The Velocity Revolution

In 1995, I was trying to solve a transportation problem.

Not involving cars. Or airplanes.

But something much smaller.

I was trying to understand how a protein gets into the nucleus of a human cell.

I was at Boston College, working on my doctoral research, studying human papillomavirus. HPV.

More specifically, HPV16, one of the types responsible for causing cancer.

My research focused on a tiny viral protein called E7.

Now, E7 is a particularly nasty little protein.

It can interfere with the molecular machinery that normally prevents cells from dividing uncontrollably.

And I was fascinated by a simple question.

How does this protein get into the nucleus of a human cell, where it can interfere with some of those critical processes?

You see, cells don't just let everything into their nuclei.

There are elaborate systems that control what gets in and what stays out.

And my colleagues and I discovered that E7 could enter through an unconventional pathway.

It was using a different route from the ones we initially expected.

Now, imagine spending years of your life trying to understand how one tiny protein moves inside a cell.

That's what science is often like.

Years of work. Countless experiments. Small discoveries that help us understand one little piece of an extraordinarily complicated puzzle.

At the time, there was no licensed vaccine against HPV.

Then, in 2006, the first HPV vaccine was approved in the United States.

And today, twenty years later, we have vaccines that can prevent infections responsible for many HPV-related cancers.

We've seen extraordinary reductions in infections with cancer-causing HPV types and in precancerous lesions.

I wasn't involved in developing those vaccines. They were based on a different viral protein.

But having spent years studying this virus, I find it remarkable to see what an entire scientific field has accomplished.

Because that's why we do science.

Not simply to understand molecules.

But to improve human lives.

And yet, after spending nearly thirty years working across biotechnology and drug development, there's something that continues to trouble me.

For every successful story like HPV, there are countless promising discoveries that never become meaningful treatments.

Why?

Why, with all our extraordinary scientific capabilities, is it still so difficult to turn discoveries into medicines?

Well, I have a theory.

And it starts with something that probably doesn't belong at a scientific conference.

A lottery ticket.

[Hold up a lottery ticket.]

Now, if I buy one lottery ticket, my chances of winning the jackpot are incredibly small.

If I buy ten, my chances improve.

A thousand? Better still.

And if I could buy millions, I'd certainly increase my chances.

Of course, drug development isn't literally a lottery.

Scientists don't randomly throw molecules at patients and hope something works.

There's enormous expertise, evidence, and reasoning behind every development program.

But in some ways, we've built an extraordinarily sophisticated version of a lottery.

We generate hypotheses.

We identify targets.

We screen compounds.

We develop molecules.

We test them.

And most fail.

So what do we do?

We generate more hypotheses.

Screen more compounds.

Develop more molecules.

We buy more lottery tickets.

And sometimes, we win spectacularly.

Antibiotics. Vaccines. Immunotherapies.

Medicines that have transformed millions of lives.

But there's a problem.

The tickets are getting extraordinarily expensive.

In 2012, a group of researchers described a phenomenon called Eroom's Law.

You may have heard of Moore's Law, which describes the remarkable progress in computing.

Well, Eroom is Moore spelled backward.

And it describes a rather uncomfortable historical trend in pharmaceutical research.

The researchers estimated that between 1950 and 2010, the number of new drugs approved per inflation-adjusted billion dollars spent on research and development had declined approximately eightyfold.

Eightyfold.

And think about everything we invented during those sixty years.

Genomics. Automation. High-throughput screening. Advanced imaging. Computational biology.

And now artificial intelligence.

We have become extraordinarily good at doing things faster.

We can generate more data, analyze more compounds, and explore more biological possibilities than at any point in human history.

Yet turning scientific discoveries into successful medicines remains incredibly difficult.

So what are we missing?

I think part of the answer is something we all learned in school.

The difference between speed and velocity.

In physics, speed tells us how fast something moves.

Velocity tells us how fast it moves and in which direction.

Imagine I'm standing here in Boston, and I want to get to New York.

You put me in a Ferrari.

I'm traveling at 150 miles per hour.

Very impressive.

But there's just one little problem.

I'm heading north.

Toward Maine.

I'm going incredibly fast.

But I'm getting farther from my destination every second.

And if I accelerate, I just make the problem worse.

[Pause.]

Now, look at the questions we ask in biotechnology today.

How quickly can we identify a biological target?

How quickly can we screen a library of compounds?

How quickly can AI design a new molecule?

How quickly can we enter clinical trials?

These are useful questions.

But perhaps we're not asking the most important one.

How do we know we're moving in the right direction?

Because doing the wrong experiment twice as fast doesn't make it a better experiment.

And advancing the wrong drug into clinical trials six months earlier isn't necessarily progress.

It may simply mean reaching an expensive failure sooner.

Throughout my career, I've become increasingly interested in a particular problem.

The difference between what works in a laboratory and what actually works in a human being.

You can have a beautiful scientific hypothesis.

Elegant experiments.

Convincing data.

And then you test that hypothesis in people.

And human biology gives you a completely different answer.

The drug doesn't work.

Or it produces unexpected toxicity.

Or it works, but only in a small group of patients.

And suddenly, years of effort have to be reconsidered.

Now, I don't think failure is the enemy here.

Failure is fundamental to science.

It's how we learn.

The real problem is when we fail too late, at enormous expense, without learning enough to understand why.

Imagine two drug development programs.

Both are ultimately going to fail.

In one, we discover the problem after eight years, having committed enormous resources.

In the other, we recognize a flawed biological assumption after eight months.

And we understand why.

Both programs failed.

But the second gave us something incredibly valuable.

The opportunity to change direction.

And that, to me, is the beginning of the Velocity Revolution.

Not eliminating uncertainty.

Not eliminating failure.

But becoming much better at learning, deciding, and changing course.

So how do we do that?

Well, here's where I'm genuinely excited about the future.

Because several technologies are beginning to converge in ways that could fundamentally change how we develop medicines.

One is a family of technologies with a rather uninspiring name.

New Approach Methodologies.

Or NAMs.

These include living human tissues, organoids, organs-on-chips, and advanced computational models.

The idea is to generate evidence that can help us better understand and predict what may happen in humans.

Now, none of these technologies perfectly reproduces the complexity of an actual person.

But they can allow us to ask more relevant questions, earlier.

And that's enormously important.

In 2026, the FDA issued draft guidance on the validation of these methods for drug development.

That doesn't mean every new model is automatically reliable or ready to replace existing methods.

It means the scientific and regulatory conversation is changing.

Because if our ultimate destination is a human being, shouldn't our evidence increasingly reflect human biology?

Think of these technologies as helping us create a better map.

But a better map isn't enough.

We also need instruments that tell us where we are.

And whether we're moving in the direction we intended.

That's where biomarkers become particularly interesting.

And especially proteomics.

Most people have heard about genomics.

Your genome tells us an extraordinary amount about your biological potential.

But proteins are among the molecules actually doing the work.

They communicate.

They regulate.

They respond to disease.

They respond to treatments.

They change as we age.

And today, we can measure thousands of them simultaneously.

In 2024, researchers studied more than 45,000 people and developed a measure of biological aging based on 204 proteins.

That measure was associated with the risk of several chronic diseases and mortality.

Think about what that means.

We are beginning to identify biological patterns that may help us understand not simply someone's current health, but aspects of where their health may be heading.

Of course, a biomarker isn't automatically proof that a treatment works.

And making a biological-age score look younger doesn't necessarily mean you've made a person healthier.

We need to validate those relationships.

But the opportunity is extraordinary.

We may increasingly be able to measure biological trajectories instead of waiting only for distant endpoints.

And nowhere is that more important than in longevity.

Imagine I come to you with an intervention that I believe could extend your healthy life by fifteen years.

Wonderful.

How do we test it?

Do we wait thirty years?

Forty?

Fifty?

We can't afford to wait decades for every answer.

We need ways to investigate whether we're changing the biological processes that matter, and whether those changes actually predict better health.

Longevity makes the challenge impossible to ignore.

We need better ways to understand where biology is going.

Not just how quickly we can conduct another experiment.

And then there's artificial intelligence.

AI may be the most powerful accelerator biomedical research has ever seen.

It can analyze enormous datasets, identify patterns, help design molecules, and generate new hypotheses.

But I think we sometimes misunderstand what makes AI valuable.

If we use it simply to generate ten times as many molecules, we may end up buying ten times as many lottery tickets.

And that brings us right back to where we started.

AI can be an extraordinary engine.

But an engine is not a compass.

The opportunity is to combine AI with better human-relevant models, meaningful biomarkers, and clinical evidence.

So we don't simply generate more possibilities.

We become better at identifying which possibilities deserve to move forward.

And which ones don't.

Now imagine what drug development could look like if we got this right.

Instead of waiting years to discover that a biological assumption was wrong, we might recognize the problem much earlier.

Instead of pursuing every promising result, we'd become better at distinguishing a promising experiment from a promising medicine.

Instead of simply measuring how much work we accomplish, we'd measure how much useful uncertainty we remove.

And when the evidence tells us we're heading in the wrong direction, we'd change course.

Not reluctantly.

Not as an admission of defeat.

But because that's exactly what a good scientific decision looks like.

That's the transition I want us to imagine.

From lottery to navigation.

And I want to be clear.

This isn't about replacing scientific curiosity with some rigid system that only pursues predictable outcomes.

Some of the greatest discoveries in medicine came from unexpected observations.

We must preserve that openness.

But once we decide to invest years of effort and enormous resources into bringing an intervention to patients, we owe it to those patients to use the best possible evidence to guide our decisions.

Because ultimately, that's who this is about.

Not molecules.

Not experimental platforms.

Not algorithms.

People.

People waiting for a diagnosis.

For a treatment.

For a little more time with the people they love.

And every year we spend pursuing the wrong direction is a year those people don't get back.

[Pause.]

I began this talk by telling you about a tiny viral protein.

Thirty years ago, I was trying to understand how that protein found its way into the nucleus of a human cell.

Today, I find myself asking a much bigger question.

How do we help extraordinary scientific discoveries find their way into human lives?

When I began studying HPV, there was no licensed vaccine.

Today, vaccines are helping prevent cancers caused by that virus.

That's what makes me optimistic about science.

We know what's possible.

But imagine how much more we could accomplish if we became better at recognizing the right directions earlier.

I've spent nearly thirty years watching us get better at moving faster.

I believe the next ten years can be about something fundamentally different.

Getting better at knowing where we're going.

We don't need to stop exploring.

We don't need to stop taking risks.

And we certainly don't need to stop making discoveries.

But we do need to stop confusing activity with progress.

Because speed alone isn't enough.

What matters is the direction in which that speed takes us.

And when we combine extraordinary scientific capability with better evidence, better decisions, and the courage to change course...

We can start turning speed into meaningful velocity.

That is the Velocity Revolution.

And I believe it has already begun.

Thank you.