Nautilus Biotechnolgy, Inc. Common Stock 2026 Q3 Earnings Call

Thanks, Marissa, and thank you all for joining us today. Last quarter, I walked you through a deliberate decision to concentrate our resources on proteoforms. This quarter, I want to talk to you about what proteoforms are and why we believe they're one of the most important opportunities in life sciences and how we're building the foundation for meaningful long-term growth. Let me start with the idea that guides everything we do. We are making functional biology measurable. For decades, proteomics has primarily been about telling us how much protein is present. While the quantity of a protein is meaningful information, it's not the whole story. Most of today's tools don't have the resolution and scale required to measure the myriad of function-driving modifications that vary across every protein molecule. Shotgun mass spectrometry breaks proteins into peptides and infers the proteins from those pieces.

Most affinity assays estimate average abundances by measuring a singular epitope on a protein. Either way, functionally distinct protein molecules are undifferentiated as the information about splice variants, truncations, and which modifications co-occur on the same molecule is lost. That matters because the complexity of human biology does not come from the number of genes we carry. Consider the banana. A banana has almost 40,000 protein coding genes. As humans, we are clearly more complex than a banana. In the early days of the Human Genome Project, scientists expected humans to have well over 100,000 genes. In fact, we only have about 20,000, half as many as a banana. The reality is that our complexity comes from the functional biology layer above the genome, and particularly from the many forms each protein can take.

Those genes are merely the instruction manual for building the machinery of life. A single gene can give rise to many distinct proteoforms, shaped by splicing, sequence variation, and by many modifications that occur after a protein is made. Each molecule is a discrete functional unit, and its form determines what it does, how it signals, who it interacts with, and where it sits spatially in the cell. The specific form a protein takes often determines what it actually does, where it goes, what it binds to, and what pathway it turns on, and which phenotypes it ultimately drives. AKT1 is a good example. It is a central switch in cancer signaling, and whether that switch is on depends on which sites are phosphorylated together on the same molecule. Two cells can carry the same amount of AKT1 and be in completely different signaling states. Tau is another example. In Alzheimer's disease, what matters is not just how much tau is present, but which modified forms accumulate.

Since proteoforms were never previously measurable at scale, it was never possible to use this critical information to understand health and disease. Proteoform resolution provides valuable, novel information for drug development and precision medicine, and it has not been accessible at scale until today. We built Iterative Mapping, the method that powers our Voyager platform, to change that. It reads millions to billions of single intact protein molecules across a massively parallel array, so it can resolve each molecule's functional state rather than breaking it apart or averaging them away. When you can see that layer of molecular resolution, you're looking at an entirely new kind of biological insight.

We believe no one else can deliver the resolution and scale our platform can. That is what we mean by making functional biology measurable, and I believe the value of this approach will become increasingly clear as our data reaches more scientists. We believe that proteoform resolution at scale will be a critical foundation for AI in biomedicine. The industry is starting to build large data sets to train the next generation of AI models, and those models are only as good as the data they learn from. We believe highly reproducible proteoform data is exactly the differentiated data they will need to learn functional biology. To accelerate our proteoform roadmap, we effectively tripled the number of people and resources focused on proteoform development, and we set a goal of developing 20 proteoform assays by the middle of 2028.

That plan is now in motion, and two things give me confidence in where we're heading. The first is the pull we're seeing from the scientific community. Second is the evidence that we can execute against our roadmap. Let me take each in turn. On the pull from the community, at this year's HUPO World Congress, the leading global proteomics meeting hosted by the Human Proteome Organization, proteoforms were such a pervasive theme that one article covering the conference in Chemical & Engineering News ran under the headline, "Proteins Don't Exist: Scenes from HUPO 2026." We also hosted a panel with three leading researchers whose labs develop and apply some of the field's most advanced protein analysis technologies.

Birgit Schilling of the Buck Institute for Research on Aging, Jochen Schwenk of KTH Royal Institute of Technology, and Bernd Wollscheid of ETH Zurich to ask, "Will proteoforms change our understanding of biology?" I want to share some of their perspectives. Dr. Schilling put it simply, "My thought is proteoforms are critical everywhere. These proteoforms influence function. They influence really everything that makes a cell work." Dr. Wollscheid also noted, "We operate under the assumption, like everybody else, there are no proteins, only proteoforms. And I think we have an opportunity to not only decode individual proteoforms, but also individual proteoform-specific complexes which have functional relevance." When we ask what customers need, they consistently want more proteoforms, more sample types, and more applications. A great example came when we announced our partnership in the SEEDS Consortium.

SEEDS is funded up to $51.7 million by the Advanced Research Projects Agency for Health, or ARPA-H. We expect to receive up to $6 million over the four-year program to apply our Voyager platform to tau in Alzheimer's disease and EWS-FLI1 in Ewing sarcoma, our first work in pediatric cancer. The data is expected to help train AI models on disordered proteins, which today's models handle poorly. Our work with both SEEDS and the Michael J. Fox Foundation are early examples of a broader opportunity, large-scale studies coupled with custom development. As we work more deeply with pharma, we expect partners to want to build assays for their own targets beyond what's on our price list. We're ready to build those assays and to help apply them across large cohorts. Turning to the second reason for my confidence in our plan, our track record of early execution.

At HUPO, we shared first data from three of our oncology proteoform assays targeting AKT1, EGFR, and P53. To put the results in context, first, let's go back to tau. We believe it delivers some of the most reproducible, detailed proteomics data in the world, both internally and with our partners at the Buck Institute, the Voyager platform delivered a median coefficient of variation, or CV, of 5.5%. That reproducibility is what makes our data trustworthy and scalable and makes it possible to build AI on top of it. After about a year of development, AKT1, EGFR, and P53 are already performing in line with tau, our most mature assay, which took roughly five years to build. Median CVs were 8.8% for AKT1, 3.6% for EGFR, and 6.6% for P53, all well below our 10% target and far better than industry norms closer to 25%. That is a remarkable result.

It tells us our core capability carries over from one target to the next, and it gives us confidence in the development of all three assays and the ones behind them. Stepping back, we have accomplished a great deal this year. We launched our Early Access Program and brought in our first Early Access customers, including Baylor College of Medicine. The Michael J. Fox Foundation is funding the development of an alpha-synuclein proteoform assay for Parkinson's disease, and that work is well underway. We recognized our first revenue from both the Early Access Program and the MJFF grant. In Q3, our tau proteoform work was published in *Nature Methods*, a peer-reviewed validation of the Iterative Mapping method, and accompanied by a News & Views article in the same issue.

In the News & Views commentary, Albert Heck of Utrecht University wrote that Iterative Mapping might be a valuable platform that could even be used clinically to assay tau proteoform landscapes, supporting diagnosis and stratification of disease subtypes, and potentially even assessing the effect of therapy. It can be adapted to measure the proteoform landscapes of other proteins. In a small cohort analyzed in the paper, tau proteoform analysis distinguished people with Alzheimer's disease from cognitively normal individuals, and the clearest signal came from proteoforms with as many as four phosphorylations on the same molecule, information a single site measurement cannot capture. Rounding out our accomplishments year to date, we moved three oncology assays into development and shared their first data. We were selected to participate in SEEDS, and we remain on track to place our first beta unit this year.