Consider the transistor, the basic unit of computer processors. The transistors can be small in size One -number nanopolitan measuring devices In size. Billions can fit On the computer chip.
Although they do not have mobile parts, they devour electricity while storing and modifying it. “These zeros and zeros are encrypted Timothy SherwoodProfessor of Computer Science at the University of California Santa Barbara. “When you do any account, what happens inside the exact processor is that there are some one That is transferred to a zeroOr a zero That transformations to one. Each time this happens, a little energy is used. “
Subscribe here to explore the big and complex problems the world faces and the most efficient ways to solve them. Send twice a week.
When you add this – via billions of transistors on chips and then billions of these chips in computers and server farms – they constitute a large and increasing share of human energy appetite.
According to International Energy Agency (IEA), data accounts and storage somewhere between 1 and 1.5 percent of global electricity demand at the present time.
As artificial intelligence and cryptocurrencies that depend on databases on an industrial scale, this participation is preparing for growth. For example, typical Google Search uses about 0.3 watts an hour While Chatgpt 2.9 watts consumes an hour. In 2024, the amount of data data center under construction in the United States 70 percent jumped compared to 2023. Some technology companies that tend to artificial intelligence have seen Increased greenhouse gas emissions And you find it difficult to achieve their environmental goals.
How much this account will need in the coming years, and it will put us The goals of climate change are far -up?
AI injecting chaos in energy demand expectations
IEA estimates that energy demand in the data center will double by 2030. MCKINSEY estimates Somewhere between three times and Quintupling. As a result, the main players in the field of technology are trying desperately to pay their energy supplies. During the past year, they were one of the largest purchases of energy sources that produced A few greenhouse gas emissions. Amazon is The largest buyer of a renewable energy company In the world. Companies like Microsoft revives old nuclear factories while also investing in the next generation of nuclear technology.
But some of these companies It is not difficult to satisfy where their strength comes from. “What we need from you,” said the former CEO of Google Eric Schmidt The Energy and Trade Committee in the House of Representatives Earlier this month, “energy in all its forms, renewable, not renewable, whatever. It should be there, and it must be there quickly.”
Indeed, energy demand extends from data centers Lifestyle for old power plants It creates a market for new natural gas stations. Iea can During the next five yearsRenewable energy sources will meet half of the additional demand for electricity from databases, followed by natural gas, coal and nuclear energy.
However, many of these energy demand forecasts are forecasts based on current trends, as well as many things change very quickly. “The first thing I will say is that there is a lot of uncertainty about how energy demand grows at the data center,” he said. Jessica TreaticProfessor at the Massachusetts Institute of Technology studies the technology and energy sector.
Below is some context that must be taken into account: Remember that databases are less than 2 percent of the total electricity demand now, and even double or three times or quintupling will continue to keep their share in individual numbers. More part of the growth of global electricity demand is preparing from developing countries that manufacture and climb the income scale. Energy use is also associated with the economy; In the recession, for example, the demand for power tends to decrease.
Climate change can also play a role. One of the largest electricity demands was simply last year It was very hotHe leads more people to operate Air conditioners. So, while artificial intelligence is important and growing energy users, it is not the only thing that changes the future of energy demand.
We also in Cambrian explosion The era of Crypto and AI companies, which means that there are many different companies that try to a variety of methods. All this experience is to increase energy use in the short term, but not all of these methods will make it. With these sectors maturing and consolidating their players, this may also reduce energy demand.
The good news is that computers are getting more efficient. Amnesty International and Hosage of encryption Graphic processing unitsThe optimal chips for the types of accounts behind these techniques. Graphics processing units have made huge leaps, especially when it comes to the ability of artificial intelligence to take new information and create conclusions.
In the past ten years, our platform has become 100,000 times more powerful Inference Work burden Joshua ParkerWho leads the company’s sustainability efforts in NVIDIA, one of the largest GPU producers in the world. “In the past two years – one generation of our products – we have become 25 times more efficient.”
NVIDIA has now created leadership leaders in the artificial intelligence race, making it one of the most valuable companies in history.
However, with the increasing efficiency of computer processors, they cost less, which may lead to more people, which compensates for some energy savings.
“It is easier to make the state of work to spread artificial intelligence, which means that the fingerprint is growing, so it is a real paradox,” said Parker. “Ultimately, this type of Si growth remains only if you actually reach a gradual zero costs. There are still energy costs and there is still a cost of the account. As much as we are moving towards efficiency, there will be a balance in the end because it is not free.”
Another factor to consider is that artificial intelligence tools can have their own environmental benefits. Using artificial intelligence to perform simulations can avoid some of the need to perform a real, slow, slow, and density of energy density. Aircraft design, For example. Amnesty International operators use Amnesty International Improving electricity distribution To integrate renewable energy sources, increase reliability, and reduce waste. I have already helped artificial intelligence in design Better batteries and Better solar cells.
Amid all this uncertainty about the future, there are still paths that can maintain artificial intelligence with efforts to reduce climate change. Technology companies need to continue to withdraw the efficiency lever. These sectors also have great opportunities to reduce carbon emissions in supply chains of these devices, and in Infrastructure for data centers. It is necessary to spread clean energy greatly.
We have already seen a number of countries growing their economies while cutting greenhouse gases. Although artificial intelligence slows some of this progress at the present time, it should not increase the long -term climate change, and it can accelerate efforts to keep it under examination. But this will not happen by chance, and will require deliberate procedures to reach the right track.
“It is easy to write the title, which says artificial intelligence will break the network, this will lead to more emissions,” Parker said. “I am very optimistic – I think this is reliable optimism – that artificial intelligence over time will be the best tool for sustainability that the world has ever seen.”
