David Baker's team at the University of Washington is using artificial intelligence to design functional proteins from scratch, a breakthrough that earned Baker the 2024 Nobel Prize in Chemistry. Yet despite this landmark achievement, significant hurdles remain before AI-designed proteins can become practical medicines. The gap between designing a protein that binds to a target and creating one that works safely in the human body reveals why Baker himself separates the hype from reality in this rapidly advancing field.

What Makes Protein Design So Difficult?

The numbers alone illustrate why protein design remained unsolved for decades. A small protein composed of just 100 amino acids can theoretically fold into 20 to the power of 100 possible sequences, yet only a vanishingly tiny fraction of those sequences actually fold into stable, functional structures. Misplacing a single amino acid residue by an angstrom, a unit of measurement roughly the width of a hydrogen atom, can mean the difference between a drug binding tightly to its target or failing completely.

For the antibody drug market, worth hundreds of billions of dollars, this precision matters enormously. Antibodies are proteins the immune system uses to recognize and neutralize threats, and they have become one of the most valuable classes of medicines for treating cancer and autoimmune diseases. Historically, designing the antibody loops, the flexible regions responsible for binding to disease targets, has been nearly impossible to do computationally.

How Is AI Changing Protein Design?

Last November, researchers in Baker's lab published a landmark study in Nature demonstrating that artificial intelligence could now design full-length de novo antibodies, meaning antibodies created entirely from scratch rather than modified from natural ones, that could bind to user-specified targets. This represented a genuine breakthrough: AI models could construct antibody loops with the precision needed for drug development.

The achievement reflects decades of foundational work. The field's early breakthroughs trace back to 1988, when researcher William DeGrado demonstrated that protein sequences not found in nature could achieve stable three-dimensional folds, challenging the long-held belief that only evolution could produce functional proteins. A decade later, Steve Mayo's team published the first computational protein design validated experimentally. Baker and then-postdoctoral researcher Brian Kuhlman expanded this further in 2003, creating entirely new protein folds that had never existed in nature.

Where Does the Hype End and Reality Begin?

When asked to separate hype from reality, Baker was direct about both the promise and the limitations.

"The reality is that we can now design proteins on a computer. The hype is that for therapeutics, there's a lot more than the basic activity of a protein binding or catalyzing a reaction. Whether de novo proteins will revolutionize medicine will require improving our understanding of the biology," Baker explained.

David Baker, Director of the Institute for Protein Design at University of Washington

This distinction matters enormously. Designing a protein that binds to a disease target is one challenge. Creating a protein that remains stable in the human body, avoids triggering unwanted immune responses, can be manufactured at scale, and actually improves patient outcomes is an entirely different problem. The gap between these two milestones has fueled industry debate over whether generating de novo medicines is even possible.

How Baker Built a Culture of Collaborative Innovation

Baker's approach to leadership offers insight into how breakthrough research emerges. His lab at the University of Washington Institute for Protein Design spans multiple floors and includes more than 100 researchers united by a shared mission. Rather than operating as isolated researchers, Baker describes his vision as a "communal brain," where diverse expertise converges to unlock new directions.

This collaborative culture shapes how the lab operates:

• Flat Hierarchy: Graduate students and postdoctoral researchers can speak up in meetings and question the work without fear, breeding a culture focused on what matters rather than rank or status.
• Personal Attention: Despite leading over 100 trainees, Baker knows each person's name, project, and expectations before meetings, maintaining individual connections at scale.
• Intentional Community Building: Weekly rituals like "chocolate hour" and social events at happy hours create informal spaces where researchers from different disciplines naturally collaborate and support one another.

Graduate student Seth Woodbury, who designs metallohydrolases for sustainability applications, noted that Baker excels at breaking down social barriers between researchers. "Once you talk to your colleagues at happy hour, it's not so scary to go ask them a question," Woodbury said. Fellow graduate student Woody Ahern added that this culture of questioning and interdisciplinary focus "breeds this culture of staying focused on what matters in an interdisciplinary way".

When Ria Sonigra, now an IPD graduate student designing programmable nanopores for molecular sensing, sent Baker a cold email from India with questions about the lab, he responded quickly and connected her with another international student to help navigate the application process. "People outside the lab may think that David can't pay attention to everyone, which is not true," Sonigra said. "He knows your project and what he expects of you before the next meeting, even if he has a hundred trainees".

What Does the Nobel Prize Recognition Mean for the Field?

Baker shared the 2024 Nobel Prize in Chemistry with Demis Hassabis and John Jumper from Google DeepMind, whose AI model AlphaFold solved the protein structure prediction problem and has become one of the most widely adopted computational tools for drug discovery. The Physics prize that same year went to Geoffrey Hinton and John Hopfield for foundational discoveries in machine learning with neural networks.

Together, these prizes marked a pivotal moment: artificial intelligence had moved beyond computer science into biology, chemistry, and physics, earning recognition as a breakthrough deemed to confer the "greatest benefit to humankind." Nearly 200 current and former members of Baker's lab gathered in Stockholm to celebrate, a testament to his scientific reach and mentorship.

Baker himself remained humble about the recognition. "My group was not the first to do protein design," he said. "The prize was given because protein design has so much promise now, and that reflects the work of the whole community." Today, early pioneers in the field, including William DeGrado, Steve Mayo, and Brian Kuhlman, continue advancing structural biology as prominent faculty members at UCSF, Caltech, and UNC Chapel Hill respectively.

How Is This Research Translating Into Real-World Applications?

The commercial potential of AI-designed proteins has attracted significant investment. Nathaniel Bennett, a former postdoctoral researcher in Baker's lab, co-founded Xaira Therapeutics in 2024 with over 1 billion dollars in total funding. The AI-focused biotech company aims to develop de novo protein medicines and counts Baker as a scientific advisor. The leadership team includes Marc Tessier-Lavigne, former president of Stanford and Chief Scientific Officer of Genentech, as CEO, along with Nobel laureate Carolyn Bertozzi, former FDA head Scott Gottlieb, and former Johnson and Johnson CEO Alex Gorsky on the board of directors.

Xaira represents one of many biotech companies Baker has co-founded over three decades. When asked about this prolific track record, Baker reflected on what matters most: "Science all becomes obsolete quickly because the field's moving! The people that you mentor are more important than any science you do. They all go on and do great things".

Baker

The challenge ahead remains substantial. While AI can now design proteins that bind to targets with precision, translating those designs into medicines requires solving problems in manufacturing, stability, immunogenicity, and efficacy that go well beyond computational design. Baker's candid assessment suggests that the real revolution in medicine will come not from the hype of AI-designed proteins, but from the hard biological work of understanding how to make those designs work in living systems.