Rebirth 08: Rise from copycat phones

Chapter 386: Crazy Price Increase of Graphics Cards

Just as Zhiyun Group first released the super-large generative AI model Yun AI, and subsequently released several open source AI models, including the generative AI model GTAI.

A huge wave of generative AI has been set off around the world!
Major technology companies have announced their own generative AI plans one after another, which has made many ordinary people sigh and wonder whether the era of artificial intelligence is really about to be popularized on a large scale.

However, within Zhiyun Group, this wave of generative AI itself is not taken too seriously.

Although Zhiyun Group released the Yun AI model and commercialized it, due to the high computing power costs, the entire Yun AI project is actually not profitable and is suffering from continuous losses.

On the contrary, Zhiyun Group pays more attention to the surge in orders for APO computing chips caused by this wave of generative AI!
In fact, selling APO graphics cards is the main reason why Zhiyun Group is promoting the generative AI wave.

After all, Zhiyun Group has always been engaged in hardware. It is different from Yihai Technology, Google, Microsoft and other companies... Even though the revenue of Yihai Technology's software and Internet business has increased year by year, it has never been the core business of Zhiyun Group.

Its core business has always been very clear, that is, various smart hardware products, such as mobile phones, notebooks, tablets, VR glasses, and now a robot.

Then there is the semiconductor design and manufacturing business.

Many of the software and Internet businesses in Zhiyun Group are actually designed to provide support for its own hardware products, and then make some money in the process. It doesn’t matter if they don’t make money.

For example, the cloud computing business is purely for the purpose of digesting and supporting its own server chips. It even specially developed cloud computers, which ostensibly sell computers but actually sell computing power, and the core is selling chips... Unfortunately, the cloud computer project failed to make any waves.

Various types of design software, such as EDA software for chips, are purely to support their own semiconductor design and manufacturing business, and are also supplied to other semiconductor manufacturers.

The APP mall business is to support the ecology of various smart terminals under its umbrella... The content ecology is the most important, and making money is secondary.

Even artificial intelligence technology is intended to add more competitive advantages to one's own hardware products...stimulate the artificial intelligence market and sell computing chips.

Therefore, even though the revenue from software and service services of Zhiyun Group is not low and has reached the level of several hundred billion US dollars, this is still not the core business of Zhiyun Group.

To be honest, Zhiyun Group is actually the largest Internet company in China. The revenue and even profits of its software and Internet services are higher than those of Yihai Technology, the largest Internet company in China...but no one still regards Zhiyun Group as an Internet company.

The same is true for Yun AI this time... Zhiyun Group has never expected to make much money from it. In fact, it is not making much money at present. Being able to maintain a balance in computing power costs is already a great success...

The reason why Zhiyun Group is promoting this wave of generative AI is very simple, which is to sell APO graphics cards and PX computing power platforms for airborne or vehicle-mounted use.

Why is it necessary to devote so much resources and energy to promoting AI and stimulating the computing chip market?

Because the gross profit margin of this thing is super high.

The GPU business unit of Zhiyun Group produces APO graphics cards for servers, as well as PX computing platforms, EYQ computing platforms, EYEQ computing platforms, and ZY computing platforms for terminals. These computing products focus on the field of machine learning, that is, artificial intelligence. Their total revenue reached US$11.6 billion last year, but their gross profit margin reached 80%, reaching US$9.2 billion.

The growth rate and high profits of the computing chip business were one of the main reasons for the substantial increase in Zhiyun Group's net profit last year.

The profit from computing chips is much higher than that from conventional civilian graphics cards. The civilian graphics cards that also belong to the GPU business unit, that is, various gaming graphics cards of the NVIDIA brand, have a gross profit of only about 50%.

What’s more critical is that the market for gaming graphics card business is close to saturation, and there is little room for large-scale growth: not many computers are being sold these days, so naturally not many graphics cards can be sold either.

------

The current operating conditions of various businesses in Zhiyun Group are as follows: the revenue and profits of the smart terminal business are gradually stabilizing, with no room for substantial growth.

The robotics business has just emerged, and although the prospects are very good, in the short term, in order to expand the market, it will not only fail to make money, but will continue to lose money.

The software and Internet business has a limited scale and it is difficult to take on the main responsibility. Many things look good, but in fact they do not make much money.

Even Yun AI is the same... It would have been good if it hadn't made losses in the first few years. If this thing could make a lot of money, Zhiyun Group would have released Yun AI long ago, let alone open source... The computing power cost is too high.

The market for conventional semiconductor products is also relatively fixed. Whether it is CPU, mobile phone SOC, gaming graphics card, various auxiliary chips, or semiconductor sensors, it is difficult for revenue and profits to show large-scale growth... There are too many competitors in this field, including Intel, AMD, Qualcomm and other American semiconductor companies, making it difficult to expand the market for these products and even more difficult to maintain high gross profits.

The above-mentioned businesses have either hit their ceiling or have not yet reached their redemption potential.

How can Zhiyun Group break the impasse, further increase revenue and profits in order to maintain the healthy operation of the entire huge enterprise and continue to invest large amounts of funds in various research and development fields? This has become a question that the Zhiyun management team headed by Xu Shenxue has to consider.

After discussion, the senior management of Zhiyun Group decided to focus on computing chips!
Because in the field of computing power chips, Zhiyun Group has an absolute leading advantage, and it is a monopoly advantage.

In the field of server GPUs, Zhiyun's APO graphics cards rely on their powerful performance and ecological support to almost monopolize the global market...with no competitors.

In the field of terminal computing power platforms, self-use products such as the EYQ platform will not be included in the discussion.

The terminal computing power platforms available on the market are mainly the PX computing power platform under Zhiyun Semiconductor. In addition, there is the EYEQ computing power platform which also belongs to Zhiyun Semiconductor.

These two computing power platforms monopolize almost all of the global assisted driving, autonomous driving and part of the robotics market.

If Zhiyun's two terminal computing power platforms are out of stock, the production lines of assisted driving and self-driving cars of global automakers will have to stop.

The reason why Zhiyun Group can achieve monopoly in these two fields is that on the one hand it relies on its hardware advantages, and on the other hand it relies on its ecological advantages.

The technical and ecological barriers established by Zhiyun Group in the field of computing chips are very high.

AMD has been playing with training cards for several years, but due to ecological issues, it still hasn't made any waves, and now AMD is lying flat.

Although there are competitors such as Qualcomm in the terminal computing power platform, they are no match for Zhiyun Group... because the computing power platform is not just about hardware, but also about the ecosystem.

For Zhiyun Group, it not only sells computing power platforms, but also provides a full set of development software for free to global customers and even individual enthusiasts.

After Zhiyun Group acquired NVIDIA, it integrated NVIDIA's GPU software development ecosystem with its own GPU software development ecosystem and CPU development ecosystem, and then formed a complete and all-encompassing software development platform in the field of artificial intelligence.

Globally, anyone who studies, develops using GPUs, or even makes games, is using Zhiyun Group's platform. Even students majoring in related fields at universities around the world are learning this platform!
This shows the huge monopoly advantage of Zhiyun Group in the field of computing power chips.

In this case, Zhiyun Group naturally wants to further expand the market capacity of computing chips to make more money!

Today, as the wave of generative AI is forming, Zhiyun Group has also released several different types of open source AI, including generative AI, AI for robots, and AI models in other fields.

This ultimately completely ignited the market demand for computing chips!
On April 19, Xu Shenxue was at the Zhiyun Group headquarters. The computing chip market was booming, and customers from all over the world were placing orders enthusiastically, which led to insufficient production capacity.

A special supply chain meeting for computing chips was held.

During the meeting, Ji Chenghe, the group's senior vice president and head of marketing, introduced the current business progress of computing chips: "At present, our APO series graphics cards, mainly APO4000 and APO4500, have received orders from home and abroad worth US$ billion. The main domestic customers are large Internet companies such as BTA and Yihai Technology, and overseas customers are high-tech companies such as Google, Microsoft, Apple, and Tesla!"

"This is only temporary. Our business department is still communicating with a large number of customers. In the next few months, we will receive more orders for data center GPUs, and the total orders are expected to reach the level of US$15 billion."

"At the same time, our terminal computing chip business has also been affected and has grown significantly, and the sales of the PX platform and EYEQ platform will increase significantly!"

"In addition, due to the explosion of the overall GPU market, our civilian GPU business has also seen growth, especially the X series graphics cards used by many digital currency fields and several other high-end graphics cards, which have seen price increases and increased orders."

"The entire GPU business, excluding the integrated GPUs of the CPU business unit and the SOC business unit, only counts the independent GPU business and the comprehensive computing chip business. Our GPU business is expected to reach a revenue of $20 billion this year!"

"The above business does not include the AI ​​series chips, EYQ series chips, and ZY series chips purchased by the group for its own use and some strategic cooperation institutions. If they are included, the number will be even greater."

The AI ​​series chips are still a confidential project in Zhiyun Group. Although the outside world knows that Zhiyun Group has a data center GPU chip for its own use, Zhiyun Group has never announced it to the public. The chip is still only provided to the group for its own use and supplied to the defense field. Even companies such as Yihai Technology, Hailan Automobile, and Nanmen Aerospace cannot get this chip.

The performance of the AI ​​series chips is so powerful that it completely exceeds the level expected in this era. Its impact is extremely significant. It not only concerns Zhiyun Group’s core competitive advantage in the field of artificial intelligence, but also a large number of secret national defense projects.

Therefore, it is impossible to sell the AI ​​series chips to the outside world, let alone export them... The authorities have long classified this series of chips as secret technology and prohibited their disclosure, let alone export.

This thing is a confidential technology. This does not mean that Zhiyun keeps it confidential, but it is confidential in the legal sense.

If someone leaks relevant technical information, people from the security agency will come to his house and invite him for tea in minutes.

The EYQ computing platform and ZY series chips are not confidential projects, but in order to maintain the core competitive advantage of his companies, Xu Shenxue only approved their own use and supply to sister companies, such as HaiLian Automobile and Xiaojiang Technology.

Moreover, HaiLian Auto also has a certain say in the EYQ chip. According to the cooperation agreement between the two parties, Zhiyun Group cannot supply EYQ chips to other automobile manufacturers.

It is also worth noting that some defense projects also use the EYQ platform extensively for development and testing during the development phase. For example, the terminal computing chip used on the prototype of the X1 fighter is the EYQ3 chip based on the -nanometer process.

However, Zhiyun Semiconductor Department has also developed a defense-specific computing power platform in accordance with the technical requirements of the cooperative institutions. The project is code-named GF Computing Power Platform. The first-generation GF1 chip is based on a ten-nanometer process and adopts extremely stringent design and packaging requirements. The chip is required to be resistant to high and low temperatures, interference, impact, radiation and many other requirements.

The chip has been successfully tested and put into small-scale production and is being used in some secret projects.

The production of this chip was even earlier than that of the S803 chip... There was no other way, as a large number of projects from other cooperative institutions were waiting for this chip.

Considering the high R&D cost and low production volume, the price of the GF1 chip will be very expensive...far exceeding the unit price of civilian products, and the unit price may reach several million yuan per chip.

Its high price is not due to high technical requirements and manufacturing difficulties, but because of low production volume, which cannot spread the huge R&D and production costs.

For a chip, the R&D cost is the same regardless of whether the demand is 100 or 1 million.

Some hardware expenses are also indispensable in the production process, such as the production of masks. Even if you produce one chip, you have to make a complete set of masks, which makes the manufacturing cost of small-volume chips very expensive.

The limited production and usage are the core factors that lead to the sky-high price of GF1 chips...not the high cost caused by more difficult technology.

However, the GF1 chip is not a chip dedicated to fighter jets. It is a general-purpose terminal computing chip that can be used in other artificial intelligence terminal fields in the future, thereby increasing demand and gradually reducing costs.

However, no matter how much it is lowered, it is impossible to lower the level of large-scale mass-produced chips.

------

Considering the limitations of self-use, the market shortage of computing chips caused by this wave of artificial intelligence will not actually affect these series of products.

What Ji Cheng and they are discussing now is mainly the various computing chips that can be exported on a large scale.

The large influx of orders in the short term has also posed a huge challenge to the production capacity of Zhiyun Microelectronics, a subsidiary of Zhiyun Group.

Ding Chengjun, head of Zhiyun Microelectronics, said: "The process technology of our computing chips is concentrated on the 14-nanometer process and the 12-nanometer process. Although our production capacity of these two processes is the highest in the world, we also need to use these two processes for a lot of products." "Because at present, the main core logic chip products used by our group for self-use and external sales are basically concentrated in these two processes, including mobile phone SOC chips, computer CPUs, server CPUs, consumer-grade GPUs, some functional chips, etc. are all designed, produced and sold around these two processes!"

"The data center GPU chips and terminal computing chips, including APO4000, APO4500, EYQ3, EYEQ3/4, ZY series chips, as well as the group's large-scale self-use AI5500 chips, have occupied a large amount of production capacity!"

"Currently, our 14-nanometer process, including the 12-nanometer process, has a total production capacity of only 280,000 wafers per month, and the production capacity is already running at full capacity!"

"Considering that the S803 chip in the S series uses a 703-nanometer process and will be mass-produced in the future, and that the shipments of the S603 chip using a -nanometer process and the S chip using a -nanometer process continue to decline, we should be able to free up about chips per month in the future and convert them to the expansion of the APO series of chips!"

"At the same time, we are upgrading the technology of Fab 8 and expect to provide 50,000 wafers of 14-nanometer process capacity in July."

"By this fall, our 14nm and 12nm process production capacity will reach the expected 330,000 wafers per month, initially alleviating the capacity shortage problem of this process."

"At the same time, we are building the third production line in the third plant. This production line is a 12-nanometer process production line with a designed production capacity of 40,000 pieces per month. It is expected to be put into production in the middle of next year."

"At the same time, we are also promoting the technological transformation of the 17th plant, expanding the production capacity of the 12-nanometer process to 30,000 pieces. The project is expected to be put into production next winter!"

"By next year, our 14-nanometer and 12-nanometer process production capacity will reach 400,000 wafers per month, far exceeding other competitors and meeting market demand."

"We believe that this production capacity has been able to meet the initial requirements. Although there are still many potential orders, it is worth noting that we are constantly advancing process technology, and with the advancement of process technology, the process nodes of many advanced semiconductor products will also follow."

"Just like the S series chips, we will promote the production of the S803 chip with -nanometer process on a large scale in the first half of this year. The production capacity of this part will reach more than pieces per month. At the same time, starting from the end of the year, we will also promote the flagship chips of the W series and the AI ​​series chips to move towards the -nanometer process."

"By then, orders for 14-nanometer and 12-nanometer chip processes will gradually decrease!"

"For this reason, we plan to maintain the production capacity of the 14nm and 12nm processes at 400,000 wafers per month!"

“And then invest more funds and technical personnel into the expansion of advanced 7-nanometer technology!”

"At the same time, we are also continuing to expand our advanced production capacity of 28-nanometer technology, because a large number of terminal devices, including many chips for automobiles, except for computing chips, have lower process requirements for other conventional chips, which are generally based on 28-nanometer technology."

"We plan to gradually increase the production capacity of the 28-nanometer process to 300,000 wafers per month."

"As for the more backward 40-nanometer and 45-nanometer processes, as well as the mature processes such as the 55-nanometer and 90-nanometer processes, we currently have no plans to expand production. Instead, we will continue to use the current production capacity, which adds up to about 400,000 pieces per month!"

"The main reason is that other domestic semiconductor manufacturers are following very quickly. They are expanding their production capacity very quickly in mature processes above 28 nanometers, causing foundry prices to continue to fall. It is no longer worth it for us to continue investing!"

In the past two years, especially after the domestic Gulf Technology's lithography machine technology matured and was supplied on a large scale, and even Zhiyun Group began to license patents for mature processes, the domestic semiconductor manufacturing industry has developed extremely rapidly.

Many domestic second-tier manufacturers have invested in the construction of mature processes above 55 nanometers, resulting in extremely rapid expansion of production capacity in this field.

At the same time, several other first-tier domestic manufacturers, such as SMIC, are stepping up efforts to expand the production capacity of 28 and 45 nanometer processes.

Then SMIC is still thinking about developing the 14nm process, but they can't do it in a short time, and it will take time... After all, this process node is the main profit node of Zhiyun Microelectronics, and the process is not licensed to others. They have to rely on their own strength to do it, which is not easy.

Looking at the entire domestic semiconductor manufacturing industry, it is basically Zhiyun Microelectronics that is impacting advanced processes in higher fields, while focusing on the two advanced processes of 14 nanometers and 28 nanometers.

Several other first-tier domestic manufacturers are focusing on the 28-nanometer, 45-nanometer, and 55-nanometer process nodes.

Several second-tier manufacturers focus on mature and backward processes such as 45-nanometer, 55-nanometer, and 90-nanometer processes.

The levels are very clear...

Under such circumstances, Zhiyun Microelectronics is more inclined to promote advanced processes and expand production capacity...Mature processes are now severely involuted and profits have dropped a lot. Zhiyun Microelectronics no longer values ​​orders for these mature processes.

Ding Chengjun introduced the situation of Zhiyun Microelectronics, while Xu Shenxue was looking at the monthly production capacity plan of 400,000 pieces for the 12-nanometer and 14-nanometer processes.

He did a quick calculation and concluded that the production capacity was generally sufficient and no additional investment was needed…If we really invest in large-scale expansion now, we will have overcapacity and insufficient capacity utilization in two years.

The focus is still on the current 10-nanometer process and the future 7-nanometer process.

As for the 803-nanometer process, the current production capacity planning is not much, only at the level of 1 pieces per month. In the early stage, it is mainly used for the production of S4 chips, and then some GF chips and EYQ chips are appropriately produced, but the production capacity occupancy of the latter two is relatively low.

This is a typical transitional process. Neither Zhiyun Microelectronics nor Zhiyun Semiconductor plans to invest too many resources in this field. After the production capacity of the S series chips is migrated to the more advanced equivalent seven-nanometer process, this process node will be used to produce other computing chips, including more advanced APO graphics card chips, etc., without worrying about waste of production capacity.

The process where Zhiyun Group has really invested significant resources is the equivalent of seven-nanometer technology!

Ding Chengjun also introduced at this time: "We at Zhiyun Microelectronics have invested heavily in the equivalent 7-nanometer process. In addition to the two production lines in the original 25th plant, the planned production capacity is 50,000 pieces per month."

"At the end of last year, Zhiyun Microelectronics officially launched the 28th Plant Plan, planning to build a chip factory in the Shenzhen base with a monthly production capacity of 50,000 pieces and a process level equivalent to the 7-nanometer process. The estimated investment is more than 7 billion US dollars."

"Eventually, we will be able to produce 100,000 chips per month using the equivalent 7-nanometer process, which will be used to meet the production needs of S series chips, W series chips, and various newer generations of AI chips!"

"However, the production capacity planning of the equivalent 7-nanometer process using DUV immersion lithography machines is not close to the limit of 100,000 pieces per month. This is mainly because the cost of using DUV immersion lithography machines to produce equivalent 7-nanometer process chips will be very expensive. Except for S series chips, W series chips and AI chips, other types of chips cannot afford such expensive costs."

"Blindly using DUV immersion lithography machines to expand the production capacity of the equivalent 7-nanometer process may lead to insufficient capacity utilization in the future!"

"In terms of production capacity planning for the equivalent 7-nanometer process, we are more looking forward to the technological breakthrough of EUV lithography equipment from Gulf Technology."

"Gulf Technology has already started manufacturing the first prototype of an EUV lithography machine. Its design parameters are used to manufacture chips equivalent to a seven-nanometer process. Although it is still in the prototype stage, we expect to be able to get a mass-produced model within three years, and then use it for large-scale mass production of the equivalent of a seven-nanometer process!"

“By that time, the cost of the equivalent 7-nanometer process can be significantly reduced and used in the production of other conventional advanced chips!”

Xu Shenxue said at this time: "The capacity planning is not a big problem. Gulf Technology has reported to me the progress of the EUV lithography project many times. Overall, it is going smoothly. There will be no problem in producing a prototype by the end of the year!"

"Because all the core technologies of EUV lithography machines are self-developed and controllable by us, the subsequent promotion of commercial mass production models will be relatively smooth, and there will be no problem in delivering them within three years!"

“It is feasible to plan the progress of the equivalent 7-nanometer process according to the progress of the EUV lithography machine!”

"As for the APO graphics cards and several other GPU products that use advanced technology, there is nothing we can do about the current limited production capacity. We need to explain to customers and ask them to wait a little longer until next year."

“The marketing department needs to do a good job with customer work!”

Ji Chenghe immediately said: "Don't worry, Director Xu, it's not a big problem. Anyway, there are no competitors now, so they have to wait whether they want to or not!"

After a meeting on chip production capacity, Zhiyun Group also notified its customers one after another.

The main purpose is to inform customers that we will be able to free up some production capacity in the near future, and there will be another batch of production capacity in the fall, and there will be even more next year, so don’t worry that we won’t be able to deliver the goods... Although the waiting period is a bit long, the goods can definitely be delivered!

At the same time, in order to achieve the greatest possible balance, Zhiyun Group also tries to balance the delivery time and quantity for major customers, rather than completely following the order schedule.

After hearing this, the customers tried various ways to get more APO series graphics cards as soon as possible, especially the most advanced APO4500 graphics cards.

If that doesn't work, the APO4000 graphics card will also work.

During this process, the business department of Zhiyun Group also gradually raised the prices of APO graphics cards. For different customers and different orders, the increase ranged from 30% to 50%, and the selling price ranged from 100,000 to 150,000.

In principle, the closer the delivery deadline is, the more expensive the price is. If the delivery deadline is relaxed to next year, then the price will be relatively cheaper and you can buy it for 100,000. If delivery is required in the near future, then it will be 150,000... and a high deposit is required.

As for spot stock...there is no spot stock, it has been sold out long ago!
Not to mention other companies, even brother companies like Yihai Technology and HaiLan Automobile cannot buy APO graphics cards in stock.

APO graphics cards are hard to come by and their prices continue to rise. At one point, the APO4500 graphics cards in the hands of card dealers were priced at over yuan each, but even these were snapped up by high-tech companies.

This matter was initially circulated within the industry, but was soon noticed by the technology media and then reported.

At this time, ordinary people discovered that the graphics cards sold by Zhiyun were so expensive...

A graphics card actually costs hundreds of thousands, this is crazy!
Some technology bloggers were so annoyed that they analyzed the cost of the APO4500 graphics card in detail... and came to the conclusion that its cost was only about yuan at most... But Zhiyun's wholesale price for this graphics card to large customers was as high as yuan per piece!

Zhiyun Group simply added a zero to the cost and sold it out... Even so, a large number of high-tech companies were still very happy and even cried and begged Zhiyun Group to sell them more APO4500 graphics cards.

Someone described Zhiyun Group's AI graphics card business like this: It could have robbed you of your money, but it gave you a graphics card for free. This guy is pretty nice...

Due to tight production capacity, the price of APO graphics cards has increased, which has also affected the prices of other graphics card products under Zhiyun Group.

For example, the price of Zhiyun's high-end gaming graphics card, the X2 graphics card in the X series of graphics cards, which can actually be barely used for low-end AI training, has increased from just over RMB 2 to more than RMB 2... Zhiyun itself did not increase the price that much. It was mainly because graphics card manufacturers were out of stock and could not get enough X chips, which eventually led to the price increase of X graphics cards.

Zhiyun Group does not directly produce graphics cards in these consumer-grade graphics card fields, but only sells GPU cores. Graphics cards are manufactured by other graphics card manufacturers.

Not only high-end ones, but even those ordinary gaming graphics cards that cost several thousand yuan have increased in price significantly... Because many mining bosses used to be able to purchase APO series graphics cards for mining, but now APO graphics cards are completely isolated from them and are simply not available. Moreover, the prices are too expensive.

So they set their sights on the X-series graphics cards, which are also very useful for mining.

Some mining bosses simply use high-performance gaming graphics cards worth thousands of dollars to mine.

This situation was passed down layer by layer, leading to a sharp price increase and shortage in the graphics card market.

Then...a large group of ordinary people started cursing!
Good job Zhiyun, I thought you were the leader in domestic semiconductors. After acquiring NVIDIA, you should be able to provide domestic players with cheap graphics cards... But I didn't expect that graphics cards would be sold at higher and higher prices.

The price of X2 graphics card has increased from over to over ... they are so shameless.

Zhiyun Group will naturally not give a positive response to this... After all, tight production capacity leads to price increases, which is good for Zhiyun Group itself.

However, in order to solve the shortage of gaming graphics cards on the market and avoid being usurped by AMD and grabbing market share!
Zhiyun Group has also taken countermeasures: all GPU products based on the 28nm process have lowered their prices across the board, which means that old graphics cards have been reduced in price!

By lowering the price of old graphics cards, we can avoid losing users... If you can't afford high-end graphics cards, you can still buy cheap old graphics cards.

Some ordinary individual users are happy about this, while others are crazy...

There are all kinds of ideas.

But one thing is certain: even if high-end graphics cards are produced in China, they will still be extremely expensive... It’s just a dream to think they can be sold at a bargain price!
(End of this chapter)