Three-Line Summary
*AI has grown beyond a single technology into a core engine driving economic growth, trade, finance and electricity.
*Against that backdrop, slowing the pace of AI development could put the brakes not only on technology but on economic growth itself.
*The task ahead may come down to how to design growth through AI alongside safety.
"We need to put the brakes on artificial intelligence, which may slip beyond human control and threaten humanity." A debate over slowing down AI that emerged abruptly last week has stirred the world. The heads of companies at the frontier of AI development — OpenAI, Anthropic and Google — raised the concern, while Nvidia and Intel pushed back, saying the technology can be adequately controlled.
That prompted a question. AI has moved beyond being simply a technology to become a central pillar driving economic growth, trade, finance and electricity. What would actually happen if the pace of AI development were slowed? In this article, we consider how realistic a choice slowing down AI is, and why it has become such a difficult question. Given the subject, this piece will range beyond energy into other fields as well.
'AI = Growth'
Many experts agree that AI's influence has moved beyond the industrial sphere. Consider the United States first, the leading AI power. ING estimates that AI-related investment accounted for roughly 36% to 44% of U.S. gross domestic product growth in the second quarter of this year, on a real basis. JPMorgan puts the share at about one-fifth. More recently, two extreme positions have clashed: Harvard University professor Jason Furman has argued that AI infrastructure investment contributed more than 90% of GDP growth, while Jan Hatzius, chief economist at Goldman Sachs, countered that the contribution is close to zero because most semiconductors are imported from Korea and Taiwan. Either way, the point that matters is this: AI already carries part of the burden of growth in the United States.
It is not only the United States. Kristalina Georgieva, managing director of the International Monetary Fund, said last month that AI investment is helping the global economy hold up better than feared amid the energy shock triggered by conflict in the Middle East.
AI has emerged as an important pillar supporting trade, finance and energy as well. Ngozi Okonjo-Iweala, director-general of the World Trade Organization, said in a recent interview with the Financial Times that AI trade is masking the effects of tariffs and all the other disruption facing global trade — adding that if AI were a bubble, such trade growth could not have been sustained. According to the WTO, trade in AI-related goods such as semiconductors, chemicals and industrial equipment rose 42% in the first quarter of this year from a year earlier, in sharp contrast with 7% growth in non-AI goods trade over the same period.
Korea is no exception to this trend. For a country enjoying an AI-driven memory super-cycle, slowing AI down would come as a bolt from the blue. Unlike the United States, Korea is a latecomer in AI, and the sense of urgency about sprinting ahead appears greater. On this point, Bae Kyung-hoon, deputy prime minister and minister of science and ICT, said it is an entirely different matter for a country already out in front to moderate the pace of AI development than for a country that must catch up while simultaneously creating new markets to slow itself down.
Money and Electricity Follow AI
The bond market is in the same position. AI capital spending is one of the factors behind the recent slide in U.S. Treasury prices: large corporate bond issuance by hyperscalers is increasing demand for funds in the bond market and adding upward pressure on long-term yields.
Could one counter, then, that moderating the pace of AI development would shrink corporate bond issuance and calm some of the turmoil in the Treasury market? As noted above, a sudden drop in AI capital spending cannot be ruled out as a negative signal that economic growth is slowing. That, too, could have a considerable impact on the Treasury market.
This series has reported repeatedly on AI's effect on energy. The International Energy Agency expects annual electricity consumption by data centers worldwide to more than double, from about 415 terawatt-hours in 2024 to about 945 terawatt-hours in 2030. By 2030, the electricity used by data centers alone would exceed Japan's current annual power consumption. That is why nuclear power generation is reviving in what is being called a nuclear renaissance, and why grid infrastructure equipment such as transformers is in short supply. The power industry, in other words, has entered a period of upheaval.
Growth and Safety Together
Taken together, slowing the pace of AI development means the brakes could come on in these areas as well. At this point, one can also ask whether moderating AI is feasible — that is, whether slowing growth is even possible. Slowing the pace of AI technology has become a far more complicated, economic choice than it might appear.
In that respect, AI technology differs to some degree from the frequent points of comparison, nuclear weapons and genetic engineering. Those technologies also had important effects on national security and industry, but they were not the core engine determining economic growth itself.
Still, there is a clear need for alternatives to address the very concern that invites comparison with nuclear weapons and genetic engineering — that AI could not only escape human control but put human lives at risk. On that score, the position Amazon has taken on the current controversy is worth examining: "We do not see this as a choice between progress and safety. Models should be released when they are ready to be used and judged safe, and that is secured through rigorous testing and strong safeguards." One could read that to mean that anyone genuinely worried about AI's risks should devote as much investment and attention to safeguards as to performance.
How to design growth and safety together. That may be the real question posed by the debate over slowing down AI. What do you think?
Related 'Petro-Electro' articles
① The 'AI Electricity Bill' Is Coming — Who Pays?
② If the Fate of Global Warming Also Rests on AI
③ Are Data Centers Only Guzzlers of Water and Power?
※ From petroleum to electricity. Energy is a key keyword for understanding the economy, industry, international affairs and the response to climate change. Subscribe to this reporter or to the [Cho Yang-joon's Petro-Electro] series for economic insight through the lens of energy.
