Swisscanto says it is too early to call AI a bubble
Gerhard Wagner of Swisscanto separates AI’s structural potential from specific overvaluations and from the challenge of turning investment into profits.
Heavy investment in artificial intelligence has brought the same market question back into view: are we seeing a durable technological transformation or a bubble? Gerhard Wagner, Head of Sustainable Equities Strategies at Zürcher Kantonalbank, the delegated manager of Swisscanto funds, says it is still too early to settle the question with the latter label.
In an interview published by Finect on 3 March, Wagner argues that AI can drive structural change, while adding an important qualification: not every company and not every part of the value chain necessarily warrants the same valuation. That distinction is more useful than an absolute answer to whether the entire sector is, or is not, a bubble.
The thesis does not remove risk
Wagner is not arguing that every AI-related investment will work. The risk he identifies is monetisation: companies are making significant infrastructure investments, and that pace could slow if uncertainty about their returns persists.
He also acknowledges that some areas may be overvalued. That is why he argues for examining valuations company by company rather than treating AI as a single, uniform category. In his account, a high share price can only be assessed alongside expectations for revenue, earnings, invested capital and competition.
That matters in a debate that is often simplified. A technology can have real applications while some companies connected to it trade on excessively demanding expectations. AI usefulness and the price the market places on each company are not the same question.
From chips to applications
Swisscanto places some of its interest in semiconductor manufacturers and in the electrification connected to data centres. But Wagner also points to a possible shift in emphasis: from infrastructure and models towards applications.
In the interview, he cites business processes, consumer products and robotics. Healthcare has a special place in his analysis, both because of the adoption of AI tools and because of its potential in drug development and other medical uses. A Swisscanto article published in February makes a similar case: the value chain does not end with computing suppliers.
This does not prove future results. It does help explain why some investors do not reduce the story to chip sales or the number of data centres. If applications produce measurable improvements in productivity, costs or products, the debate moves from technical capability to value creation.
Infrastructure matters too
AI expansion depends on electricity, grids and data centres. The International Energy Agency report Energy and AI notes that there is no AI without electricity for data centres and examines both energy demand and possible AI uses within the energy system.
That context explains Swisscanto’s interest in electrification. Infrastructure is not a minor detail in this story: it shapes costs, timelines and deployment capacity. It also requires a distinction between software promises and the physical investments that allow models to operate at scale.
A cautious reading
The claim that it is too early to call AI a bubble is Wagner’s market thesis, not a certainty or a recommendation to buy assets. His own argument recognises geopolitical risks, uncertainty about profitability and valuations that should be reviewed case by case.
For readers, the lesson may be simpler than choosing a label. It is worth asking what problem an AI application solves, who pays for it, what infrastructure it requires and which assumptions are already embedded in a company’s share price. The technology may create real opportunities without making every market expectation a certainty.
Sources
Sources for this piece
This piece draws on 3 primary source(s), gathered during reporting.
This article was produced with artificial intelligence under human editorial oversight.
