Stock markets and AI: from unbridled enthusiasm to existential questions

The first version of ChatGPT was unveiled in November 2022, creating buzz around AI-themed stocks and kicking off a rally. Investors flooded into data-centre hardware makers as well as the hyperscalers¹ that were ordering those data centres. It was heady days for AI stocks around the world – until the first signs of a rift began to appear in October 2025. Hardware manufacturers (such as for data-storage devices and microchips) continued their upwards march, while software developers and hyperscalers began to run out of steam. For software developers, investors were worried about the direct, ruthless competition from AI, and for hyperscalers, the concern was whether these companies would earn a return on their data-centre investments – trillions of dollars enriching hardware manufacturers first and foremost.

Then in early February 2026 came the “Anthropic moment.” Anthropic – a competitor of OpenAI and its ChatGPT chatbot – launched an AI agent designed specifically for the legal profession. Other AI agents soon followed. Investors were stunned by how efficient these models were and got their first glimpse of AI’s potential to dismantle reams of white-collar jobs. They duly slashed their holdings in knowledge-based services companies, which were perceived as being most at risk in an AI world and whose stock prices had been boosted by pumped-up valuations of intangibles. That month, South Korea’s KOSPI index – heavy on hardware firms – outperformed the S&P 500 by over 20%, as the US index was dragged down by the sell-off in knowledge services. As a result, fears of an AI bubble finally materialised through a simple yet abrupt rotation among these subsectors, which pushed back the timing of AI’s grande finale on the global equities stage.

At that point, concerns about an AI bubble gave way to concerns about the technology’s macroeconomic impact. Highly convincing articles about the potential for job destruction hit the market. One by Citrini Research² outlined a bleak scenario for 2028, describing a world thrown out of kilter by overly efficient AI: “The companies most threatened by AI became AI’s most aggressive adopters… The collective result was catastrophic. Every dollar saved on headcount flowed into AI capability that made the next round of job cuts possible.” In another compelling article, an economist³ esteemed for his vision and independence described how capitalism could come to an end if capital made the supply of labour infinite: “In this situation, capital becomes labour.” He wrote that if AI enhanced workers’ capabilities, that would increase the value of labour and push wages up. But if AI replaced workers’ minds and eventually their hands, that would solve the labour shortage problem by putting the value of labour at zero. We would therefore need to think up a new way of organising society in which robots paid the taxes that subsidized humans’ inactivity. Marx seemed to have already predicted this situation in “The Fragment on Machines.” Nearly two centuries later, speaking at Davos, Elon Musk said: “You can’t have work that has to be done by only some people and amazing abundance for all.” The debate was therefore on between augmented intelligence (which is desirable) and replaced intelligence along the lines of what we saw when Jack Dorsey⁴ laid off 40% of his workforce on 27 February 2026. Never had it seemed so important for business owners/decision-makers to take a responsible approach to employment, and never had the need for a clear vision and strong political will seemed so dire. After all, creating a new system of distribution in order to reorganise the way in which society operates is no mean feat! We were pulled body and soul into an AI world, which we imagined would be filled with humanoid robots either serving at our feet or on Big Brother’s payroll.

Then on 28 February, Israel and the US launched strikes against Iran. Now the smell of oil and gunpowder fills the air. The link between fossil fuel and energy-hungry AI is essentially a bridge between the prior world and this new one. While AI seems to be creating deflationary pressure, it also has an inflationary counterpart. The flames in the Middle East should shake us out of our AI tunnel vision. Every major technological revolution has given rise to deep-seated fears. Each time, the fear in the beginning was about net job destruction, since it’s easier to see the jobs that will be lost than to envisage the ones that will be created. And each time, more jobs ended up being created than destroyed. Each revolution also came with specific fears: a number of “experts” in the mid-19^th century wrote that the human body couldn’t withstand railcar speeds of 60 km/h for an extended period of time. And in the early 20^th century, sometimes hysterical warnings were issued about the dangers of electricity. Today, what is the specific fear associated with AI? Is it that power will be concentrated in just a few hands, that software and robots will control our lives, that people will stop thinking for themselves, or that human labour will disappear?

What we can do as fund managers is pick out the AI winners and losers (current and future) at each step of the technological revolution under way and identify how it will affect the economy, in order to best serve our clients. To a certain extent, we can also encourage companies to act in a socially responsible manner amid the structural shifts taking place.

Businesses, governments, and sole proprietors – whose ranks could swell as a result of AI – will need to be creative, dig up old practices, and create a system of parallel universes, much like in Japan where an extremely competitive, open economy sits alongside a sheltered economy for those segments of the population less able to compete. Looking further out, the expected productivity gains from AI and its robots will come at just the right time to respond to a shrinking workforce and aging population. Nature, sometimes helped by an artifice, works wonders.