Nearly thirty years old, established in science and technology

Chapter 23 Just like a child holding gold in the busy city

After a busy day, Ji Hong and Cheng Hui went to bed early. Downstairs, a floor away from them, Miao Wenlu and Professor Zhang Mingzhen were talking about Ji Hong.

"I can understand his caution. His current state is like a child holding gold. He knows the dangers of the downtown area, so he hides it.

"However, you have to face it after all. You can't hide it forever, right? So he scraped off some gold particles little by little and sold them. But if he sold more, everyone would know that he had a big pile of gold at home.

"He is very cautious, thoughtful and considerate. He is indeed safe within the rules, but he is afraid that someone will not abide by the rules!"

"Is it so exaggerated?" Miao Wenlu didn't understand technical things, but at this moment she obviously knew what Zhang Mingzhen was talking about, which was Ji Hong's large model: "It has a big impact?"

“Essentially, this is a war!” Zhang Mingzhen mused: “To this day, I believe everyone can see that artificial intelligence is already an important tool that can influence or even change the future.

"Some fierce people even put forward the argument that 'computing power is national power, and it is technological nuclear weapons in peacetime.' Although this argument is fierce, it is not unreasonable to a certain extent.

"You can tell from Americen's reaction that all their previous restrictions were only around chip design and manufacturing companies, and never involved purchases.

“In other words, the purpose of restricting us is to make us give up research and development. As long as you don’t do research and development, we can sell you anything.

“But have you noticed that the treatment of AI chips is completely different—it’s basically all-round pursuit and interception.

“High-performance computing cards are directly banned from sale, and even the gaming graphics card Nvidia RTX4090 has been affected to a certain extent.

"For this reason, they even did not hesitate to let the corresponding market push Huawei's second choice - Shengteng. This is why they would rather kill the enemy by one thousand and lose eight hundred by themselves. This can be seen from this.

"Why? What it limits is our cloud AI computing power. Because of the lithography machine, Ascend's production capacity is limited and difficult to increase in a short period of time.

"The total domestic high-end AI computing power will be affected by this and will be difficult to improve in the long term. And artificial intelligence training is a field that takes one step at a time, and the impact is not insignificant."

"Why haven't I heard you talk about this before?" Miao Wenlu saw Zhang Mingzhen talking so excitedly in one breath, and comforted her: "There is always a way."

"Yes!" Zhang Mingzhen was also confident: "The high-end computing power is not enough, won't we use mid-range or even low-end computing power to increase scale?
“In the field of supercomputing, it was not possible to buy high-end CPUs from Intel back then, but even with the size of the stack, didn’t we take the first place in the world?
"What they can't do and think no one can do it, we have done it more than once, so the AI ​​field will definitely fail? Where do they get the confidence?
"Moore Thread's kilo-card intelligent computing cluster was our original attempt." Zhang Mingzhen paused for a while before continuing: "But now, I think we may have a better choice."

With that said, he pointed to the roof of the building. Obviously, the better choice was Ji Hong.

"Then he is no longer a child holding gold in the busy city. This is equivalent to holding nuclear weapons in the world."

Miao Wenlu also realized the extraordinary peculiarity of this matter. In this kind of game, sometimes the opponent does not follow the rules very much:
"Then get ready..."

"Don't our school intend to participate in the development and pre-selection of some equipment for the [Chang'e-8 mission of the fourth phase of the lunar exploration project]? You also know the strength of the school, and participation is the most important.

"But after meeting Ji Hong, I had a new idea, which might work." Professor Zhang Mingzhen said thoughtfully.

"Are you going to help him, or are you going to let him help you? If this happens, your school will obviously benefit more, right?" Miao Wenlu knew that her lover's school, including himself, had always been involved. Some peripheral projects in aerospace have also been selected, but the core technology is not even included, which is also a joke.

"What are the advantages and disadvantages? We have the qualifications and at least have the opportunity to be shortlisted." Professor Zhang Mingzhen is actually right. Unless this type of project is an open call, it is a little more core and only relies on Juaner intelligent technology. It’s really hard to touch the door:
"Once selected, you can directly establish a relationship with the Aerospace Electronics Technology Research Institute. That unit is a military industry in the military industry. If Juaner Intelligent Technology has the opportunity to become a strategic supplier, then think about it...

“From the perspective of AI capabilities, equipment such as planetary exploration vehicles and work robots used to be program-controlled. If we use advanced AI, especially end-test AI that can operate with low power consumption in extreme environments, the competitiveness will definitely be unparalleled. .

“Based on the strength of our school, compared with other universities and research institutes, it is indeed inferior. After all, it is just a 211, with limited strength and even more limited funds.

"But if we form a consortium with Juaner Technology, at least when it comes to AI, no one in the world will be afraid of anyone! We have the qualifications and Jihong has the technical capabilities. This can be considered mutually beneficial!

"Go to sleep. I've made an appointment with this guy to visit the laboratory tomorrow. I'll ask him about his intentions then."

……

"Professor Zhang, of course I have no problem. It is my duty to contribute to the aerospace industry."

This project is a long-term project. In addition to research and development, it will also undergo long and rigorous testing. The actual application is estimated to be several years later.

Of course Ji Hong agreed wholeheartedly. Such an opportunity was hard to come by, so there was no reason not to agree.

There are two things like making money or not. As an aerospace equipment supplier, this style has gone up in an instant.

"Now I just have this idea, and the school may not support it." Professor Zhang Mingzhen took a precaution:

"However, it shouldn't be a big problem, because there is really no hope for us to do it ourselves. As long as you let the school see the strength of Juaner smart technology, then this thing can be done.

"The main project in the early stage is Moore Technology's kilo-card intelligent computing cluster project. I see that Mr. Zhang is very enthusiastic and will confirm it as soon as possible if it is suitable. To be honest, he asked me to talk to him about the project, and I brought it to him, and he decided to do it himself! "

It is true that one family does not belong to the same family. The two couples have similar personalities and are very interesting. Zhang Mingzhen directly explained everything to Ji Hong.

"Well, I will consider it." Ji Hong also smiled. Moore Technology is indeed a natural collaborator of Juer Intelligence at this stage.

Ji Hong really wants to develop the chip himself, but it must be a new type of GPU chip that is fully suitable for [thinking-like model] training or a next-generation CPU. However, at present, the time is not mature, and he does not have a complete plan. Still in the stage of sorting out my thoughts.

As for the traditional GPU, there is no need to do it again - the core of everything at Juaner Technology at this stage is speed, quick entry, quick effect, and quick influence.

And if you end up buying a traditional GPU, not to mention the investment, even if someone is willing to invest, it will take at least a year and a half. For Ji Hong, this is not cost-effective.

(End of this chapter)