Charlotte de Kerpoisson:
Hello and welcome to this podcast by BNP Paribas Wealth Management. My name is Charlotte de Kerpoisson.
2026 so far has seen a remarkable degree of dispersion in stock performance between stocks and sectors considered to be long-term winners of artificial intelligence, on the one hand, and stocks and sectors that are at risk from AI-driven disruption, on the other.
An example of this dispersion is global memory semiconductor companies, which have risen 65% year-to-date on strong anticipated demand from AI-focused data centres, while US software companies have fallen by more than 25% over the same period.
To discuss this topic in more detail, I'm delighted to be joined by Edmund Shing, Global Chief Investment Officer.
Hello, Edmund.
Edmund Shing:
Hello, Charlotte.
Charlotte de Kerpoisson:
Are investors overreacting to every announcement of new sector-specific features at AI Large Language Model companies, such as Anthropic, given the tremendous uncertainty around the risk of AI disruption to established players in software and other industries?
Edmund Shing:
Yes, well, I think probably investors are overreacting at the moment because there is enormous uncertainty, both at the economic level and at the company-specific level. What will AI really change in the medium to long term? And obviously, when you price that into the earnings expectations for companies, whether in the software industry or other, there's a huge impact over time. The problem is much more about the uncertainty. At the moment, we just don 't know. We don't know how big the impact is going to be, how long it will take to have this impact. But clearly, investors today are very nervous about the perceived impact. And so they're shooting first and asking questions later a little bit.
This is particularly true when you look at the reaction of the software sector, which, as you've mentioned, formerly a high-flying sector, now is judged to be a prime victim of AI disruption in the short to medium term. My actual feeling is many software companies will probably net-net profit from implementing AI in their software models. Actually, this could be a benefit to them, certainly in productivity terms, but also potentially in revenue terms as well. And yet today, investors are extremely skittish. So I think yes, it's probably an overreaction.
Does that mean I'd be a buyer right now? No. Because it's very difficult to catch a falling knife. And when you see this negative momentum, it's usually best to step aside and wait until we can be sure a bottom has been formed.
Charlotte de Kerpoisson:
In view of this uncertainty surrounding AI, are investors now moving away from the technology sector towards more industrial sectors, more traditional industrial sectors, which are holding up against the threat from AI, so the so-called "HALO," Heavy Asset, Low Obsolescence sectors, such as mining, oil & gas, and building & construction?
Edmund Shing:
Yes, absolutely. This has actually been true for some time, in fact, Charlotte. If we look through 2025, some of the best-performing sectors were not at all the tech sectors or anything related to tech, but much more related to, as you've mentioned, these raw materials-focused sectors, mining in particular. But in the last few months, as you mentioned as well, we've seen the oil and gas sector certainly picking up. Because whatever happens, you need those raw materials, you need the copper, you need the rare earth metals, both for technology and defense purposes. Of course, in terms of energy, you need oil and gas as well as nuclear and renewable energies in order to power the data centres that are being constructed today to run these AI Large Language Models.
So very much we see the bottlenecks for development in technology and AI at the front end of the chain, very much in the raw materials and the energy supply. And you're absolutely right. At the same time, these heavy asset industries are at a low risk of getting disrupted by AI. They should benefit from AI, the boom in AI, the energy demand of AI. But they are at a very low risk of being disrupted by it. So absolutely, I think that this HALO effect is here to stay.
I think that investors are rebalancing their portfolios, at a sector and even at a regional level, away from such dominance in the US and such a heavy weighting towards US large-cap technology companies towards other sectors and other regions. And this, I think, is a rotation, both a sector and a regional rotation, that will continue over time.
Charlotte de Kerpoisson:
Within Financials, does this argue for a preference for banks, especially European banks, versus insurance companies, given the risk of AI disruption to the insurance sector?
Edmund Shing:
I think it does, because I do think that particularly property and casualty insurance companies, that is home insurance and car insurance, do benefit from what we call "sticky" customers, people who cannot be bothered, like myself, who cannot be bothered to hunt around for a cheaper car or home insurance policy every year. You simply renew and you pay the higher premiums that are asked from the insurance company.
However, in a world where AI-driven agents can do the hard work of comparing insurance policies and pricing on your behalf and just delivering you the results easily and telling you which insurance policy you should be buying from whom and why, I think that erodes this barrier to entry, this erodes this "laziness" premium that insurance companies enjoy when they reprice your insurance premia on your homes and your cars higher every year. I think that can introduce much more competition, put much more price competition into the sector.
So for that reason, absolutely, I think banks, particularly European banks, are in a much better place today to cope with AI disruption because they will benefit from implementing AI in their back-office and middle-office processes to improve productivity, to improve profitability and potentially to reduce staff over time, whereas insurance companies clearly could be disrupted much more by this increased competition, particularly over pricing going forwards.
Charlotte de Kerpoisson:
Should investors continue to focus on investing indirectly in the AI mega-theme, so for example, in essential metals and energy infrastructure necessary for defense and technology, rather than investing in the US technology mega-caps?
Edmund Shing:
Absolutely yes. If we look at the performance of, for instance, the energy infrastructure theme in the US, even since November last year, it has continued to go from strength to strength. The same can be said of essential, for instance, rare earth metals or even copper miners and the copper price. Since November, these have gone from strength to strength. At the same time, if we look since November, the technology mega-cap names, the Magnificent Seven in the US, like Amazon and Google, have materially underperformed since November of last year. Even Nvidia, which has been the star of the Seven, even when it reported excellent quarterly results far above analyst expectations, did not manage to generate a very positive price reaction in the wake of these strong results.
So even there, it seems that the market is getting used to Nvidia delivering excellent news quarter after quarter and is not prepared anymore to mark the prices higher. I think that tells you there is a sea change in terms of sentiment around the US technology sector and around these Magnificent Seven names in particular.
So for me, absolutely, stay with the commodity end of the spectrum and stay with the electricity infrastructure end of the spectrum, i.e. the picks and shovels, as opposed to investing in the companies, such as the Magnificent Seven, who after all, are going to be spending in Capex terms, something like $660 billion this year, which is a huge amount, more even than what they spent last year.
Charlotte de Kerpoisson:
Do you think that the current AI disruption panic may give rise to attractive buying opportunities in selected, beaten-down companies where the risk to growth and profitability from AI is exaggerated? And how can investors seize these buying opportunities?
Edmund Shing:
I do think that ultimately there is a bit of exaggeration going on. A lot, even in software companies, actually are not going to suffer anything like as much as people imagine today or as people are worried about today. With a lot of software companies, I actually think the AI will be a net benefit. They'll be able to integrate AI technology, AI models, and algorithms into their existing software to improve the software, particularly the experience for the final customer. And that in many ways could both enhance productivity at the company but also deliver a better product for the client over time and actually could therefore enhance the product offering and the stickiness of the client.
I think there are all manner of services, both IT and otherwise, which are being marked down heavily today because of the fears of AI disruption, where the reality is very simply that customers will remain sticky. When you're used to a software package like Word, you're not going to change instantly to another one just because something might be a bit cheaper. Because you've already learned that specific software package. You don't want to relearn a different package. And I think there is that barrier to entry, which is very much underestimated, underpriced at the moment in, for instance, the software companies today.
Now, the problem is that not all software companies will be the same. Some will be disrupted by AI. In other ways, others will probably benefit from AI. Sifting the winners from the losers is not going to be easy. It's not something you can buy necessarily at a sector level because there will be some winners but clearly some losers at the same time. So I think it does mean you need to rely on specific expertise of those companies and of the business models to figure out which types of business models will resist or even be improved, and which other ones are at the blockbuster risk of obsolescence in the medium term.
How do you take advantage of that? I'm afraid to say it's very likely to be at the stock-specific level. You will have to buy the stock-specific names. An alternative approach, which is perhaps more interesting, is to buy into hedge funds and alternative UCITS funds that are so-called "market neutral" and employ long-short strategies. That is, they buy long exposure to some companies, expecting them to do better, but equally well, they short other companies, expecting them to suffer. Those hedge funds and strategies could really benefit from what we call this wide dispersion between the winners and losers.
The idea is if you can buy more of the winners and you sell more of the losers, without taking an exposure to the overall market, you can benefit from the widening dispersion performance between these two groups. So that's a way, without trying to pick the companies yourselves, to buy into a fund that can benefit from this increasing dispersion over time.
Charlotte de Kerpoisson
Thank you, Edmund.
And thank you to our audience for following this podcast. Please like share and subscribe to our weekly podcasts and visit our website for our investment themes for 2026 and strategy research. Goodbye.