I see it all the time. A founder comes to me with a brilliant idea for an AI diagnostic tool. They’ve got a killer algorithm, a slick pitch deck, and a team of PhDs. They think they’re going to change the world. And maybe they will. But they’re missing one huge, glaring problem that will kill their startup before it even gets off the ground.
Nobody talks about the data moat problem.
Most founders think building a great AI model is enough. They're wrong. Here's the uncomfortable truth about what it really takes to succeed in the brutal, regulated world of healthcare AI.
For the last ten years, I’ve been in the trenches of Silicon Valley. I’ve built and sold two companies, and I’ve invested in over 200 startups, including some of the biggest names in AI like Anthropic, OpenAI, and Scale AI. I’ve seen what works and what doesn’t. And I can tell you that in healthcare, the model is the easy part. The hard part is the data.
The Data Moat Illusion
Everyone in AI talks about building a “data moat”—an exclusive dataset that gives you a competitive advantage. In theory, it’s a great idea. In practice, especially in healthcare, it’s a fantasy. The data you need is locked away in hospitals, fragmented across a dozen different EMR systems, and protected by a fortress of regulations like HIPAA.
I remember a few years ago, I invested in a startup that was building an AI to detect early signs of Alzheimer's from brain scans. The team was brilliant. Their model was incredibly accurate in the lab. They were sure they were sitting on a billion-dollar company. But when they tried to get real-world data to train and validate their model, they hit a brick wall. Hospitals wouldn’t share their data. The legal and administrative hurdles were insane. The company burned through its seed funding in less than a year and quietly died.
It was a painful lesson. But it taught me something important: in healthcare AI, you don’t win by having the best model. You win by having the best data acquisition strategy.
The Counterintuitive Path to Success
So how do you solve the data moat problem? It’s not about trying to build a bigger, better moat. It’s about finding a different way to cross the river.
Here are a few of the strategies I’ve seen work:
Go niche. Instead of trying to solve a huge, sexy problem like cancer detection, focus on a smaller, more specific clinical need. I know a startup that’s building an AI to optimize scheduling for operating rooms. It’s not as glamorous as curing cancer, but it’s a real problem that hospitals are willing to pay to solve. And because the data they need is more contained, they’ve been able to build a real business.
Befriend the clinicians. Don’t just talk to the hospital administrators. Talk to the doctors and nurses who will actually be using your tool. They’re the ones who understand the clinical workflow. They’re the ones who can champion your product internally. And they’re the ones who can help you get the data you need.
Get creative with data. Don’t just think about getting data from hospitals. Think about partnerships with medical device companies, patient advocacy groups, or even other startups. I’ve seen companies have success with synthetic data, but you have to be careful. The FDA is still figuring out how to regulate it.
Embrace the regulations. Don’t think of HIPAA and the FDA as obstacles. Think of them as a roadmap. If you can build a product that meets their standards, you’ll have a huge advantage over everyone else. It’s a high bar, but it’s also a powerful differentiator.
A Story of Success
It’s not all doom and gloom. I’ve also seen companies that have gotten it right. One of my portfolio companies, which I won't name for confidentiality, is in the AI radiology space. They didn't start by building a model. They started by spending a year in hospitals, just watching radiologists work. They learned their pain points, their frustrations, and their needs.
Then, they built a tool that wasn't just an AI model, but a complete workflow solution. It integrated with the existing EMR system. It had a simple, intuitive interface. And it solved a real problem for the radiologists, saving them hours of tedious work each day.
Because they had built that trust and provided real value, the hospitals were willing to work with them on data. They were able to get the data they needed to train their model and get FDA clearance. Today, they’re one of the fastest-growing companies in the healthcare AI space.
The Takeaway
The AI revolution in healthcare is real. It’s going to change everything. But the companies that succeed won’t be the ones with the fanciest algorithms. They’ll be the ones that are relentless, creative, and humble enough to solve the data problem.
So if you’re a founder in this space, my advice is this: stop obsessing over your model. Start obsessing over your data. It’s the only thing that matters.
Frequently Asked Questions
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.