Rebirth 08: Rise from copycat phones

Chapter 639 The First YANC Computing Center
On October 9th, Xu Shenxue took a special plane to Xiyang City in the southwest of China.

After Xu Shenxue's private plane arrived at the local airport, the head of the local branch of Zhiyun Group and a group of management personnel were waiting at the airport.

Not only the head of the local branch of Zhiyun Group, but also the local representatives were waiting for Xu Shenxue, the big spender.

Xu Shenxue doesn't come to Xiyang City often, but every time he comes, he brings an extremely large investment... that's all money.

Before long, several police motorcycles drove out first to clear the way, followed by several police cars, and then a large convoy.

In addition to several buses carrying accompanying staff, there were also more than twenty Aqua Blue SLC luxury sedans or Aqua Blue SEV large luxury SUVs.

Many of them are bulletproof versions.

With such a large convoy of police cars leading the way, even an ordinary person could tell at a glance that some important person was coming.

On the way, Xu Shenxue listened to a briefing from the head of the Xiyangcheng branch, which mainly focused on the operation and construction of the local artificial intelligence computing center.

Xu Shenxue traveled a long way to this place for a very simple purpose: to inspect the large-scale computing center cluster of Zhiyun Group in the area, and more importantly, to see the construction progress of the first YANC system.

Xiyang City, located in the southwest region, has attracted many companies to build data centers in the area due to its abundant power resources and relatively cheap electricity prices.

However, the largest investor in the area is Zhiyun Group... Zhiyun Group has invested in and built a large number of data centers of various types here.

Many years ago, Zhiyun Group deployed a large-scale data center here to support data for software and internet businesses, massive amounts of user data from Zhiyun smart terminals, and especially cloud storage data.

Although Zhiyun Group is a company that researches, manufactures, and sells high-tech hardware products, its software and internet businesses are also quite large. The sheer volume of data from its mobile phone users and app store is enormous.

To serve users around the world, Zhiyun Group has not only built a large number of data centers in China, but also in many countries overseas... Its overall scale is enormous.

In addition to building its own data centers, Zhuyun Group later ventured into the cloud storage business in order to indirectly sell its own server CPUs, RAM, and flash memory chips, thus indirectly selling hardware products by selling computing power.

Therefore, as the cloud storage business progressed, Zhiyun Cloud Computing Technology Co., Ltd., a subsidiary of Zhiyun Group responsible for various data centers and cloud computing business segments, became the world's third largest cloud computing company, providing large-scale cloud computing services to domestic and a large number of overseas enterprise users.

The above refers to data centers in the traditional sense!

The truly massive and costly facility within the Zhiyun Group is the artificial intelligence computing center.

In particular, the scale of AI computing power center construction in recent years has been extremely large. Large-scale AI computing power cluster centers have been successively deployed and built in many cities across the country, and the AI ​​computing power scale of the group has reached 600 EFLPOS.

Xiyang City is a key location for Zhiyun Group's strategic deployment in the field of artificial intelligence computing power centers... for no other reason than that the power supply in this area is relatively abundant.

Artificial intelligence computing centers are extremely power-intensive. For example, if an APO5000 or APO6000 graphics card is used as the core, a data center with 100 EFLPOS of computing power would consume approximately 100 billion kilowatt-hours of electricity. This figure is roughly the same as the current power consumption of other companies, such as Google and other overseas companies.

In the past few years, Zhiyun Group has mainly built integrated computing centers that combine superconducting quantum computers and GPUs. The computing cost is several times lower. The annual power consumption of 100 EFLPOS computing power is about 30 billion kilowatt-hours, which is only about one-third of that of other companies in the same industry.

Even though it has a greater advantage in computing power-to-energy ratio than its peers, the scale of AI computing power of Zhiyun Group is too large... so the power consumption is still terrifying.

In this context, the new generation of YANC computing centers, which utilize neuromorphic computing and superconducting quantum computing technologies, along with advanced GPU technology, to form a comprehensive computing center integrating these three elements, becomes extremely important…

The new YANC computing center has further reduced its computing power energy consumption, and the annual power consumption of 100 EFLPOS can now be controlled at 10 billion kilowatt-hours!
Using this technology, Zhiyun Group can build 100 EFLPOS computing power with an annual power consumption of 10 billion kilowatt-hours, while competitors such as Google would need to build 100 billion kilowatt-hours of computing power per year... The difference between the two is tenfold.

This tenfold difference brings more than just a tenfold difference in electricity costs... more importantly, it enables the use of limited power supply to obtain ten times the computing power!

Well, although electricity is expensive, it's nothing to these high-tech companies.

The problem is that there's no way to obtain more power supply in a short period of time. Building power plants takes time and requires huge investments... China has strong infrastructure capabilities to support it, and it's willing to spend money on power construction, as well as a large number of professional and technical personnel. Every year, it can generate more electricity than other countries can dream of. However, even so, it can't keep up with the huge power demand of the Zhiyun Group... This lousy company's new power consumption is often tens of billions of kilowatt-hours, or even hundreds of billions of kilowatt-hours.

When Zhiyun Group communicated with the power company, they proposed that they would need at least one trillion kilowatt-hours of electricity supply over the next ten years... Such terrifying energy consumption would leave many people speechless.

You're just a small company, yet you consume more electricity than the entire country of Japan. That's ridiculous!

The terrifying power consumption of artificial intelligence has led to a problem for Zhiyun Group: it's no longer about manufacturing more chips, but about not having enough power... So they have been trying to find ways to utilize the limited power supply to obtain more computing power.

If even the mighty Zhiyun Group is in such a predicament, then other companies like Google are even worse off... Even if they have the money to buy a massive number of APO graphics cards, they don't have enough power to use them.

Without power, what's the point of playing with artificial intelligence?
For the sake of electricity, Zhiyun Group even came to Xiyang City, a relatively remote place, to make large-scale investments... simply because the area has abundant power resources.

Traditional data centers weren't anything special, and Xu Shenxue had seen plenty of older-generation AI computing power, so he only glanced at these facilities briefly.

However, Xu Shenxue attaches great importance to the YANC computing center, which has already started construction!

"This is the first YANC system computing center built by our company. As it is the first YANC system computing center, it uses a lot of new technologies. To be on the safe side, we designed it to be relatively small in scale. Its designed computing power is 150 EFLPOS, but its power consumption is only 33% of the previous generation of integrated computing centers with the same computing power, and only 10% of the GPU computing center."

"Not only does it have high computing power, but most importantly, it consumes little power!"

The AI ​​centers of Zhiyun Group are all quite large. In the early years, the AI ​​computing centers with GPUs as the core were often between several EFLPOS and twenty or thirty EFLPOS in size. Of course, the early AI computing centers used early AI series chips, which consumed a lot of electricity.

A computing center with 20-30 EFLPOS has an already terrifyingly high power consumption.

With the advent of superconducting quantum computers, the computing power of the designed computing centers has increased significantly. Zhuyun Group began to deploy integrated computing centers that combine superconducting quantum computers and GPU cores. During this period, the computing power of superconducting computing centers was generally between tens of EFLPOS and one hundred EFLPOS.

The new generation of YNAC has a larger computing power scale due to its low energy consumption, reaching a scale of two to three hundred or even four to five hundred EFLPOS!
The power consumption of these computing centers from different eras is actually quite similar... A YANC computing center with two or three hundred EFLPOS does not consume more power than an earlier computing center with only ten or so EFLPOS. However, to be conservative and to experiment with a large number of new technologies, the first YANC computing center that Xu Shenxue sees is relatively small in scale... Of course, this is only relative.

Compared to the scale of computing centers of other companies, it is still a behemoth!

Wei Renzhi, Vice President of Zhiyun Group and General Manager of Zhiyun Computing Technology Co., Ltd., gave Xu Shenxue a detailed introduction to the new generation YANC computing center that is still under construction.

"To maintain secrecy, especially regarding the confidentiality of superconducting quantum computers and neuromorphic computing, the construction of our entire computing center was jointly undertaken by our group's internal engineers and long-term partner companies' fixed construction teams."

“Starting last year, we began preparing the site and buildings needed for the factory for this new project. The entire project covers a large area and consists of five large computer rooms!”

"During the construction process, in order to reduce power consumption, we collaborated with heat dissipation and cooling suppliers to develop a new generation of data center cooling systems by providing some key technologies and financial support."

"Because the new generation of data center cooling systems uses many new technologies, according to our design goals, the overall energy consumption of the project can be reduced by about eight percent compared to the previous generation of cooling solutions."

"Although the new cooling system has a higher initial investment due to the use of more new technologies, its higher cooling efficiency will result in a lower overall cooling cost in the long run, which can effectively reduce the overall computing power cost."

Whether it's a traditional data center or an AI computing center, the biggest cost after it's built and put into operation is electricity... And the huge electricity consumption comes from the system operation itself, while the other huge energy consumption is heat dissipation.

Chips and various devices generate heat during operation... A small amount is not a problem, but this is an extremely large computing center, and the heat generated by the entire system is terrifying, which requires various methods of cooling and heat dissipation...

Therefore, reducing heat dissipation energy consumption is also an important way to save electricity costs and reduce computing power costs.

Wei Renzhi continued, "In addition to adopting a newly designed data center cooling system, we have also upgraded the power supply system, adopting a newly designed uninterruptible power supply!"

"We will use a brand-new uninterruptible power supply system to prevent operational risks to the computing center caused by unexpected power outages!"

"In the field of energy storage, we have adopted the solid-state battery energy storage system from Hailan Battery, a subsidiary of our sister company, Hailan Automotive Group. It has advantages such as high energy density, no risk of fire, and long-term energy storage!"

"In addition, there is a new generation of supercapacitor systems to provide instantaneous high-power power supply. After all, our entire system consumes a lot of power. In the event of an unexpected power outage, we need to provide instantaneous high-power power, which is where the supercapacitor system comes in!"

“We have made significant technological upgrades across the entire power management field, and the construction of the YANC computing center is also an important opportunity to verify the reliability of these new systems and find potential technical problems.”

Although the headquarters of Zhiyun Group had built a YANC system test machine, it was only used to verify YANC. The various experiments were limited to the YANC system itself. However, to truly build a large-scale YANC system like the one in front of us and to ensure the stable operation of the YANC system, a large number of external devices and subsystems are needed to provide assistance. These subsystems and external devices may not be considered cutting-edge technology, but they are still very important.

Ensuring power supply and heat dissipation while minimizing computing costs presents significant technical challenges and tests the comprehensive capabilities of China's industrial sector... This is something that Zhiyun Group alone could not accomplish on its own.

Because it involves too many fields.

Even excluding superconducting quantum computers and GPUs, the core technology of the neuromorphic computing center also involves optical transmission modules—massive amounts of data need to be transmitted and exchanged through optical fibers.

Then there's copper cable transmission... A large amount of information exchange cannot be entirely conducted using fiber optics; much of it still requires traditional copper wires for information exchange.

Then there are the switches in the network equipment, which are also very important.

Next is the cooling and heat dissipation module, which is a systematic engineering project. Heat dissipation itself is not very technical, but how to achieve heat dissipation with less electricity and save power consumption is very technical.

Finally, there is power supply and management, which can be further divided into high-voltage DC power supply modules, uninterruptible power supply modules, backup battery modules, high-power power supply modules, conventional power grid equipment, generators, etc.

The devices that support the operation of the YANC system alone fall into more than thirty major categories, which are further subdivided into hundreds of types, involving hundreds of supply chain companies.

Even excluding the YANC core system, the external support subsystems of such large-scale artificial intelligence computing centers are highly technical. To be precise, building a large-scale, low-energy computing center requires a very high level of technology... most companies can't handle it.

Of course, the most crucial part of the entire computing center is the YANC system; that's the soul of it!
To deploy this computing center, Zhiyun Group will utilize a large number of YCL-51 superconducting quantum computers, APO6000/160GB graphics cards, neuromorphic computers, as well as a new generation of WZ series server CPUs, enterprise-grade high-speed memory, and flash memory.

The entire system is highly technologically advanced, encompassing superconductivity, quantum computing, advanced semiconductors, and artificial intelligence. Significant technological breakthroughs are required in all four fundamental fields to integrate these technologies.

Xu Shenxue is quite satisfied with and even looking forward to the YANC computing center under construction... Although the computing center is still relatively conservative, with a computing power of only 150 EFLPOS, this is just the beginning.

Before long, larger YANC computing centers will be built in other places, and the scale of each computing center will be larger, with single-machine computing power reaching three hundred or even four or five EFLPOS, which is possible... depending on the local computing power supply level.

If a place has abundant power resources and sufficient power supply, then the construction scale can be larger; conversely, the construction scale can be smaller.

Regardless, the scale of cloud computing power for artificial intelligence will usher in a period of significant growth.

The goal of building 2,000 EFLPOS may be achieved in less than three years.

At the same time, the future goal of 5,000 EFLPOS will be easier to achieve!

The massive improvement in overall computing power, even if the core AI algorithm technology remains at its current level, will enable AI technology and its applications to experience explosive growth.

Not to mention that the YANC system has another core function: superior low-precision computing capabilities, making it more useful for performing fuzzy calculations and other similar operations.

This means that Zhiyun Group's AI Research Institute can use the YANC system to train smarter big data models!
This is a role that senior executives of Zhiyun Group, and Xu Shenxue himself, value even more.

While Xu Shenxue was inspecting Xiyang City, a foreign social media website suddenly reported that Zhiyun Group had achieved a key technological breakthrough in the field of quantum computing and had already used superconducting quantum computers on a large scale in its computing center!

This news initially went largely unnoticed, but it was quickly reprinted and reported by major media outlets, thereby attracting the attention of the global technology industry!
What's going on? Zhuyun Group has secretly developed a superconducting quantum computer and is even using it on a large scale?
What is the specific level of their superconducting quantum computer? What kind of effect will it have on the field of artificial intelligence?
Most importantly, how did Zhiyun Group develop a superconducting quantum computer and put it into practical application?
This is not normal!
(End of this chapter)