Ultima Genomics is a biotechnology company that develops high-throughput DNA sequencing platforms designed to make genome sequencing more affordable and scalable.
What has changed in sequencing technology broadly speaking, and concretely at Ultima Genomics in the past few years?
Three years ago, when we were still in stealth, our premise was that the same dynamic that has driven information technology for decades would play out in biology: you need more information, you need more data. When you can produce data more cost-effectively, people do not simply spend less for the same amount. They want more, creating what we call hyper-elastic growth, where each time cost drops by two, demand in gigabases rises by more than two. The goal was to bring a Moore’s law-like dynamic to biology by scaling the reading of DNA, biology’s source code.
Since then, that premise is no longer really disputed, and AI has been a massive driver of demand. At Ultima, we launched our platform two years ago, made a major upgrade a year ago, and by the time this is published we will already have launched our second-generation platform: half the footprint, half the runtime, and twice the output of our current system. Since we last spoke, we increased output substantially and we are doubling again, because we see the hyper-elastic effect in practice.
What are the major megatrends driving demand for sequencing right now?
First is the shift from “just the genome” to “genome plus.” A blood draw or sample contains more information than inherited variants: epigenetics, how the genome is methylated and folded, and proteins. The UK Biobank is a good example: they already had genomes, and about a year and a half ago they decided to add proteins. Even with proteins, the final readout can still be done through sequencing, because proteins are tagged with DNA.
Second is oncology, and especially MRD—minimum or molecular residual disease. What has really taken off is using blood-based sequencing to ask whether cancer is responding, whether it is gone, or whether it is coming back. That is different from sequencing a tumour once for therapy selection; after the tumour is removed, you are searching for rare signals in a large background, which requires much more sequencing and high accuracy.
Third is the rise of AI models trained on large-scale cellular data to predict how cells behave under different conditions.
The idea is to measure enough cells that have been “perturbed” in different ways, read out their state—classically via single-cell RNA-seq—and use that data to train models that can predict how a cell will respond to drugs, CRISPR knockouts, temperature, stress, and other inputs.
Illustrative are initiatives like the Chan Zuckerberg-backed “one billion cell” effort, where standardised single-cell measurements are generated at massive scale and sequenced on Ultima’s platform, alongside other large single-cell projects using Ultima sequencing to help train AI-driven discovery models.
How do you view the shift from a “static” genome to a “dynamic” genome?
People used to think: sequence your inherited genome once and you are done. But what is becoming clearer is that there is a dynamic component—cells divide and accumulate changes. Cancer is a clear example, but immune cells also change as they respond. More broadly, disease can have signatures in DNA, in gene expression, and in proteins.
So the trajectory is toward sequencing more regularly to identify disease and guide decisions, not only sequencing once for inherited information. In that sense, the field is moving from a static view to a more dynamic view of what sequencing can capture over time.
Why are depth and accuracy so critical in MRD, and how does Ultima address that?
In MRD, tumour DNA is mixed into a lot of normal DNA in the blood, so you need much deeper sequencing to find the signal. You also need accuracy because you cannot confuse a sequencing error for a true biological finding when the thing you are looking for is extremely rare.
We launched ppmSeq—parts-per-million sequencing. Because of the way we flow our chemistry one base at a time, for us to make a base-substitution error we would have to make at least two mistakes, and the probability of that is very low. That gets you to an error rate below one in a million, which is highly impactful for earlier, more confident MRD detection.
MRD has been considered difficult to use routinely. How wide is adoption now, and what obstacles remain?
In the U.S., adoption is growing rapidly - faster than pretty much any other molecular test in recent history—so it is being widely adopted and expanding.
The first barrier was reimbursement: you have to show value, not just clinical interest. More and more data show MRD can help guide treatment decisions, including right after surgery, and reimbursement is expanding as evidence grows. For global adoption, the key obstacle is cost: tests can be many thousands of dollars, and MRD is sequencing-hungry, so reducing sequencing cost is part of what we need to keep driving, alongside clinical partners making testing more accessible.
Taking a step back, how does Ultima Genomics differentiate itself from competitors such as Illumina?
We differentiate through scalable architecture and pace of improvement, and through an open, best-of-breed partnering approach. Some competitors aim for a one-stop-shop model. We focus on tight integration with partners so customers can choose the best components—protein platforms, single-cell platforms, analysis tools—while still getting a seamless experience.
What types of partnerships are you prioritising now?
We prioritise partnerships that make adoption easier for customers and partnerships that add more information. On the “genome plus” side, that includes proteins, epigenetics, and single-cell, so a customer can build a richer biological picture with integrated workflows.
The other priority is interpretation. Sequencing is becoming less of a constraint as throughput rises, but turning sequencing into insights requires connecting data to medical records and other information sources at scale, and extracting understanding with analytics and AI. In my view, the interpretation layer is the biggest gap right now.
Can you tell us more about the free sequencing programme you ran last year?
The main programme was called “Count on Us,” a wordplay because many sequencing applications are fundamentally counting applications—RNA-seq is counting, many proteomics workflows become counting, and those are a natural fit because they require a lot of sequencing.
It was inspired by the major industry conference at the end of February last year, when many scientists were distracted by budget uncertainty and funding confusion. We offered a defined amount of sequencing for free for academia to help labs through that period. It totalled trillions of reads, was extended due to strong demand, and it both supported researchers and introduced many groups to our technology.
How much closer to mass or global adoption do you think your second-generation platform brings you?
We already had some global growth with the first-generation system. We have instruments in Asia and in Europe, and in several places in both. With the second-generation platform, there is also more flexibility. It has an ultra-high-throughput version and a medium-throughput version, so yes, we think it will help drive placements across the world.
At the same time, global adoption takes more than us. We will make sequencing affordable and make the machine easy to install, smaller, and faster. But it also needs partners in oncology, like the companies developing MRD tests, and partners in data interpretation to make a genome valuable. We are seeing more and more partners coming in. Between machines that are easier to deploy and partners that are encouraged, I believe that if you do this again three years from now, the impact of the genome in regular healthcare will be much further along. We are 20-something years after the first human genome and it is still often edge cases—rare diseases, sick babies, people with tumors and cancers—so it is still early days. But the momentum is building.
Where do you see Ultima Genomics in three years’ time?
I think it is a years-and-decades journey. It is the same mission, just further along. It is like the semiconductor industry: it spends decades growing, and still, even last year, it was growing like crazy again. When were computers “good enough”—when I got my first PC, when I got my first mobile phone, or now when we are using AI? It keeps evolving in waves.
Genomics will go through waves too, and I expect that in three years we will be on another wave of adoption. The mission will be the same, but adoption and impact will be further along.
