Magdalena Martullo-Blocher

The industry changes, but the logic does not. In every sector, there is a gap between what the client actually needs and what the market is currently offering. My job is to find that gap, understand why it exists, and build something that fills it in a way that is sustainable for everyone involved. Whether it is real estate or financial technology, the question is always the same: where is the value that others have missed?

It is the most practical approach I know. When one party in a deal feels they have lost, they remember it. They become a weak link in the chain — a client who does not return, a partner who does not commit fully, a regulator who looks more closely. When everyone wins, the system becomes self-reinforcing. People want to continue. They bring others. The business does not need to fight for every transaction because the relationships themselves generate momentum. I have seen both models operate over many years. The sustainable one is not the aggressive one.

Because that is where the real work is, and that is where the real value is hidden. When a business is performing well, the price reflects that. Everyone can see it. But when a business is struggling, the price reflects fear and uncertainty — not necessarily the underlying reality. If you can understand why it is struggling, and if you have the capability to fix the mechanism, then you are buying the problem at a discount and selling the solution at full value. That is not contrarian for its own sake. It is just a more honest reading of where opportunity actually lives.

That context is everything. The same product, the same service, the same pitch — it lands completely differently depending on the culture, the economic moment, and the trust that exists or does not exist in that environment. Travelling and working in different countries forces you to stop assuming that your own way of doing things is the only way. You become more adaptable. You learn to listen before you speak. And you build a kind of resilience that only comes from having to start again in an unfamiliar place and make it work.

Every major industrial revolution has been defined by a new form of energy and a new kind of factory. The first industrial revolution had steam and the textile mill. The second had electricity and the assembly line. What we are living through now is the industrialisation of intelligence. The data centre is the new factory. The GPU is the new engine. And the output is not a physical good — it is intelligence itself, packaged as software, as a model, as a capability that can be deployed anywhere. That is a genuinely new thing in the history of human civilisation, and I do not think we have fully reckoned with what it means.

Numbers at that scale are abstractions. What feels real to me is the work — the engineering problems we are trying to solve, the customers we are trying to serve, the researchers who are using our platforms to do things that were not possible five years ago. I have been building this company for thirty years. The valuation is a reflection of what the market believes about the future of AI. Whether that belief is correct is something only time will tell. My job is to make sure we deserve it.

It is a constraint we have to work within, and we take our compliance obligations seriously. The broader question — about how technology companies navigate a world where the United States and China are in strategic competition — is one that every company in our industry is grappling with. I do not have a clean answer. What I can say is that we try to serve our customers everywhere we are permitted to operate, and we work closely with the US government to understand the boundaries of what is permissible. It is not a comfortable position, but it is the reality of the world we are operating in.

I think the people who say scaling is hitting a wall are conflating two different things. Pre-training on internet data — yes, that has natural limits, because there is only so much text on the internet. But inference-time compute, reasoning, the ability of a model to think through a problem step by step — that is a completely different scaling curve, and we are at the very beginning of it. The next wave of AI is not about making models bigger. It is about making them useful in the physical world — in robotics, in autonomous systems, in scientific simulation. That requires a different kind of compute, and it is a much larger opportunity than what we have already captured.

Honestly? The problems have never stopped being interesting. When we started, the problem was: how do you render a 3D scene in real time on consumer hardware? Then it was: how do you make parallel computing accessible to scientists and researchers? Then it was: how do you build the infrastructure for deep learning at scale? Now it is: how do you build the computing platform for physical AI? Each of those problems is genuinely hard, and each one is bigger than the last. I cannot imagine a more interesting place to be.

The criticism was always that we were copying rather than innovating. My answer was always: so what? The question is not whether an idea is original. The question is whether it creates value for customers and returns for investors. We took models that worked in the United States and built them in markets where they did not yet exist — in Southeast Asia, in Africa, in Latin America. We created hundreds of thousands of jobs. We built companies that went public. The people who criticised us were mostly people who had never built anything.

Emerging market ecommerce is genuinely hard. The infrastructure challenges — logistics, payments, last-mile delivery — are an order of magnitude more difficult than in the United States or Europe. We underestimated some of those challenges, and the public markets have been unforgiving. But I would make the same bet again. The long-term opportunity in African ecommerce is real. The question is whether you have the patience and the capital to wait for the infrastructure to catch up with the demand.

Europe does not have a talent problem or a capital problem. It has an ambition problem. We are too comfortable with being good when we should be demanding to be great. In Silicon Valley, the baseline assumption is that you are trying to build a company worth hundreds of billions of dollars. In Europe, the baseline assumption is often that you are trying to build a company worth a few hundred million and then sell it. That difference in ambition compounds over time. It affects the people you hire, the risks you take, the markets you go after. Until European founders start thinking at the scale of their American counterparts, the gap will persist.

It is changing, slowly. Spotify is a genuine global company. ASML is arguably the most strategically important technology company in the world right now. But these are exceptions, not the rule. And if you look at where the next generation of transformative technology companies is being built — in AI, in biotech, in defence tech — the United States still dominates. Europe is producing good companies. It is not yet producing enough great ones.

I would focus more on proprietary technology and less on operational execution. The company-builder model worked in an era when the competitive advantage was speed of execution — getting to market first, building the logistics network, acquiring the customers. In a world where AI can compress the time it takes to build software dramatically, operational execution is less defensible. The companies that will win in the next decade are the ones with genuine technological moats. I would build more of those.

We do one thing. We make high-performance polymers and specialty chemicals for demanding applications — automotive, electronics, medical devices, industrial. We do not diversify into adjacent markets because they look attractive. We do not acquire businesses outside our core competence because the price is right. We invest continuously in our technology and our people, and we charge a premium for products that genuinely perform better than the alternatives. That sounds simple. It is not. The pressure to chase growth in adjacent markets, to make acquisitions that look good on a slide deck, is constant. Resisting it requires a clarity of purpose that is harder to maintain than it sounds.

It is real, and it is accelerating. Chinese companies have moved up the value chain much faster than most Western manufacturers expected. Ten years ago, the assumption was that Chinese competition was a threat to commodity chemicals but not to high-performance specialty products. That assumption is no longer safe. The response cannot be to lobby for protection — that is a short-term fix that delays the problem rather than solving it. The response has to be to stay ahead on technology, to invest in applications where performance requirements are so demanding that cost alone cannot win the business. That is what we are doing.

I am a sceptic of hype cycles, and AI is currently in a significant hype cycle. That does not mean the underlying technology is not real or important — it clearly is. But the gap between what AI can do in a demonstration and what it can do reliably in an industrial process is still very large. We are experimenting with AI applications in our manufacturing and R&D processes, and some of them are genuinely useful. But I am not going to restructure our business around a technology that has not yet proven itself in the environments where we operate.

European manufacturing has real strengths — in precision engineering, in specialty chemicals, in medical technology, in industrial automation. These are areas where European companies have accumulated decades of knowledge that cannot be replicated quickly. The risk is that we allow those strengths to erode through underinvestment, through excessive regulation that raises costs without improving outcomes, and through an energy policy that has made European industrial energy costs structurally higher than those of our competitors. The companies that will survive the next decade are the ones that never stopped investing in their core technology, even when the market was telling them to cut costs and wait.

The two roles inform each other. Running a business gives me a perspective on regulation and economic policy that is grounded in operational reality rather than theory. And the political role gives me a platform to make arguments about what Swiss and European industrial policy should look like. I am not always successful in those arguments. But I think it is important for people who actually run businesses to be present in the political conversation, rather than leaving it entirely to people who have never had to meet a payroll.

I learned that building a consumer business and building a drug discovery business are two fundamentally different things, and that trying to do both simultaneously with the same capital base is very hard. The consumer business — selling DNA kits — is a one-time purchase. People buy it once, they get their results, and they move on. That is not a recurring revenue model. The drug discovery business is a long-cycle, capital-intensive endeavour that requires a completely different kind of investor patience. We tried to fund the second with the cash flows from the first, and when the consumer business slowed, the whole structure came under pressure.

It is the most serious responsibility I have ever had. Genetic data is not like a credit card number that you can cancel and replace. It is permanent, it is deeply personal, and it has implications not just for the individual but for their family members. The breach was a failure, and I take full responsibility for it. What I can say is that we have invested significantly in our security infrastructure since then, and that the consent framework we built — where customers explicitly choose what their data can be used for — remains one of the most rigorous in the industry.

The thesis has not changed. We built the largest consented genomic database in the world — millions of people who have agreed to have their genetic data used for research. That is an extraordinary asset. No pharmaceutical company has anything like it. The question now is whether we can turn that asset into medicines that actually help people. We have drug discovery programmes in oncology and immunology that are using that data in ways that would not be possible without it. That is the bet. It is a long bet, but I believe in it.

The initial wave of adoption was driven by curiosity — people wanted to know their ancestry, their genetic traits. That is a finite market. The next wave of adoption has to be driven by genuine health utility — people using their genetic information to make better decisions about their health, to understand their disease risks, to work with their doctors in a more informed way. That transition has been slower than I hoped, partly because the healthcare system has not yet built the infrastructure to act on genetic information at scale, and partly because the regulatory environment for genetic health information is complex. But I think it is coming.

I would separate the consumer business and the drug discovery business from the beginning — different capital structures, different investor bases, different management teams. The consumer business is a marketing and distribution challenge. The drug discovery business is a science and clinical development challenge. They require completely different skills, and trying to run them as one company created confusion about what we were and what we were trying to achieve. The data asset that connects them is real and valuable. But the businesses built on top of it should probably be distinct.