Betting on the rise of artificial intelligence requires monitoring the uranium market

The future depends on the ability to generate more electricity in light of the energy-intensive consumption of artificial intelligence

Will a crumbling electricity distribution network kill our AI ambitions?

Artificial intelligence can do anything, and it will change the features of the global economy, and bring us that productivity boom that we have waited for so long. While it does all our boring work, it will provide us with the free time that economists have predicted for decades, and save lives through early diagnosis by “smelling” chemicals linked to diabetes and cancer.

It will also drive our self-driving cars, predict and prevent flight delays, and create new varieties of climate-resistant rice. Rather, it is possible that he will perform a task that no amount of money or number of people will be able to accomplish. Saving the UK’s National Health Service (NHS).

The list goes on, but you get the point. Elon Musk said in a post on his “X” platform that by next year: “Artificial intelligence is likely to be more intelligent than any single person, and by 2029 it will be more intelligent than all humans together.”

Artificial intelligence raises energy demand

The main problem is that artificial intelligence is not a form of magic, and it cannot perform its miracles or do anything without the presence of electricity, or even with large quantities of it. No one knows exactly how much electricity AI will need, but a recent estimate cited by many puts it at approximately Sweden’s annual consumption, or 0.5% of total global electricity consumption.

Another way to capture this consumption is to look at the size of data centers. The latest report on data centers issued by commercial real estate firm Jones Lang LaSalle notes that it is “common” today for developers to announce data centers with new designs that operate at 10 times the capacity that was previously available. You’ve only been using it for a few years.

Although we cannot precisely determine how much electricity AI will consume, we know that it will require a greater increase in generating capacity and transmission capacity. This is the current situation in light of the two ongoing changes: An attempt to reduce distribution network emissions through the use of renewable energy, and efforts to further increase the proportion of electricity in energy consumption to reduce the resulting emissions as well. The European Union expects electricity demand to rise by 60% by 2030.

The need to increase generating capacity

This generating capacity is not yet available. For example, electricity generation in the United States has been almost stable since the beginning of the twenty-first century. Michael Semplast of JP Morgan Asset Management said in a recent report that this is not a big problem; “Improving energy efficiency offsets the electricity needs of a growing population.”

However, the emergence of artificial intelligence and its increasingly life-changing uses, its surprisingly rapid adoption – GBT chat has become the fastest growing application in the history of the Internet – and its insatiable demand for energy are changing all that. Now generating electricity in the United States is a major problem. Meeting demand in America, and in the United Kingdom, where the distribution network is also deficient, is linked to building a new world in generating electricity and spending billions on the distribution network necessary to transmit that energy. However, due to the increase in high-voltage DC transmission lines, which enable the transmission of renewable energy from its usually remote locations, there has been no significant development.

Semblast added that, for example, in the United Kingdom, National Grid acknowledged that there is a “crisis” in connecting electricity generation projects. In the United States, generating capacity at the beginning of the third millennium rarely exceeded 100 gigawatts in electrical interconnection requests to the regional distribution company PJM Interconnection. The generating capacity currently approaches 250 gigawatts. However, not all of these projects will be approved and completed, but you understand that. The question here is: Will the distribution network kill our ambitions in artificial intelligence?

Distribution network stability

There are solutions. The first is Musk’s choice; For artificial intelligence itself to find a solution that has not occurred to us yet, perhaps generating electricity from nuclear fission or fusion, or even through transmitting solar energy from space using giant mirrors is closer than we imagine. AI may bring new efficiencies to the distribution network, stabilizing it by providing better forecasts of renewable capacity and intermittency, and improving demand and maintenance forecasts, reducing downtime.

It is also quite logical that increased productivity in other countries thanks to artificial intelligence will offset the increased demand it causes. A report prepared by Google and Boston Consulting Group claims that artificial intelligence can reduce global emissions by between 5% and 10% by 2030.

Legalization of consumption

The second solution is the regulatory option; That is, preventing artificial intelligence from consuming a large amount of electricity. A US government report released last month noted that the rise of artificial intelligence and artificial general intelligence (AGI) is having “significant implications for democratic governance and global security,” and the impact is not good.

The report proposed a relatively wide range of measures to reduce consumption, one of which is to criminalize training artificial intelligence models using computing power that exceeds a relatively reasonable minimum. While the new artificial intelligence law approved by the European Union – among many other rules – imposes requirements that force various artificial intelligence systems to adhere to complete transparency regarding energy consumption.

Nuclear energy is a sustainable solution

The last solution is the nuclear option; Regulators or electricity companies should not be surprised by the rapid rise in demand for electricity and the huge investment plans, after the insistence on switching to the use of electricity in all fields.

However, the adoption of nuclear energy on a large scale – certainly with the help of artificial intelligence – and if it is located close to consumption sites, may solve all supply problems, and most electricity transmission problems as well. There is no longer a need to establish small nuclear reactors in the North Sea, miles away. the coast.

Analysts at Doomberg believe that this will represent the final push to achieve a nuclear renaissance, which represents the only sustainable way to produce electricity that is commensurate with our drive to steadily develop our computing power, to achieve the clear goal and “all possible goals.”

In a possible sign of what’s to come, Amazon Web Services (AWB) announced that its new data center in Pennsylvania will run directly on electricity generated by an existing nuclear power plant nearby.

The AI ​​revolution may coincide with an electricity revolution, which is something uranium investors should keep an eye on.

Exclusive to Bloomberg