The computing power network is coming!

In March this year, our country’s daily Token calls exceeded 140 trillion, an increase of more than 1,000 times compared with the beginning of 2024. As the costs of using AI models keep rising, many practitioners say their burden has increased.

Looking back at the early days of mobile internet, mobile browsing speeds were slow and data charges were high. But with the rapid rollout of 4G and 5G networks, data costs kept coming down. What was once a scarce commodity has now become a daily necessity.

So, will a similar development path be repeated again in the AI industry? The answer is: very likely. This is because, this time again, the country has stepped in.

This month, the State Council’s executive meeting mentioned the need to “strengthen the planning and construction of water networks, new power grids, computing power networks, the next-generation communication networks, urban underground pipe networks, logistics networks, and so on,” placing computing power networks and public infrastructure such as water and electricity on an equal footing with one another. At the same time, the same wording was proposed in the Political Bureau meeting of the CPC Central Committee last month. Earlier, the “15th Five-Year Plan” also included a specific statement: “We will build a multi-level computing infrastructure system and a nationwide integrated computing power network.”

In terms of budget, the National Development and Reform Commission said that this year, the total investment in the “Six Networks” and related fields is expected to exceed 7 trillion yuan.

What exactly is a computing power network?

Simply put, a computing power network is like a “computing version of the national power grid.” Just like our electricity supply doesn’t require every household to install generators, because the large power grid is built and managed centrally, a computing power network connects data centers and supercomputing centers across the country into a single network and operates under unified scheduling. When needed, we can directly “purchase computing power” over the computing power network and pay on demand, so costs naturally come down.**

In the future, we may be able to enjoy conveniences similar to “computing power vouchers,” or computing power benefits like packages for water use, electricity use, or data usage.

Token is the basic unit for measuring information processing in AI. The total amount of Token consumed reflects the depth and breadth of the deployment of innovative AI applications such as intelligent agents, while Token cost reflects the efficiency of algorithmic computing power. As technology continues to mature and computing power supply keeps improving, the cost of using Token per unit will gradually decrease.

The construction of computing power networks is not only aimed at meeting public needs; from the perspective of economic development, it is just like building high-speed rail or 5G networks—it has strong “multiplier effects.” It can not only boost investment, but also help upgrade the entire industry chain, including artificial intelligence, big data, and the industrial internet.

Making AI into tools that everyone can use well

Tools you can afford

Precisely this is the value of computing power networks in the future

Source: Xinhua News Agency

Risk warning and disclaimer

        The market carries risks; investment involves caution. This article does not constitute personal investment advice, nor does it consider the special investment objectives, financial conditions, or needs of any particular user. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their specific circumstances. Invest accordingly at your own risk, and you will bear responsibility for any consequences.