The future belongs to organizations that can generate human data at scale.
In the long struggle to close the gap between laboratory promise and human healing, a Boston biotech called Xellar Biosystems has secured $50 million to pursue a fundamental reorientation of drug discovery — away from animal proxies and toward computational models built on human biological data. Founded in 2022, the company is wagering that organ-on-chip systems, automation, and artificial intelligence, woven together, can do what decades of pharmaceutical research have not: make the path from molecule to medicine faster, cheaper, and more humane. It is, at its core, a bet that biology — like language before it — can be understood at scale once the right data exists to illuminate it.
- Drug discovery's core wound remains open: most promising compounds still fail in human trials because animal models cannot faithfully replicate human biology, costing the industry billions and patients time they do not have.
- Xellar's closed-loop system — organ-on-chip tissues, high-throughput automation, and machine learning — is designed to generate the kind of rich, multi-dimensional human biological data that has been the missing ingredient, not more computing power.
- The $50 million Series A and A+ round signals that investors believe the bottleneck is solvable, and that the next wave of pharmaceutical productivity will come from companies replacing animal testing with human-relevant, AI-powered alternatives.
- The company is now racing to expand automated data generation, build out its AI and computational biology teams, and develop virtual cell technologies that could let researchers forecast drug outcomes before a single human trial begins.
Xellar Biosystems, a Boston-based biotech founded in 2022, has closed a $50 million funding round to scale a platform it calls 3D Bio Intelligence — a system built to generate human-relevant biological data at the speed modern drug discovery demands.
The problem it is addressing is old and costly. Despite advances in AI and computational biology, the pharmaceutical industry still leans heavily on animal models and simplified lab assays to predict whether a drug will work in humans. These are imperfect proxies. They miss the complexity of human biology, which is why so many drugs that succeed in the lab collapse in clinical trials. Xellar's argument is that the real bottleneck is not algorithmic — it is the absence of scalable, high-quality human biological data to learn from.
The company's answer is a three-part closed-loop system: miniaturized organ-on-chip tissue models that mimic real human organs; laboratory automation and high-content imaging capable of running thousands of parallel experiments; and machine learning that transforms the resulting data into predictive models of drug behavior. Multi-omics analysis — measuring genes, proteins, and metabolites simultaneously — adds further depth to what the system can see and understand.
The ambition, as CEO Xin Xie frames it, is to move biology from something researchers passively observe into something they can systematically predict. The company is building biological foundation models and virtual cell systems — computational representations of human tissues that could forecast drug efficacy and toxicity before any human trial begins.
The $50 million will go toward expanding automated data generation, strengthening AI and computational biology teams, and accelerating virtual cell development. The broader context is a biotech industry under pressure: development costs are rising, failure rates remain stubbornly high, and the next productivity leap is widely expected to come from human-relevant alternatives to animal testing. Xellar's wager is that biology, like language or images before it, becomes modelable at scale once the right data and architecture exist — and that winning that wager could reshape not just how drugs are discovered, but how long and how much it costs to bring them to patients.
Xellar Biosystems, a Boston-based biotech company founded in 2022, has closed a $50 million Series A and A+ funding round to scale what it calls a 3D Bio Intelligence platform—a system designed to generate human-relevant biological data at the speed and scale that modern drug discovery demands.
The problem Xellar is trying to solve is straightforward but stubborn. Drug discovery remains brutally expensive and slow. Despite decades of advances in artificial intelligence and computational biology, the industry still relies heavily on animal models and simplified laboratory assays to test whether a potential drug will work in humans. These methods are imperfect proxies. They fail to capture the full complexity of human biology, which means many drugs that look promising in the lab fail in human trials, wasting years and billions of dollars. The bottleneck, Xellar argues, is not computing power or algorithmic sophistication—it is the lack of scalable, high-quality human biological data to train on.
Xellar's approach combines three technologies into a closed-loop system. First, organ-on-chip models: miniaturized, three-dimensional human tissue systems grown in the lab that mimic the structure and function of real organs. Second, laboratory automation and high-content imaging that can run thousands of experiments in parallel and capture detailed data about what happens inside these tissues. Third, machine learning and computational biology that analyze the resulting data to build predictive models of how drugs behave in human biology. The company also performs multi-omics analysis—measuring genes, proteins, and metabolites simultaneously—to understand biological processes at multiple levels of complexity.
Unlike traditional preclinical testing, which observes what happens when you expose a tissue or animal to a drug, Xellar's system is designed to predict what will happen. By continuously generating high-dimensional biological data under conditions that approximate the human body, the company is building what it calls biological foundation models and virtual cell systems—computational representations of human cells and tissues that can be queried to forecast drug efficacy, toxicity, and clinical outcomes before a single human trial begins.
Xin Xie, Xellar's founder and CEO, framed the vision in stark terms: artificial intelligence alone will not transform drug discovery. The future belongs to companies that can generate human data at scale. Organ-on-chip systems, automation, and AI must function together as a self-reinforcing cycle. The goal is to move biology from something researchers passively observe into something they can systematically understand, model, and predict.
The $50 million will fund three priorities: expanding the company's automated biological data generation capabilities, hiring and strengthening its AI and computational biology teams, and accelerating development of virtual cell technologies that bridge wet-lab experiments and computational modeling. The timing reflects a broader shift in biotech. As drug development costs continue to climb and failure rates remain high, investors and pharmaceutical companies are betting that the next major productivity gains will come from companies that can replace or augment animal testing with human-relevant alternatives powered by machine learning.
Xellar's bet is that biology, like language or images before it, can be modeled at scale once you have enough high-quality data and the right computational architecture. If that bet pays off, the company could reshape not just how drugs are discovered but how long it takes and how much it costs to bring a new therapy to market.
Notable Quotes
AI alone will not revolutionize drug discovery. The future belongs to organizations that can generate high-quality human data at scale.— Xin Xie, Founder and CEO of Xellar Biosystems
The Hearth Conversation Another angle on the story
Why does drug discovery need a new data infrastructure? Isn't the problem just that we need better AI?
AI is necessary but not sufficient. You can have the most sophisticated algorithms in the world, but if you're training them on data from mouse models or oversimplified cell cultures, you're still working with a fundamentally incomplete picture of human biology. Xellar's insight is that the bottleneck is the data itself—we need human-relevant biological information at scale, and we don't have it yet.
What makes organ-on-chip different from just growing cells in a dish?
Organ-on-chip systems are three-dimensional and dynamic. They mimic the structure and microenvironment of real tissues—the way cells interact with each other, how nutrients flow, how mechanical forces shape behavior. A flat cell culture is like studying a city by looking at a photograph. An organ-on-chip is closer to watching the city actually function.
So Xellar is automating the creation of these tissues and then analyzing them with AI?
Exactly. The closed-loop part is crucial. Automation lets them run thousands of experiments in parallel and capture detailed imaging data. Machine learning finds patterns in that data that humans would miss. Those patterns feed back into the next round of experiments, which generates better data, which trains better models. It's a flywheel.
What would it mean if this actually works at scale?
Drug development timelines could compress significantly. Toxicity and efficacy could be predicted before human trials, which means fewer failures, lower costs, and faster paths to patients who need new medicines. It also means fewer animals used in testing, which matters ethically and practically.
Is this a replacement for human trials or a complement?
A complement, at least for now. You still need to test in humans eventually. But if you can predict which drugs will fail before you invest hundreds of millions in clinical trials, you've solved a massive problem. The goal is to make the human trials more targeted and more likely to succeed.