#106 Powering the Future: Nuclear Fusion Meets AI
Today, Sridhar Krishna explores whether nuclear fusion can be the answer to AI’s energy requirements, followed by this week’s curated note by Lokendra Sharma that elaborates on the implications of cyberwar.
Technopolitik: Nuclear Fusion in the Age of AI
— Sridhar Krishna
Artificial Intelligence (AI) has taken the world’s imagination by storm, and governments, business enterprises and the public at large are debating the use of AI, the opportunities, the risks and the regulations required to ensure AI serves the public good.
There is another very critical issue with AI that is less understood by many who are discussing the implications of AI and that is the energy cost of AI.
In January 2024, Sam Altman, CEO of OpenAI, warned the participants at the World Economic Forum at Davos that AI systems were consuming way more energy than expected earlier and that there was a need for a breakthrough in this space before AI could really be deployed at scale.
Generative AI-based queries, according to research published in Nature Magazine, consume four to five times more energy than a conventional web search and need enormous amounts of water to cool their processors. In the midst of all the hype around AI, the environmental cost of AI should not be ignored.
According to Roland Berger, a global consulting firm, training a single AI model can consume as much energy as 626,000 KWH.
AI’s energy usage today is only 2-3% of the technology sector’s energy usage today, but this is growing very rapidly.
While the industry is looking at more energy efficiency, there is no doubt that existing power grids will fail with large-scale AI deployment. Companies deploying AI need more captive power generation. Companies like Amazon and Microsoft are betting on nuclear energy to meet the clean energy needs of the upcoming decades. Billions of dollars are being invested in nuclear energy, specifically in nuclear fusion.
Nuclear fusion is capable of producing 4 million times as much energy as burning coal to produce energy and the by-product is helium, an inert gas.
There are many obstacles to be overcome before nuclear fusion is commercially available but many milestones are being crossed with the renewed interest in this technology.
High Temperature requirement - For the electromagnetic repulsion between two hydrogen atoms to be overcome, temperatures as high as 100-150 million degrees centigrade is to be achieved.
Maintaining the temperature - Then this temperature is to be maintained for a period of time
Plasma stability –the behaviour of plasma is not well understood and the stability of the plasma is to be maintained to ensure the reaction proceeds without disruption. Scientists are studying the behaviour of mayonnaise to understand plasma better.
Raw material - While the deuterium isotope of hydrogen is available in plenty in seawater, the tritium isotope that is also required for nuclear fusion is very scarce.
Net energy gain - It is only recently that nuclear fusion reactions have generated more energy than was put in to start the reaction but there is a long way to go for Q to move from 1.2 to a viable 10.
Cost efficiency - nuclear plants have very high capital costs and economic viability is another big bridge to cross.
However, the promise of a clean and abundant energy source in the future has taken the imagination of the AI companies and these companies are today the biggest investors in nuclear fusion. This is accelerating the progress in this technology. Over $7.1 Billion has flowed into this space over the last 12 months, but this may still be inadequate, according to the Fourth Annual Fusion Industry report, to meet the requirements for delivering results the AI industry and the world at large is seeking from nuclear fusion.
In conclusion, AI shows tremendous promise as a disruptive technology but will require vast amounts of energy that conventional sources will fail to meet sustainably. Nuclear fusion could be the answer and there is renewed interest due to the hype caused by AI but the just is out on whether this interest is enough to move nuclear fusion from the laboratory to commercial viability in the near future.
Thanks for reading Technopolitik! Subscribe for free to receive new posts.
Technology has become important not just in our everyday lives, but has also become an arena for contestation among major powers including India. Through this survey, we intend to understand and assess what people think about how India should navigate high-tech geopolitics.
Please take a few minutes to complete this survey: https://bit.ly/technopolitik_survey
Cyberwar is here, but does it matter?
— Lokendra Sharma
It will soon be 1000 days of a war that began with the Russian invasion of Ukraine on 24 February 2022. While ‘cyberwar is coming’ since at least 1993 as John Arquilla and David Ronfeldt would like us to believe, there was particularly quite a lot of euphoria surrounding cyber operations in the initial days of the Russia-Ukraine war. Cyber operations have been mounted by both sides during the course of the last nearly 1000 days, including, for example, the recently reported ‘Russia-linked cyber campaign’ which targets ‘Ukrainian draft-age men with information-stealing malware.’ But the cyber dimension of the Russia-Ukraine war has not met the euphoric expectations — at least not like what drones have been able to accomplish. As Carnegie and Stimson have highlighted in their late 2022 and early 2024 outputs, respectively, cyber operations have been less effective than anticipated. A journal article by Marcus Willett in Survival in 2022 supported the same line of argument.
Theoretical and analytical arguments aside, is there any empirical evidence for the (in)effectiveness of cyberwar in the Russia-Ukraine conflict, which has been ongoing in some form or another for more than a decade? The following seminal paper published in 2017 by the Journal of Conflict Resolution has some answers:
Kostyuk, N., & Zhukov, Y. M. (2017). Invisible Digital Front: Can Cyber Attacks Shape Battlefield Events? Journal of Conflict Resolution, 63(2), 317-347. DOI: https://doi.org/10.1177/0022002717737138
This paper Kostyuk and Zhukov looks at ‘1,841 unique, mostly low-level, cyber attacks from August 27, 2013, to February 29, 2016’ and finds that ‘cyber attacks did not affect battlefield events in Ukraine.’ They find little relationship between cyber attacks and physical violence during war. They also briefly evaluate the Syrian conflict as well and find consistent results.
On the relationship between cyber and kinetic operations, they note: ‘At present, interaction between cyber and kinetic operations is similar to that between airpower and ground operations in World War I— when armies began to use aircraft for reconnaissance but had not realized their full potential to shape battlefield outcomes.’ While this may hold in the Russia-Ukraine context, Israel has recently demonstrated the effectiveness (notwithstanding the heavy collateral damage) of mounting a cyber kinetic attack when it exploded pagers and walkie-talkies in Lebanon.
Kostyuk and Zhukov do acknowledge certain shortcomings in their empirical analysis, including the limitation that their findings may not apply to large-scale wars, such as between the US and China or the US and Russia. This is because ‘no armed conflict has yet provided researchers with the data needed to evaluate this possibility.’
Despite the limitations, their finding that ‘cyber attacks are not (yet) effective as tools of coercion in war’ seems to be holding even after seven years (from the publication of their paper), especially in the Russia-Ukraine case. Would it continue to hold for the aforementioned actors? Would Israel’s cyber kinetic operation remain an aberration, or does it provide a glimpse into future cyber operations? We may have the answers to these questions sooner than we expect.
What We're Reading (or Listening to)
[Opinion] Solar Power: A New Opportunity for India-Taiwan Ties, by Anushka Saxena and Rakshith Shetty
[Podcast] Voter Sentiments in the Kashmir Valley, by Karishma Mehrotra and Carl Jaison
[Opinion] Why India and Pakistan should come together against the smog, by Adya Madhavan
Great write-up, Sridhar.