What are the reasons for the slowdown in AI growth?

Citing data from Boston Consulting Group, Barron's stated that data centers will consume 7.5% of the total electricity available in the US by 2030. Data centers built for AI systems can consume hundreds of megawatts of electricity per facility, meaning the power supply system will no longer be able to keep up with the rapid growth in the number of data centers.

The energy demands are enormous to power AI servers.

The analysis also indicates that from 2022 to 2030, the energy consumption of data centers in the US will increase from 126 to 390 terawatt-hours, an amount of electricity sufficient to power 40 million American households.

Estimates from 650 Group suggest that the volume of server systems supplying AI needs will increase sixfold from last year to 2028, reaching 6 million units. Gartner forecasts that the average power consumption of an accelerator server will increase from 650W to 1,000W.

Energy consumption will increase not only due to the growing number of servers but also due to conditions. Efforts to optimize energy costs through the application of AI in energy consumption control will partially mitigate this trend, but will not completely solve the problem. The transition to liquid cooling for server systems will be inevitable in most cases. According to Super Micro, data center operating costs can be reduced by more than 40% by eliminating traditional air cooling systems in favor of liquid cooling.

The problem is further exacerbated by the uneven development of the regional power supply network. Additionally, not all locations can efficiently transmit generated electricity to the sites of high-energy consuming systems. Experts believe the US generates enough electricity to support AI development, but the distribution network is problematic.

Source link: https://thanhnien.vn/toc-do-tang-truong-cua-ai-bi-kim-ham-vi-ly-do-nao-185240315100546606.htm