"It's currently under urgent construction and is expected to be operational in about half a month," said Lei Bus, the general manager of Geek Technology.
Geek Technology is currently building the world's largest supercomputing center to facilitate the next-generation third-generation AI Xiao Ke.
Of course, this third-generation AI Xiao Ke doesn't actually have much connection to the previous two generations, although they are all called AI Xiao Ke.
This is because the technical route taken by this third-generation AI Xiao Ke is actually the generative AI technology route of ChatGPT from his previous life.
After all, Gao Nian came from 2024 and naturally knows the prospects of ChatGPT-type generative AI technology routes.
Therefore, when Gao Nian wanted to develop a stronger and smarter AI.
After much deliberation, Gao Nian still chose to follow ChatGPT's generative AI technology route.
After all, there were many types of AI technology in his previous life, but ultimately only ChatGPT's generative AI technology route became a hit for a reason.
So, considering that countless scientists and companies in his previous life had already paved a proven path with blood, sweat, tears, and money.
Gao Nian would naturally choose the simplest, already proven path.
However, because Gao Nian was just an ordinary worker in his previous life, he didn't know the specific technical principles and technical details of ChatGPT at all.
Although he had seen popular science videos from Bilibili (B Site) content creators like "Neolithic Park" and knew the history of various types of AI and generative AI technologies.
He also knew the general principles of the Transformer model, but transitioning from general principles to practical implementation was still a huge challenge.
However, no matter how big the challenge, Gao Nian had to complete it, because if Geek Technology wanted to make purely domestically produced computers rise, they had to develop their own AI ChatGPT!
Because relying solely on the second-generation AI Xiao Ke to analyze game data.
And then use the Geek Super Game Engine to model an identical game scene at a 1:1 ratio was, to be honest, somewhat difficult.
After all, the essence of the second-generation AI Xiao Ke is just an intelligent voice search engine, and although it is very powerful in this era.
It is still quite lacking compared to the AI of his previous life, with a significant difference in intelligence and creativity.
So, it is really difficult to have the second-generation AI Xiao Ke create an exact replica of a game using the Geek Super Game Engine at a 1:1 ratio based on a game.
Therefore, before developing the Super Game Engine, Geek Technology had also prepared in advance to develop the third-generation AI Xiao Ke.
What was this preparation?
Naturally, it was to organize a large workforce to prepare a large amount of training data in advance to facilitate future AI training.
On one hand, they purchased a large number of Intel CPU processors and used Phoenix 1090 graphics cards to build the world's largest supercomputing center.
As for where this world's largest supercomputing center was built?
This world's largest supercomputing center was not built in Yue Guang City, nor was it built in Shen City, but rather in Rong City in Chuan Province.
Because Guang Province has a developed economy and its industrial production capacity is the best in the country, with many industrial products in the world being produced in Guang Province.
This has also led to a surprisingly high demand for electricity in Guang Province, and AI itself is a big consumer of electricity.
In his previous life, experts from the United States conducted statistical analysis based on the global AI power consumption and growth over the years.
They found that the power consumption of AI is increasing year by year, especially after 2022, with the explosive popularity of AI ChatGPT.
Companies and research institutions around the world have also invested a lot of money in the AI field, which has led to a sharp increase in the power consumption of AI.
If the situation continues to grow like this, then in 2030, the total power consumption of AI will reach 5% of the total power consumption!
Although it is a bit exaggerated for AI to consume 5% of the electricity, it is still possible to shrink it down to about or .
And ~ may not seem like a big number, but in reality, this is talking about the global power consumption.
And on a global scale, there are actually only a few countries that can afford to play with AI.
After all, if you want to play with AI, you need to invest huge resources to build supercomputing centers and data centers.
After that, you have to spend expensive prices to buy graphics cards and CPUs, and after all that is done, the huge amount of electricity consumed is another big problem.
Therefore, AI is not something that some small and poor countries can afford to play with, only those countries with abundant electricity resources and funds can afford to play with it.
Among them, China is undoubtedly a leader, so China will naturally become the world's supercomputing hub,
providing AI services to the world.
In this case, because it not only serves the domestic market, but also the global market.
In the end, China's AI power consumption will definitely exceed of the proportion, and breaking through 5% is not impossible.
Therefore, Geek Technology must be prepared to consider in advance.
They cannot allow their AI Xiao Ke to frantically compete for the electricity of industrial production, leading to the forced reduction or stagnation of industrial production scale.
So, after much deliberation, Gao Nian decided to put the world's largest supercomputing center in Rong City.
After all, that is the hinterland of China, easy to defend and difficult to attack, with significant strategic significance.
In addition, there are a large number of renewable electricity resources in the northwest direction, and using renewable electricity resources to support the operation of AI can greatly reduce environmental protection pressure and protect the environment.
So, Gao Nian finally placed the location of the supercomputing center in Rong City, and it is expected that Rong City will become the world's largest supercomputing hub in the next few decades.
As for the specific location of the supercomputing center in Rongcheng, the location is on the shore of Xinglong Lake.
This place had become the hub of the Chuan Province Supercomputing Center in his previous life, which shows the superiority of its geographical location.
Finally, it is close to the Science City of his previous life, and the surrounding environment is beautiful, so it is naturally the best choice.
At this time, the construction time of this supercomputing center has actually been nine months. It can be said that it started from the time Gao Nian decided to develop a purely domestically produced computer.
At this time, it took nine months to build, and he didn't expect that it would still be more than half a month late before it could be initially put into use.
"I understand, but we can't wait for the supercomputing to be completed before starting the research and development of the third-generation AI Xiao Ke. We must do a good job in preliminary research in the future," Gao Nian said of his decision.
"Yes, I will make corresponding arrangements," Ni Guanghai, the director of the Technology Research and Development Department, agreed.
Although Gao Nian is actually leading the research and development of the third-generation AI Xiao Ke, and Professor Qiu Chengtong and Academician Ni Guangnan are assisting in the research and development.
But the work of Ni Guanghai, the director of the Technology Research and Development Department, is not easy. He must do a perfect job in logistics and not delay the progress of research and development.
"What is the decision of the Xia Academy of Sciences regarding our proposal to jointly develop the third-generation AI Xiao Ke?"
