The winds have shifted. Nuclear energy is back in the game. Once thought of as a hazardous and unclean source of energy, it is now the green energy of choice for many of the big tech companies. In 2024, Microsoft invested in reviving a closed power plant in Pennsylvania, whereas Google and Amazon have both been investing in the creation of small modular reactors.
So, why the shift? These big tech companies need energy, they need it to be green and it must be reliable. Following public pressure, many of these companies have made climate pledges which they are at considerable risk of not meeting. Amazon has claimed it would reach net zero by 2040, Google declared it would do the same by the end of 2029, and Microsoft said it would remove more carbon than it emits by 2030.
While wind and solar energy sources have thus far been the preferred investment choices, they have the drawback of being weather-reliant. On a day with no wind or sun, the only energy available is the limited amount that can be stored in batteries or some other form of storage. Nuclear energy, on the other hand, is available around the clock.
Tech companies are not the only ones turning towards nuclear energy. The US government, particularly under the Biden administration, championed a number of nuclear energy projects during their term. While the US has the highest number of active nuclear plants than any other country (94), in the last couple of decades Democrats opposed the building of nuclear plants due to “safety, environmental and economic concerns”. Therefore, this widespread shift towards nuclear energy may mark the beginning of what some have deemed the ‘nuclear renaissance’.
AI is hungry for energy Despite this shift towards nuclear power, many of these big tech companies will struggle to meet their climate goals due to the increased energy pressures presented by the rise of AI. AI technologies require far more power than other forms of technology, such as web searches or social media.
This highlights how AI energy demands are of an entirely different magnitude. Many of these tech companies consume more energy than it takes to power entire countries. These energy demands will only increase as these companies compete to build larger AI models, requiring even bigger data centers.
With the increased energy demands created by AI, we begin to see statements of tech companies shamelessly backtracking on their climate goals. Microsoft recently announced that its carbon emissions are now 30% higher than in 2020; Google revealed that its carbon emissions have gone up by 48% in the last five years; and while Amazon has managed to decrease its carbon footprint by 3%, its investments in AI will begin to present some challenges.
Will access to energy hold the US back? In the race to AI dominance, one of the biggest threats to a US victory is access to energy. In one of our previous blogs we discussed the race to AI dominance between the US and China. While the US may be currently ahead, their lead may depend on being able to rapidly scale up their energy production.
The US suffers from “skills gap and labor shortages in construction and engineering, a complex permitting process trapping projects in years of bureaucratic review across multiple agencies and high costs of capital”. This has resulted in a significant backlog for many data centers to be connected to the grid. In fact, in Virginia (a data center hotbed) the waiting time can be up to seven years.
China is currently being held back in its AI development by US trade embargos on semiconductor chips. Yet, as alternative Chinese-built chips begin to emerge on the market, China will strive to catch up in the race. With chains unbound, China is unlikely to suffer from the same energy expansion problems as the US. China has a remarkable ability to fast-track infrastructure projects. In the time that the US built three nuclear power plants, China managed to build thirty-six.
The future of energy Just as access to oil has dominated the last 20 years and more of geopolitics, creating conflicts and sealing economic fates, will access to renewable energy hot spots now do the same? Will countries blessed with geothermal energy, reliable sunshine or ample wind, now hold the same political sway? As demand for renewable energy increases, with big tech companies competing to secure access, we should be prepared to see a change in the political power associated with energy resources.
It may be possible to reduce the resource strain (caused by an increasing demand of energy) by placing data centers in more remote areas. Youlian Tzane has argued that in the past the “industry norm has been to build around economic hubs rather than sources of renewable energy”, but that this should change so that the data centers can make use of renewable energy sources like hydro or geothermal power.
Unfortunately, the urgency for energy resources has caused some companies to turn to polluting energy sources like coal, notwithstanding the impact that this could have on their climate goals. The International Monetary Fund has suggested that a carbon tax may dissuade big tech companies from straying from their carbon goals. While I do believe that CO2 emissions, particularly by mass-energy consumers, should be discouraged, as a stand-alone action I don’t think it will resolve the problem.
The long and bureaucratic system of approving clean energy projects must be streamlined. Azeem Azhar suggests that just as the US established the importance of semiconductor chips through the CHIPS and SCIENCE Act which mandate export controls aimed at boosting the AI economy, so too should a similar focus and urgency be applied to upgrading America’s electrical infrastructure. If we create a demand for renewable energy sources through carbon taxes and similar instruments, without first solving the obstructions which prevent us from creating new renewable energy sources, we may merely increase the price and make it inaccessible to many. Governments across the world should begin to consider the very real risk of AI's demand for energy raising prices, making energy poverty both an important moral and political issue.
Written by Celene Sandiford, smartR AI