AI has reshaped entire industries, from software development to entertainment, yet its impact in healthcare hasn’t followed the same explosive trajectory. Despite years of hype around AI-powered drug discovery, diagnostics, and clinical tools, the returns have often disappointed, leaving investors and operators wondering: What’s really holding biotech back?

In this clip from our episode with AI biopharma leader Werner Lanthaler, we revisit the state of AI in healthcare three years after our first deep dive. What’s changed? What hasn’t? And what needs to happen for AI to finally deliver on its promise in biotech?

The answer, according to Werner, comes down to two things: data quality and business models.

Here it is in video format if you prefer to watch:

🚀 AI Is Transforming Other Industries. So Why Not Healthcare?

Over the past few years, AI has fueled massive gains in social media, coding, and creative industries. These sectors may have faced turbulence, but one thing is undeniable: money is flowing in. The impact is visible, immediate, and commercial.

But in healthcare? Not so much.

Despite being the most data-producing industry on the planet, biotech has not experienced the same surge in AI-driven investment, especially when it comes to AI drug discovery and AI-enabled diagnostics.

Why?

According to Werner, the industry fell into a familiar trap: overpromising and underdelivering.

“We promised we would be so much faster and so much more efficient in the clinic, and then we still failed in the clinic with AI-predicted compounds.”

The challenge wasn’t the algorithms. It was the biology, still notoriously complex, and the datasets, often fragmented, uncurated, or simply not large or consistent enough to produce reproducible results.

🔄 The Shift: From Hype to Practical, Data-Driven AI in Healthcare

The good news? We may finally be entering the next phase.

A new generation of AI-enabled biotech and diagnostics companies is emerging, one that focuses on:

• Properly curated datasets
• Clear applications with measurable outcomes
• Business models that actually capture value

As Werner puts it:

“The first generation overpromised. The second generation is coming with proper databases, proper application of AI, proper curated data, and finally with business models that make sense.”

This shift is critical because even the best algorithm is useless without a clear answer to a simple question:

How does this create value?

For years, many AI-biotech companies struggled here. They had strong technical assets but unclear revenue logic. That’s now changing.

🧬 Why Shared Molecular Patient Databases Could Be a Breakthrough

One of the most compelling ideas discussed in the episode is the creation of shared molecular patient databases, datasets that are not locked inside single companies but structured to be used by multiple drug developers.

Why does this matter?

Because no single biotech company will ever work on 100+ targets at once, this means siloed data leads to underutilisation and poor returns.

A collaborative, multi-tenant approach could unlock:

• Higher data productivity
• Better target validation
• Faster candidate selection
• Better economics for everyone involved
  

Several companies (Flatiron, Owkin, Tempus, and others) have taken steps in this direction, but the industry still lacks a scalable, widely adopted model.

“If only one company uses a patient database, the returns will never be good enough. But when multiple companies can build on the same curated datasets, productivity becomes absolutely doable.”

This is the kind of infrastructure shift that transformed other industries, and could finally unlock AI’s full potential in biotech.

💡 The Real Opportunity: AI + Healthcare at Scale

Healthcare is not entertainment. It’s not social media. It’s not e-commerce.

It’s more complex, but also more meaningful.

We are only scratching the surface of what’s possible when we apply AI to:

• Clinical imaging
• Diagnostics
• Surgical assistance
• Precision medicine
• Drug discovery pipelines
• Disease-specific molecular data
• Longitudinal patient records

“We’re not even at 1% of what AI can do in healthcare.”

But to tap into this potential, AI in biotech must evolve beyond clever algorithms and become operational, scalable, and commercially viable.

That means:

• Better data
• Better business models
• Better integration into clinical and development workflows

Once these pieces come together, a new era of AI-driven drug development and diagnostics becomes possible.

⭐ Why Now Is the Time to Build in Biotech AI

If the first wave of AI companies in healthcare was defined by bold ambitions and mixed results, the second wave will be defined by:

• Pragmatism
• High-quality data
• Clear paths to value creation
• Shared infrastructure
• Clinical relevance

For founders, investors, and operators, this moment feels like a reset and an opportunity.

The question now is not whether AI will change healthcare.
It’s how quickly, at what scale, and who will build the foundational models, datasets, and businesses that drive that change.