Nvidia announced on Thursday that it will partner with EPRI, the power industry’s R&D organization, to use AI to solve the problems facing electric grids. Perhaps ironically, the problem is caused primarily by the increased demand for electricity from the AI itself.
The Open Power AI Consortium, which includes many utilities and high-tech companies, says it will use what is known as a domain-specific AI model to devise new ways to tackle the problems the power industry is expected to face in the coming years. This model will be made available to academic and industrial researchers with open source.
As AI is increasing the need for computing power, the electricity industry is facing surge in demand from data centers in the US and elsewhere. According to the International Energy Agency, electricity demand is expected to increase by 4% per year over the next few years.
In addition to Nvidia and EPRI, the consortium includes PG&E, Con Edison, Constellation Energy, Duke Energy, The Tennessee Valley Authority, and Enowa from Neom’s Energy and Water Company. Technically, Microsoft and Oracle are both members.
To stay ahead of the trend, tech companies are competing to ensure their generation capabilities as their power has been converted from simple line items to competitive advantage.
Over the last year or so, tech companies have consistently signed new contracts. They are spread primarily to renewable energy projects, primarily spurring the low cost, modularity and speed of deployment of solar.
For example, Microsoft recently added 475 megawatts of solar power to its considerable renewable portfolio. Last year he became an anchor investor in the $9 billion renewable development project run by Acadia, and in collaboration with Brookfield Asset Management at the beginning of the year, he deployed 10.5 gigawatts of renewable electricity in the US and Europe, and is expected to be online by 2030.
But while new power supplies may be the most obvious answer to losing power shortages, that’s not the only one.
One recent study has shown that by reducing use of grid peak demand, including shifting tasks that are not sensitive to low periods of grid peak demand, which can unlock 76 GB capacity in the US, and accounting for around 10% of peak demand demand in the US.
It could probably be the kind of solution that this new consortium will explore.
Source link
