Consider the transistor, the basic unit of computer processors. Transistors can be small, to the size of the single digit nanometers. Thousands of millions can fit in a computer chip.
Although they do not have mobile parts, they devour electricity while storing and modify information bits. “One and zeros are encoded as these high and low voltages,” said Timothy Sherwood, a computer professor at the University of California Santa Barbara. “When you do any calculation, what is happening inside the microprocessor is that it is something one That the transition to a zeroor a zero That transitions to one. Every time it happens, a little energy is used. “
Register here to explore the big and complicated problems facing the world and the most efficient ways to solve them. Sent twice a week.
When he adds that, through the billions of transistors in chips and then the billions of these chips in computers and server farms, form a significant and growing proportion of humanity’s energy appetite.
According to the International Energy Agency (IEA), computing and data represent between 1 and 1.5 percent of world electricity demand at this time.
With the growth of artificial intelligence and cryptocurrencies that depend on industrial scale data centers, that participation is prepared to grow. For example, a typical Google search uses approximately 0.3 watts-operas, while a chatgpt consultation consumes 2.9 watts-operas. In 2024, the amount of capacity of the data center under construction in the US.
How much more electricity will this calculation need in the coming years, and will stop reaching our climate change goals?
AI is injecting chaos into energy demand forecasts
The AI estimates that the energy demand of the data center will double by 2030. McKinsey estimates somewhere between a triple and a quintuado. As a result, the main technological players are desperately trying to prop up their food. Around the last year, they have been some of the largest buyers of energy sources that produce few greenhouse gas emissions. Amazon is the largest corporate buyer of renewable energy in the world. Companies like Microsoft are even reliving the old nuclear plants while investing in the next generation of nuclear technology.
But some of these companies are not demanding about where their power comes from. “What we need from you,” said the former CEO of Google, Eric Schmidt, to the Chamber of Energy and Commerce Committee at the beginning of this month, is “energy in all forms, renewable, non -renewable, whatever.
Already, the energy demand of the data centers is extending a line of life to the old coal power plants and is creating a market for new natural gas plants. The IEE estimates that in the next five years, renewable energies will meet half of the additional electricity demand of data centers, followed by natural gas, coal and nuclear energy.
However, many of these energy demand forecasts are projections based on current trends, and well, many things are changing very quickly. “The first thing I will say is that it is only much uncertainty about how the energy demand of the data center will grow,” said Jessika Trancik, a teacher to study the technological sector and energy.
Here there is a context to take into account: remember that the data centers are less than 2 percent of the general demand for electricity now and even duplicate, triple or quintuplicate would still maintain their participation in the unique digits. A majority of the global growth of electricity demand is prepared for countries in industrial development and upload the income scale. The use of energy is also linked to the economy; In a recession, for example, energy demand tends to fall.
The weather could be able to play a role too. One of the biggest drivers of the demand for electricity last year was simply that it was so hot, which led more people to light the air conditioners. So, although AI is an important and growing energy user, it is not the only thing that alters the future of energy demand.
We are also in the era of the Cambrian explosion or the cryptographic and AI companies, which means that there are many different companies that test a variety or approaches. All this experimentation is an energy use in the short term, but not all of these approaches will. As these sectors mature and their players consolidate, that could also reduce energy demand.
The good news is that computers are becoming more efficient. Ai and graphic processing units of AI and Crypt, chips optimized for the types of calculations behind these technologies. GPUs have made mass performance leaps, particularly when it comes to AI capacity to take new information and generate conclusions.
“In the last 10 years, our platform has become 100,000 times more energy energy for exactly the same inference workload,” said Joshua Parker, who leads corporate sustainability efforts in Nvidia, one of Largt’s GPU producers in the world. “In the last two years, a generation of our product, we have become 25 times more energy efficient.”
Nvidia has now established a dominant advantage in AI’s career, which makes it the most valuable companies in history.
However, as computer processors become more efficient, they cost less to execute, which can lead people to use them more, compensating some of the energy savings.
“It is easier to make the AI of the business case, which means that the footprint is growing, so it is a true paradox,” Parker said. “Ultimately, that child of exponential growth only continues if it reaches the incremental costs of zero. There are still costs for energy and there is still cost for calculation. As much as we lead towards efficiency, it will do it.”
Another factor to keep in mind is that the tools of AI can have their own magnary benefits. The use of AI to perform simulations can avoid part of the need for real tests of the real world. Network operators are using AI to optimize electricity distribution to integrate renewable energies, increase reliability and reduce waste. The AI has already helped to design better batteries and better solar cells.
In the midst of all this uncertainty about the future, there are still paths that could maintain the expansion of the AI aligned with efforts to limit climate change. Technology companies must continue to achieve the efficiency liver. These sectors also have great opportunities to reduce carbon emissions in the supply chains of these devices and in the infrastructure for data centers. Implementing much more clean energy is essential.
We have already seen several countries grow their economies while reducing greenhouse gases. While AI is slowing down part of that progress at this time, you can’t change long -term climatic, and could accelerate efforts to keep it under control. But he won snacks by chance and will require a deliberate action to get on the way.
“It is easy to write the head that says that AI will break the grid, will lead to more emissions,” Parker said. “I am personally very optimistic, I think this is a credible optimism, that AI over time will be the best tool for sustainability that the world has seen.”
]