Know How to Fold 'Em

Why AlphaFold Changed More Than Protein Structure

Every generation of biotechnology has its defining platform.

Recombinant DNA.

PCR.

Monoclonal antibodies.

Next-generation sequencing.

CRISPR.

Artificial intelligence.

Most technologies improve existing workflows.

A few fundamentally change how science is performed.

AlphaFold belongs in the second category.

When DeepMind announced that its artificial intelligence could predict protein structures with extraordinary accuracy, many people understandably focused on the immediate scientific achievement. Protein folding had challenged researchers for more than half a century. Suddenly, an AI system could solve many of those problems in hours instead of months or years.

That breakthrough deserved every headline it received.

But I believe the larger story wasn't about protein folding.

It was about decision making.

Every Scientific Decision Has a Cost

Drug discovery has always been a game of probabilities.

Which target should we pursue?

Which protein matters most?

Which experiment should we run next?

Which molecule deserves another round of optimization?

Which project should receive another $10 million?

Every one of those decisions carries enormous financial consequences.

Historically, uncertainty has been one of the largest expenses in pharmaceutical R&D.

AlphaFold demonstrated that artificial intelligence could meaningfully reduce one important source of that uncertainty.

Not by replacing scientists.

By helping scientists ask better questions sooner.

The Best Platforms Don't Replace Scientists

One misconception surrounding AI is that it replaces expertise.

The opposite is proving true.

The most valuable AI platforms amplify human judgment.

AlphaFold didn't eliminate structural biologists.

It gave them a dramatically better starting point.

Medicinal chemists still design molecules.

Biologists still validate mechanisms.

Clinical scientists still determine whether therapies improve patients' lives.

AI simply allows each discipline to begin from a more informed position.

That distinction is critical.

Platform Technologies Create Multipliers

The true value of AlphaFold isn't that it predicts protein structures.

Its value comes from everything that prediction enables.

Target identification.

Drug design.

Protein engineering.

Antibody discovery.

Enzyme optimization.

Synthetic biology.

Even entirely new therapeutic modalities.

The platform creates value repeatedly.

Every new discovery builds upon the last.

That is what separates transformational platforms from useful tools.

The Real Revolution Is Learning Faster

Drug discovery has never suffered from a shortage of ideas.

It has suffered from slow learning.

Every experiment teaches us something.

Unfortunately, experiments are expensive.

Time-consuming.

Resource-intensive.

Artificial intelligence changes that equation.

Platforms like AlphaFold compress learning cycles.

Scientists spend less time determining what is possible and more time determining what actually works.

That distinction has profound implications for every therapeutic area.

Discovery Science Meets External Innovation

Platform technologies also change how companies partner.

Historically, pharmaceutical companies licensed molecules.

Increasingly, they license capabilities.

The most attractive platforms are not simply producing one therapeutic candidate.

They are creating repeatable systems that generate better therapeutic candidates over time.

That changes the conversation from:

"Can we license this asset?"

to

"Can this platform improve our entire discovery organization?"

Those are fundamentally different investment decisions.

The AI Era Raises a New Question

As AI becomes increasingly embedded throughout biotechnology, another challenge emerges.

Trust.

Every prediction generated by an algorithm influences future scientific decisions.

Those decisions consume capital.

Guide experiments.

Shape portfolios.

Ultimately affect patients.

Confidence in AI cannot come from elegant demonstrations alone.

It must be earned through evidence.

Transparent methodologies.

Independent validation.

Reproducible results.

Experimental confirmation.

The strongest AI platforms will not be those making the boldest claims.

They will be those consistently demonstrating that scientists make better decisions because the platform exists.

Commercialization Begins Earlier Than We Think

One lesson I have learned across business development, licensing, and commercialization is that platform success rarely depends solely on scientific performance.

It depends on adoption.

Can researchers integrate it naturally into existing workflows?

Can pharmaceutical companies validate the outputs?

Can leadership justify enterprise deployment?

Can regulators understand how it contributes to decision making?

Can business development teams confidently explain the value proposition?

Commercialization does not begin after discovery.

It begins the moment a platform enters another scientist's workflow.

Knowing How to Fold 'Em

The title of this article is, of course, a nod to Kenny Rogers' famous advice about knowing when to hold 'em and knowing when to fold 'em.

Drug discovery requires a similar kind of judgment.

Every program eventually faces difficult decisions.

Advance.

Pause.

Partner.

Pivot.

Terminate.

Artificial intelligence won't make those decisions for us.

But it may dramatically improve the information we use to make them.

That may prove to be AlphaFold's greatest legacy.

Not that it predicted protein structures.

But that it demonstrated how intelligent platforms can fundamentally improve scientific decision making.

In biotechnology, the next decade won't belong to companies with the largest datasets or the fastest algorithms.

It will belong to those that know how to fold science, artificial intelligence, human expertise, and strategic partnerships into platforms that consistently reduce uncertainty and accelerate innovation.

Those are the platforms that will shape the future of medicine.