Perspectives on the next wave of AI
Anna Adlarson, Venture & Growth investor at Kinnevik, shares her thoughts on the sectors where AI’s potential remains untapped, what sets the Nordics apart as a breeding ground for tech startups, and some of the most common misunderstandings about AI investing.
What are some sectors/opportunities where you believe the possibilities of AI remain particularly slept on?
Overall, we are still very early in the AI transformation. Most of what we see in the market today is AI being applied to fairly simple, horizontal productivity problems, often by plugging into the large, general-purpose models that are openly available. That’s important, but it’s only the first layer.
We think some of the most attractive – and most underappreciated – opportunities lie in tackling much more complex, domain-heavy problems. Biotech is one clear example, where companies like Recursion and Enveda are using AI to systematically accelerate drug discovery rather than relying on trial-and-error. In industrial technology and manufacturing, players such as Encube are starting to apply AI to optimise processes, speed up innovation cycles and improve quality in ways that were simply not feasible before. And in energy, companies like Aira illustrate how AI can be used for the energy transition – optimising consumption, managing distributed assets, and enabling new business models around electrification.
What these sectors have in common is that they are data-rich, operationally complex, and often highly regulated. That combination creates both high barriers to entry and the potential for very durable advantages once a company has built the right data assets, models and customer relationships. We believe that’s where a large part of the next wave of AI value will be created – and where the market is still not fully pricing in the long-term potential.
What makes the Nordic and European ecosystem distinctive when it comes to building AI companies?
On the talent side, the Nordics are exceptional. You typically see a combination of very strong technical foundations and a creative, product-oriented mindset. That’s a powerful mix when you’re trying to build AI-native companies that are not just research projects, but real, scaled businesses.
A lot of this comes from deep institutional roots. The region has world-class technical universities – such as DTU, KTH and Aalto – that have spent decades building strength in mathematics, computer science and engineering. Just as importantly, there is now a critical mass of global technology companies with significant AI capabilities in the region: Spotify, Klarna, Novo, Sana, Lovable, among others. These companies act as training grounds where engineers, product leaders and operators learn how to build and deploy AI systems at scale, and then leave to start the next generation of companies.
You also can’t underestimate the role of long-term, forward-looking public policy. The Nordics invested early in universal broadband, computer access and integrating technology into school curricula. There’s a great anecdote in that Sebastian Siemiatkowski, Klarna’s CEO, first learned to code through such a government-sponsored programme in Sweden. That type of systemic investment over decades is now paying off in the form of a deep, relatively egalitarian pool of AI talent and founders. Combined with Europe’s strengths around regulation, privacy and trust, it creates a distinctive foundation for building globally competitive AI businesses.
Which misconceptions about AI investing do you think need to be challenged?
The first misconception is the idea that the large model providers will “eat everything” and capture all the economics. We don’t think that’s how it plays out. In many important categories, the quality and usefulness of AI systems depend far more on access to specialised, high-quality data and deep integration into a specific workflow than on marginal differences in the base model. That creates room for very valuable companies at the application and vertical layers – especially where they can build proprietary datasets and embed themselves into mission-critical processes.
The second misconception is that growth at all costs is the only thing that matters. We believe investors should pay just as much attention to defensibility, unit economics and the evolution of industry structure. Infrastructure players will, quite rationally, try to move up the stack over time to capture more margin. At the same time, many companies that look impressive today are effectively thin wrappers around commoditising capabilities, without real moats in data, distribution or product. Those businesses may not be around in a few years.
For us as long-term investors, the key is to distinguish between “AI as a feature,” where competition and pricing pressure will be intense, and AI-native companies that are building durable advantages in data, workflows and customer relationships. That’s where we think sustainable value – and returns for our shareholders – will be created.
