How are bond markets responding to the relentless advances in artificial intelligence? Economists from the Massachusetts Institute of Technology admit they are ‘very, very surprised’ by the reaction.
Between 2023 and 2025, bonds reacted to every new breakthrough in AI – such as a new version of a specific chatbot – with an increase in price and a decrease in yield. This applied to long-term government bonds, inflation-linked bonds and corporate bonds alike. Furthermore, this reaction was not transitory but lasting.
These economists are startled because the general narrative suggests that AI will deliver significant, widespread economic growth. Such growth should typically correspond with rising yields on long-term bonds.
Bond markets sceptical of scarcity claims
Elon Musk, chief executive officer of SpaceX and Tesla, has argued that AI will end the age-old problem of scarcity. But the bond market’s reaction suggests that AI is unlikely to fundamentally transform the economy into a state that, for the first time in history, practically overcomes the problem of scarcity, nor lead to any other similar extreme.
In short, bond markets do not believe that AI will trigger extreme, widespread disinflationary or deflationary pressures. However, Kevin Warsh, the likely incoming head of the US Federal Reserve, is betting on strong AI-driven disinflation. He believes the productivity surge caused by AI will be so significant that it will lead to a marked reduction in inflationary pressures.
This would allow the central bank to lower its benchmark interest rates – a move desired by President Donald Trump – as a means of mitigating pressures associated with the planned reduction of the Fed’s balance sheet.
AI winners and losers
What bond markets are most likely signalling is that AI will deepen inequality, the effects of which will dampen economic growth. In one part of the economy, productivity will increase substantially, perhaps unprecedentedly, leading to exceptional profitability – primarily for AI leaders like Amazon, Alphabet, Meta and partially Microsoft. From this perspective, the current massive investments by these firms into AI infrastructure and data centres make good sense.
However, a significant remainder of the economy appears likely to be adversely affected by AI. There is a risk of widespread layoffs and rising unemployment, which will limit the consumption capacity of a substantial portion of the population enough to significantly slow overall economic growth. This will create pressure in the form of ‘bad disinflation’ or even ‘bad deflation’ – driven not by technological progress and falling unit costs, but by a broader weakening of demand.
The pro-inflationary pressure of stronger economic growth (driven by AI-enabled profitability) will thus be countered by two deflationary forces: the aforementioned ‘bad deflation’ and the ‘good deflation’ caused by the reduction in unit production costs.
In aggregate, these two deflationary pressures will be stronger than the pro-inflationary ones. Consequently, the net impact of AI will be ‘good’ disinflationary, though not in the extreme form envisioned by Musk.
While part of the economy will benefit immensely from AI, another part will be significantly harmed. The damage to the latter could trigger strong calls for massive redistribution from ‘AI winners’ to ‘AI losers’, a sentiment that is likely to find a sympathetic political response. Demands for measures such as universal basic income are also expected to intensify.
The portion of the economy ‘defeated’ by AI will weaken growth to such an extent that bond markets today are not reacting to AI shifts as if they were a catalyst for clear, significant and widespread economic expansion.
Dealing with demographics
Finally, the bond market reaction suggests that the economic growth and productivity gains caused by AI will not be strong enough – due to these adverse side effects – to solve the problem of demographic ageing. In the eyes of bond markets, AI may be a major breakthrough, but not one capable of supporting both a newly created army of the unemployed and a growing retired population. This reality will dampen overall consumption so significantly that the pressure of ‘bad disinflation’ (driven by insufficient demand) will be a dominant factor.
Moreover, the larger the unemployed portion of the population, the stronger the natural pressure for massive redistribution. This unemployed segment could grow even larger the longer current immigration patterns persist.
This is most likely why, at this year’s World Economic Forum in Davos, figures like Larry Fink, CEO of BlackRock, and Alex Karp, CEO of Palantir, essentially argued that countries that did not allow mass immigration are better prepared for AI. Fink and Karp are among those who will undoubtedly rank among the ‘AI winners’. They already realise that mass immigration today represents even stronger future pressure to share their AI profits with the ‘losers’. Consequently, they adopted a relatively anti-immigration rhetoric in Davos that would have been unthinkable there just a few years ago.
Lukáš Kovanda is Chief Economist at Trinity Bank.
Interested in this topic? Subscribe to OMFIF’s newsletter for more.