A team at Stanford recently posted a preprint describing a system they called “Virtual Biotech”: a coordinated squad of AI agents organized to mirror a real drug discovery company, complete with a virtual Chief Scientific Officer, specialized scientist agents, and over 100 tools for querying biomedical databases.

For their headline demonstration, they deployed over 37,000 agents in parallel, each one tasked with annotating a single clinical trial, linking therapeutic targets to genomic and single-cell transcriptomic features. The resulting dataset spans 55,984 trials. The analysis turned up what the authors call previously unreported associations: drugs targeting cell-type-specific genes were 40% more likely to advance from Phase I to Phase II, 48% more likely to reach market, and showed 32% lower adverse event rates.

In another case study, the system pulled together genetics, transcriptomics, and clinical data on B7-H3 in lung cancer and landed on an antibody-drug conjugate strategy, the same bet several pharma companies are already running in the clinic. It also flagged liabilities and differentiation angles. The whole thing reportedly cost $46 in API credits and took less than a day.

The Virtual Biotech is one lab’s preprint, but it lands in what we recently likened to a “gold rush”—agents are being deployed across clinical operations, translational biology, antibody design, and regulatory workflows. Major pharma companies are in an apparent compute arms race, stacking GPU clusters and billion-dollar AI partnerships within months of each other. Startups backed by hundreds of millions are launching agent-focused platforms. NVIDIA’s Jensen Huang even went so far as to declare agentic AI “the new computer” at this year’s GTC.

Whether the implementations match is another question. In a recent experiment, researcher Liang Chang asked—”Can AI make better decisions than pharma executives?” and sent AI agent teams back to a pivotal 2012 decision in oncology, the BMS vs. Merck biomarker strategy that ultimately decided the Keytruda-Opdivo war, and found that both Claude and GPT independently recommended the same path BMS took. The path that lost.

The agents produced rigorous analysis, identified the exact competitive threat, and still followed the consensus. As Chang put it: “AI can give you the best possible analysis. It can’t give you the courage to go against it.”

What can AI agents do today, where are they falling short, and why is everyone building them?

Subscribe now

🤖 Why agents, and why now?

A historical detour. The term “agent” gets used loosely enough in AI marketing that it might be worth tracing from its original meaning. The ideas behind it were actually tested long before today’s language model AI existed. The fundamental idea behind an agent is a feedback loop where a system perceives its environment, observes changes and adjusts in response.

Norbert Wiener and W. Ross Ashby worked on this in the 1940s, doing cybernetics. Their framework kept coming back to one idea that effective control depends more on the quality of the feedback loop than on the sophistication of the controller. Even a simple device like a thermostat qualifies: it doesn’t need to be smart, it needs a clean reading and a reliable switch.

For roughly three decades after that, the dominant AI paradigm assumed the opposite: that intelligence requires building an internal symbolic model of the world and then reasoning over it. Sense the environment, build a representation, plan against it, act. This was sometimes called GOFAI (“Good Old-Fashioned AI,” John Haugeland in 1985), and it produced systems that could play chess and prove theorems but later couldn’t walk across a room without tripping.

By the late 1980s, Rodney Brooks at MIT was building robots that dispensed with internal world models entirely. These had layered behaviours (avoid obstacle, follow wall, seek light) that composed into complex action without any central planner.

His argument against the symbolic AI mainstream was that intelligence doesn’t live inside the agent. It comes from the agent’s relationship with the environment. In “Elephants Don’t Play Chess” (1990), he wrote:

The world is its own best model—always exactly up to date and complete in every detail.

A simple agent in a well-structured environment beats a complex one in a poorly structured one.

Through the 1990s, multi-agent systems became a formal subfield concerned with how to coordinate many autonomous software agents, each with limited capabilities, so that useful collective behaviour emerges. The Belief-Desire-Intention models taken from philosophy and applied to software gave individual agents beliefs about the world, desires they wanted to achieve, and intentions they committed to. Swarm algorithms showed that coordination could arise without any agent understanding the whole and air traffic simulations demonstrated the approach at scale.

When returning our attention to the modern day version of LLM-based AI, let’s remind ourselves that, at its core, a large language model predicts text. Fittingly enough, it got good at this through human feedback during training.

And here is where the circle closes. We spent decades scaling the internal capability of AI systems, built the largest, most capable text-prediction machines in history, and the moment we try to make them do things in the world, act on observations, use tools, adjust to what happens next—the oldest insight in the field loops right back on us. To make an LLM good at acting (and, perhaps, closer to intelligence), we are back to feedback loops.

💭 Agents today

The current, fashionable incarnation of this idea is an LLM with access to tools.

Instead of a chatbot that answers questions, an agentic system acts. It breaks a goal into subtasks, calls external tools at each step (e.g. databases, APIs, code execution environments, other agents), and, ideally, carries context across the chain without losing the thread.

In biotech and pharma, this can map onto things like target identification, literature mining, data extraction, safety profiling, trial design, and regulatory documentation, all run through separate teams, separate tools, and separate institutional memories. An agentic system can serve as connective tissue across these, operating at a speed and parallelism higher than any single team.

Why now? A few reasons behind the current momentum:

1. Context windows grew large enough that models can now hold meaningful complexity in a single reasoning chain.

2. Tool-use capabilities matured: previously, every agent needed custom connectors to every data source, which has now moved closer to plug-and-play.

3. Inference cost dropped. According to Epoch AI, the price of achieving a given level of model performance has been falling by 10x to 900x per year, depending on the benchmark. GPT-4-level performance that cost $20 per million tokens in late 2022 now runs at roughly $0.40.

4. Open-source agent ecosystem exploded. From orchestration frameworks like LangChain, CrewAI, and AutoGen to full personal-agent runtimes like OpenClaw, the barrier to building an agentic system dropped considerably. In many cases, there’s no need to wire everything from scratch.

As a timely demonstration from pure machine learning recesses, the other day, Andrej Karpathy (former head of AI at Tesla and one of the original OpenAI researchers) open-sourced a minimal setup where an AI agent modifies code, runs a five-minute ML experiment, checks if the result improved, keeps or discards, and loops. Running on a single GPU node (that’s still a lot of compute), he left it iterating for two days and came back to ~20 improvements that all held up.

Karpathy called it ‘wild’ as he’d spent two decades doing exactly this kind of iterative neural-net tuning manually, and his very first naive attempt with the agent already beat what he considered a well-tuned project. His read on where it leads is that every frontier lab will do this, spinning up agent swarms that collaborate to tune models at increasing scale and “...humans (optionally) contribute on the edges.”

⏪ Last we checked

When we surveyed the landscape of AI agents in biotech last spring, the honest summary was this:

• Early-stage, fragile, mostly academic.

• A handful of systems had demonstrated interesting capabilities like TxAgent (Harvard), BioDiscoveryAgent (Stanford), SpatialAgent (Genentech), and Causaly’s knowledge graph agents.

• None were in production, tool chains were brittle, costs were steep, and there was no regulatory framework for any of it.

The field was caught between two realities of impressive demos on one side, and on the other, as always, the irreducible complexity of biology. But things changed quite fast.

⏳️ What changed?

The most visible change is what’s happening with the hardware and the scale of investment, starting with infrastructure.

🔹 Eli Lilly went live with LillyPod in late February with over 1,000 GPUs delivering 9,000+ petaflops. That followed a $1 billion co-innovation lab with NVIDIA announced in January. Through Lilly’s TuneLab platform (with access to models trained on ~$1B Worth of proprietary drug discovery data), select models will be available to biotech partners via federated learning where partners train on Lilly’s models with their own data, without transferring it.

🔹 Going even higher on compute, Roche just announced the deployment of over 3,500 GPUs across the U.S. and Europe, the largest announced GPU footprint in pharma. Genentech’s Aviv Regev framed it around Roche’s “Lab-in-the-Loop” strategy, which she said they have pursued for more than five years. An NVIDIA pre-briefing offered some early concrete numbers: nearly 90% of eligible small molecule programs at Genentech now integrate AI, and at least one molecule was designed measurably faster.

🔹 Extending into the instrument layer, Thermo Fisher and NVIDIA announced a strategic collaboration at start of this year to develop AI-native laboratory workflows and instrumentation.

But the hardware is ahead of the results. Lilly’s Diogo Rau told CNBC last October that AI-assisted benefits would likely materialize around 2030. At the LillyPod inauguration, he was cautious: “The hype is actually a serious threat to the research itself. Because if the hype becomes the story, then we’re all going to be disappointed.” The infrastructure is there, but the public record still contains far more detail on compute scale than on named downstream outputs. Roche has described at least one molecule whose redesign was accelerated.

That’s the infrastructure investment. What about actual use?

Pharma appears to see the first value of agentic AI in fixing data and workflow mess, not in autonomous discovery:

1. In the Owkin/STAT survey, 37% called implementation “very important,” but only 3% said it was the number one priority. More importantly, respondents put data challenges first at 41.6%, ahead of early discovery at 28.7%.

2. Deloitte‘s September 2025 survey of 100 U.S. healthcare technology executives found a similar pattern from the budget side: 61% were already building agentic AI initiatives or had secured funding, and 85% planned to increase investment over the next two to three years.

3. The Microsoft-NEJM AI report, focused on health systems, found actual deployment even thinner—just 3% of 30 surveyed organizations, with 43% still in pilots.

The numbers above show that deployment is thin, but there are already a few visible examples.

➕ AstraZeneca

AstraZeneca published one of the first honest accounts of putting an agentic system into a real pharma pipeline. Their paper describes ChatInvent, a conversational interface for drug discovery that evolved from a single-agent proof of concept into a multi-agent architecture. The paper is notable for what it says about how things break: every LLM upgrade required at least a week of prompt re-tuning and could change agent behavior unpredictably. The supervisor agent would silently mangle inputs, sub-agents would sometimes refuse tasks they were perfectly capable of handling. The multi-agent system was faster and cheaper than the single-agent one, although it also introduced more errors.

➕ IQVIA

At GTC 2026, IQVIA launched a unified agentic platform built with NVIDIA that bundles over 150 specialized agents for clinical, commercial, and real-world evidence workflows. The collaboration with NVIDIA dates back over a year; one of the earlier agents, a clinical data review orchestrator first shown at GTC Paris in mid-2025, uses automated checks and sub-agents to catch data issues early, cutting the review cycle from seven weeks to two. The initial release covers trial start-up, target identification, data review, market landscaping, and field sales preparation, with more agents expected in Q4.

➕ Daiichi Sankyo

Daiichi Sankyo has been using AI built with BCG to personalize responses to patient and HCP queries inside its Veeva-based systems across Europe and Canada. It's a more commercial deployment than AstraZeneca's experiment or IQVIA's agents, with content generation and protocol writing on the roadmap for 2026.

➕ Visions, and Others

There’s recent Insilico Medicine and Eli Lilly paper worth flagging that belongs in a more of a ‘vision’ category for now. Their February “From Prompt to Drug” paper describes a fully autonomous pipeline where a central reasoning controller coordinates specialized AI agents across target discovery, generative chemistry, automated synthesis, and clinical planning in a single closed-loop workflow. A scientist types a prompt; the system orchestrates the rest. The authors acknowledge the end-to-end vision “may seem far beyond what is possible today,” and argue the individual building blocks already work at a smaller scale.

➕ And then, there’s the FDA

The FDA also deployed agentic AI capabilities for its own staff in December 2025, though the rollout has been bumpy. The agency’s earlier gen AI tool, Elsa, was seen fabricating nonexistent studies and misrepresenting research, with employees describing it as unreliable for anything beyond meeting notes. All of this against the backdrop of over 1,000 staff cut from the drug review center and multiple missed approval deadlines. The regulator is experimenting with the same tools it will eventually have to regulate, and running into the same problems.

📝 A recent review in Drug Discovery Today suggests agentic AI is already valuable in drug discovery, just not where most of the headlines are pointing.

The paper covers eight case studies from companies including Potato, Plex Research, and Coincidence Labs, several with specific quantitative benchmarks. The gains that hold up are in the operational middle of discovery: literature synthesis, protocol generation, assay design. Potato’s Tater agent took the design cycle for a qPCR assay from one-to-four months down to under two hours, although empirical validation was still needed.

The agent architectures in these systems mirror how discovery teams already work— supervisor delegates to specialists, shared context, iterative refinement—just without the multiweek meeting cadence. The authors note that current benchmarks capture whether an agent got the right answer but not whether the reasoning behind it was sound, and that early-stage results like cell-level inhibition don’t guarantee downstream translation.

One gap: every case study reports time savings, none report what the infrastructure costs to build and run. But the broader view is that the real value right now is compression of the coordination overhead between steps that already work on their own, not autonomous science.

📌 What does this add up to?

The infrastructure is real, the investment is committed, and a handful of systems are actually running in production workflows. But the gap between the hardware announcements and the named scientific outputs is still wide, and the honest accounts show that multi-agent systems in messy real-world pipelines are notoriously fragile. The most grounded read right now is that agentic AI in biopharma is just right on cusp of leaving the “interesting demos” phase but well short of the “reliable infrastructure” phase.

🧪 Agents doing science

A single LLM asked to check its own work tends to agree with itself. In one medical study, some frontier models (in 2025) complied with illogical requests up to 100% of the time. Most multi-agent systems built over the past year have had to engineer around this. The solutions vary, but they all converge on the same idea of creating friction.

• Google‘s AI co-scientist uses what it calls a “generate, debate, and evolve” framework where agents propose hypotheses, other agents critique them, and the survivors get refined through iteration. In a collaboration with Stanford, two of three co-scientist-recommended drugs for liver fibrosis showed “significant anti-fibrotic activity” in human liver organoids.

• DeepMind‘s Aletheia has a generator-verifier-reviser loop where agents produce solutions, check them for flaws, and correct or discard faulty reasoning.

• Stanford‘s Virtual Biotech (the one we opened with) assigns a dedicated reviewer agent that evaluates outputs and pushes back. The same lab is also trying out the opposite approach with a generalist single Biomni agent that skips the team structure entirely, composing its own workflows across 25 biomedical subfields.

• FutureHouse, an Eric Schmidt-backed nonprofit in San Francisco building what it calls an “AI Scientist,” went wider: a single up-to-12-hour run executes up to 42,000 lines of code across 166 data-analysis agent rollouts and reads roughly 1,500 papers across 36 literature-review agent rollouts, coordinated through a structured world model. It reported seven discoveries, four of them “novel,” and three that independently reproduced unpublished human findings.

Even though friction helps, it doesn’t solve the problem entirely. Just from the systems above: Kosmos’s reports were rated about 79% accurate by independent scientists, but FutureHouse’s own team notes the system often chases statistically significant but scientifically irrelevant findings. Aletheia ran through all 700 open Erdős problems in a week, its verifier flagged 212 as potentially correct, human experts confirmed 63 as technically valid, but only 4 resolved genuinely open questions. Sakana AI, about a year ago, produced what it called the first fully AI-generated paper to pass peer review, but the caveat here is that it was a workshop submission, humans selected which generated papers to submit, and the paper was withdrawn.

Model providers are, of course, ambitious and optimistic.

🔹 OpenAI told MIT Technology Review that building a fully automated AI researcher is now its explicit priority with an “autonomous research intern” by September, a full multi-agent system by 2028. Its chief scientist Jakub Pachocki described a future where a “whole research lab” exists inside a data center. Doug Downey at the Allen Institute for AI calls the prospect “exciting” but cautions that multi-step scientific work compounds error, and chaining tasks makes success less likely across the whole sequence.

🔹 Anthropic is focusing less on a standalone autonomous researcher and more on embedding Claude into existing scientific workflows through partners like HHMI’s Janelia campus and the Allen Institute, while extending into research infrastructure and biopharma R&D through its Claude for Life Sciences rollout.

So while some push toward replacing the process entirely, others look to situate the tools inside an already existing human/institutional process.

Continuing with limitations—when agents are all instantiations of the same underlying model (or similar models), their “disagreement” is bounded by shared priors, shared training data, and shared failure modes. They’re unlikely to catch each other’s systematic blind spots, and only catch surface-level inconsistencies—a so-called ‘tyranny of the majority’ where homogeneous agents converge on shared errors unwittingly. None of them can encounter surprise.

A real experiment can falsify a hypothesis in a way no agent in the loop anticipated. Multi-agent critique can’t replicate that. Conveniently, a real science lab provides that ‘for free’, as some friction with reality is inherent to it.

♻️ The lab-in-the-loop

Let’s recall Rodney Brooks when he wrote that “the world is its own best model.” A wet lab isn’t quite the world, but it’s a lot closer to it than AI agents debating themselves virtually. Ross King‘s Robot Scientist ‘Adam’ was already doing something close to this in 2009 by formulating hypotheses, running physical experiments, interpreting results, and confirming novel gene functions in yeast without a human in the loop.

Lila Sciences’ CTO Andrew Beam frames a certain bottleneck we’ve now reached: AI advanced fastest in domains where results are “easy to verify,” like mathematics, where proofs can be checked mechanically. Science doesn’t offer that shortcut, so verification means running an experiment. Beam posits that boosting the throughput of experiments describing the physical world will provide the critical data stream for the next generation of AI models.

The approach works best when three conditions align: a large combinatorial space, automatable chemistry, and a fast quantitative readout.

A decent example of this is LUMI-lab—a self-driving platform for ionizable lipid discovery recently published in Cell. A foundation model pretrained on 28 million molecular structures proposes candidates, robots synthesize and test them, and the results feed back in. One design-make-test-learn cycle every 39 hours. Over ten rounds, the system evaluated over 1,700 lipids, and by round ten more than half exceeded the transfection efficiency of MC3, a clinical-grade benchmark. The top compound achieved 20.3% gene editing in mouse lung epithelial cells via inhalation, reported as a new bar for inhaled CRISPR delivery. Humans still handle hardware errors and interpret edge cases, but the experimental loop itself runs unattended.

Not every lab-in-the-loop system aims for full autonomy: Le Cong’s (Stanford) and Mengdi Wang‘s (Princeton) LabOS keeps the researcher in the loop and augments them instead—AI agents connected via smart glasses and robots read experimental context and assist in real time, extending their earlier CRISPR-GPT work into the physical lab.

Ginkgo Bioworks and OpenAI report they connected GPT-5 to Ginkgo’s cloud laboratory and optimized cell-free protein synthesis across six iterative rounds over six months, testing over 36,000 reaction compositions. They say the system reduced production cost by 40% relative to prior benchmarks. One important caveat is that the results were demonstrated on a single protein (sfGFP), and when tested on twelve additional proteins, only half were even detectable. Ginkgo is already selling the AI-improved reagent mix commercially.

Roche’s “Lab-in-the-Loop” strategy, Lilly’s integration of agentic AI with robotic biomanufacturing, and Lila Sciences’ autonomous labs are all aimed at this kind of continuous computation-experiment cycle. The barrier to entry is also dropping because cloud lab platforms like Strateos and Emerald Cloud Lab let smaller teams plug into robotic infrastructure without building their own.

📌 Reality checkpoint

The gap between what these systems can do and what they’re being described as doing is still wide. The idea of lab-in-the-loop is more trustworthy than pure virtual debate, but it only works when the problem is shaped right. Most biology either isn’t shaped right or is hard to shape. The more durable near-term bet is probably the less glamorous one: embedding these tools inside existing scientific institutions rather than replacing the process completely, accepting that human judgment stays load-bearing for now, and letting the autonomy expand incrementally as reliability earns it.

⛓️‍💥 What doesn’t work

Now, back to limitations.

• Reliability. A February 2026 paper from Princeton and Cornell evaluated 14 agentic models and found that nearly two years of rapid capability gains have produced only modest improvements in reliability. Agents that can solve a task often fail on repeated attempts under identical conditions, with outcome consistency scores ranging from 30% to 75%. All three major providers clustered together. Scaling up didn’t uniformly help: larger models improved calibration and robustness but actually hurt consistency, showing more run-to-run variability. An agent that passes a benchmark may behave differently each time you run it on the same input.

• Architectural narrowness. A systematic evaluation of six drug discovery frameworks (Wijaya, Feb 2026) found all six locked into the same pattern: LLM reasons over text, calls APIs. That works for literature review and SMILES-based molecular design. It breaks when we need what drug discovery actually requires: model training, reinforcement learning, simulation, in vivo data integration, multi-objective optimization. The bottleneck isn’t really model knowledge (frontier LLMs reason about peptides competently) but the fact that no framework exposes those capabilities. There’s also a resource assumption baked in misaligned with small biotech realities: all six frameworks assume large-pharma data volumes, cluster-scale compute, and specialized teams.

Having several LLMs critique each other’s reasoning is the most discussed mitigation. The evidence is growing, and it’s mixed.

• Estornell and Liu (NeurIPS 2024) formalized the core problem as “tyranny of the majority”: when most agents share a misconception, minority agents conform rather than push back. The echo chamber follows from models sharing training data, priors, and failure modes.

• Wynn and Satija (2025) went further, showing that debate can actively degrade performance—models shifted from correct to incorrect answers by favoring agreement over challenging flawed reasoning, even when the stronger model outnumbered weaker ones.

• Wu et al. (2025) confirmed the pattern from a different angle: in controlled experiments, intrinsic reasoning strength and group diversity drove debate success, while structural tweaks — depth, turn order, confidence reporting — did little. You cannot scaffold your way past weak reasoning.

• Mixed-vendor teams help. Yuan et al. (Feb 2026) showed that assembling agents from different model families consistently outperformed single-vendor teams in clinical diagnosis, catching blind spots that homogeneous teams reinforced. But this is diversifying error profiles, not eliminating error.

📑 The echo chamber has an upstream version—Marinka Zitnik, associate professor of biomedical informatics at Harvard, noted in a recent Fay Lin’s GEN feature that 95% of all life sciences publications focus on roughly 5,000 of the most well-studied human genes. An agent trained on that literature will generate hypotheses that cluster around the same targets, not because the model lacks reasoning ability but because the knowledge base is lopsided.

Diversifying the model vendor doesn’t fix a skewed training signal. What does, at least partially, is tying agents to data modalities the literature underrepresents—single-cell sequencing, molecular structures, longitudinal clinical trajectories—which is another way of saying: back to the lab.

And there’s more:

• Tool fragility. In AstraZeneca’s case, every LLM upgrade required at least a week of prompt re-tuning, supervisor agents mangled inputs, sub-agents refused tasks they could handle, and multi-agent setups introduced more errors than single-agent ones. A single unexpected API response crashes a multi-step chain. Recovery is ad hoc.

• Error compounding. An agent that’s 95% accurate per step drops below 60% over a ten-step chain.

• Preclinical speed is not total speed. AI compresses early discovery timelines. It does not compress clinical trials, patient enrollment, regulatory review, or biology itself.

Presently, agents are getting better at talking about science faster than they’re getting better at doing it.

🔒 What can go wrong

While ‘what doesn’t work’ is about epistemological and performance failures like reliability, narrowness, hallucination, echo chambers, regulatory gaps (failures in a benign environment)—there’s also ‘what can go wrong’ adversarial failure—what happens when someone is actively trying to break or exploit the agent?

Agents with access to clinical data, lab automation systems, and regulatory documents present an attack surface that we are only starting to reckon with. Cisco’s State of AI Security 2026 report found that only 29% of organizations felt prepared to secure agentic deployments.

Anthropic reported that in mid-September 2025 it detected what it described as the first documented large-scale cyberattack executed without substantial human intervention, targeting roughly thirty entities across sectors including finance and chemical manufacturing through manipulated Claude Code.

In pharma, where a compromised agent could alter experimental protocols, misroute regulatory filings, or leak proprietary compound data, the consequences are sector-specific and hard to bound. The now-(in)famous “move fast and break things” motto that somewhat works in consumer software carries a different risk profile here.

🔭 Looking ahead

The tools evolved, deployments are growing (if thin), and many are building their own or adding on agents. The AI infrastructure commitments are serious, and so is the gap between what’s been announced and what’s been shown to work. Somewhere between a gold rush and a correction there are a few things to look out for:

• Regulation. This January, the FDA and EMA jointly identified ten principles for good AI practice across the medicines lifecycle, spanning work from early research through post-market activities. The EU AI Act’s high-risk provisions are now coming into force, with healthcare AI in scope. But neither touches agentic AI specifically. Autonomous agents that plan, chain tools, and act across multi-step workflows present a different challenge that the current frameworks haven’t caught up to.

• First regulatory submission with agentic contributions. At some point, someone will file an IND where agents meaningfully contributed to the evidence package, target selection, data analysis, or safety profiling. When a regulator has to evaluate that and decide what counts as adequate documentation of what the agent did and why, there will be a conversation around audit and accountability.

• Interoperability standards. Seed-funded by Genentech, The Pistoia Alliance is building agent-to-agent communication protocols for life sciences. Although most companies aren’t yet at the stage where cross-vendor agent communication is the binding constraint.

• Talent. The scarcest resource in agentic AI deployment is people who combine AI engineering with life sciences domain knowledge and quality systems experience. Pistoia Alliance polls rank skills shortage as the second-biggest barrier to AI adoption in pharma, behind resistance to change. Many are trying to build these hybrid teams from scratch while simultaneously running pilots.

• Consolidation & Stratification. The field is splitting between companies with proprietary biological data and those building on public data alone. This matters because data moats increasingly determine which AI/agent systems can produce differentiated outputs.

The field is moving fast enough that a survey like this one dates quickly. There is a lot of inflated optics surrounding AI, and the current cycle is agents—so in the crossfire of major forces and infrastructure investments, that’s worth keeping in mind when gauging the reality.

The underlying tension between what these systems can do and what biology actually requires will stay for a while. We’ll be watching for named partnerships and plans, of course, but more so (and mainly) for tangible outputs and reproducible results.

As always, if you're working on any of this or watching it from the inside—we'd love to hear what you're seeing, leave a comment!

Leave a comment

⚠️ The most immediate one is patient recruitment. Far back in 1979, the father of clinical pharmacology Louis Lasagna observed that the pool of eligible patients shrinks by 90% the moment a trial opens, only to reappear once it closes. Lasagna’s Law remains as relevant as ever: according to a 2022 article, 11% of trial sites enrol zero participants and nearly 90% of trials face meaningful delays. With Phase II and III trials costing roughly $40,000 per day, the financial toll is brutal.

⚠️ Another issue is clinical attrition. Research from VU Amsterdam found that between 2012 and 2019, the share of trials successfully completing each phase declined steadily—particularly at Phase II. In the first half of 2024, nearly a third (32%) of trials were discontinued at Phase II—a 56% rise compared to pre-pandemic levels. Combined with stagnant rates of Phase III initiation over that same decade, the picture is one of a systemic bottleneck: trials that begin are increasingly unlikely to see the finish line.

⚠️ Rare disease research presents its own distinct challenge. As the FDA’s Rare Disease Evidence Principles note, shrinking patient populations make it progressively harder to generate reliable efficacy data through conventional designs—especially placebo-controlled trials, where enrolling enough participants to reach statistical significance can be close to impossible.

⚠️ Apart from operational intricacies, there is the ethical dilemma. Randomized controlled trials remain the gold standard for evaluating new therapies, but randomization isn’t always defensible. When an effective treatment already exists, assigning patients to a placebo raises serious moral questions, e.g. HIV cure trials with the antiretroviral treatment interruption.

The question, then, is whether parts of the control process can be simulated rather than physically recruited.

Digital twins are emerging as a compelling response. Last October, Sanofi Ventures invested in a digital twin platform developer QuantHealth, bringing its total funding to $30M. In 2025, the FDA announced plans to phase out animal testing requirements for monoclonal antibodies in favor of human-relevant methods, including AI-driven computational models—with the EMA moving in the same direction. Both industry and regulators, it seems, are taking this technology seriously.

👥 How it Works

A digital twin is a virtual replica of a physical object, continuously updated with real-world data so it mirrors the original’s behavior in real time. The concept, first applied by NASA in the 1960s, has since migrated from engineering into healthcare.

The applications are wide-ranging: optimizing industrial processes as Eli Lilly did to boost production of their GLP-1 drugs, predicting equipment failures, streamlining supply chains, and accelerating product development.

In a clinical trial patients are generally divided into two groups, also known as arms. The intervention arm receives the experimental treatment; the control arm receives a placebo, standard-of-care treatment or sham, serving as the baseline against which results are measured. Randomized controlled trials (RCTs) are the gold standard because randomization minimizes bias, but that randomization isn’t always flawless.

The traditional workaround of external controls drawn from historical trials, health records, or registries all carry their own limitations. For instance, data like those don’t include underrepresented groups or don’t account for placebo effect due to their observational nature.

Digital twins go a step further: using AI models augmented with historical data, they generate individualized predictions of how a patient might respond under different treatment scenarios. When used to simulate outcomes for patients who do not receive the experimental therapy, these models can produce a synthetic control arm.

These trial-level twins build on a foundation of patient-specific digital twin modeling (virtual replicas of individual physiology shaped by genomics, imaging, and clinical history) which we covered in our earlier overview of biological and patient-specific twins.

In clinical trials, an AI-powered digital twin typically operates in three steps:

• Build virtual patients — AI integrates biomarkers, imaging, genetics, and real-world evidence to generate synthetic profiles capturing the full variability of real populations.

• Run simulated trials — virtual cohorts replace placebo groups or test experimental therapies in silico, probing efficacy and safety without exposing patients to unnecessary risk.

• Optimize continuously — trial parameters like dosing and sample size are continuously refined in real time, anchored by validation against real-world data.

A less computationally demanding synthetic control arm approach uses AI-generated patient data based on registries, and real-world evidence but unlike DTs not modelling it on a particular individual. The appeal is sharpest in rare diseases, where finding enough eligible control patients is often impractical. The FDA, EMA, and NICE have all endorsed the approach, and it’s gaining traction: recent Phase II/III myeloma and lymphoma trials have leaned on external control data, and in at least one case (blinatumomab for acute lymphoblastic leukemia), a synthetic control arm helped support accelerated regulatory approval. AstraZeneca used over 300M synthetic patient records to advance its clinical trials, allegedly saving up to $100M per drug in development.

Synthetic control arms built from historical data have already supported label expansions and accelerated approvals with alectinib, blinatumomab, palbociclib among them. AI-generated individualized digital twins, however, have not yet served as primary evidence in a completed approval.

🦾 An Industry Arm

Read more

Shortly prior to last Christmas the European Commission published a proposal of a European Biotech Act, a strategic initiative aimed at setting up a regulatory framework to strengthen the life sciences sector across the EU. The document has been mostly positively received by the sector leaders as a needed step towards fostering local biotech innovation. Brussels isn’t stopping there. Commission President Ursula von der Leyen pitched EU-Inc, a pan-European company structure meant to solve what many see as the EU’s core startup problem of navigating 27 different bureaucratic regimes. The proposal would let one register in 48 hours, fully online and in English.

The Case for Urgency

Why does it matter? Europe gave the world its first blockbuster pharmaceutical (Aspirin in 1899) and just 30 years ago produced half of all new treatments globally. Today, that share has fallen to just one in five. Even though the EU biotech industry has grown twice as fast as the overall union’s economy over the last decade, it struggles to convert the world’s top science into commercially viable products.

Europe holds a comparable share of the top 10% most-cited biomedical research to the US and China, yet lags significantly behind in venture investment — a gap caused by underdeveloped private equity markets and fragmented, complex regulatory frameworks. The disparity is also visible in listing trends, with 66 of the 67 EU companies that went public over the past six years choosing foreign stock exchanges.

To address this, the Biotech Act includes measures like:

Read more

Transformer architectures trained on biological data are beginning to predict drug response, generate therapeutic hypotheses, and identify which patients are likely to benefit from which treatments (part of a broader push that includes early attempts at virtual cell models) tasks that previously required years of wet-lab iteration. Some of that work is still early, though a handful of results have made it far enough through validation to be worth paying attention to.

• Google Research, Google DeepMind, and Yale spent much of 2025 scaling C2S-Scale, a language model that reads single-cell RNA data as text; the 27-billion-parameter version, released in April, came in October with wet-lab validation of a model-generated hypothesis about making immune-”cold” tumors visible to T cells.

• A collaboration between Microsoft Research, Providence Health, and the University of Washington took a complementary approach: GigaTIME, published in Cell in December, routinely converts pathology slides into virtual immune-protein maps, surfacing over 1,200 significant associations across 14,256 patients.

• At Davos in January, Demis Hassabis now put Isomorphic Labs‘ first trials, primarily oncology candidates, at end of 2026; the company followed this month with IsoDDE, a general-purpose drug design engine that reportedly doubles AlphaFold 3’s accuracy, already deployed across its oncology programs.

Not all of it is language-model work.

Read more

Just a day earlier, MIT Technology Review published its annual 10 Breakthrough Technologies list. This year, three of the highlighted technologies were in genomics: personalized gene editing, embryo scoring, and gene resurrection. From there, it seems like genomic applications are moving more into the mainstream technology discourse.

Another just-in data point from a few days ago is a report out of San Diego, where Element Biosciences says its newly announced VITARI benchtop sequencer can deliver a whole genome for $100, positioning it as a lower-cost alternative to Illumina’s high-throughput systems.

Looking at these and many of last year’s developments, genomics come into view as an integrated technology wave that extends from data generation to interpretation, intervention, and biological reconstruction.

With those early-2026 pings as a starting point, let’s do a selective pass through a few genomics patterns that seem to be carrying momentum into 2026.

Read more

At first glance, these headlines span different companies and medical areas. But they share a common thread: each centers on advanced therapeutic modalities (ATMs)—a new generation of medicines that go beyond the limits of conventional drugs.

According to BCG, eight of the top ten best-selling biopharma products in 2025 are new-modality drugs, and the global pipeline value for these therapies has reached $197B. ATMs are becoming a more established part of the industry and are noticeably contributing to its growth.

Reject Tradition, Embrace Modernity

Read more

This article is brought to you by our writer Anastasiia Rohozianska; also, check out our regular BiopharmaTrend.com contributor, Dr. Louise von Stechow, who writes on AI, biotech strategy, rare disease therapeutics, and emerging modalities (will also be publishing here soon!).

Subscribe now

Each year, the day has a different theme. This year it’s “Synergizing AI, Social Science, STEM and Finance: Building Inclusive Futures for Women and Girls,” which makes it natural to first ask where women generally stand in science, AI and finance today before looking at those who are now shaping these systems.

The numbers show that progress has been slow: women still account for roughly one-third of researchers worldwide and about 35% of STEM graduates, a proportion UNESCO data indicate has barely shifted in a decade. Inside the AI pillar that is rapidly reshaping drug discovery, healthcare and finance, women are estimated to hold around a quarter of AI jobs and less than 15% of senior roles, while a global synthesis of usage studies finds women about 20% less likely than men to engage with generative AI tools in the first place. At the same time, a UN analyses suggest a higher share of women’s jobs than men’s are exposed to AI-enabled automation, particularly in clerical and administrative roles, making it more likely that women experience AI as something that reshapes or displaces their work rather than an arena where they set agendas and build systems.

We usually cover the intersection of advanced biology, digital technologies, and emerging therapeutics—so for this occasion, we focus on STEM through the lens of life sciences.

Today, genomics, cell signalling, neurotrophic pathways and infectious-disease therapeutics are standard components of drug discovery pipelines, but much of this infrastructure traces back to scientists whose careers were shaped by structural barriers and delayed recognition, which is why an international observance focused on women and girls in science is not just symbolic context for an AI-era discussion, but part of the story: modern biomedicine is built on work that women often completed without full recognition.

• Rosalind Franklin’s X-ray crystallography on DNA and viruses was central to solving the double helix and later structural virology, but her role was widely recognized only decades later.

• Barbara McClintock’s maize genetics revealed transposable elements and genes as mobile (“jumping genes”), now fundamental to genomics and epigenetics, yet her Nobel Prize came more than 30 years after her first reports.

• Working under fascist racial laws that barred Jewish scientists from universities, Rita Levi-Montalcini identified nerve growth factor, the first growth factor and a basis of modern cell and neurobiology.

• Tu Youyou’s isolation of artemisinin created a new class of antimalarial drugs that have saved millions of lives and remain standard malaria treatment, while her contribution stayed largely unknown outside China for many years.

Below, we turn to women who are defining research, product, and investment priorities across AI and life sciences, rather than appearing only as those whose jobs, data, or care are shaped by these systems.

Daphne Koller

Founder & CEO at insitro

Daphne Koller, PhD, is a computer scientist and entrepreneur who moved from foundational work in probabilistic graphical models and Bayesian machine learning at Stanford into building an AI-native drug discovery company as founder and CEO of insitro. Koller’s academic career spans key contributions at the interface of AI, computer vision, and computational biology, recognized with a MacArthur Fellowship, the ACM Prize in Computing, and election to the US National Academy of Engineering and National Academy of Sciences. She also co‑founded Coursera and later co‑founded Engageli; today she remains engaged in digital learning while focusing her primary efforts on applied biology and therapeutics at insitro.

At insitro Dr. Koller oversees a platform that couples high‑throughput human cellular systems with machine learning models optimized for target discovery and molecule design. The insitro Human (ISH) platform builds iPSC‑derived disease models using human genetics and functional genomics, while complementary systems such as the POSH (Pooled Optical Screening in Human cells) platform combine pooled CRISPR perturbations, high‑content imaging and self‑supervised deep learning; together these platforms generate multimodal, multi‑omics datasets that feed cell‑level ML models.

ChemML, insitro’s small‑molecule design engine, integrates proprietary binding and ADMET data with physics‑based in silico screening, DNA‑encoded libraries and active‑learning medicinal chemistry to design and optimize small‑molecule therapeutics. With the acquisition of CombinAbleAI and the launch of the TherML platform, insitro now operates a single AI system that spans small molecules, oligonucleotides, antibodies and other complex biologics, explicitly optimizing for efficacy, developability and safety.

insitro’s preclinical portfolio is backed by major pharma alliances and in-house metabolic programs, with milestones including a novel ALS target from a Bristol Myers Squibb partnership, a Gilead deal in NASH worth up to $200M per target, and an AI-driven metabolic disease discovery program feeding internal efforts and a collaboration with Eli Lilly.

Alice Zhang

Founder & CEO at Verge Genomics

Alice Zhang, is the CEO and co-founder of Verge Genomics, an AI-enabled drug discovery company that uses human patient data to develop drugs for ALS, Parkinson’s, Alzheimer’s and related diseases. She trained in molecular biology at Princeton and spent five years in the UCLA-Caltech MD/PhD program before leaving to start Verge, and she has been recognized on Forbes 30 Under 30 and MIT Technology Review’s Innovators Under 35 lists while serving on the board of the California Life Sciences Association and remaining active as an angel investor in tech-enabled bio companies.

Under Zhang’s leadership, Verge has built CONVERGE, a proprietary “all-in-human” discovery platform that combines one of the field’s larger multi-omic patient tissue datasets with machine learning to find and validate targets directly in human disease biology.

Public description cites more than 61 terabytes of integrated human data supporting an end-to-end platform from target discovery through early clinical testing. The stack includes patient-derived CNS omics, computational target-prioritization models with reported higher validation rates, and digital biomarker tools that capture mobility, respiratory, sleep, and speech data via in-home sensors and AI-based analysis.

On the capital side, Zhang has led Verge through an oversubscribed $98M Series B led by BlackRock with Eli Lilly, Merck GHI, and others. The company also signed a three-year collaboration with Eli Lilly around up to four ALS targets with up to $694M in milestones, and a four-year, multi-target rare neurodegenerative and neuromuscular disease collaboration with Alexion/AstraZeneca that includes up to roughly $840M in milestones and royalties, alongside an Alexion equity stake.

Suchi Saria

Bayesian Health founder & director of the Machine Learning and Healthcare Lab at Johns Hopkins

Suchi Saria is a computer scientist and AI researcher at Johns Hopkins, where she leads the Machine Learning, AI & Healthcare Lab and founded Bayesian Health, a clinical AI company. Her roles span computer science, medicine, and health policy. Trained at Mount Holyoke and Stanford (PhD under insitro’s Daphne Koller), she’s recognized for work in machine learning and computational healthcare, with honors including a Sloan Fellowship, MIT’s Innovators Under 35, and WEF’s Young Global Leader.

Dr. Saria’s core platform combines methods, data, and governance, with her Johns Hopkins lab developing AI systems that learn from noisy EHR data to drive diagnostic and treatment tools. This work spans theory, deployment, and policy, and is commercialized through Bayesian Health, founded in 2018, as an adaptive clinical AI platform integrated into hospital systems. The research and product have drawn support from agencies like NSF, DARPA, FDA, NIH, and CDC, providing access to large datasets and regulatory-facing initiatives.

Saria also co-founded the Coalition for Health AI (CHAI), bringing together over 1,500 stakeholders, including Mayo Clinic, CVS Health, major cloud providers, the FDA, and the White House OSTP, to establish standards for evaluating healthcare AI. She also serves on the National Academy of Medicine’s AI Code of Conduct working group and the editorial board of the Journal of Machine Learning Research.

Saria co-authored TREWScore and TREWS sepsis models, which use real-time ICU data to identify septic shock risk hours before organ failure. TREWS, deployed via Bayesian Health’s platform, was evaluated in a large multi-site Nature Medicine study covering over 760,000 patient encounters, showing an 18.2% relative drop in sepsis mortality. Since its 2023 rollout, Johns Hopkins reports a similar 18% mortality reduction across dozens of hospitals, faster diagnoses by nearly two hours, and significant platform expansion. The same system is now used for other conditions like clinical deterioration and pressure injuries.

Together with Daphne Koller and Anna Penn at Stanford, Saria co-developed PhysiScore, a predictive tool that uses clinical data to assess health risks in premature infants. Beyond her academic and clinical work, she also served as an investment partner at AIX Ventures, where she evaluated and supported early-stage AI startups.

Regina Barzilay

MIT Computer Science & Artificial Intelligence Laboratory (CSAIL) professor & Jameel Clinic faculty lead, Phare Bio scientific advisor

Regina Barzilay is a School of Engineering Distinguished Professor for AI and Health in MIT’s Department of Electrical Engineering and Computer Science, a core member of CSAIL, and the AI faculty lead for the MIT Jameel Clinic for Machine Learning in Health, where she focuses on machine learning for clinical decision support and drug discovery.

With a background in natural language processing, she is a prominent figure in applying deep learning to oncology and chemistry, recognized with a MacArthur Fellowship, the AAAI Squirrel AI Award, and election to the US National Academies of Engineering and Medicine and the American Academy of Arts and Sciences.

Dr. Barzilay’s group leads some of the most widely cited imaging-based cancer risk models. Mirai is a deep learning system that reads screening mammograms and produces a personalized risk score for up to five years, designed to work across different scanner types and to handle missing clinical covariates. According to the Jameel Clinic, Mirai has now been validated on >1.5 million mammograms across 43 hospitals in 14 countries, and is being deployed through an international hospital network backed by Wellcome Trust.

Sybil, a companion model for lung cancer screening, predicts six-year risk from a single low-dose CT scan, with performance detailed in the Journal of Clinical Oncology. It forms part of a broader imaging model stack built atop radiology infrastructure and informs screening policy through the Jameel Clinic’s AI Hospital Network.

Barzilay co-leads the Jameel Clinic’s AI antibiotics program with Jim Collins, where deep learning models trained on assay data are used to screen large chemical libraries. Their Cell study identified halicin, a novel antibiotic active against drug-resistant pathogens in mouse models. She also advises Phare Bio, a non‑profit company spun out of the Barzilay/Collins antibiotics work, and is a senior co-author on the open-source Boltz models for therapeutic design, including BoltzGen, which generates protein binders for challenging targets.

Jen Asher

Founder & CEO at 1910

Jen Asher (Nwankwo) is the founder and CEO of 1910, an AI-native biotech company in Boston focused on small and large molecule drug discovery using multimodal data, high-throughput lab automation, and advanced AI models.

Dr. Asher holds a PhD in Pharmacology and Experimental Therapeutics from Tufts, founded 1910 Genetics to reflect a molecular-level approach to disease, inspired by the year sickle cell was first identified in the U.S. Before launching the company, she held roles in preclinical drug discovery at Eli Lilly and Novartis, worked in management consulting at Bain, and led business development at a health-tech startup and Transparency Life Sciences.

She leads 1910’s multimodal, multi-AI agent system Input-Transform-Output (ITO) platform, which combines proprietary data sources like simulations and lab assays with automated labs and secure learning across sites. The platform supports both small-molecule and biologics R&D and includes models like CANDID-CNS, which reportedly outperforms industry benchmarks for predicting brain-blood barrier permeability, and PEGASUS, which helps design cell-permeable macrocyclic peptides (a class of ring-shaped molecules that can slip into cells and bind broad, hard-to-drug protein surfaces) by integrating massive assay data with physics-based simulations.

1910 Genetics has a five-year commercial agreement with Microsoft to integrate its ITO platform into Azure Quantum Elements, offering co-discovery, co-engineering, and platform-as-a-service models to global pharma and biotech partners.

Accenture, now a strategic investor, is co-packaging the platform as an enterprise AI layer for biopharma R&D. The company has raised $26 million in seed and Series A funding from M12 (Microsoft’s venture fund), Playground Global, Sam Altman, FoundersX, and others, with additional strategic capital implied through the Accenture deal.

The five women spotlighted in this piece are only a small part of a much broader growing cohort shaping how AI enters biology and medicine.

• At ETH Zurich, bioethicist Dr. Effy Vayena is an important voice on digital health and data governance, including co-chairing the WHO expert group on ethics and governance of AI for health.

• At MIT, Dr. Marzyeh Ghassemi’s Healthy ML group builds machine-learning systems for healthcare that are robust, private and fair, tackling distribution shifts and bias in real clinical data.

• At UC Berkeley, Dr. Emma Pierson, core faculty in the Computational Precision Health program, develops data-science and machine-learning methods to study inequality and healthcare.

• Harvard’s Dr. Finale Doshi-Velez leads the Data to Actionable Knowledge lab on human-AI decision making, accountability and regulation.

• At EPFL, Dr. Charlotte Bunne works at the interface of machine learning and cell biology, including co-authoring Cell roadmap on AI-based virtual cell models.

• At Recursion, recent CEO Dr. Najat Khan leads an AI-native biopharma stack built around the Recursion OS platform, which uses large-scale biological and chemical datasets to support a partnered and internal pipeline in oncology, rare diseases and neuroscience, with work also spanning immune-mediated indications.

Together with founders like Daphne Koller, Alice Zhang and Jen Nwankwo, and researchers such as Regina Barzilay, they point to an emerging architecture in which women could be at the center of aligning AI, social science and biomedicine with real-world impact.

UNESCO’s “Closing the Gender Gap in Science” Call to Action, launched in 2024, was explicit about how much work remains. That is why pipeline initiatives targeting girls and young women in AI, coding and computational thinking are more than feel-good side stories. Programs such as Girls Who Code, Black Girls Code, AI4ALL and Technovation Girls already reach large global cohorts of girls, especially from underrepresented communities, combining coding and AI content with entrepreneurship, mentoring and problem solving around real-world challenges.

From Franklin, McClintock, Tu Youyou, and Levi-Montalcini to today’s AI-native founders, ethicists and methodologists, the story of biology has been repeatedly rewritten by women working against structural headwinds; the International Day of Women and Girls in Science is the yearly checkpoint on whether this next wave of AI-enabled biology will finally be different in who designs it, who funds it and who is allowed to benefit.

Originally posted on BiopharmaTrend.com

In DealForma’s figures cited by CEO Chris Dokomajilar, deal flow between large-cap biopharma and Chinese biopharma accelerated in 2024-2025. In 2025, big pharma completed 18 in-licensing and asset purchase deals (just one in 2020) from Chinese companies with $50M+ upfronts, totaling $57.3B in deal value and $3.9B in upfront cash and equity. By 2026, China continues to emerge as a major source of globally licensable, clinical-stage biotech assets, backed by an increasingly complete innovation stack, even as new policy constraints complicate cross-border data flows and outsourcing.

In late January, speaking at the Asian Financial Forum in Hong Kong, executives from Merck and Amgen pointed to China as a likely early approval market for fully AI-designed drugs. Merck China president Marc Horn suggested that 2026 could mark the shift from AI-assisted discovery to compounds designed end-to-end by AI entering regulatory pipelines, citing China’s patient datasets, clinical execution, and the government’s recent “AI Plus” policy push. Amgen’s chief medical officer Paul Burton pointed to a similar timeline, seeing 2026 as a year when AI-driven and human genetics–led discovery could begin translating more directly into drug candidates.

For perspective, among recent big pharma deals involving Chinese companies, this year’s JPM week had AbbVie’s $5.6B partnership with RemeGen around a bispecific oncology asset. Looking back at just 2025, Pfizer licensed a bispecific from 3SBio with $1.25B upfront, AstraZeneca entered a multi-year $5.3B AI-enabled small-molecule discovery collaboration with CSPC Pharmaceuticals, and GSK’s x Jiangsu Hengrui agreements included $500M upfront and up to about $12B in potential milestones.

From Generics to Innovation

Read more

Read more

Invasive & Minimally Invasive BCIs

Brain-computer interface (BCI) systems are being explored and used as a way to restore lost motor, speech, or sensory functions, particularly in patients with paralysis or neurodegenerative conditions. They work by placing electrodes on or in the brain to capture high-resolution neural activity, which is then translated into actions like moving a cursor, generating speech, or triggering stimulation.

Typically, BCIs include implanted pulse generators and wireless connections to external processors, which decode brain signals such as spikes or local field potentials from targeted brain areas, then use trained algorithms to translate those activity patterns into outputs such as cursor motion, text, or stimulation commands.

In 2025, several programs moved into multi-center or early pivotal territory:

• Neuralink extended its PRIME program into Great Britain with the GB-PRIME study at UCLH and Newcastle, evaluating the fully implantable N1 interface in patients with motor neuron disease and spinal cord injury, and reporting the first UK patient controlling a computer within hours after surgery. The same implant was used at home by ALS patient Brad Smith to control a motorized Insta360 webcam, demonstrating extended real-world use beyond cursor control.

• Paradromics received FDA IDE approval for its Connexus system to start the Connect-One early feasibility study, targeting speech restoration and computer control in people with severe paralysis via a high-bandwidth, fully implantable BCI. The Connect-One trial is designed around speech restoration as a primary endpoint rather than generic cursor control.

• Precision Neuroscience advanced its thin-film Layer 7 cortical interface. The 1,024-electrode subdural array, FDA-cleared as a temporary mapping device, was profiled in first human recipients as a minimally invasive, high-density platform that sits on the cortical surface rather than penetrating tissue.

• CorTec’s Brain Interchange BCI system reached first-in-human use in a stroke patient as a fully wireless, closed-loop implant capable of recording and stimulating cortex in real time, positioning it as a European competitor in implantable neuromodulatory BCIs.

• Synchron introduced an updated version of its endovascular Stentrode BCI that integrates Nvidia AI and the Apple Vision Pro headset to let people with severe paralysis control digital and physical environments using neural signals. Later, Synchron publicly demonstrated a person with ALS using its implanted Stentrode to control an iPad entirely by thought by converting neural motor-intent signals into native iPadOS inputs.

Read more

There are other speakers highlighting the promises of AI for solving aging. Anthropic CEO Dario Amodei said at 2025 WEF that if AI dramatically accelerates biological research, doubling the human lifespan by around 2030 isn’t unrealistic because it could compress “100 years of progress” into 5–10 years. Such claims are controversial, but they reflect a real trend: AI is impacting both basic geroscience and emerging longevity medicine. Before delving deeper into the intersection of AI and longevity, let’s overview the history of this field before machines came.

Nothing Lasts Forever

Aging is the gradual, time-dependent decline in the physiological functions required for survival and reproduction. Unlike age-related diseases (such as cancer or heart disease), the defining features of aging are shared by all individuals within a species.

As an integral part of life, aging has caused a multitude of philosophical disputes throughout history, tracing back to 350 BCE when Aristotle first tried to explain senescence, viewing it as a ‘natural illness’. However, conventional aging research started much later, in the 20th century.

Read more

Terray’s work in diffusion methods prompted a broader reflection on generative AI in biology. Today, most conversations and publications center on Transformer-based systems, especially large language models (LLMs) and other foundation models (FMs). LLMs make up a major subset of FMs, but whereas language models are trained primarily on textual data like natural language, code, or biological sequences, foundation models extend the paradigm to additional modalities, including images, audio, video, and even multimodal combinations.

Recent meta-reviews in biomedical NLP collectively catalog nearly 300 LLM instances across hundreds of studies. Foundation models are also proliferating, with over 200 tools developed since 2022 in drug discovery alone. In contrast, the literature on diffusion models for biological and chemical applications remains comparatively modest. So far, there have been only a handful of reviews capturing the diffusion generators. Yet despite lower popularity, diffusion architectures are carving out a meaningful and distinctive role in biotech research and industry.

Before diving deeper into their role in biomedicine, let’s briefly review how diffusion models work in general.

Read more

This transformation is happening thanks to numerous advances in the biology of aging, including the emergence of age-specific biomarkers or “aging clocks,” senolytics, cell reprogramming, and many other promising directions.

But it is also enabled by the rapid advent of foundational technologies, such as artificial intelligence (AI), which has opened doors for more sophisticated biology modeling, better understanding of aging processes, and potentially new creative ideas for addressing humanity’s biggest current limitation: the inevitable decline and end of life.

Below, I would like to highlight several recent advances in aging research that can redefine the landscape of therapeutics development in this area, both R&D and business-wise. This is based on learnings from this year’s 12th Aging Research and Drug Discovery Meeting (ARDD2025) in Copenhagen, and my coverage of last year’s ARDD2024 event.

Is the first anti-aging drug class on the horizon?

During this year’s Aging Research and Drug Discovery Meeting, Andrew Adams, Group Vice President of Molecule Discovery at Eli Lilly, presented a thought-provoking case for GLP-1 receptor agonists as potentially the first class of longevity therapeutics, with both mechanistic relevance and clinical scale to be able to impact human healthspan on a global population scale.

Traditionally indicated for type 2 diabetes and obesity, GLP-1s such as semaglutide and tirzepatide have shown robust, multi-system benefits: reducing the progression from prediabetes to diabetes by up to 94%, significantly lowering cardiovascular events (MACE), and producing weight loss with downstream metabolic and inflammatory effects.

Adams emphasized the relevance of these outcomes in delaying the onset of chronic, age-related diseases, thus compressing morbidity, a core objective in longevity medicine. He further highlighted early signals suggesting GLP-1s may exert positive effects on vascular function, cognitive decline, and even psychiatric and addiction-related conditions, though he acknowledged that more evidence is needed in these emerging indications.

Importantly, he framed these drugs within a broader shift from “sick care” to proactive, preventive healthcare, suggesting that long-acting delivery formats, such as RNA-based or gene-editing approaches, could further extend their reach and adherence in aging populations.

While he did not claim GLP-1s reverse aging or extend maximum lifespan, Adams posed a critical question: are GLP-1s the first true longevity drug class, considering their scalable, evidence-based impact across age-related disease trajectories?

A similar perspective was reinforced during the same conference by Dr. Lotte Bjerre Knudsen, Chief Scientific Advisor at Novo Nordisk and the scientific mind behind semaglutide. She opened her presentation with a striking and unambiguous title slide: “Semaglutide as a Proven Longevity Medicine,” as Alex Zhavoronkov, Co-founder and CEO of Insilico Medicine, writes in his LinkedIn post.

Now, zooming back from the burgeoning area of GLP-1s with all the promise of their therapeutic expansion into age-related diseases and even aging itself, let’s talk about a broader issue for the aging research field: how to cure something that is not a disease?

Finding the missing link between aging research and clinical development

Aging may be the biggest risk factor for chronic disease, but it still isn’t recognized as a disease itself, which creates a major bottleneck for companies attempting to go after anti-aging therapies directly. Without a formal indication, there’s no clear regulatory path, no accepted clinical endpoints, and little incentive for venture capitalists to fund such “vague” endeavors. This situation, undoubtedly, inhibits the progress of the longevity field.

To get around this, co-authors of a recent paper published in Aging, Alex Zhavoronkov, Dominika Wilczok, Feng Ren, and Fedor Galkin, put forward a compelling idea: instead of trying to treat aging head-on as a business model, treat diseases that biologically mirror it. Their work suggests that some age-related diseases are far more representative of the aging process than others, and that these can serve as practical, high-fidelity proxies for testing geroprotective drugs.

To find the best candidates, they built a scoring system that ranks diseases based on how closely they align with the hallmarks of aging—processes like telomere shortening, mitochondrial dysfunction, cellular senescence, and deregulated nutrient sensing.

Not only does IPF score high across nearly every hallmark, but it also progresses quickly. It has well-defined clinical endpoints, making it uniquely suited for real-world trials of anti-aging therapies. In essence, IPF offers the aging field something it’s been missing: a tractable, measurable, and regulatorily viable way to bring geroprotectors into the clinic, even if indirectly.

This alignment is not just theoretical; it has already begun to influence drug development strategy. For example, Rentosertib, developed by Insilico Medicine and currently in a Phase 2a clinical trial, inhibits TNIK and modulates Wnt/β-catenin signaling to reduce epithelial senescence and fibrosis.

The well-known senolytic combo dasatinib and quercetin is being tested for its ability to selectively clear senescent cells in IPF lungs. Danazol and cycloastragenol aim to restore telomerase activity, targeting telomere attrition, a hallmark strongly implicated in familial IPF.

And upstream, interventions like resveratrol and caloric restriction mimetics are being explored for their effects on sirtuins and epigenetic regulation.

For companies focused on longevity, IPF trials provide not only a clinically relevant path to market but also a powerful signal generator for anti-aging efficacy. A positive outcome in IPF can serve as both regulatory validation and scientific proof-of-concept, enabling indication expansion into other fibrotic or aging-aligned diseases such as chronic kidney disease, liver cirrhosis, or atherosclerosis. This approach may help work around the regulatory bottleneck of targeting “aging” directly and anchor aging drug development in diseases already recognized by regulators and payers.

Finally, the third big idea in aging research centers around a somewhat unintuitive topic in this context: the gut microbiome.

Do bacteria in our gut hold the key to healthspan and longevity?

At ARDD 2024, Scotch McClure, CEO of Maxwell Biosciences, a Texas-based AI-driven preclinical stage biotech, proposed a striking thesis: microbial imbalance (dysbiosis) may not just drive disease, it may actively accelerate biological aging. And reversing it, he argued, may offer a direct intervention point for extending healthspan.

The human body hosts over 39 trillion microbial cells, weighing approximately 1.5 to 2 kilograms, with the gut microbiome alone containing over 1,000 species of bacteria. Their collective genetic material outnumbers human genes by 150:1. When balanced, this ecosystem regulates immune response, metabolism, and tissue repair. But when disrupted, via pathogens, antibiotics, or poor diet, dysbiosis sets in, triggering chronic inflammation, immune dysregulation, and systemic aging.

McClure linked dysbiosis directly to all 12 hallmarks of aging, including genomic instability, telomere attrition, epigenetic alterations, mitochondrial dysfunction, cellular senescence, and loss of proteostasis. He cited evidence that dysbiosis impairs stem cell function, deregulates nutrient sensing, and accelerates immune decline (immunosenescence), leading to early onset of frailty, cognitive decline, and chronic disease.

In an exclusive interview for BiopharmaTrend, Scotch McClure said:

“The human microbiome expresses over 200x the human cell genetic expression. So the major lever in epigenetics is the microbiome. Therefore, aging should be at least partially treated as a communicable disease. Pathogens are a key driver of inflammaging, immunosenescence, stem cell depletion, heart disease, cancer, diabetes, and all-cause mortality. It is actually quite silly that we haven’t already categorized pathogens as a key driver of aging”.

To counter this, Maxwell Biosciences is developing a synthetic analog of LL-37, a broad-spectrum antimicrobial peptide naturally produced by the immune system. While LL-37 is effective against bacteria, viruses, and fungi, its clinical potential is limited by instability and rapid degradation in vivo.

Using their proprietary CLAROMER® platform, Maxwell has created synthetic peptoids, poly-N-substituted glycines, that mimic LL-37’s structure and function, are stable at room temperature, require no cold chain logistics, show no resistance development in pathogen exposure studies, and do not disrupt the healthy gut microbiome.

These peptoids disrupt microbial membranes, including biofilms, while preserving host tissue. The company’s lead candidate will be delivered as a nasal spray, targeting the nasal-brain axis, an emerging interface between microbiota, inflammation, and neurodegeneration.

The broader ambition is to build a “synthetic immune system”: modular, tunable peptoid-based therapeutics that restore innate immunity, clear chronic pathogens, and reduce inflammation — addressing the microbial drivers of aging at their source.

The company plans its first clinical trials in 2026 and aims to expand into health‑span and aging applications. If successful, this approach could reframe aging as not solely an internal biological clock, but as a process modulated by persistent, low-grade microbial stress, and ultimately, treatable with precision-designed immune molecules.

I also heard echoing ideas about the microbiome’s much bigger role in organism health during the recent Drug Discovery Innovation Program Conference (DDIP 2025) in Barcelona, the event I covered for my LinkedIn newsletter. Mainly, in his presentation, Dr. Rafik Fellague-Chebra, Global Medical Director at Novartis Oncology, argued that the microbiome is not a passive bystander in cancer but an active modulator of treatment outcomes.

Citing recent clinical data, Dr. Fellague-Chebra noted that antibiotic use can reduce the efficacy of immunotherapy by up to 50%, while fecal microbiota transplants (FMTs) from responders can restore treatment response rates. He described this tumor-resident microbial ecosystem, the “oncobiome,” as a key modulator of immune evasion, drug resistance, and pathway-level signaling via microbial metabolites affecting targets like mTOR and AKT.

These insights are in line with findings presented by Maxwell Biosciences at ARDD2024, and together these perspectives underscore a shift toward a new paradigm where microbiome profiling and modulation could become the future of precision medicine and therapeutic design.

EvolutionaryScale emerged in 2023 after Rives, along with Tom Sercu and Sal Candido, left Meta’s AI protein group (FAIR) during the company’s “year of efficiency” (there are, again, plans to cut 600 AI jobs after a $14.3 billion Scale AI investment and hiring spree this summer). Backed by the likes of Amazon and Nvidia, the team raised $142 million to develop large-scale generative models for protein design and became known for the ESM family of protein language models (PLMs) trained directly on amino-acid sequences.

Its flagships, ESM3 and ESM Cambrian, extended this work to fully generative modeling of protein structure and function. ESM3, trained on 2.7 billion proteins, has already been used to design molecules like the novel green fluorescent protein variant, esmGFP, said to represent roughly 500 million years of natural evolution.

CZI’s Biohub folds this hire into its broader “virtual biology” plan, setting out four scientific challenges: building an AI-based model of the cell, advancing imaging, instrumenting inflammation, and using AI to reprogram the immune system, with the Virtual Immune System as one of the flagship projects. The ES team is brought in “to help advance this initiative.” In the VIS roadmap, the molecular-interactions axis explicitly calls for protein language models that “can learn the universal grammar of immune recognition and enable the rational design of novel receptors.”

With that, let’s step back and look closer at what protein language models are, what kinds of applications companies are building them for, and where pharma is already involved.

Read more

Last year, OpenAI partnered with Formation Bio and Sanofi as well as signed agreements with Moderna, Eli Lilly; followed by a Thermo Fisher Scientific deal in 2025. At the same time xAI advertises Grok for Government with support for science and healthcare purposes, while DeepSeek gains adoption across Chinese hospitals.

Today we’ll look at LLMs entering biomedical workflows, examine what these systems can do in lab- and clinic-adjacent tasks, and how hybrid designs aim to mitigate common failure modes.

Read more

This May, researchers from the Centre for Genomic Regulation (CRG) reported the first use of generative AI to design DNA sequences that regulate gene expression within living mammalian cells. The team focused on synthetic enhancers—short DNA “switches” that control when and where genes are activated. In their proof-of-concept, they tasked the AI with generating enhancer sequences able to switch on a fluorescent reporter gene in specific mouse blood cells, while remaining silent elsewhere. The ~250-base-pair sequences were synthesized, delivered into cells by a virus (notably, they were integrated at random locations), and performed as intended in healthy mouse blood cells.

What CRG demonstrated fits into the typical synthetic biology workflow—where we define a functional specification, generate design candidates, synthesize them, and test them in cells. DNA elements are treated as modular parts composed into higher-order circuits and systems. Borrowing principles from classical engineering like standardization, abstraction, iteration, synthetic biology turns cellular programs into designable constructs.

These days, AI and machine learning are being applied all over the life sciences, and we’ve been tracking this closely across recent updates. Before diving into what’s happening in AI-SynBio today, let’s zoom out and see how the field emerged.

Historical Throughline

💠 1961–1999: The Origins.
Although the term “synthetic biology” was first introduced in 1912 by Stéphane Leduc in his work on the physico-chemical basis of life and spontaneous generation, the field’s origins are often traced to 1961, when François Jacob and Jacques Monod introduced the lac operon model in E. coli, showing that cells use regulatory circuits to respond to their environment. This framed biology as a system of logic and control. Early achievements included Meselson’s discovery of restriction enzymes in 1968, Boyer and Cohen’s recombinant DNA technology in 1973 — which paved the way for the first production of a synthetic protein, human insulin, in E. coli by Riggs and Itakura in 1978—and Kary Mullis’s invention of PCR in 1983. At the start of the decade, Barbara Hobom reintroduced the term ‘synthetic biology’ describing genetically modified bacteria with recombinant DNA. Advances in sequencing, genome mapping, and “omics” in the 1990s generated vast cellular catalogs; at the same time computational biology revealed networks of genes and proteins as structured, modular systems.

Read more

Despite the extraordinary effort, the odds of success remain bleak. Only about 1 in 10 drug candidates entering clinical trials ultimately achieve regulatory approval, with failures most often linked to safety issues or insufficient efficacy. Even high-throughput screening (HTS), once celebrated as a breakthrough, delivers a discouraging hit rate of just 2.5%. Such low yields amplify delays, inflate costs, and exhaust resources.

Artificial intelligence (AI) and machine learning (ML) are emerging as powerful alternatives meant to accelerate discovery, improve prediction accuracy, and break the limitations of traditional methods. Importantly, the story of AI in drug discovery has evolved alongside advances in computer tech by building on decades of incremental progress:

• 1960s: The drug discovery field took its first computational step with the development of the QSAR (Quantitative Structure–Activity Relationship) method. The groundwork for QSAR was laid by Corwin Hansch and his colleagues in 1962 when they researched the correlation of molecular properties with biological activity.

• 1980s: The release of the CHARMM program in 1983 enabled general molecular simulation. In the meantime DOCK developed by Kuntz’s group in UCSF became the first molecular docking software.

• 1990s: Molecular modelling software platforms like Schrödinger made computer-aided drug design widely accessible, embedding computational tools into the pharmaceutical workflow. Additionally, open-source alternatives like GROMACS appeared (University of Groningen in 1991).

• 2010s: Deep learning catalyzed a wave of AI-first biotech startups like Recursion and Insilico Medicine. In June 2025 Insilico released the Phase IIa results for an AI-designed compound Rentosertib against idiopathic pulmonary fibrosis, which is believed to be a considerable milestone for a solely AI-inspired drug candidate up to date.

• 2020s: In CASP14, DeepMind’s AlphaFold achieved a median Global Distance Test (GDT - the main CASP metric for prediction precision evaluation) score of 92.4 —an unprecedented leap in accuracy. In 2024, the Nobel Prize in Chemistry recognized Demis Hassabis and John Jumper (for AlphaFold) and David Baker (for computational protein design).

Over a decade ago, strategists at Big Pharma noted the possibility of a broader AI’s potential for R&D transformation, considering progress in deep learning at the time, and started testing grounds for wider adoption. One of the earliest examples took place back in 2012, with Merck tapping into the online data science community Kaggle to crowdsource solutions for a core drug discovery challenge: predicting biological activity of molecules, both on-target and off-target. The 60-day competition awarded $40,000, with the top prize going to a team led by George Dahl (University of Toronto) for their use of neural networks and deep learning—a hint at how these methods would reshape drug R&D.

Today, we will follow stages of the drug discovery pipeline as outlined in the January 2025 Nature article by researchers from Wenzhou Medical University. We’ll look at each stage in detail and highlight how leading biopharma companies are leveraging AI across the entire drug development lifecycle.

The pipeline unfolds across six critical phases: target identification → discovery → preclinical/clinical → regulatory → post-market. The first three stages represent unique challenges and opportunities for innovation, and AI is increasingly shaping how the pharmaceutical industry approaches them.

Target Identification

Finding the right molecular target,usually a protein or nucleic acid, has always been a bottleneck in drug discovery. Classic approaches like pull-down assays or genome-wide screens work, but they’re slow and costly..

AI is taking over the target detection by uncovering hidden molecular patterns and disease links that traditional tools miss.

• NLP models like word2vec have been used to map gene functions in high-dimensional space, boosting sensitivity when data overlap is sparse.

• Graph deep learning takes this further by combining network structure with deep models to identify key targets and explain its reasoning, a good example of which is CGMega—a GNN-based tool for cancer gene module dissection.

• Platforms like Insilico’s PandaOmics show what’s possible: by linking omics data with biomedical literature, it flagged TRAF2- and NCK-interacting kinase as an anti-fibrotic target, leading to a new inhibitor (INS018_055).

In 2025, AstraZeneca accelerated its AI-driven oncology strategy with a clear focus on target identification and validation. In April, it launched a $200M partnership with Tempus AI and Pathos AI to build a multimodal foundation model to uncover novel targets and accelerate therapeutic development.

Read more

In July, Deloitte released a report Trends Shaping Biopharma in 2025, highlighting two big changes.

• The first is that data and AI have become the new competitive advantage. With healthcare data growing rapidly and AI tools advancing, pharma companies can no longer just react to patient needs. They must predict them, provide proactive support, and design personalized treatments. Many firms are already bringing patient services in-house, and more than half of industry leaders admit their business models need to be updated.

• The second change is the rise of smarter supply chains and manufacturing. Geopolitical risks and supply chain disruptions are pushing companies to adapt, and technologies like digital supply networks, cloud platforms, digital twins, and advanced analytics are starting to reshape how drugs are produced—making the process faster, leaner, and more resilient.

  In this article: Alphabet/Google — Amazon — Apple — Microsoft — NVIDIA — Meta — Intel — IBM — Samsung — Oracle — Big Tech, Biopharma and Society

Tech companies, with their infrastructure, data power, and global reach are well positioned for the transformation biopharma needs. What draws them to healthcare is data. Biopharma holds one of the richest but least used data resources—genomic sequences, clinical trial results, real-world evidence, and patient feedback. Combined with AI and machine learning, this data could fuel breakthroughs in drug discovery, diagnosis, prognosis, and personalized medicine.

And tech giants started investing into biopharma a while ago. Amazon’s $3.9B acquisition of One Medical in 2023 signaled its healthcare ambitions. In July 2025, the White House’s “Make Health Tech Great Again” event highlighted the process: more than 60 healthcare and technology players — including Apple, Amazon, Anthropic, Google, and OpenAI — pledged to build a next-generation digital health ecosystem. Networks committed to CMS interoperability standards, health systems promised to ease patient data use, and EHR vendors vowed to streamline exchange.

All the signs indicate tech giants are here for a while, so let’s look at the key steps notable players have taken in biopharma this year.

Alphabet/Google

A recent InvestorsObserver analysis of Alphabet’s 13F equity positions suggests ~40% are in healthcare/biotech, which makes the company one of the most active investors in U.S. healthcare. The company strives for global AI leadership, considering healthcare and biotech as one of the major fields to fulfil this goal.

In March 2025 Google introduced TxGemma, a suite of “open” AI models aimed at advancing drug discovery. Scheduled for release later this month, the models are part of the company’s Health AI Developer Foundations program. TxGemma is designed to process both natural language and molecular structures—ranging from small molecules to proteins and other therapeutic entities. By enabling predictions about critical drug properties like safety and efficacy, Google expects the platform will accelerate the long, expensive, and high-risk process of drug development.

With this announcement, Google enters a crowded and increasingly competitive field of AI-powered drug discovery, which gathered 87% of all alliance investment in 2025. While the technology holds promise, with the 30% estimate for new drugs discovered with AI, results so far have been mixed. Several AI-designed drugs have failed in clinical trials, despite successes like Rentosertib - an IPF drug developed by Insilico Medicine which recently reached 2a phase trials.

Google’s most notable drug discovery player is Isomorphic Labs, the DeepMind spinout founded in 2021 and led by Demis Hassabis. The London-based company has raised this year $600M in its first external funding round, as it works to translate its AI drug-design platforms into clinical programs. President Colin Murdoch has said the company is “getting very close” to starting human trials, initially focusing on cancer therapeutics. Additionally, in April DeepMind and Isomorphic made AlphaFold3 (protein structure prediction tool) available for non-commercial use.

Note: In October 2024 the Nobel Prize in Chemistry went to David Baker for computational protein design, and jointly to Demis Hassabis and John Jumper for AlphaFold.

Drawing on AlphaFold’s capabilities, Isomorphic has already signed partnerships with Novartis and Eli Lilly, though it reported losses of £60M in 2023 due to heavy R&D spending and hiring.

Meanwhile, Alphabet’s other major healthcare initiative, Verily (previously Google Life Sciences), is navigating its own transformation. The life sciences arm plans to convert from an LLC to a C-corp as it prepares for a fresh funding round. CEO Stephen Gillett told employees the restructuring is intended to improve investor appeal, though no financing has been confirmed. At the same time, staff were informed that their equity had been revalued at roughly 80% below 2024 levels, reflecting what Gillett described as a gap between earlier valuations and current earnings. Nevertheless, the work is still ongoing and in early 2026 Verily plans to release Lightpath - an AI-backed chronic care solution aimed to help members with weight loss, e.g. providing support during and after GLP-1 use.

Amazon

Read more

We might be entering a new era of space exploration as nations and private companies are racing to push the limits of what lies beyond Earth. China made the Tiangong space station fully operational in 2022; NASA has advanced with the Artemis program, launching its Space Launch System on an uncrewed mission as the first step toward a permanent lunar base and, eventually, crewed missions to Mars; SpaceX drew wide coverage in 2023 with the first orbital flight-test attempt of Starship—a reusable spacecraft built to carry heavy payloads into orbit and one day ferry settlers to Mars. India, too, is carving its place in human spaceflight: ISRO’s Gaganyaan mission is entering its final phase, now set for launch in 2027.

🌌 Microgravity

One of the unique features of the space environment is the microgravity condition. In 2020, the Center for the Advancement of Science in Space (CASIS, the nonprofit that manages the ISS National Lab) and the University of Pittsburgh’s McGowan Institute for Regenerative Medicine co-hosted the Biomanufacturing in Space Symposium. Held virtually, the event brought together leading experts in tissue engineering, regenerative medicine, and space-based research to explore how the ISS could be utilized to improve biomanufacturing. The event marked an initial move toward building a roadmap for the space-based biomanufacturing market.

Participants identified and prioritized three major areas of opportunity for R&D:

1. Disease modeling using microphysiological systems (tissue chips) and organoids

2. Stem cells and stem-cell-derived products

3. Biofabrication

What’s special about microgravity? One illustrative example comes from Merck’s research on Keytruda. By leveraging the International Space Station (ISS) for crystallization studies, Merck achieved remarkably uniform 39 μm particles, compared to the irregular 13-102 μm range typically produced on Earth. This improved consistency is beneficial for drug formulation due to improving manufacturing efficiency and delivery methods. Similarly, a promising therapy for Duchenne Muscular Dystrophy (DMD, a devastating muscle-wasting disease) was developed from a protein crystal studied aboard the ISS. TAS-205, an HPGDS inhibitor informed by ISS protein crystallography data, entered Phase 3 but was discontinued in July 2025 after missing co-primary endpoints.

Beyond protein crystallization, microgravity alters cell growth, differentiation, and tissue formation. In stem cells, microgravity reshapes the cytoskeleton, extracellular matrix, and gene expression; for example, human iPSC-derived cardiomyocytes in space showed altered calcium handling and 2,635 differentially expressed genes, while blood-derived stem cells lost stemness markers and differentiated earlier into bone. The promise of space-based stem cell research is underscored by a recent Mayo Clinic experiment, launched last month aboard the SpaceX Dragon to the ISS, which investigates how bone-forming stem cells interact with the signaling protein IL-6.

Cancer research shows that microgravity drives re-differentiation: lung cancer stem cells lost stemness and underwent apoptosis, while colorectal CSCs increased CD133/CD44 double-positive populations.

These conditions also promote scaffold-free 3D spheroids and organoids, made to more accurately model tumors and improve drug testing. A great example of organoid use in disease modeling is the NIH’s Tissue Chips in Space initiative, led by NCATS in partnership with NASA and the ISS National Lab in 2017. The program investigates how organs function under the unique conditions of microgravity. By 2021, kidney tissue chips (developed by Nortis; later acquired by Quris-AI) had already flown twice to the ISS, providing valuable insight into how kidneys respond to toxic and pharmacokinetic stress. These models allow researchers to observe drug effects that might remain hidden during conventional preclinical testing.

In regenerative medicine, microgravity enables engineering of bone, cartilage, vasculature, skin, liver, and heart tissues with enhanced differentiation compared to Earth. For instance, rabbit MSCs in microgravity bioreactors formed cartilage expressing collagen I/II and aggrecan, while vascular progenitors displayed improved angiogenic potential. Additionally, in September 2023, Redwire announced that it had successfully 3D bioprinted the first human knee meniscus in space using its upgraded BioFabrication Facility aboard the ISS. The tissue, cultured for 14 days in Redwire’s Advanced Space Experiment Processor, was returned to Earth on the SpaceX Crew-6 mission for analysis. Building on this success, since late August 2024 Redwire has been equipping its bioprinting efforts with advanced 3D bioprinters supplied by the Finnish company Brinter AM Technologies.

📖 History Brief

The Space Race ignited in the heat of the Cold War, sparked by the Soviet Union’s 1957 launch of Sputnik 1—the world’s first artificial satellite. In response, the United States established NASA the very next year, determined to match and surpass its rival’s extraterrestrial achievements. Just a few years later, human spaceflight became a reality: Yuri Gagarin’s historic orbital mission in 1961 and Alan Shepard’s pioneering Mercury flight set the stage for a new era of exploration. Alongside these milestones, biomedical research into life beyond Earth gained momentum. NASA’s early Mercury (1958–1963) and Gemini (1965–1966) programs pushed the limits of human endurance in space, proving tolerance to microgravity, validating spacewalks, and extending mission durations. These foundational steps culminated in humanity’s giant leap—the Apollo 11 lunar landing in 1969.

But the leap from short missions to long-duration spaceflight brought a new set of challenges. When the US launched Skylab, its first space station, in 1973, astronauts encountered profound physiological difficulties: prolonged existence in microgravity led to bone demineralization, muscle atrophy, and cardiovascular deconditioning—even with carefully designed exercise programs. These conditions created a unique opportunity to study aging, disease progression, and therapeutic interventions in fast-forward.

To neutralize these effects, researchers tested countermeasures such as lower body negative pressure devices, which provided useful data on how the cardiovascular system adapts in microgravity, even though concerns about long-term health risks remained. The Apollo–Soyuz mission in 1975 also marked an important shift: for the first time, American and Soviet crews worked together in space, exchanging medical monitoring practices and setting the stage for later international cooperation in protecting astronaut health.

The Space Shuttle Era (1981–2011) opened an entirely new chapter for biomedical research by transforming microgravity into a powerful experimental tool. Scientists could now probe musculoskeletal physiology, cardiovascular regulation, and immune function in ways impossible on Earth, all while supporting astronaut health during longer missions. Shuttle flights also deepened collaboration with Russia’s Mir space station, creating a bridge for joint biomedical studies. Cooperation reached new heights with the 1998 launch of the ISS and the arrival of its first permanent crew in 2000.

Since then, the ISS has become humanity’s primary laboratory for studying life sciences beyond Earth.

📚 Space Flavors of Biology

In January 2024 Aaron J. Berliner with the team from the Center for the Utilization of Biological Engineering in Space (CUBES) released a comprehensive overview “Domains of life sciences in spacefaring: what, where, and how to get involved” in npj Microgravity. There, authors define three major fields integrating space research and biology—Astrobiology (AB), Bioastronautics (BA), and Space Bioprocess Engineering (SBE).

➕ Astrobiology

Astrobiology investigates the origins, evolution, and distribution of life, searching for habitable environments beyond Earth and signs of biology on other worlds. Mars, with evidence of past liquid water, remains a prime candidate, while icy moons like Jupiter’s Europa and Saturn’s Enceladus, with subsurface oceans and geysers, seem as well quite promising. Curiosity and Cassini missions have uncovered organic molecules, suggesting that the building blocks of life may be widespread across the cosmos. Astrobiology also emphasizes planetary protection, preventing Earth microbes from contaminating alien environments and safeguarding Earth from potential extraterrestrial life.

Spaceflight studies aboard the ISS have shown that bacteria like Salmonella and Serratia marcescens can become more virulent in microgravity, highlighting both risks for astronaut health and the importance of countermeasures.

Apart from microbes, astrobiology considers the possibility of intelligent life and its societal impact. Ultimately, this subfield advances our knowledge of life on Earth while preparing us for safe exploration and potential settlement beyond our planet.

➕ Bioastronautics

Bioastronautics is the study of how spaceflight affects living systems, with a focus on human health and performance in extraterrestrial environments. It addresses the challenges of long-duration missions while developing technologies to safeguard crews. Prolonged exposure to microgravity and radiation can cause bone loss, cardiovascular strain, immune dysfunction, and vision problems like Spaceflight-Associated Neuro-Ocular Syndrome. Isolation and confinement add further risks, including stress, depression, and cognitive decline.

To mitigate these effects, bioastronautics develops countermeasures ranging from exercise and radiation shielding to advanced air and water recycling. Model organisms play a central role: NASA’s Rodent Research program and JAXA’s Mouse Habitat Unit provide vital data on mammalian health in space, and zebrafish, fruit flies, worms, and plants contribute insights into how gravity shapes biology and how agriculture could sustain future crews. As noted above with Salmonella and Serratia marcescens, spaceflight alters microbiomes and can boost virulence in pathogens.

Addressing these risks requires not only medical strategies but also innovations in spacecraft and habitat design. Incorporating microbial control measures (e.g. advanced air and water filtration systems) reduces the likelihood of harmful microorganisms spreading in closed environments. This area of research lies at the heart of international initiatives like the European Space Agency’s MELiSSA (Micro-Ecological Life Support System Alternative) and NASA’s CUBES (Center for the Utilization of Biological Engineering in Space).

➕ Space Bioprocess Engineering

The idea of biotechnology as essential for space was first highlighted in the 1992 National Academies report Putting Biotechnology to Work. Today, with deep-space missions on the horizon, Space Bioprocess Engineering (SBE) is emerging as a defined discipline. SBE integrates synthetic biology and bioprocess engineering to design, build, and manage biological systems that sustain astronauts when resupply from Earth is limited. Unlike bioastronautics, which studies the effects of spaceflight on life, SBE develops the technologies that make long-term living in space possible.

Core SBE goals include in situ resource utilization (ISRU), loop closure (LC) for recycling, in situ manufacturing (ISM), and food and pharmaceutical synthesis (FPS). Efforts range from ultra-efficient carbon and nitrogen capture to programmable biomanufacturing for foods, medicines, materials, and even self-healing structures. Central to this are resilient platform organisms—microbes and plants engineered to thrive in extreme environments. Some examples: Arthrospira platensis (cyanobacteria for nutrients and pharmaceuticals), Cupriavidus necator (bioplastics), Methanobacterium thermoautotrophicum, and higher plants like lettuce and potatoes.

Challenges remain in safety, containment, and reliability, and NASA’s Decadal Survey (2023–2032) calls for bold investment. Its proposed BLiSS campaign (Bioregenerative Life Support Systems) seeks to harness biology for food, air, water, and waste management—making sustainable offworld habitation possible.

Let’s look at some of the companies involved in integrating biotech with space travel and research.

Read more

By the mid-2010s, that changed. Advances in WSI, falling storage costs, and FDA approval sparked the rise of digital pathology—scanning slides for easier sharing, annotation, and remote consultation. Computationally minded researchers quickly recognized the value of these datasets, and startups emerged, often spun out of collaborations with hospitals and labs holding vast archives of digitized slides.

How it started

Pathology history begins with the careful observations of ancient doctors, who recognized the importance of tracking how diseases developed even without understanding their causes. The 17th-century BC Edwin Smith Papyrus, written in hieroglyphs, described cases like skin ulcerations but could not explain their origins.

Centuries later, Hippocrates proposed the influential Humoral Theory, suggesting illness arose from imbalances in four bodily fluids—a framework that guided medicine until the 17th century. In the Middle Ages and Renaissance, figures like Antonio Benivieni advanced the field by systematically recording autopsy findings, marking a shift toward pathology as a distinct discipline.

The invention of the microscope in the late 16th century, refined by Robert Hooke, took on new significance in the 19th century when Rudolf Virchow used it to establish that diseases originate at the cellular level. Improvements in tissue preparation (fixation, embedding, staining) helped shape modern histopathology. The 20th century brought integration with other sciences, as advances in immunology, chemistry, and molecular biology deepened understanding.

Antibody discovery enabled immunohistochemistry, allowing for the precise detection of proteins within tissues and more accurate diagnoses. The invention of PCR in 1983 pushed diagnostics forward by capacitating genetic material to be amplified from very small samples. By the 21st century, pathology had gained the ability to examine single cells, offering unprecedented precision in detecting disease and guiding prognosis.

How it’s going

Over time researchers started trying to integrate digital solutions into pathology. In 1965, Judith Prewitt and Mortimer Mendelsohn from Upenn performed a first computerized analysis of microscopy images of cells and chromosomes. Over three decades later, 1999 was marked by the introduction of the whole slide imaging (WSI). WSIs are created by scanning glass microscope slides to produce a high resolution digital image, which is later reviewed by a pathologist to determine the diagnosis. Fast forward to 2017 and Phillips received a milestone FDA approval for its IntelliSite, the first WSI system for primary diagnosis in surgical pathology.

Essentially, WSI analysis can be encoded as the image recognition problem, which made digital pathology a fertile soil for the involvement of machine learning algortihms.

AI arrival in WSI diagnostics

The use of AI for a wide range of diagnostic tasks involving whole-slide images (WSIs) has grown in recent years. A comprehensive meta-analysis by McGenity et al. in npj Medicine published last year provides plenty of insights in this area.

According to McGenity, while AI has shown promise across multiple disease types, its most notable successes have been in applications to cancer. A landmark early study by Bejnordi et al. (2017) evaluated 32 AI models developed for detecting breast cancer metastases in lymph nodes as part of the contest on computer-aided diagnosis in histopathology using WSI (CAMELYON16 grand challenge). The best-performing model achieved an area under the curve (AUC) of 0.994, achieving near-human performance in this controlled setting.

More recently, Lu et al. (2021) trained an AI model to predict the tumour site of origin in cases of cancer of unknown primary, achieving an AUC of 0.80 for top-1 accuracy and 0.93 for top-3 accuracy on an external test set. AI has also been applied to predictive tasks, such as estimating 5-year survival in colorectal cancer patients and determining mutation status across multiple tumour types. Other reviews have investigated AI applications in liver, skin, and kidney pathology, with certain models demonstrating strong diagnostic performance.

Time for Foundation Models

Interest in LLM foundation models (like ChatGPT, DeepSeek, and Grok) exists for a practical reason: one pretrained system can be adapted to many tasks at lower marginal cost. For digital pathology, the workable path is using them as the language-and-logic layer atop whole-slide image models.

In March 2024, Faisal Mahmood, a computer scientist at Harvard Medical School, released UNI—a self-supervised vision transformer (ViT-L) for computational pathology, pretrained with Meta’s DINOv2 (2023) on Mass-100K, a dataset of 100M+ patches from 100k+ diagnostic H&E WSIs spanning 20 tissue types. UNI addresses the limits of prior encoders trained on smaller, less diverse datasets like TCGA. It was evaluated on 34 varied computational pathology tasks—regions of interest- and slide-level classification, segmentation, retrieval, and few-shot learning—covering challenges from cancer grading to rare disease subtyping.

On benchmarks OT-43 and OT-108 (43 and 108 cancer types), UNI showed strong scaling laws with data/model size and outperformed state-of-the-art encoders (ResNet-50, CTransPath, REMEDIS), especially for rare cancers. The model was recently updated to UNI 2 with an expanded training set - more than 200 million images and 350,000 slides. To support the model, in June 2024, Mahmood’s team launched PathChat, an AI-copilot that combines the UNI model with a LLM and is fine-tuned on nearly one million medical Q&A pairs. Licensed to Modella AI in Boston, it can analyze pathology images, generate reports, and has received FDA breakthrough-device designation.

Similar to UNI, Mahmood’s model CONCH (Contrastive Learning from Captions for Histopathology) — demonstrated superior performance to other models in classification tasks, such as cancer subtyping, according to the researchers. For instance, it could identify cancer subtypes carrying BRCA gene mutations with over 90% accuracy, whereas competing models generally performed no better than random chance. Trained on over 1.17 million images, CONCH was also capable of classifying and captioning images to generate visual representations of patterns found in particular cancers. However, its accuracy in these multimodal tasks was lower than in classification. In direct comparisons, CONCH consistently surpassed baseline methods, even when only a small number of data points were available for downstream training.

Several research teams have created their own foundation models for pathology. Microsoft’s Prov-GigaPath, for example, was trained on 1.38 billion image tiles from >171,000 slides collected from 28 cancer centers across the United States to perform tasks like cancer subtyping and pathomics. Using real-world data and a two-stage self-supervised learning process, it achieves state-of-the-art performance on 25 of 26 benchmark tasks across cancer subtyping, mutation prediction. For instance, in EGFR mutation prediction on TCGA dataset Prov-GigaPath demonstrated +23.5% AUROC and +66.4% AUPRC vs. next-best model REMEDIS.

Alternative tool mSTAR (Multimodal Self-taught Pretraining), released in July 2024 by computer scientist Hao Chen and his team at the Hong Kong University of Science and Technology, is the first pathology foundation model to integrate three modalities—pathology slides, pathology reports, and gene expression data. It leverages >26,000 slide-level multimodal pairs from >10,000 patients across 32 cancer types (over 116 million pathological patches) to identify metastases, subtype cancers and perform other tasks. Similarly to PathChat, Chen’s group developed their own AI-chatbot - SmartPath, now in hospital trials in China, where it is being tested against pathologists in diagnosing breast, lung, and colon cancers.

All that being said, French-based Bioptimus decided to think bigger and in May 2025 they released H-Optimus-1, the largest open-source pathology foundation model. It is a 1.1-billion-parameter Vision Transformer trained self-supervised on >500,000 whole slides spanning 50 organs and ~800,000 patients from 4,000 clinics. Compared with its predecessor H-Optimus-0, it raises average AUROC—a score that tells how well a classifier separates true-positives from false-positives (1.0 = perfect, 0.5 = chance)—on nine mutation-and-biomarker tasks from 0.835 to 0.856 and nudges the HEST gene-expression correlation from 0.413 to 0.422. The model’s embeddings support tumour detection, metastasis screening, mutation prediction (e.g., KRAS, BRAF, MSI) and survival modelling, and can be fine-tuned for laboratory-specific workflows.

The Market Players

Read more