In recent trading weeks, the Hang Seng China Enterprises Index saw a substantial increase of over 4%, leading global marketsWithin this context, the Hang Seng Tech Index outperformed with an increase of over 6%. The rapid development of Artificial Intelligence (AI) in China has continued to expand, with major international investment firms consistently releasing reports that upgrade their ratings on Chinese assets.

However, the narrative surrounding AI's growth doesn't yet account for the larger implications in the long termAs we delve deeper into the potential of AI, it becomes evident that its future is intrinsically linked to energy consumptionThis idea is reminiscent of the observations made by British economist William Stanley Jevons, who noted that improvements in steam engine efficiency led to increased coal consumption rather than a decreaseThis phenomenon aligns with the fundamentals of microeconomic demand theory: when something becomes cheaper, its demand often escalatesThus, as AI technologies become more affordable, the energy consumption required for their operation is expected to grow, leading to potential shortages in energy resources.

Yet, the current market discussions do not extend this far into the futureThe pressing issue today isn't a lack of electricity but rather that most individuals find themselves unable to effectively utilize AI technologies, often encountering poor user experiencesFor instance, when a user poses a question to an AI, they often face instances of spinning icons as the server remains busy processing requests.

The question arises: why is this the case?

The majority of AI companies are heavily focused on a singular goal—developing Artificial General Intelligence (AGI). AGI is characterized by its ability to complete at least 80% of the tasks that a human would typically perform, ultimately freeing individuals from tedious and labor-intensive chores.

Most resources within AI enterprises are devoted to training models for AGI

Advertisements

Consequently, there often isn't enough capacity to provide ordinary users with free access to these technologiesA service that is free might attract users, but it entails significant costs to maintain—hence, the servers frequently experience overloads.

To address this growing problem, society must allocate substantial computational powerThis means investing in high-performance chips that can deploy AI models effectively, assisting users in their quest for informationUsers themselves do not require chips or models; they are simply in search of answers to their queriesHence, it can be argued that the core of the AI industry revolves around the sale of computational power.

To illustrate, consider the necessity of building an enormous centralized kitchen that prepares food for everyonePeople wouldn’t need to hassle with their own kitchens; instead, they would simply order takeoutIn this analogy, purchasing computational power is akin to ordering food.

Amidst these developments, a significant transformation is taking place within the domestic operators.

Telecommunication operators, historically focused on selling data plans that included 4G and 5G services, invested billions annually to enhance internet speeds through the construction of 5G towers to ensure comprehensive network coverage.

However, today, the demand for internet speeds and coverage has become saturatedMany people may not fully grasp the implications of 4G and 5G technology.

To clarify, 4G technology boasts a download speed of 100 Mbps, while 5G can reach 1 Gbps, and the anticipated 6G promises an astounding 10 Gbps! Each innovation offers a tenfold increase in performance.

Despite these advancements, even bandwidth-intensive applications like streaming 4K movies typically require only 10-20 Mbps, indicating that 4G is presently sufficient for most users

Advertisements

This disparity reveals that the enhancements provided by 5G are rarely leveraged, leading to a surplus of both data and internet speed.

Consequently, over the past few years, telecom operators have steadily reduced their capital expenditures, ceasing the construction of new towers, and discussions surrounding 6G technology have dwindled, essentially due to this surplus in network speed.

As society reflects on its current reality, it has become increasingly apparent that the true bottleneck is not in speed but in computational powerThe role of telecommunication operators is evolving from merely selling data plans to providing computational resourcesThis transition involves acquiring advanced chips, such as the Ascend 910B, to establish computational capacity platformsIn this shift, operators are poised to become the foundational infrastructure providers for the AI era.

It is important to highlight a key difference from the past: utilities such as water, electricity, and internet services are typically managed by state-owned entities with minimal competitionIn contrast, computational power operates in a highly competitive marketBesides telecom operators, large internet firms and various capable enterprises can enter this space and act as power providers.

This intensified market competition might erode profit margins, but it also catalyzes rapid technological innovation and enhances societal valueCurrently, we are witnessing a phase of frenetic investment, where capital markets are encouraging a focus on investment scale over immediate profitability.

As we forge ahead, there is no turning backThe race in the technological arena has begun, akin to the Olympics, marking a new era of competition for advancement.

Advertisements

Advertisements

Advertisements