He data center electricity consumption It continues to rise, especially due to the expansion of generative AI and conventional Artificial Intelligence. According to the forecasts for this type of installations, over the years next two years your consumption will increase by 160%which according to Gartner will make the 40% of data centers dedicated to AI existing in 2027 they will have operational problems due to the availability of energy.
The consulting firm estimates that the annual power that data centers will need to be able to run servers that will increasingly be optimized for AI will reach 500 Terawatios per hour in 2027. It is 2.6 times more than what they consumed in 2023.
The shortage of energy for data centers that is proposed for the near future will also lead to an increase in the Price of electricity, which will be passed on to users of generative AI systems, with an increase in the costs of working with models. great language.
Furthermore, in the near future, the number of new data centers, as well as the growth of generative AI, will be conditioned by the availability of energy to operate these facilities. Therefore, it is advisable to determine the risks that the energy problems that may exist would affect products and services.
Apart from this, companies should, according to Gartner, assess their future plans, anticipating high energy costs, and negotiate long-term contracts for data center services, with reasonable prices for energy. Companies will also need to take into account significant cost increases when developing plans for the launch and development of new products and services, as well as looking for alternative approaches that require lower energy consumption.
In the long term, new technologies will arrive for the development of improved battery energy storage, or other clean energies, which will help to achieve sustainability objectives again, as well as reduce problems related to energy shortages for centers of data.
In view of the situation, Gartner recommends that companies reevaluate their sustainability objectives in terms of CO2 emissions for the coming years, due to the future requirements of data centers, and the availability of energy sources. In it generative AI application developmentthey should focus on using a minimum amount of computing power, and review the feasibility of other options, such as edge computing and smaller language models.
