There are no prizes for guessing the number one topic of discussion at conferences this year, regardless of industry. AI and its ramifications – from construction and data centres to software and healthcare – have dominated discussions between companies and investors to an extent I have rarely seen. But we increasingly wonder whether the conversation focuses on the right places.
There is often an underlying assumption that AI is a theme accessible only to US-focused investors, or perhaps those looking at emerging markets and Korea.
While European investors cannot access every layer of the AI stack, there are still plenty of opportunities. In fact, from our perspective even some of the ‘losers’ could be more attractive than many of the high-profile ‘winners’ elsewhere.
The opportunity stack
A useful starting point when assessing an industry's long-term profitability is to count its competitors. Market share stability is another indicator: frequent changes in leadership can suggest low barriers to entry and, consequently, less attractive returns over time.
We see the model providers at the top of the AI stack through this lens. Google, Grok, Meta and potentially OpenAI and Anthropic are opportunities for our US colleagues. But are we jealous? Not really. Given the number of competitors and the speed at which successive models displace their predecessors, we see little evidence of durable competitive advantages.
These businesses may become profitable, but we question whether they will generate attractive returns on the enormous capital they are investing. Meanwhile, cheaper models may prove good enough to erode pricing power further.
Next come the data centres. Some appear highly profitable today, depending on their accounting policies. But abundant capital, numerous competitors and limited evidence of barriers to entry make sustained high returns uncertain. Today's profits may simply reflect demand temporarily outstripping supply.
Then come the chip designers and manufacturers, including Nvidia and TSMC, and eventually we reach the companies that make the machines that go into the factories that produce the chips. This is where we get interested.
European investing suffers from a perception problem. Investors often associate the region with struggling carmakers or stagnant GDP growth. Neither reflects the full opportunity set.
Take ASML, Europe's largest company. It makes the sophisticated lithography equipment used to print patterns onto silicon wafers and is the sole supplier of the most advanced patterning systems. This dominant position gives it considerable pricing power and the potential to earn attractive returns on capital.
The appeal of ASML, Technoprobe, BE Semiconductor and our other semiconductor equipment holdings is similar. They operate in concentrated niches, hold leading market positions in advanced technologies and are already profitable.
Their substantial R&D investment also creates a reinforcing advantage: market leaders can often invest more in developing the next generation of products, helping them maintain their dominance.
European AI plays may not grab attention, but they certainly have a lot going for them.
Software stickiness
European software is another area we find particularly interesting. The stock market increasingly appears to believe that AI will make software effectively free, undermining the value of established software businesses. We suspect the real world is rather less frictionless.
Anyone travelling through the UK in early September will appreciate this. A failure in the NATS air traffic control systems on 8 September reportedly caused around 2,000 flight cancellations, with disruption continuing for days.
Parts of the underlying technology date back decades, and a major upgrade has been discussed for years, with implementation not expected until the 2030s. If nothing else, this illustrates how difficult it can be to replace critical systems.
A less extreme example is SAP, Europe's largest software company. Its enterprise resource planning (ERP) software underpins the operations of many major businesses. Yet adoption of its cloud platform has been remarkably slow.
More than a decade after its introduction, around half of SAP's customers have reportedly yet to begin migrating, despite the benefits.
These examples highlight a fundamental characteristic of mission-critical software: switching costs are high, operational risks are significant, and businesses are understandably reluctant to disrupt established workflows or sensitive data.
Technological change may therefore happen far more slowly than its theoretical potential suggests.
If that remains true, some software companies whose valuations have fallen sharply could prove substantially undervalued, presenting compelling investment opportunities.
Europe, then, offers a rich hunting ground for AI-related investment. We see both genuine AI winners and businesses potentially mislabelled as AI losers. Both could deliver attractive long-term returns.
Thomas Brown is manager of the Premier Miton European Opportunities fund. The views expressed above should not be taken as investment advice.