Sam Altman, CEO of OpenAI, the company that created ChatGPT, issued a warning in January at the Davos Forum: the artificial intelligence (AI) industry is about to cause an energy crisis. The new generation of generative AI will consume much more energy than expected, he told the world’s top leaders and entrepreneurs, to the point of putting global energy networks in check. “There is no way to get there without drastic changes,” he snapped.
That “drastic change” he was referring to is the so-called advanced nuclear energy, a term that includes pocket reactors and nuclear fusion, both still in the experimental phase. Several companies have focused on this alternative, which would provide them with energy autonomy and greater cost control. The Biden Administration does not view it with bad eyes. Energy Secretary Jennifer Granholm met in March with several technology companies, including Amazon, Google and Microsoft, to explore imaginative ways to supply them. One of the topics discussed was the fit of small nuclear reactors in data centers, those extensive warehouses full of processors running day and night.
The latest estimates say that 8% of the world’s energy is already dedicated to AI, which is consumed by powering the processors on which the models are trained and the systems are hosted. That figure, as Altman predicts, will soon fall short, as users are added and new versions of ChatGPT, Gemini or Copilot emerge, which will require more and more computing power. “I’m glad he said that in Davos. “I have seen constant minimization and denial about the environmental costs of the AI industry since I started publishing about it in 2018,” wrote Kate Crawford, one of the leading researchers on the AI footprint.
Las big tech have already taken the first steps towards the nuclear age, a declining energy source in the West (more reactors are dismantled than are built) with some major exceptions: the US, France, the United Kingdom and several Eastern European countries. . Companies, for their part, conceive it as a way to ensure a stable and lasting supply of energy in a context in which the supply is not enough. Senior Google executives said The Wall Street Journal considering the possibility of signing purchase agreements with developers of small nuclear reactors. “I think nuclear energy, especially the most advanced, is making a lot of progress,” Maud Texier, Google’s global head of clean energy, told the American newspaper. Company sources do not confirm to EL PAÍS that the nuclear route is an option for the future, although they do not deny it either. Google recently signed an agreement with Microsoft and Nucor to “accelerate advanced clean energy technologies,” including “advanced nuclear.”
In October of last year, Microsoft closed purchase agreements (PPA) with the American company Helion Energy through which it will receive energy obtained from nuclear fusion starting in 2028, a technique that is even more theoretical than practical and that, unlike fission, does not produce radioactive waste. Asked by this newspaper about its strategy in the nuclear field, the company refers to a document from December of last year titled Accelerating a carbon-free futurein which it is made clear that “advanced nuclear energies”, as well as traditional reactors, are one of the pillars on which Microsoft’s green turn will pivot, although there is no talk of deadlines or dates.
AWS, Amazon’s cloud computing division, has recently purchased a large data center in the United States located next to the country’s sixth largest nuclear power plant, which supplies it with 100% of its energy at a fixed price. “To complement our wind and solar projects, which depend on weather conditions to generate energy, we are also exploring innovations and technologies and investing in other sources of clean, carbon-free energy. The agreement with Talen Energy [la empresa dueña de la citada central nuclear estadounidense] for carbon-free energy is a project that goes in that direction,” company sources tell EL PAÍS.
Silicon Valley’s nuclear flirtation
The idea that nuclear energy is the salvation of AI is catching on among the Silicon Valley jet set. Sam Altman is one of his great supporters. He is so convinced of the future of the proposal from Helion Energy, a pioneer of nuclear fusion, that he has invested $375 million in it. He also chairs a startup, Oklo, that aims to design and manufacture nuclear fission reactors like those used today, but much smaller (the so-called SMR, short for small modular reactor).
Bill Gates is another of the technology tycoons with interests in SMRs. His TerraPower company is working on a sodium nuclear reactor, an experimental variant that, if successful, promises to be 25 times cheaper than nuclear fission.
Meta’s chief generative AI engineer, Sergey Edunov, said a few months ago that only two large nuclear reactors would be needed to cover the entire global energy demand projected for 2024 in terms of AI, including powering already operational models and training. of the new ones.
Does the nuclear route have a route? “There are no advances on the horizon that would allow for the immediate deployment of SMRs, which are currently in the initial prototyping phase in numerous countries. This option would only be viable if we talk about a period of decades,” says engineer Heidy Khlaaf, a specialist in evaluation, specification and verification of complex computer applications in safety-critical systems. Some countries, such as the United Kingdom, France, Canada or the United States, have plans to develop this type of facilities, but not before 20 years.
Khlaaf is especially concerned that Microsoft has put generative AI to work on the paperwork to achieve nuclear licenses, a process that can take years and cost millions of dollars. “This is not a box-ticking exercise, but a process of self-assurance. “To view these regulatory processes as mere cumbersome paperwork says a lot about your understanding, or lack thereof, of nuclear safety,” he snaps.
Is it realistic to trust the future of AI to nuclear fusion? Helion Energy’s most optimistic estimates say that in 2029 it will be able to produce enough energy to supply 40,000 average homes in the United States. It is estimated that ChatGPT already consumes the equivalent of 33,000 homes today.
Why so much energy consumption?
The emergence of AI has shaken the global energy table. Most of the consumption associated with generative AI models occurs before they are used, during the development phase. training. This is a key process in the development of deep learning models that consists of showing the algorithm millions of examples that help it establish patterns with which to predict situations. In the case of large language models, such as ChatGPT, the system is intended to conclude that the series of words “The sea is colored” has a high probability of being followed by the word “blue.”
Most data centers use advanced processors called GPUs to train AI models. GPUs require a lot of energy to operate, about five times more than CPUs (conventional processors). Training large language models requires tens of thousands of GPUs, which need to operate day and night for weeks or months.
“Large language models have a very large architecture. A machine learning algorithm that helps you choose who to hire might need 50 variables: where the candidate works, what salary they have now, previous experience, etc. The first version of GhatGPT has more than 175 billion parameters,” illustrates Ana Valdivia, professor of AI, Government and Policy at the Oxford Internet Institute.
Once the model has been trained, it is necessary to host and exploit the data on which it works. This is also done in data centers, which have to operate day and night.
What is the total consumption of AI? How much energy is dedicated to training and feeding the most used models? Companies don’t publish that information, all we have are estimates. For example, Google’s Gemini Ultra model, one of the most advanced today, required 50 billion petaflops to train, according to a recent report from Stanford University. To achieve that computing power with commercial computers (although supercomputers are used in these tasks) about 10,000 billion (10 to the power of 16) computers would be needed. The cost associated with this training was $191 million, largely attributable to the energy consumed.
A single AI model can consume tens of thousands of kilowatt-hours. Generative AI models, such as ChatGPT, can have 100 times greater consumption, according to estimates by technology consulting firm IDC.
Apart from the power of the systems themselves, consumption also goes to the cooling systems of the processors. The most common techniques include electric ventilation and the use of water to cool the environment and machines. This latter system is beginning to cause problems in places with water scarcity, although the most modern techniques involve the use of closed circuits that minimize losses of water resources.