. Mr. Hu, please ask someone to make a feasibility analysis report, and I will go to Didu to get the land."
"Here!" Mr. Hu said with joy and took out a document!
Wang Yi's mouth twitched: "..."
"Mr. Hu, have you been ready?"
Mr. Hu chuckled: "Aren't I worried about insufficient production capacity?"
"Okay, I'll leave the 28nm process to you, and I'll take care of the fab expansion!"
Wang Yi smiled and turned over the feasibility analysis report. It was done very professionally and all aspects were considered.
It was even mentioned whether the second phase of construction would have any impact on the first phase of the factory.
Xingyi Wafer Fab is built according to seven-level earthquake protection, and will not be affected by earthquakes of four or five levels.
In addition, the first phase and the second phase are separated by greenery and roads, as well as the office building, dormitory building and canteen of the first phase wafer fab, which is the first phase wafer fab R&D center and factory building.
To put it bluntly, the second phase of the project is several kilometers away from the first phase of the factory and will not cause any impact.
Wang Yi flipped through the report: "Mr. Hu, I'll send it up as soon as possible, and we'll take pictures as soon as possible. I'll leave the fab to you, and I'll celebrate for you tonight!"
"Thank you, Chairman!" Mr. Hu was in a good mood.
Following Wang Yi is nourishing. If you have any requirements, just ask them and Wang Yi will satisfy them.
If you want one fab, you can get two!
In the afternoon, Wang Yi returned to Xingyi Technology Park and went straight to the semiconductor department.
"Chairman, the specific performances of Kunpeng 702, Kunpeng 700 and Pterosaur 3000 have all been measured!"
Williams smiled and pulled out two cell phones:
"On the left is the original Xphone 1pro, equipped with NVIDIA Tegra 3. The other configurations on the right are exactly the same as the left, except that the processor has been replaced by Kunpeng 702!"
"After our testing, compared with the NVIDIA version of Xphone 1, the CPU performance of Kunpeng version of Xphone 1 is improved by 33%, the GPU is improved by 102%, the power consumption is reduced by 31%, and the temperature is also reduced by 5 degrees. It can be said that it is an overall improvement."
Wang Yi was silent: "..."
I can only say that Nvidia Tegra 3 is too cool. Apart from being the world's first quad-core mobile phone chip, it can't do anything else.
Especially the GPU performance is terrible.
"How does it compare with Qualcomm APQ 8064?" Wang Yi asked the key.
Kunpeng 702 beats Nvidia Tegra 3. As expected, the key is to compare with Qualcomm.
Williams took out two more mobile phones: "Dr. Wang, on the left is the original xphone 2, equipped with Qualcomm APQ 8064, and on the right is the xphone 2 equipped with Kunpeng 702."
With that said, Williams opened the video recorded for the test just now:
"Look, the CPU of the Kunpeng version of the Xphone 2 is almost the same as the Qualcomm version, and the GPU performance is slightly better by 2%. The power consumption is 1% higher than that of Qualcomm, and the temperature is 1.3 degrees higher than that of Qualcomm."
"In general, thanks to the leading generation V8 architecture, the performance of Kunpeng 702 with 40nm process and Qualcomm APQ 8064 with 28nm process is the same!"
Wang Yi waved his hand: "It's not that we are equal, but that we have an advantage."
"Don't forget, we have optimized the APQ 8064 version of xphone 2 in place. As for the Kunpeng 702 version of xphone 2, we have only simply optimized it and have not carried out in-depth optimization!"
Williams suddenly realized: "Yes! Kunpeng 702 without in-depth optimization can be on par with Qualcomm APQ 8064. Once Kunpeng 702 is deeply optimized, it can completely lead Qualcomm! It's hard to say if it is more, a 10% lead is not a big problem."
"Coupled with our integrated Wing Loong 3000 baseband, which supports mobile 3G, it has more advantages than Qualcomm's solution!"
"That's right, this time Kunpeng 702 is completely successful. Well done, let's celebrate tonight!" Wang Yi was in a good mood.
Huawei's K3v1 failed, K3v2 overturned, and Kunpeng 702 was a one-time success, and it is still far ahead. It can only be said that the people poached from Texas Instruments are very powerful.
After all, when it comes to making chips, Texas Instruments has always crushed Qualcomm and Samsung.
Texas Instruments' 512M dual-core chip is comparable to Qualcomm's 1G dual-core chip, so its strength can be imagined.
Now that Qualcomm is killing it, it's just a baseband problem.
Williams was not proud at all: "Chairman, we have also tested the Kunpeng 700."
With that said, Williams opened a new test video:
"Kunpeng 702 has a main frequency of 1.7G, and Kunpeng 700 has a main frequency of 1.3G. In terms of CPU, Kunpeng 700 is about 30% weaker than Kunpeng 702, and is slightly 3% stronger than the CPU of Nvidia Tegra 3."
Wang Yi fell silent: "..."
Is the Kunpeng 700’s CPU already surpassing the Nvidia Tegra 3 with its remaining HP?
Are you kidding me?
This is the testable result.
After a second thought, Wang Yi understood.
Both are 40nm processes.
NVIDIA tegra 3 has a 1.4G frequency and A9 architecture.
Kunpeng 700 has a main frequency of 1.3G. Although it is 0.1G lower than the 1.4G of Tegra 3, the A53 architecture is a little more advanced than the A9 architecture.
Taken together, the CPU of Kunpeng 700 with remaining HP is slightly stronger than Nvidia Tegra 3.
"Wouldn't the GPU be even more inferior?" Wang Yi couldn't help but ask.
Williams nodded: "Yes, Chairman, the GPU of Kunpeng 700 is also Mali-T604, but one core is shielded to become a three-core GPU. Therefore, the GPU performance of Kunpeng 700 is about 30% lower than that of Kunpeng 702, but
It’s 40% more powerful than Nvidia Tegra 3!”
What can Wang Yi say?
All I can say is that NVIDIA is garbage.
For example, the GPU performance of NVIDIA Tegra 3 is 100, the GPU performance of Qualcomm APQ 8064 is 200, and the Kunpeng 702 quad-core GPU is 210.
Kunpeng 700 blocks one core, and the remaining three-core GPU also has 140, which is 40% ahead of Nvidia Tegra 3!
"In this way, Kunpeng 700 beats NVIDIA Tegra 3, and Kunpeng 702 is directly ahead of APQ 8064!"
After some testing, Wang Yi thoroughly understood the actual performance of the two processors.
All I can say is surprise!
However, using such a powerful Kunpeng 700 on a thousand-yuan phone is equivalent to using Nvidia Tegra 3 on a thousand-yuan phone. It is completely wasteful and even takes over the competition.
The performance of a thousand-yuan phone cannot be too strong, otherwise the flagship phone will not be sold well.
The previous Redmi generation was released in the second half of next year, equipped with MediaTek MT6589T, which is equivalent to the Snapdragon 400. It is low-end, far inferior to the Nvidia Tegra 3, and even worse than the Kunpeng 700!
With that said, you can adjust your strategy!
Xphone 1pro's Nvidia Tegra 3 was replaced with Kunpeng 700, and the GPU performance increased by 40%
The Qualcomm APQ 8064 of xphone 2 has been replaced by Kunpeng 702! The overall improvement is 10%, and it also supports mobile 3G.
In this way, the gap between the two generations of flagship phones can be widened and the generation gap can be distinguished.
As for next year's unbounded thousand-yuan machine, the Kunpeng 700 is already too powerful. We will continue to castrate it and make it into the Kunpeng 500...
In the future, high-quality chips will be made into Kunpeng 702, and inferior chips that cannot run high frequencies or have a broken GPU core will be made into Kunpeng 700.
If it's even worse, make it into Kunpeng 500 and give it to the Unbounded Thousand Yuan Machine.
In this way, the three grades are separated.
What if there are not so many products?
It doesn't matter, just reduce the main frequency production and the yield rate will be higher.
As for the Xingyi Tablet, we also need to change the core, but we have to change our strategy.
Xingyi Tablet X1pro was replaced with Kunpeng 700, and the GPU performance increased by 40%, but it was still not enough.
After all, the third generation iPad was a failure. Half a year after its launch, Apple will launch the fourth generation iPad. The GPU performance of the A6X is so powerful that the Kunpeng 700 is no match for it. The only solution is the Kunpeng 702.
Therefore, in addition to the Xingyi Tablet X1 pro, it is also necessary to launch a Xingyi Tablet X1 pro equipped with Kunpeng 702.
But the name needs to be changed, to Xingyi Tablet X1 Ultra!
As for the Xingyi Tablet X1 equipped with Qualcomm dual-core, it is too rubbish and its sales have gradually declined. Only people who like eight-inch small screens will continue to purchase it, and production will be gradually reduced.
From now on, we will mainly recommend the Xingyi Tablet x1 pro equipped with Kunpeng 700, and the Xingyi Tablet X1 Ultra equipped with Kunpeng 702!
As for the 2nd generation Xingyi Tablet, it will be released next year and will be directly equipped with the 28nm Kunpeng 901!
Mainly a suppressor generation, with unrivaled performance!
After thinking about this clearly, Wang Yi looked at George and Zhang Chao: "The outstanding performance of Kunpeng 700 and Kunpeng 702 shows that the A53 architecture is very successful, and it also shows that our direction of integrated baseband is very correct."
"George, Kunpeng 901, which you are responsible for, follow this idea and get it done as soon as possible!"
"Okay, Chairman. With the experience of Kunpeng 702, add two A53 cores, and upgrade the GPU, it will be Kunpeng 901."
Wang Yi nodded: "Also Zhang Chao, you and George will study together. The key is to integrate the baseband. It is best to directly integrate the 4G baseband. The pressure falls on the old man."
Pang Liguo waved his hand: "Chairman, we have finished the 40-nanometer 3G baseband, and we have also started the 28-nanometer 4G baseband. It is not easy, but it is not a big problem. There is hope early next year!"
"Okay, very good. If Kunpeng 902 can integrate 28nm 4G baseband, it will really catch up with Qualcomm, or even surpass Qualcomm!" Wang Yi was filled with emotion.
Now Kunpeng 702 has better performance than Qualcomm APQ 8064. The only disadvantage is that it does not support 4G!
The Qualcomm 800 early next year will be 28 nanometers and integrate 4G baseband.
Kunpeng 902 is also 28nm and integrates 4G baseband. Even if it is a few months later, it can surpass Qualcomm Snapdragon 800.
"Chairman, what should I do?" Williams suddenly asked.
Everyone was stunned.
Yes, they all have work to do. As the leader, Williams took care of the Kunpeng 700 and Kunpeng 702, but he had nothing to do.
Wang Yi smiled lightly: "William, of course you are indispensable. Your task is to develop the next generation of new flagships, Kunpeng 905 and Kunpeng 906!"
"Kunpeng 905, 28nm process, does not integrate baseband, uses eight-core A53 architecture, clocked at 2.0G. For GPU, it can be upgraded to Mali-T658 mp6 six-core, targeting Qualcomm's new flagship and Apple A7 processor in the second half of next year.
"
"Kunpeng 905 integrates 4G baseband, which is Kunpeng 906 SOC, and you are responsible for it."
"Okay, Chairman!" Williams was in a good mood. This is a big project.
It’s just that at the beginning of next year, after George finished Kunpeng 901, it will be free!
However, you can enter PC chips!
Xingyi notebooks will also be available next year, but the difference is that Wang Yi does not plan to use Intel or AMD CPUs, nor does he plan to use Nvidia graphics cards, but will develop them all by himself.
Like Huawei and Qualcomm, they upgraded from mobile phone chips to PC chips.
Maybe this road is not easy to walk, but you have to walk it.
Even the system does not use Windows, but the self-developed XOS system.
Of course, for the convenience of users, XOS + Windows dual systems will be adopted in the initial stage, allowing users to choose.
Later, when the XOS system becomes larger, the Windows system can be gradually cut off.
For this reason, PC processors and graphics cards must be laid out and developed in advance.
You can't wait until it's blocked and then improvise, it's really too late.
In this way, after George finished developing Kunpeng 901, he went to develop PC chips.
From now on, Williams and Zhang Chao will be responsible for the research and development of mobile phone SOC, while Pang Liguo will continue to be responsible for the research and development of baseband.
George is responsible for the research and development of PC chips.
As for the graphics card...
This is more difficult. You must first break through the problem of 0 to 1.
The simplest thing is that Huawei will be able to make PC chips ten years later but cannot make high-end graphics cards.
Without him, the difficulty level of the graphics card is too great, which is far from being comparable to that of mobile phone GPUs.
Otherwise Nvidia would not have been brilliant for so many years.
In the previous life, Jingjiawei went through eight years of research and development until 2014, when it finally released the JM9 series of graphics cards, breaking the gap in domestic graphics cards.
Although the performance is a bit poor, it is still the best in China.
Years later, Moore's thread appeared again...
But it still did not drive the rise of domestic graphics cards.
There is no way. The technical barriers to graphics cards are too high, far exceeding mobile phone SOCs.
If Xingyi Technology wants to develop graphics cards, it can only poach people from NVIDIA and AMD.
There is nothing we can do about this.
Wang Yi plans to steal Lao Huang's home and dig up a core expert to make Xingyi Technology's graphics card.
If you don't poach big talents and just rely on yourself to manage things, it won't be difficult for you in less than four or five years.
Technological development has always been a step-by-step process. If you want to leapfrog, you will have to find a master who is already familiar with the technology.
Just like Wang Yi wants to make a mobile phone chip, without the big guys like Williams from Texas Instruments, let alone a 40-nanometer chip, I'm afraid even a 65-nanometer chip wouldn't be possible.
Only by standing on the shoulders of giants can we see further.
The same goes for graphics cards.
Wang Yi has a plan, graphics card experts should dig in advance!
What the US emperor likes to do most is to block graphics cards.
And the graphics card is even more important.
Not only are consumer electronics equipment needed, but AI large models are also needed!
For Wang Yi, the graphics card had to be purchased.
In response, Wang Yi called Williams aside:
"William, next year I plan to enter the PC market and start the research and development of PC chips. What suggestions do you have?"
"PC chip..." Williams instantly understood Wang Yi's ambition and knew that Wang Yi wanted to win not only the mobile phone market, but also the PC market.
"Chairman, in fact, Texas Instruments originally made PC chips, but lost to Intel. Now it can make PC processors, which is an opportunity to avenge its past."
Williams looked a little complicated.
Newcomers like George to Texas Instruments were not familiar with Texas Instruments' PC consumer chips when they joined the company, so they didn't understand that history.
But Williams, the old guys from Texas Instruments, still had to endure the humiliation all the way.
"Well, what Texas Instruments likes to do the most is to get up early and catch a late set!"
"The same is true for PC chips, and the same is true for mobile phone chips."
"I think back then, the first 8-bit computer CPU was made by Texas Instruments, but it turned out that the bet was in the wrong direction, and Intel overtook it..."
"There are also mobile phone chips. Texas Instruments has dominated the market for decades, and Qualcomm is the younger brother."
"Nokia has hundreds of millions of mobile phones every year, all of which use Texas Instruments chips."
"Who would have thought that Nokia began to decline, and Texas Instruments' largest customer was stretched thin."
"Qualcomm has no moral integrity and uses patents to talk things over. As a result, Texas Instruments' mobile chip business is completely over."
Wang Yi also knows this period of history.
Texas Instruments chips sell for $39 each, but they only have CPU, GPU, DSP, etc., and no baseband. You need to buy Qualcomm's baseband.
Qualcomm baseband sells for $49 directly!
When the manufacturer heard about it, he found that the price was even more expensive than the chip, and he couldn't accept it at all.
Qualcomm turned around and said, don't worry, we also sell packages, Qualcomm processor + Qualcomm baseband, a set only costs 49.9 US dollars!
As you can imagine, most manufacturers directly abandoned Texas Instruments chips and chose Qualcomm SOC.
Not long after, Texas Instruments’ mobile chip business suddenly collapsed!
Similarly, Nvidia's mobile chips are also finished in this way.
Buying a baseband and getting a chip is not just a talk, it directly killed the two giants.
Williams sighed and suddenly said: "Director Wang, now that the mobile phone chip department has gradually matured, why not leave it to George and Zhang Chao. I will go to the PC chip department!"
"???" Wang Yi was startled. Did this arouse Williams' fighting spirit? Want to do Intel and avenge his previous shame?
Isn't that the case?
Williams sighed: "Director Wang, to be honest, when I joined Texas Instruments, I was initially responsible for the PC chip business. However, it was no match for Intel. Later, Texas Instruments changed direction and focused on mobile chips, and I followed suit.
Changed business."
Wang Yi understood. This was a pity for Williams.
Such technical experts will have obsessions.
Just like when we play games, if we can't pass a level, or if we kneel down in the rankings ten times in a row, we always want to win the game!
Otherwise, I wouldn’t be able to sleep at all!
"And I have a few close friends. Some of them switched jobs and went to Intel, but they didn't do well. Let's see if I can recruit them."
Wang Yi nodded: "Okay, then I'll leave the PC chip to you. As for Kunpeng 905 and Kunpeng 906, I'll leave it to Zhang Chao!"
"From now on, George will be responsible for the research and development of Kunpeng 901 and 902, and Zhang Chao will be responsible for the research and development of Kunpeng 905 and 906."
In fact, 901 and 902 are products of the same generation, with the same CPU and GPU. The difference is that 901 does not have a baseband and will be used for tablets and TVs. 902 is an SOC with integrated baseband and will be used for mobile phones.
Similarly, Kunpeng 906 is just the baseband integrated in Kunpeng 905!
In this way, each of the two people is responsible for a generation of flagship chips, and they can advance alternately.
As for who is in charge of mobile chips?
It just depends on who is better between the two. Only when you are under pressure will you be motivated!
Williams has extensive connections and experience, so it would be perfect for him to develop PC processors.
"By the way, William, the graphics card also needs to be repaired." Wang Yi said meaningfully.
Williams understood instantly: "Well... I also know a few friends at NVIDIA..."
"Leave it to you!" Wang Yixiao said.
There is nothing better than asking Williams to poach Lao Huang's people.
In this way, Williams will be in charge of the computer CPU, and then he will hire a big guy from NVIDIA to be in charge of the graphics card!
The business map will gradually become complete.
Later, Wang Yi approached George and Zhang Chao: "Let's make some final adjustments. The research and development of Kunpeng 901 and Kunpeng 902 will all be handed over to George."
"George, you first get the Kunpeng 901 without integrated baseband, and then integrate the 4G baseband to make the Kunpeng 902 SOC!"
"Okay, Chairman." George agreed readily.
Wang Yi looked at Zhang Chao: "Zhang Chao, leave Kunpeng 902 to George. You will take over the next generation of Kunpeng 905 and Kunpeng 906 projects."
"Yes!" Zhang Chao responded. As a newcomer taking the lead, the pressure was also increasing.
Among the three, Zhang Chao is the youngest, but he has great abilities and the greatest potential, and is highly favored by Wang Yi.
Maybe it's not as good as George and Williams now, but it's hard to say in two years.
To attack the frontline, it depends on ability, experience, talent and explosiveness.
The younger you are, the more creative you are, and the more exciting you are!
What the two of them didn't understand was that since both of them were responsible, what would Williams do?
Resign?
Is Mr. Wang trying to kill the donkey?
You just got Kunpeng 700 and Kunpeng 702, are you going to force Williams to leave?
No way!
Williams saw what the two were thinking and said with a smile:
"Next year we will focus on PC chips, and the chairman wants me to take the lead and make PC processors! The mobile chip business will be left to you two, come on!"
"That's it!" The two of them suddenly realized.
"Thank you, teacher, we will live up to your expectations!" George said with a smile.
I am very grateful to Williams for his kindness and support.
If Williams hadn't led him to change jobs, he would have been laid off like other colleagues at Texas Instruments, and he still can't find a job...
There is no other way. Texas Instruments has too many mobile chip R&D personnel and the salaries are high. Wang Yi cannot accept them all and will only hire the best.
"Okay, that's it for today's work. Call everyone, let's go to the hotel to celebrate!"
Wang Yi waved his hand and took Semiconductor and everyone out.
The vehicle has already been prepared downstairs.
The same is true at Xingyi Wafer Fab.
I am destined to never come home until I get drunk tonight.
Sure enough, I drank until early in the morning and was sent back by the arranged person.
However, Apple Chairman Cook, who is far away in the United States, stood on the stage with high spirits and opened Apple's new product launch conference.
Chapter completed!