NVIDIA has a confusing idea to reduce the energy consumption of AI
Several electricity companies and technological companies have gathered within the Open Power AI Consortium in order to use AI models to reduce the high energy consumption of this technology. These models will be accessible to university researchers and industrialists in open source.
Reduce consumption of AI … with AI!
The energy sector faces the growing demand for data centers in the United States and in the rest of the world. The situation is not likely to improve With the Stargate project at 500 billion euros. AI has major needs in terms of computing power: according to the International Energy Agency, demand would increase by 4 % per year in the coming years. Almost double the figures for 2023.
In addition to NVIDIA and EPRI, the Open Power AI Consortium has PG & E, CON EDISON, Constellation Energy, Duke Energy, Tennessee Valley Authority, Enowa, which is Neom's energy and water company. Microsoft, of which Openai seeks to emancipate himselfand Oracle are also members.
Faced with significant energy demand, companies secure their production capacities. Now, electricity has a competitive advantage. Last year, the companies of the Open Power AI Consortium concluded new contracts, in particular for renewable energy projects against the low cost of solar, its versatility and its speed of deployment.
For example, Microsoft added 475 solar energy megawatts to its renewable energy portfolio. In 2024, the Redmond firm became one of the main investors in a project to develop renewable energy of $ 9 billion, managed by Acadia.
Earlier in the year, Microsoft also announced a collaboration with Brookfield Asset Management to deploy 10.5 gigawatts of renewable energy in the United States and Europe. The set is planned to be operational by 2030.
Another solution according to a study: by reducing the use during peaks of demand on the network, in particular by moving the poorly urgent tasks at times of low necessity, it would be possible to release a capacity of 76 gigawatts in the United States. About 10 % of peak demand in the country.
