Reborn in 2011, 985 missing from two copies

Chapter 718 Semiconductor War, 14-nanometer breakthrough!

In addition, Xingyi Technology emphasizes pragmatism and competence, and does not engage in the unhealthy practices of seniority-based promotion and flattery.

As long as you have the ability and are willing to work hard, you can find a stage to realize your ambitions at Xingyi Technology, enjoy a fair promotion system, and have a bright future, which naturally attracts more talents to join.

With the addition of these pragmatic core talents, Xingyi Technology's development speed is getting faster and faster.

In addition, Wang Yichao's advanced overall vision enabled each business unit to make great strides and dominate its region.

The situation is much worse for BAT (Baidu, Alibaba, Tencent). Their core, pragmatic talents have all left, leaving only a bunch of sycophants, seniority-based, and scheming mediocre individuals.

When it comes to scheming and intrigue, these people are formidable and undisputed. But when it comes to getting things done, conducting research and development, and expanding markets, they are utterly incompetent.

As a result, the efficiency of BAT (Baidu, Alibaba, Tencent) has become increasingly lower, and the development of each business unit has become increasingly worse.

The situation is completely the opposite.

At the end of February, Wang Yi attended Xingyi Technology's month-end summary meeting again.

Wang Yi rarely attends regular meetings anymore; each business unit holds its own meetings, makes its own plans, and discusses its own issues.

However, Wang Yi still needs to personally oversee and control the overall situation at key meetings such as the end-of-month summary meeting and the quarterly meeting.

The faster a train travels, the more severe the derailment will be in the event of an accident.

The faster Xingyi Technology develops, the more Wang Yi has to pay attention.

Yu Longxing, president of the graphics card division, took the lead in reporting: "Chairman, all of our new products have been selling like hotcakes a month after the New Year's shopping festival launch event."

"The best-selling product is the RTX 960, with monthly retail sales exceeding 170 million units, OEM channel sales of 50 units, and GPU chip sales of 40 units. The total monthly sales reached 260 million units."

"Secondly, the RTX 950 achieved monthly retail sales of over 150 million units, OEM channel sales of 40 units, and GPU chip sales of 30 units. The total monthly sales reached 220 million units."

"RTX 970 ranked third, with monthly retail sales exceeding 132 million units, OEM channel sales of 45 units, and GPU chip sales of 25 units. Total monthly sales reached 202 million units."

"RTX 980 ranked fourth, with monthly retail sales exceeding 43 units, OEM channel sales of 10 units, and GPU chip sales of 15 units. Total monthly sales reached 68 units."

"Throughout February, our four graphics cards were number one in global sales, consistently outselling Nvidia!"

"Very good, congratulations to Xingyi Graphics Card Division!" Wang Yi clapped, and everyone else clapped along.

The achievements of Xingyi's graphics card division are truly remarkable!

Even more impressive than Xingyi's CPU division.

Although the Starry CPU is highly competitive and sells well globally thanks to its powerful integrated graphics.

However, thanks to its established reputation as a CPU giant, Intel did not falter and its sales remained strong.

Both sides have their strengths, so they are evenly matched.

In the graphics card field, under Wang Yi's strategic planning, Xingyi graphics cards have achieved a crushing advantage with precise product segmentation and a full range of strategic positioning. Coupled with Nvidia's own self-destructive failure, Xingyi graphics cards have achieved a crushing advantage!

"However, we can't be careless. Nvidia isn't stupid; after such a crushing defeat, they're bound to make a move! They'll definitely retaliate!" Wang Yi changed the subject:
"This overwhelming advantage won't last long. It's expected that Nvidia will launch a series of measures to turn the tide by the end of March or the beginning of April. Are you prepared?"

Yu Longxing nodded: "Yes, Chairman. First of all, we've increased production capacity and been shipping like crazy these past few months. Whether it's graphics card manufacturers, computer manufacturers, or retail channels, as long as there are orders, we deliver them as quickly as possible to maximize sales!"

"At the same time, we are also advancing the research and development of new graphics cards."

"According to our information, Nvidia is preparing to release the GTX 950 ahead of schedule, with an expected launch in April. The RTX 950 can handle it. Our strategy is to ship a large quantity of GTX 950s before they are released, and then lower prices to undercut them after their launch!"

Wang Yi nodded: "Okay."

Yu Longxing continued, "I also received news that Nvidia is preparing to use two million of its unsold, defective GTX 970 GPUs to downclock them into GTX 960 Ti graphics cards, aiming to counter our RTX 960! They are expected to be available by the end of March."

"To this end, we also cut down the algorithm of the RTX 970 GPU chip to make it into the RTX 960 Ti, which features a full 4GB of video memory, but it will not be available until April."

"In March, we can only ship as many RTX 960s as possible as quickly as possible."

Wang Yi nodded again: "Okay, even if we earn less money, we have to go all out against Nvidia! But we can make a fuss about the GTX 960 Ti, like this... like this..."

Yu Longxing laughed: "That's a good idea, haha. Nvidia turned the disastrous GTX 970 into a GTX 960 Ti and only cared about clearing out its own inventory and enjoying itself, ignoring major customers and old users. They deserve this disaster!"

"Not bad, if you handle it well, the GTX 960 ti won't be able to cause much trouble!"

"Yes, this is an opportunity."

Yu Longxing then changed the subject: "In addition, NVIDIA's perfectly repaired GTX 970 will also be available for retail in March. Without the defective GTX 970, it will also have some impact on our Xingyi RTX 970..."

“Well, that’s also a problem.”

Wang Yi nodded: "But what about Nvidia's repaired GTX 970? Are they planning to sell the reference design themselves? Or will they continue to cooperate with graphics card manufacturers?"

"According to our intelligence, Nvidia originally intended to continue cooperating with graphics card manufacturers such as ASUS to promote the retail non-reference GTX 970 without any issues. However, ASUS and other manufacturers demanded that Nvidia take responsibility for the previous GTX 970 issues, which Nvidia is currently refusing to do."

"In this situation, the two sides had a falling out, and the graphics card manufacturers were too lazy to place orders for the GTX 970 GPU chip, which is guaranteed to be reliable."

"Nvidia will probably have no choice but to sell its own reference designs, but not the single-fan version. Instead, it will follow our example and release dual-fan or triple-fan versions."

Wang Yi nodded: "That's possible. If so, there's still a chance to make a move! You know what I mean."

Yu Longxing smiled knowingly: "I know, Chairman. This is another opportunity for us. Next, Xingyi Semiconductor will not only suppress Nvidia's new graphics card products, but also strategically suppress them!"

Wang Yi laughed and said, "Nvidia is arrogant and treats its big clients like innocent victims, refusing to take responsibility. We, on the other hand, treat our clients with kindness, treat our big clients as friends, and take full responsibility!"

"This time, we're not just going to beat Nvidia, but also Nvidia's major customers!"

"Yes!"

"By the way, how's the 14nm process going?"

Other graphics cards are fine, but the key is Nvidia's flagship graphics card, the GTX 980 Ti, coming in June!

That's the most important thing.

Flagship graphics cards represent the pinnacle of technology.

No matter how good the mid-to-low-end products are, if the flagship models aren't up to par, they still can't hold their own.

The key to competing with Nvidia's GTX 980 Ti lies in whether the 14nm process can be successfully broken through!
Wang Yi looked around but didn't see Old Hu anywhere.

Wang Yi was used to this.

Professor Hu focused on the research and development of semiconductor processes, spending most of his time in the R&D facility and rarely leaving except for special meetings.

Samsung has now made a breakthrough in its 14nm process and is currently in risk production. The yield rate is hard to say, but it is probably not high.

The progress at Xingyi Semiconductor has become quite delicate.

Williams said with a smile, "Chairman, I went to the R&D plant last week, and Mr. Hu has achieved very good results."

"Although Mr. Hu said that the 14-nanometer process is not yet ready, the yield rate in the last test has already reached 75%!"

"75%! You call that not good enough?"

Wang Yi smiled: "Old Hu is as modest as ever!"

In fact, for this kind of cross-generational technology, a yield rate of 50% is considered a success.

With a further improvement of 50%, the yield rate of 14nm will reach 85% in less than two months, at which point mass production can begin.

But Mr. Hu was different; even with a 70% yield rate, it still wasn't enough.

In Mr. Hu's view, if it's less than 85%, then it's unacceptable.

Although Samsung's 14nm yield rate may only be 70%, or even less than 70%.

"Is the Kunpeng 920 fully developed?"

George nodded: "Research and development is complete, Chairman. We're just waiting for the 14-nanometer process to begin mass production and tape-out. In addition, we've also developed the Kunpeng 920X!"

"The Kunpeng 920X uses the same architecture as the Kunpeng 920, but it removes communication modules that are not needed in tablets and game consoles, freeing up space to increase the main frequency and heat dissipation, thereby achieving more powerful performance. It can be considered a high-frequency upgrade of the Kunpeng 920!"

"The Kunpeng 920 X has also been successfully developed and is just waiting for the breakthrough of the 14-nanometer process to be completed and tape-out."

The Switch project has been underway for over a year and was actually completed long ago, but it has not been released yet due to insufficient processor performance.

Even the 28nm Kunpeng 910C is underperforming.

The Kunpeng 910 is fine as a mobile phone processor, but it falls short as a processor for a game console.

If the Switch were just a console, it would be simple; a Starry Night RTX 950M graphics card would suffice, providing enough performance. But the Switch is more than just a console; it's also a handheld!

Power consumption and battery life are both important issues. The RTX 950M has too high power consumption, and the battery life will collapse and it won't be able to last.

Therefore, the Switch project will remain on hold until the Kunpeng 920 X is released!

The Kunpeng 920 X, 14nm, features a quad-core A53 + quad-core A57 processor and is a high-frequency version with strong heat dissipation. It boasts powerful performance and significantly lower power consumption than a graphics card, making it a perfect fit for the Switch.

At that time, some AAA games can be ported over.

Especially since the Switch also uses the XOS system!

If the Switch sells well, a series of AAA game developers will actively adapt their games to the Switch in order to make money, which is equivalent to adding the XOS system.

Even if the US crushes it, and Xingyi computers can no longer use the Windows system in a few years, they can be completely replaced by the XOS system.

At that time, the XOS system will not only have mainstream software, office software, and design software, but also AAA games!

Then the XOS system is fully operational!
Even without the Windows system, the XOS system is powerful enough to support all systems, giving users peace of mind.

This is also part of Wang Yi's grand strategy.

The HarmonyOS system that Huawei previously promoted for its computers was not compatible with many games, especially AAA games, which were mostly unavailable, making it awkward to use.

Wang Yi promoted the Switch, attracting AAA game developers to actively adapt it to XOS before it was blocked by Windows. This was a proactive move to prepare for this day.

"Now that the Kunpeng 920 and 920X have been successfully developed, the 14nm yield rate has reached 70%, and the yield rate in March reached over 85%. There are no problems at all, and it can fully enter the mass production stage."

Wang Yi was in high spirits: "How's the progress of our 14-nanometer wafer fab?"

Xingyi Technology has too many things to do. When it was first planned, Wang Yi was fully involved, but the subsequent progress was controlled by other presidents and vice presidents. Wang Yi was not aware of the latest developments.

"Chairman, the S1 large-scale wafer fab in Shenzhen was completed at the end of last year, and the production lines are almost fully installed. Once the yield rate of the 14-nanometer process is high, we can start debugging the production lines, conduct trial production, and then mass production."

"This wafer fab has a monthly production capacity of 10 wafers, which can produce approximately 6000 million 14-nanometer chips per month!"

"By mid-year, the Shenzhen S2 large-scale wafer fab will also be able to start production, with a monthly output of 10 wafers and 6000 million 14-nanometer mobile phone chips. Of course, if it is producing graphics cards, the output will not be that much."

"There is also the Shenzhen-listed packaging and testing plant, which was completed and delivered in the third quarter of last year and is now in production."

"Once 14-nanometer production begins, Shenzhen Stock Exchange will be able to achieve mass production of wafers and the entire chip packaging industry."

"very good."

Wang Yi was very satisfied with this progress.

If the soldiers and horses are not moving, the grain and grass will go first.

Two years ago, before the 14-nanometer process had even begun research and development, the Shenzhen-listed wafer fab had already been planned and started construction.

With the 14-nanometer process about to break through, a large 14-nanometer wafer fab with a monthly production capacity of 6000 million chips has been built!
The production line is all installed, and the timing is perfect.

Over the past few years, all of Starway Technology's previously planned 28-nanometer wafer fabs have been completed and are now operating at full capacity.

Not only has it achieved self-production of all its own mobile phone chips, TV chips, computer chips, and automotive-grade chips, but it also sells a large number of them to third-party companies and has a large surplus of production capacity to manufacture for third-party companies.

For example, Huawei's mobile phone chips and base station chips, ZTE's base station chips, Datang and Spreadtrum's baseband chips, etc., are all manufactured by Starway Semiconductor.

Once the 14nm process is in mass production, even more customers will place orders!

In the early stages, production capacity was naturally maximized to produce its own 14-nanometer chips and sell them in large quantities to external customers.

Then they can reduce production of some of their own chips and outsource manufacturing to third parties.

In addition, silicon wafer plants and photoresist plants have been built, but have not yet started production.

The main reason is that although Xingyi Technology has made breakthroughs in silicon wafer technology, the production cost is relatively high, similar to purchasing from third parties, so it has not yet started production.

Photoresist has also made remarkable progress. High-end photoresist can now be produced, but it is still not advanced enough and has a high cost. Chips produced using high-end photoresist have a relatively low yield rate, and further research and development are needed.

Therefore, neither large silicon wafers nor photoresist have been put into production yet. Instead, it is more appropriate to wait until the technology breaks through further and the cost is lower before starting production.

This strategy can also lull the US into a false sense of security, making them mistakenly believe that Xingyi Semiconductor has no expertise in this area and can be easily strangled and controlled.

Once the US crushes the competition and completely blocks the production of large silicon wafers and photoresists, Xingyi Technology will be able to quickly start production.

Although it's not the most advanced, it's sufficient and won't leave us vulnerable to technological bottlenecks.

Of course, if the demolition takes place a year later, then the large silicon wafer production technology can be further improved, and the cost can be significantly reduced. Photoresist technology can also be further advanced.

If they crush it in two years, they have a real chance to become one of the world's top teams!
What Xingyi Technology lacks now is time!
The most difficult thing for domestic products is the breakthrough from zero to one. Now that the breakthrough from zero to one has been achieved, the remaining breakthrough from 1 to 99 is just a matter of time.

"George, contact Mr. Hu. Next time we do 14nm trial production, bring the Kunpeng 920 and Kunpeng 920X along for tape-out," Wang Yi said.

Samsung has already conducted risk production of its 14nm process, and Star Semiconductor cannot afford to fall behind!
As for TSMC, let them continue with their 20nm process!
"Chairman, that's exactly what I was thinking. I'll take the team to see Mr. Hu in a bit, and we expect to start trial production again on the 5th of next month!" George said with a smile.

Although the yield rate of the 14nm process was only 75% last time, it might reach 80% or even 85% when we start trial production again on the 5th of next month!

Once a batch of Kunpeng 920 and 920X processors are produced, a series of tests can be conducted.

As for the Kunpeng 920V with enhanced picture quality, there's no rush. The Xingyi TV 3rd Generation and Xingyi Tablet 2 Ultra, equipped with the Kunpeng 910V, have just been launched, so we'll have to wait at least half a year before making any moves.

The 14nm process currently has a yield rate of 75%. After another test on March 5th, it is expected to reach approximately 80%-85%, at which point the Shenzhen S1 wafer fab will be able to begin commissioning and trial production.

Mass production will officially begin the month after next, in April...

Initial production capacity will definitely be limited, so naturally it will be prioritized for the Kunpeng 920 SOC mobile chip. The Kunpeng 920X will probably only be produced in small batches for the Switch, and the excess capacity will be allocated to the Xingyi Tablet 3.

Once sufficient 14nm production capacity is available, the Kunpeng 920V with enhanced picture quality will be launched for the Xingyi TV 3rd generation and Xingyi Tablet 3 Ultra!

Priority is important; everyone should take turns.

"Chairman, we've also developed the Starry Sky RTX 980 Ti based on 14nm..." Yu Longxing said with a smile.

The meaning is obvious; they also want to ride this wave and get their chips manufactured.

Upon learning that Xingyi Semiconductor's 14nm process was about to be implemented, Yu Longxing immediately began the research and development of the RTX 980 ti.

Based on the RTX 980, we upgraded it by increasing the number of cores, increasing the clock speed, and upgrading to 14nm, which resulted in the RTX 980 Ti.

Without the 14nm process, Yu Longxing doesn't believe the RTX 980 Ti can outperform the NVIDIA GTX 980 Ti.

However, with the 14nm process, the 14nm RTX 980 Ti has a good chance of crushing the 28nm GTX 980 Ti!

Process advantages!

"Contact Old Hu first to see if it's appropriate. If it is, proceed with it. If not, wait a bit longer."

Wang Yi said, "After all, computer GPU chips are much more difficult to produce than mobile phone chips, and the yield rate is much lower."

Mobile phone SOC chips are small in size and contain multiple modules such as CPU, GPU, baseband, and ISP. They are flexible in design, each module is relatively small, and the internal design is relatively complex, making them easier to manufacture.

A computer GPU has only one GPU module, but its internal design is very complex, its core is much larger, its power consumption is higher, its requirements for manufacturing process are more demanding, its production difficulty is greater, and its yield rate is lower.

To give an analogy, making chips is like taking an exam.

Although the mobile SOC test covers five subjects at once, each subject is at the middle school exam level, and the total test time is five hours.

The computer GPU exam only has one subject, but that subject is as difficult as the college entrance exam, and it requires 20 hours of practice questions!

For example, in the last test, the yield rate of mobile phone chips was 75%.

If you replace the computer's GPU chip, the yield rate might only be 30%!

"Don't worry, Chairman Wang, Mr. Hu will decide on this. I'm not in a hurry."

"Well, take your time."

Wang Yi is not in a hurry for the RTX 980 ti.

This can't be rushed, let's get the Kunpeng 920 SOC sorted first!

At worst, we can let Nvidia's GTX 980 Ti be sold for two months in June.

It doesn't matter.

(End of this chapter)