Episode 458: Novartis CEO Vas Narasimhan on Drug Development & AI's Role in Disease Treatment

You know, we feel like we're in a really strong place. We're coming out of a pretty significant transformation starting in 2018, and even a few years before that, we set on this journey to refocus Novartis as a pure-play medicines company. We have over 200 years of history with many twists and turns, starting in dye-making, moving into chemicals, agrochemicals, and then medicines. But in 2018, we decided we needed to become a pure-play medicines company to have the biggest impact. It took us six years to achieve, with about $130 billion in transactions, turning one company into four: Haleon (with GSK), Sandoz, Alcon, and Novartis as the standalone innovative medicines company. We're now having our best financial performance possibly in the company's history, showing this was the right move. Looking ahead, we have a great pipeline with exciting things to look forward to.

That's a great question. Innovative medicines companies often stay as conglomerates because of the challenge of patent expiries; we have to replace all of our sales every 14 to 17 years. Diversification used to soften the innovation cycle, but it led to suboptimal capital deployment and loss of focus on core innovation. In 2018, we realized there was so much innovation happening in early science that we couldn't split our capital across all these businesses, so we decided to focus down. It was a long journey, but it was the right one—focusing on what we're great at and where we want to deploy our capital.

Culture is ultimately what a leader leaves; it's the legacy that continues to drive performance for decades. In 2018, we shifted from a top-down culture to one we call 'inspired, curious, and unbossed'—inspired by our purpose, curious about the world, and unbossed with the ability to pursue work with a servant leader mindset. It was a big task with 135,000 people in over 100 countries. We've come a long way, and now as a more focused company, it gets easier. Culture is about starting the great conversation. For example, when we rolled out 'unboss,' there was lots of debate and different interpretations, but those discussions made our culture stronger. We want people to feel empowered within guardrails, and leaders to create clarity and remove obstacles. It's a challenge to define 'unboss' in China versus the US, but the spirit in the company has been transformed over these years.

Early in your career, you build your value system. I was fortunate to work with pioneers like Paul Farmer, Richard Cash, and Jim Kim, and work on big public health programs like the first antiretroviral treatment program in Africa. It instilled in me a deep passion for public health and using technology to make big jumps in quality of life. I also learned that you have to inspire large groups of people to make big things happen. It was tough back then—HIV/AIDS was a death sentence, malaria was hard to treat—but it gives me motivation to see how far we've come. Novartis is now ranked number one in the Access to Medicines Index, and we're having a massive impact on malaria and access to medicines. That has carried through my career.

The artificial intelligence boom is a big question, and we're investing in these technologies, but it's still early days. I just came back from Davos, where AI in drug discovery was a big topic, with many big statements made. The tricky thing is that AI can accelerate many steps in drug development, but we won't really know if it designs better medicines until five to seven years from now, because every AI-generated medicine has to go through animal testing, preclinical testing, and clinical testing in humans. There's also the issue that we only understand about two to four percent of what's going on in the human body, and AI projects based on what we know; the 96% we don't know could all be wrong. Even tools like AlphaFold rely on synthetic data. So we need to approach with humility. I hope AI doubles the success rate and takes two to four years off timelines, but we should have reasonable expectations and learn as we go.

In our core process of generating productivity, AI is certainly having an impact. Aside from fundamental drug discovery, it can move a medicine faster into the clinic, run clinical trials more quickly, identify physicians for recruitment faster, and generate documents more efficiently. That impact is probably six months to a year on overall timelines, and it may compound over time. We're trying to embed AI in every step of the R&D process to accelerate drugs through our pipeline.

We've put in place a clear AI code of ethics within the company, so we're clear on how we will and won't use AI, ensuring patient data is protected and bias doesn't propagate. I think we're doing the right things, but we have to keep looking at it as systems get more powerful. It will also be interesting to see how regulators like the FDA think about AI-generated data. But I feel good that we have good guardrails in place.

Isomorphic Labs is an extraordinary organization. Their founder Demis Hassabis won the Nobel Prize, a testament to their impact. Isomorphic Labs is an outgrowth of Google DeepMind, and with AlphaFold they were the first to predict the structure of all 20,000 known human proteins. That opens up possibilities to use AI to find drugs that modulate proteins. We picked three protein targets that are hard to drug with traditional approaches and asked if AI can solve it. We're seeing positive trends and hope to expand the collaboration. If we can start drugging the 'undruggable' targets, we open whole new areas of medicine. But it will be a few years until we know if it works in animals and humans.

It could be substantial. Currently, from an early drug idea to getting into humans can take three to five years. With AI, we might shorten that by half. Over time, we hope it increases the probability of success, reducing failures, which is also a huge gain.