Exploring the Impact of Artificial Intelligence on Global Trade

The impact of AI on international trade and investment will undoubtedly result in winners and losers, but it is the secondary consequences that could be particularly intriguing.

To understand them, start with two premises: First, AI services consume a lot of energy, not all of which is green. Second, many countries regulate either the use of artificial intelligence or the introduction of goods and services derived from artificial intelligence, for example the creation of new medicines or teaching techniques.

Let’s look at each of these factors in turn.

Sending a query to ChatGPT consumes a lot of energy, one estimate says 10 times more than a Google search. Currently, large language models are limited enough that this is not a significant factor in overall energy consumption. But as the use of artificial intelligence services increases, the energy burden increases. Countries that have expensive energy or that do not allow energy consumption to increase much due to climate or regulatory reasons tend to import their artificial intelligence services from energy-rich countries.

In the future, energy-rich regions can be Spain and Morocco with solar power, South Korea with cheap nuclear power, and any number of countries that are pioneers in nuclear fusion. These countries can end up being significant exporters of information produced by artificial intelligence. They may get their AI feed from the US, but they specialize in cheap computing and data transfer. And some areas of the Americas may also join this list, especially if they are well-suited to solar and hydropower.

To be clear, the US will be exporting a lot of AI services through companies like OpenAI, Google, Meta and Anthropic. But the US isn’t as good at building affordable infrastructure, putting it at a disadvantage in the AI revolution and sharing many of the gains overseas.

It remains to be seen whether there will be greater profits from selling the original source code or the more derivative electric-powered, infrastructure-based AI calculations. Regardless, this is a potential economic and national security risk for the United States. It may end up with a strong lead in the source product, but fall far behind in the production (“manufacturing”, one might say) of the final AI outputs.

This potential problem can be alleviated by making it easier and cheaper to permit and build electricity production. Any US state that does this—and perhaps some do—could become a real economic powerhouse. Many US institutions would likely buy their AI computing domestically rather than from a foreign power, if only for data security reasons.

Another important factor in shaping international trade stems from domestic regulations regarding possible products of artificial intelligence services. Consider this scenario: AI services suggest several plausible drugs to treat some disease. Still, the United States has fairly strict restrictions on the approval of such drugs.

So a trading opportunity arises when some countries specialize in testing products for AI services. It is already the case that many large pharmaceutical companies conduct drug trials in Africa, where costs are lower and regulation is looser. The scope for regulation is greatly expanded, and the end result is that countries willing to take up regulatory opportunities will attract more foreign investment.

Or how about this related scenario: Advanced AI suggests that some teaching techniques are better. Many countries may be too bureaucratic to take advantage of such options quickly. It is not difficult to imagine that some smaller nations, especially those governed with fewer checks and balances, would be quicker to implement changes. What if Singapore adopts new educational innovations before they spread to Western Europe or California?

It is not just that education in Singapore is improving. Singapore can use those innovations to develop products for export, such as effective online training. Nations that develop or tolerate innovations produced by artificial intelligence will also become more important exporters.

Can you imagine a future where the US remains the leader in AI innovation, but the world relies on Singapore or Uruguay for many real products (and energy sources)? What if it is easier for these states to install nuclear fusion or experiment with AI-derived social and economic innovations? Over time, these countries may become more important allies.

More generally, many larger nations may seek smaller partners with more institutional flexibility as part of a new set of AI-driven economic and perhaps military alliances. Who is more dependent on whom?

In any case, as with many other human endeavors influencing AI, international trade will never be the same again.

Elsewhere in Bloomberg’s opinion:

• The death of globalization has been greatly exaggerated: Daniel Moss
• Bigger threat to US trade than Chinese chips: Tim Culpan
• Your future AI has multiple personalities: Parmy Olson

Want more Bloomberg opinions? Subscribe to our newsletter.

This column does not necessarily reflect the opinion of the editorial board or of Bloomberg LP and its owners.

Tyler Cowen is a Bloomberg Opinion columnist, professor of economics at George Mason University, and host of the Marginal Revolution blog.