Last week, Bill Gates broke ground on a nuclear power plant in Wyoming, the U.S., with a $1 billion investment. On the 16th local time, he said he would make additional investments: "I have invested more than a billion, and I will invest billions more."
The reason why Bill Gates continues to invest in this nuclear power plant is largely because the development of artificial intelligence (AI) will bring huge pressure on electricity consumption. "The data centers we are going to build will increase the electricity load by up to 10%." He said that the rise of various household heating devices such as electric vehicles and electric heat pumps has pushed up the demand for electricity in the United States, and "the emergence of data centers is even worse. Therefore, large technology companies are studying how to provide more electricity so that these data centers can meet the explosive growth of AI needs."
According to the International Energy Agency (IEA), a query using OpenAI's chatbot ChatGPT consumes 2.9 watt-hours of electricity, while a Google search only requires 0.3 watt-hours, which is only about 1/10 of the former. By 2026, electricity consumption related to data centers, cryptocurrencies, and AI may increase to 620 to 1,050 terawatt-hours. In 2023, Germany's total electricity consumption will be 465 terawatt-hours, and Japan's total electricity demand will be 870 terawatt-hours.
In an interview with the Financial reporter, Mr xiao, partner of EY's strategy and transaction advisory services, said that the large amount of electricity required for the expansion of AI computing power has become a major problem that cannot be avoided at present, but it is not unsolvable. In the future, we can start from technological breakthroughs, resource sharing and the use of new energy.
European and American governments feel the pressure of AI power demand
歐美政府感受到AI電力需求壓力
0
As AI infrastructure such as data centers proliferate, AI-related electricity demand will grow rapidly. According to research by investment bank Goldman Sachs, currently, global data centers account for 1% to 2% of total electricity consumption, but by 2030, this proportion may rise to 3% to 4%, which is doubled.
In Europe, the Goldman Sachs report said that by 2030, the region's data center electricity demand will be equal to the current consumption of Portugal, Greece and the Netherlands combined. An analysis by the Electric Power Research Institute (EPRI) found that by 2030, data centers will consume between 4.6% and 9.1% of the United States' electricity.
Earlier this month, the U.S. House of Representatives Energy and Commerce Subcommittee on Energy, Climate, and Grid Security held a hearing on the theme of "Powering AI." The minutes of the meeting show that the committee believes that it is necessary to treat related issues as national security issues. The current U.S. power grid faces huge challenges in meeting growing demand, and its reliability and stability are worrying. The United States urgently needs strategic energy planning. Intermittent renewable energy alone is not enough to meet the high reliability requirements of data centers and AI technology, so it is necessary to focus on a diversified power generation portfolio including energy storage and stable dispatchable power generation resources. In addition, the United States needs to invest heavily in the construction and upgrading of energy infrastructure to support the growing demand for AI and data centers.
In Europe, some countries have begun to impose more requirements on the construction of local data centers due to concerns that the huge energy usage of data centers will put too much pressure on national climate goals and power grids.
Germany passed a new Energy Efficiency Act in October last year, which specifically stipulates the efficiency of power use, renewable energy supply, and waste heat use in data centers. For example, from 2024, 50% of electricity consumption must be supplied by renewable energy, and from 2027, this proportion will be increased to 100%.
Last year, Equinix, one of the world's largest digital infrastructure companies, Vantage Data Centers, and EdgeConneX, a data center operator, were rejected by Ireland for planning permission to build new data centers in Dublin. In fact, since 2022, the Irish Electricity Supply Commission (EirGrid) has stated that due to the sharp increase in electricity demand, they will restrict the construction of new data centers, implement stricter approval procedures for new projects, and give priority to projects with less impact on the power grid.
去年,全球最大的數(shù)字基礎(chǔ)設(shè)施公司之一易昆尼克斯(Equinix)、數(shù)據(jù)中心運(yùn)營(yíng)公司萬(wàn)特奇數(shù)據(jù)中心(Vantage Data Centers)和“邊緣連接”(EdgeConneX)申請(qǐng)?jiān)诙及亓中陆〝?shù)據(jù)中心的規(guī)劃許可均被愛(ài)爾蘭拒絕。事實(shí)上,自2022年起,愛(ài)爾蘭電力供應(yīng)委員會(huì)(EirGrid)就表示,由于電力需求的急劇增加,他們將限制新的數(shù)據(jù)中心建設(shè),對(duì)新項(xiàng)目實(shí)施更嚴(yán)格的審批程序,并優(yōu)先考慮對(duì)電網(wǎng)影響較小的項(xiàng)目。
0
Local governments in and around the Dutch capital Amsterdam have also suspended the issuance of construction permits for new data centers to ease grid load and manage energy demand.
Technology giants: do it yourself and live a comfortable life
科技巨頭:自己動(dòng)手、豐衣足食
0
Technology giants such as Microsoft, Google and Amazon are at the forefront of AI development, and their cloud computing business units are supported by clusters of data centers around the world, which also means a lot of energy consumption.
Recently, U.S. Energy Secretary Jennifer Granholm revealed that the Biden administration is in talks with major technology companies, asking them to invest in climate-friendly electricity production to meet their own growing needs.
"If tech companies want clean electricity from the grid, they should generate their own electricity," Granholm told the media. "We have been discussing this with data companies. Large data companies have committed to achieving net zero emissions and want a stable supply of electricity generated by clean energy."
xiao said that in order to meet the electricity demand and energy-saving pressure brought by AI, in terms of the use of new energy, clean energy such as solar energy and wind energy can be used to power data centers, reduce dependence on fossil fuels, and reduce carbon emissions.
Tech giants have also already begun to lay out their plans. According to Amazon's official website, the company has more than 500 solar and wind projects around the world, and invested in more than 100 last year alone, becoming the world's largest corporate buyer of renewable energy for the fourth consecutive year. Its investment portfolio is now enough to power 7.2 million American households each year. The company said that by 2025, all of its global electricity will be powered by renewable energy, including Amazon Web Services' data centers.
Microsoft is betting on nuclear energy. In May last year, the company signed a power purchase agreement with Helion, a private American company specializing in the development of nuclear fusion technology. In January this year, Microsoft also hired a director of nuclear technology to develop atomic reactors to power its data centers. Microsoft's goal is that by 2025, all data centers and facilities will be 100% powered by new renewable energy that matches annual power consumption.
Google's parent company Alphabet announced its 24/7 Carbon Free Energy (CFE) goal in 2020, which is to use carbon-free energy for all business operations by 2030. Last year, Google opened its first data center in Jinsai City, Chiba Prefecture, Japan. In May this year, the company announced the signing of two new solar power purchase agreements (PPA) in Japan to support the construction of new solar projects and add 60 megawatts of clean energy capacity to the Japanese power grid.
In addition to starting with power generation, xiao Fusheng told the Financial reporter that technology companies can also try to share the resources needed for AI development. For example, through the leasing model, the existing graphics processing unit (GPU) resources can be maximized to alleviate the problem of computing power shortage and reduce the initial investment cost. If feasible, cloud computing platforms can also be used to provide AI computing power, reducing the need for enterprises and research institutions to deploy local infrastructure and achieving resource sharing.
In this regard, AI chip developer Nvidia has stated that the latest Blackwell GPU has four times the training performance of the previous generation product Hopper and 25 times the energy efficiency.
However, xiao Fusheng said that these solutions are not easy to achieve. "Technological breakthroughs require a lot of human, material and capital support, and the commercialization and continuous performance improvement of new chips need to be solved. Resource sharing requires ensuring the reliability and cost-effectiveness of computing power leasing services, and at the same time, it is necessary to establish an effective resource sharing mechanism. The use of new energy requires overcoming the intermittent and unstable nature of renewable energy, and achieving intelligent and efficient energy management, while also considering compatibility and transition issues with existing energy infrastructure." But he believes that "the development of new technologies always requires a process."