10x CTO on Chan Zuckerberg, Arc Institute ties and industrializing single-cell biology

In many ways, the first half of 2025 hasn’t been kind to the biotech sector’s growth trajectory. Headwinds from constrained research funding and shifting market dynamics have led to strategic recalibrations for many, including instrument makers like 10x Genomics. But the company’s Chief Technology Officer Michael Schnall-Levin is optimistic about the bigger picture.

“I think on the fundamentals, it’s actually a pretty exciting time,” Schnall-Levin said. “There’s been obvious excitement around AI, particularly in biotech, and we’re really seeing that these AI methods are powerful. In almost every field using AI, we see the same rule: more data often means better results.”

That data-centric optimism is playing out in two major partnerships announced this year. In February, the Chan Zuckerberg Initiative launched its Billion Cells Project using 10x’s Chromium GEM-X technology alongside Ultima Genomics’ UG 100 sequencing platform to create a single-cell atlas. Just two months later, the Arc Institute announced it would tap both 10x’s Chromium Flex platform and Ultima’s sequencing technology to expand its Virtual Cell Atlas beyond 300 million cells.

When the news was announced in late April, Arc said the partnership with 10x and Ultima Genomics would help “generate high-quality, perturbational single-cell data at scale,” in the words of Arc Executive Director Silvana Konermann.

“Back in 2017 we made a million cell dataset to showcase how scalable the technologies were, and people were like ‘Oh my God, a million cells, that’s crazy,'” Schnall-Levin recalls. “Now a billion cells is not just theoretical; it’s feasible and not outlandishly expensive.”

Arc’s goal, he explains, is to build a “virtual cell model.” “You basically collect lots of data, perturb lots of different genes, measure them all at the individual cell level, and then you have this incredible dataset,” he said. “Combined with AI tools, you can query that in ways that could really inform drug discovery, target identification, optimal experiment planning.”

In the following Q&A (lightly edited for brevity), Schnall-Levin discusses the economics of massive-scale experiments and the shift from million to billion-cell studies, how AI and automation are reshaping scientific workflows, the company’s latest product developments across its three core platforms, and why 10x, true to its name, maintains its philosophy of pursuing only breakthrough advances that deliver tenfold improvements.

On the ‘industrialization’ of science

R&D World: There’s growing talk about “industrializing science,” meaning things like bigger batches, automated workflows, scaling up dramatically. How do you see this trend playing out?

Schnall-Levin: It’s fascinating. I think we’ll see a mix of people doing smaller scale, very quick, creative projects alongside these larger mega projects. Both will continue to have huge value, but there are really interesting opportunities to automate and scale.

There’s the obvious scaling of technology and automating lab components to reduce manual labor. But the tremendous opportunities are in automating data analysis, not just numerical analysis, but bringing in scientific expertise. We’re already seeing this with customers. A group at Genentech built SpatialAgent on top of large language models, creating AI scientist agents that choose what genes to measure and how to analyze data. It’s not a total replacement for people yet, but it’s a major accelerator. They reported certain tasks being done better by these agents than most experts, taking 20 minutes instead of a week or two.

If you think about components that could really accelerate and reduce the total cost of doing experiments and getting insights, that’s quite exciting. You could see orders of magnitude improvement is in that downstream analysis part with emerging AI tools, although it’s still early days. A typical experiment might involve a few weeks of planning, maybe a week of running the experiment, and then six months of analyzing data and doing follow-up experiments.

Breaking the per-cell cost barrier

R&D World: You mentioned massive cost reductions in life sciences research. Can you provide more examples of that?

Schnall-Levin: We’ve released new assays that allow much larger scale single-cell experiments. A single kit can now run 5 million cells at less than one cent per cell in the highest throughput configuration. If you think about doing a billion cells, that would be $10 million of our reagents. Real money, but not insane amounts to run.

R&D World: Given the funding constraints hitting biotech, how are labs responding to these cost pressures?

Schnall-Levin: It’s one of three areas we focus on in product development: new capabilities, increasing usability, and scaling up. In other words, doing more for the same money or the same for less money. Even in a constrained funding environment, when you can run experiments that were previously impossible or prohibitively expensive, that opens up entirely new research directions. We’ve seen people use these mega-scale kits for large perturbation studies using CRISPR to knock down genes in parallel across millions of cells.

Platform evolution: Building on core technologies

R&D World: Any entirely new platform announcements since we spoke last year?

Schnall-Levin: We’re still focused on those three platforms: Chromium, Visium, and Xenium. There have been no entirely new platforms since last summer, but we’ve launched new products within those platforms.

On Chromium, we released new assays that allow much larger scale single-cell experiments. A single kit can run 5 million cells at less than one cent per cell in the highest throughput configuration. We’ve also released automation solutions, taking some of the labor out by automating parts of the Chromium workflow. We’ve seen people use this for large-scale perturbation studies using CRISPR to knock down genes in parallel in lots of cells.

For Visium, we’re about to release a new assay that allows high-resolution spatial analysis with different kinds of RNA measurements. Right now, our products are really great for differential gene expression, but the new product will allow customers to do broad sets of species, isoforms, to discover SNPs, VDJ sequences.

The last big one coming out is a multi-omics solution on Xenium: being able to do RNA and protein at the same time from the same exact tissue section. RNA is way more scalable and easier to do technologically, but protein is really the functional molecule in the cell and the classic marker people look at. If you can combine those, you get tremendous power from typing all the cells from the RNA side while doing downstream functional measurements of protein on that same tissue.

The R&D investment story

R&D World: I noticed on your website a mention of being backed by “10 years, over 2,200 patents, and >$1.5B in R&D investment.” Those are eye-popping figures.

Schnall-Levin: The company was founded in summer 2012 and really got going in early 2013 when I and the first team joined. We spent the first few years building out the core platform without having a product on the market. We launched the first product in 2015 and the first single-cell product at the beginning of 2016. It’s been almost 10 years since then.

The first product was just single-cell gene expression with extremely limited performance compared to today, and only worked on pristine, fresh samples. Since then, we’ve added this broad set of multi-omic readouts: gene expression, surface proteins, intracellular proteins, CRISPR perturbations, immune receptor molecules, epigenetics. It’s been a giant explosion of multi-omics capabilities.

How AI could change scientific discovery

R&D World: How do you see AI changing scientific discovery beyond just automating workflows?

Schnall-Levin: There’s an insane body of scientific knowledge in this massive literature, and there might be some random fact somewhere that could correlate with something you find in your experiment. Right now it’s like: what are the chances someone finds that? You have these experts who read literature, go to conferences, and someone comes up to them like “Hey, I noticed you mentioned this gene. Have you ever read that paper from Russia in 1970?” That happenstance can drive discoveries.

Some of these AI tools could be incredible. They can read every single publication that’s ever been written and help make those connections. Some could be massive savings, but some could be making discoveries that just wouldn’t have been made otherwise, finding small connections hidden in tremendous amounts of unstructured knowledge. Then people can spend more time thinking at a higher level about strategy and what problems they’re trying to solve, rather than digging through literature.

The “10x” mindset

R&D World: Can you remind me of the backstory for the 10x name?

Schnall-Levin: One of our co-founders, Ben Hindson, worked at Lawrence Livermore National Lab, and his advisor there said “If you’re not working on problems that are going to make a tenfold difference in something important, then it’s not worth working on.” It’s an argument to not spend time on incremental things, but to do things that are really impactful.

We actually had a very different name initially and needed to change it. The idea was: that’s our ambition. We want to work on these big problems. So let’s capture that in the name. Almost anything you try to do is kind of hard in some ways. So it’s better to focus your energy where you feel like what you’re doing is going to be impactful. When stuff gets hard, you’re like “Well, this is why I’m doing this. This is why I’m pushing through the challenges.”