Moderna, the American biotech company that became a household name by producing one of the first mRNA COVID-19 vaccines, is once again in the spotlight.

On the day the outbreak made global news, Friday, May 8th, Moderna’s stock surged nearly 12%. By Monday, it was up another 7–9% in premarket trading. Then investors did the math — hantavirus doesn’t spread easily between humans, the WHO assessed public health risk as low, and analysts at Evercore ISI wrote bluntly that they saw “no meaningful revenue opportunity.” The stock gave back its gains by the close. A 20% weekly rally, then a retreat. The classic pattern: panic bids into a platform company, followed by the market remembering that a platform still needs products and demand…

But something real surfaced beneath the noise and stock trading rollercoaster. Moderna confirmed it had been working on an mRNA hantavirus vaccine, research that began well before anyone had heard of the Hondius. The stock, I think, spiked because of the belief that this company can quickly come up with solutions to new potential pandemics, and always has something in the drawer…

This belief is based on ribonucleic acid (RNA), or more specifically, messenger RNA or mRNA, the key information-delivery molecule in our body.

In this issue, we will explore why RNA is different from every other category of medicine ever built, how it was central to the most rapid fortune creation in pharmaceutical history, a world of new opportunities that can still unfold for it, the geopolitical power play in biotech, and the institutional wreckage that unfolded in recent years around mRNA technology.

Why is RNA so special?

In discussions about biology and life, we usually focus on DNA, deoxyribonucleic acid, a famous double-stranded structure that encodes all our genes.

Yet, an arguably much more promising molecule for many medical applications exists next to DNA — it is ribonucleic acid (RNA), a molecule that is very similar to DNA but has only one strand instead of a double-stranded structure, and a slightly different molecular alphabet.

RNA comes in many forms — messenger RNA, transfer RNA, small interfering RNA, ribosomal RNA, and others. The one that concerns us here is messenger RNA, or mRNA: the temporary message that carries instructions from DNA to the protein-building machinery.

Zooming out a bit, our body runs on a straightforward molecular information chain. DNA is the permanent archive of information, the master copy of every instruction your cells will ever need. mRNA is the working copy, i.e. a temporary message that carries a specific instruction from the archive to the cellular machinery, a “printer for proteins” called the ribosome. Proteins are the result, the molecular components that do the actual work: building tissue, fighting infection, running metabolism, etc.

A bestselling science writer, Dr. Siddhartha Mukherjee, explains this beautifully in his book “The Gene: An Intimate History, “ which I'd highly recommend.

Now, for most of pharmaceutical history, drug companies have targeted proteins as the most straightforward way to influence biology, that’s what a typical medical pill does.

You swallow a pill that contains a small molecule, the substance travels through the bloodstream, and it interferes with a “misbehaving” protein. Say, Statins block a cholesterol-producing enzyme. Ibuprofen blocks an inflammation enzyme, etc. The approach works, but it is a “one size fits all” strategy. The molecule goes everywhere in the body, hits its target but also hits things you didn’t intend (which leads to toxicity or side effects), and wears off quickly, so you take another pill tomorrow.

Gene therapies aim at the other end of this spectrum: the DNA itself.

If you can rewrite the master archive, you can potentially fix the problem permanently, in the genes. Correct the initial blueprint, and it never triggers production of the wrong protein in the first place. Elegant in theory. In practice, it means making a permanent change to someone’s genetic code, which raises the stakes enormously. If something goes wrong, you can’t undo it. The regulatory bar is high, and the cost can be staggering. For instance, some gene therapies, like Zolgensma, are priced above $2 million per patient…

In contrast, RNA-based therapies and vaccines sit in the middle, having the best of both worlds.

In the case of mRNA vaccines, they are the message itself, written to do a specific job on the printer. The entire idea is to intervene at the messaging stage, between the blueprint (DNA) and the machine that builds proteins (ribosome), and introduce a new message to produce a protein the body wouldn’t make on its own.

And here’s the key property: the message shreds itself when the job is done.

mRNA degrades in the body within hours to days. It tells your cells what to build, they build it, and the instruction disappears. That impermanence is the entire strategic advantage!

Why? Because RNA gives you the precision of gene therapy, you can instruct the body to make essentially any protein, including ones it wouldn’t produce on its own, but with the reversibility of a conventional drug. If something goes wrong, you stop dosing, and the effect fades.

Crucially, because RNA is essentially a sequence of four chemical letters in different arrangements, designing a new one is closer to writing a line of code than to synthesizing a chemical compound. Change the sequence, change the instruction.

The four-billion-year-old platform

Now, using RNA to modulate biology, especially after the COVID pandemic, is often hailed as a revolutionary pharmaceutical idea, but it may be the oldest trick in biology, after all.

In May 2025, a team led by chemist Matthew Powner at University College London published a study in Nature showing that RNA can chemically link to amino acids (the building blocks of proteins) under mild conditions in water, without enzymes, without cells, without any biological machinery at all. The chemistry works at neutral pH, in environments like pools, lakes, or wet shorelines. It even works in ice!

What the researchers demonstrated is a plausible mechanism for how, roughly four billion years ago, RNA could have begun constructing the first crude proteins, essentially bootstrapping the very relationship between genetic information and functional molecules that every living cell on Earth depends on.

Whatever the final verdict on the origin of life turns out to be, RNA remains the most ancient messaging system in our biology, potentially older than DNA replication, older than the first cells. Worth keeping in mind when we talk about mRNA technology as something radically new…

Anyway, everything that follows — the fortunes, the patent wars, the geopolitical contests — flows from the RNA’s molecular logic.

The fastest fortune ever made… and lost

Before January 2020, Moderna was a company that most people outside the biotech industry had probably never heard of. Founded in 2010, it had spent a decade building an mRNA platform and burning through investor capital. It had never brought a single product to market. Its market capitalization at the end of 2019 hovered around $5–6 billion, respectable for a clinical-stage biotech, but surely modest by pharmaceutical industry standards.

Then the COVID-19 pandemic hit.

On January 11, 2020, Chinese scientists published the genetic sequence of SARS-CoV-2. Within 48 hours, Moderna’s team had designed an mRNA vaccine candidate. They did not need cell cultures or egg-based production or years of iterative chemistry… They took the viral sequence, identified the spike protein, wrote the corresponding mRNA instruction, and had a candidate ready for manufacturing. The first clinical batch was shipped to the National Institutes of Health for testing on February 24, 42 days (!) after the sequence was published. Those in drug discovery/biotech know that it is almost instantaneous, compared to a typical multiyear timeline.

That speed had a decade-long backstory, I’d like to revisit.

Katalin Karikó, a Hungarian-born biochemist, had been trying to make mRNA work as a therapeutic tool since the late 1980s. For most of that time, nobody cared. At the University of Pennsylvania in the 1990s, she applied for grant after grant and was rejected. In 1995, the university gave her a choice: stop working on mRNA or accept a demotion. She was being treated for cancer at the time, and she took the demotion.

What kept her going was a serious problem with the mRNAs: the body’s immune system treated synthetic mRNA as a foreign invader — it triggered inflammation and destroyed the molecule before it could deliver any instructions. Together with her collaborator Drew Weissman, an immunologist she met at a photocopier in Penn’s medical school, Karikó spent years testing chemical modifications to RNA’s building blocks. In 2005, they found one that worked: swapping uridine for pseudouridine let synthetic mRNA slip past the immune system undetected.

It was the foundational insight behind every mRNA vaccine that would follow.

Ironically, no major journal wanted the paper, and after it was published, almost no one cited it, and no wave of funding materialized. In 2013, Karikó left Penn, she said the university told her she was “not of faculty quality” — and joined a small German startup called BioNTech. In 2023, she and Weissman received the Nobel Prize in Physiology or Medicine. Every mRNA vaccine injected during the pandemic traces back to that chemical modification…

Anyway, what followed the rapid introduction of the mRNA vaccine during the COVID-19 pandemic was, perhaps, the most compressed wealth-creation event in pharmaceutical history. Moderna's revenue went from $803 million in 2020 to $18.5 billion in 2021 and $19.3 billion in 2022, almost entirely from a single product. Its market capitalization briefly approached $200 billion. A company that had never sold a drug before the pandemic was suddenly generating more revenue than some century-old pharmaceutical giants!

BioNTech, Moderna's German counterpart, followed a parallel trajectory. In partnership with Pfizer, it developed Comirnaty, the other mRNA COVID vaccine, and briefly became one of Europe's most valuable biotech companies.

And then the wave broke.

As vaccination rates declined and the pandemic shifted from emergency to endemic, the revenue collapsed. Moderna’s annual revenue fell from $19.3 billion in 2022 to around $6.8 billion in 2023, then to roughly $3 billion in 2024. Its market cap dropped from nearly $200 billion at peak to under $15 billion at its lowest. BioNTech followed a similar descent. Pfizer launched a $3.5 billion cost-cutting program, closed facilities, and laid off thousands.

During the boom, investors likely priced these companies as platforms, the idea being that the same technology that produced a COVID vaccine in weeks could produce vaccines and treatments for cancer, rare diseases, autoimmune conditions, and essentially anything else. The platform thesis was that it was not a one-product company, rather, a new paradigm for making medicine.

When the COVID revenue disappeared, however, the market perception changed, expecting product and demand beyond the initial vaccine success, which has not followed the way the market anticipated. That question remains open, and the answer will determine whether the mRNA boom was a one-time windfall or the beginning of something much larger.

A (potentially) bigger play

These days, Moderna and Merck are developing a personalized cancer vaccine called intismeran autogene (also known as mRNA-4157 or V940). The concept is striking in its ambition: a doctor takes a biopsy of a patient’s tumor, sequences it to identify the unique mutations driving a specific cancer, and Moderna designs a custom mRNA instruction set, encoding up to 34 of the patient’s tumor’s distinctive markers, which is then manufactured and injected to train their immune system to recognize and attack those exact cells.

In January 2026, the companies reported five-year follow-up data from the Phase 2b KEYNOTE-942 trial. The combination of the personalized vaccine with pembrolizumab (Keytruda, a leading immunotherapy drug) reduced the risk of melanoma recurrence or death by 49% compared to pembrolizumab alone. The benefit held steady from the three-year mark to the five-year mark, which is a signal that the immune memory triggered by the vaccine may be durable.

These are promising results, of course. But the power story is in the new paradigm of manufacturing and the economic system it implies.

Traditional pharmaceutical economics are built on mass production.

You develop one molecule, prove it works, manufacture millions (billions) of identical doses, and distribute them globally. Economies of scale are the foundation of the entire business model, and that’s how a pill can cost pennies to produce.

A personalized cancer vaccine inverts this logic entirely. Each dose is designed for one patient, so manufacturing is bespoke.

The production timeline is measured in weeks, not years, but the production run is measured in single doses, rather than millions. If the Phase 3 trials (now underway in melanoma, lung cancer, bladder cancer, and kidney cancer, with over a thousand patients enrolled) confirm the Phase 2 results, the question would not be limited to “does it work?” It effectively shifts to “who builds the manufacturing infrastructure for millions of custom products per year?”

How do you price a treatment that is, by definition, impossible to genericize? And can health systems designed for mass-market pills absorb something this fundamentally different?

Early estimates suggest personalized cancer vaccines could be priced around $200,000 per patient. Moderna and Merck have said that if the Phase 3 melanoma data are positive (readouts are expected as early as 2026), they could seek approval and launch as early as 2027. If that happens, it will create a new category of manufacturing that has no real precedent in the pharmaceutical industry.

The RNA molecule is only half the story of the mRNA revolution

Now here’s the part of the RNA story that will determine who actually controls this platform. The thing is, RNA molecules, on their own, are useless as therapeutics or vaccines.

They’re fragile, easily destroyed by enzymes in the bloodstream, and can’t cross cell membranes on their own. If you inject naked mRNA into someone’s arm, it degrades within minutes and accomplishes nothing.

The solution that made the entire mRNA vaccine era possible is the lipid nanoparticles, or LNPs. These are tiny bubbles made from carefully formulated mixtures of fat-like molecules that encapsulate the RNA, protect it from degradation, and ferry it into cells.

The groundwork was laid over four decades ago by Pieter Cullis, a physicist-turned-biochemist at the University of British Columbia, whose lab pioneered methods for packaging delicate molecules inside lipid carriers and delivering them into cells intact.

Without LNPs, there are no mRNA vaccines, no mRNA cancer therapies.

Arbutus Biopharma and its affiliate Genevant Sciences hold foundational LNP patents. For years, they pursued Moderna through courts worldwide, filing infringement lawsuits across five international jurisdictions, targeting 30 countries, with a separate US case heading toward a jury trial in Delaware. Then, in March 2026, six days before that trial was scheduled to begin, the parties announced a historic $2.25 billion global settlement. Moderna will pay $950 million upfront in July 2026, with an additional $1.3 billion contingent on an appellate ruling. If the full amount is paid, it will be the largest disclosed patent settlement in pharmaceutical history.

That number tells you everything about what the delivery system is worth. The public debate during the pandemic was about vaccine safety and mandates. The private fight, the one that just produced a $2.25 billion settlement, was about who owns the molecular packaging.

And it’s not over: the litigation against Pfizer/BioNTech’s Comirnaty is still active, with a favorable claim-construction ruling for Genevant/Arbutus issued in September 2025. Comirnaty represents roughly two-thirds of global COVID mRNA vaccine sales to date.

The mRNA sequence is easy to design. Without the delivery technology, it’s just information with no way to become medicine. The chokepoint of the entire mRNA platform is the delivery. That chokepoint has geographic consequences…

During COVID, mRNA manufacturing was concentrated almost entirely in the US and Western Europe. China and India, dominant in generic drugs and conventional vaccines, were largely shut out.

Both are now catching up, albeit slowly, likely with a long-term focus in view. China approved its first domestic mRNA vaccine in 2023, but with the end of zero-COVID, its factories sit idle, and demand has evaporated. India's Gennova Biopharmaceuticals developed an indigenous mRNA vaccine and is now working on a self-amplifying RNA platform for Nipah virus with CEPI funding, but it's still preclinical.

In Africa, BioNTech has invested roughly $150 million in a modular manufacturing facility in Kigali, Rwanda. Moderna announced a $500 million plant in Kenya, though the project has stalled. The WHO set up a technology transfer hub at Afrigen Biologics in Cape Town, which has developed its own mRNA vaccine candidate entirely on the continent. Real money, real facilities, but BioNTech's Kigali plant runs on BioNTech's proprietary technology and will manufacture BioNTech's products. A factory is not the same as an R&D platform…

The capacity to produce mRNA medicine is slowly spreading across the globe; however, the capacity to design it, control it, and decide what gets made remains concentrated in a handful of Western companies… for now.

The trust factor

The speed of mRNA vaccine development was arguably the most remarkable technical achievement in the history of pharmaceuticals. From a published viral sequence to a manufactured vaccine candidate in a matter of days. From trial to emergency use authorization in under a year. These timelines were previously thought to be unlikely.

For a significant portion of the public, that speed was precisely the reason not to trust it, however…

Ironically, mRNA’s greatest scientific strength — programmability and speed of candidate design — became its greatest political vulnerability.

Regulatory systems designed for decade-long drug development timelines were asked to validate something built in months. Public health communication infrastructure, designed for slow-moving campaigns about established interventions, was asked to explain molecular biology to billions of people simultaneously while a crisis was unfolding. The gap between what the technology could do and what institutions could credibly explain was enormous. The institutions failed to manage the mRNA technology perception crisis because they were designed for a slower world.

The result is a trust deficit that now functions as a perception tax on the entire RNA platform. The tax is even quantifiable. A study of over 35,000 adults across nine countries found that the novelty of mRNA technology alone reduced the odds of vaccine acceptance by 14.2% compared to conventional vaccines — same disease, same efficacy data, different platform, measurably lower trust.

A 2023 global survey found that nearly half of respondents trust traditional protein-based vaccines more than mRNA vaccines. And the spillover extends beyond mRNA: MMR vaccination coverage among US kindergartners dropped from 95.2% to 92.7% between 2019 and 2024, leaving over 280,000 additional children unvaccinated and contributing to the highest US measles count since the disease was declared eliminated in 2000.

A KFF survey found parental skepticism toward vaccines had risen from 22% to 27% in a single year, with perceived politicization as the primary concern.

In 2021, Moderna hired its first-ever chief brand officer from Ogilvy Health, launched the global "Welcome to the mRNAge" campaign in April 2023, explicitly to counter vaccine hesitancy by explaining the science behind the platform.

Moderna's chief brand officer Kate Cronin told Marketing Week in 2023: "We lost relevance" — a remarkable admission from a company that had been, three years earlier, one of the most recognized names on Earth. In 2024, CEO Stéphane Bancel personally took over marketing during a leadership transition.

Zooming out of Moderna’s post-pandemic roller coaster, it seems like every future RNA therapeutic, including personalized cancer vaccines, cholesterol drugs, rare disease treatments, etc, might be introduced into a public discourse shaped by the COVID-era polarization. This is not merely a communications problem that can be solved with better messaging, but more like a structural cost baked into the economics and politics of mRNA products for the foreseeable future.

What comes next

The bigger picture emerging from this story is that mRNA is not just the COVID vaccine. It is a broad programmable platform for writing temporary instructions to the human body, potentially for various therapeutic applications, but also a quick instrument to prepare for future pandemics, and the race over who controls it, who profits from it, and who gets access to it is just beginning.

Here’s what to watch.

Cancer treatment may become personalized manufacturing. If the Moderna/Merck trials deliver, a patient’s tumor gets sequenced, and a bespoke vaccine is synthesized in weeks. That would go way beyond an incremental improvement on existing oncology; it would be a different category of medicine — one patient, one product — and no health system on Earth is currently built to pay for it or deliver it at scale.

Pandemic response may shift from years to days. The mRNA platform can generate a vaccine candidate within days/weeks of receiving a pathogen’s genetic sequence. We saw this with COVID. The constraint is not so much the technology, but more so the operational and manufacturing readiness, as well as public perception management.

Diseases that were never economically viable to target suddenly become feasible. Designing a new mRNA sequence costs almost nothing — it’s computational. The expensive part is manufacturing and clinical trials. That means rare genetic conditions, tropical diseases, pathogens that primarily affect poor countries — categories that traditional pharma has largely ignored because the market was too small — could become viable if the funding and manufacturing bottlenecks are solved.

The question of who controls the platform will shape who has access. The $2.25 billion LNP settlement, the idle Chinese factories, BioNTech’s Kigali facility running on proprietary technology, the stalled Moderna plant in Kenya — these are early moves in a long contest over whether mRNA medicine becomes broadly accessible or remains concentrated in the hands of a few Western companies and the nations they operate in.

None of this is guaranteed. The platform could stall on failed clinical trials, regulatory gridlock, or a trust deficit that proves deeper than anyone expected. The cancer vaccine data could disappoint. The geopolitical competition could fragment the platform rather than expand it.

But the underlying RNA tech capability is real, it is proven, and it is not going away. A molecular class that may have started life on Earth four billion years ago is now being reprogrammed in laboratories to address some of the hardest problems in medicine. Science is moving at the speed of information while everything else — the money, the politics, the institutions, the trust — is still catching up.

But the entire paradigm was essentially a copilot.

You chat with a tool, it makes things faster, you get more done (or not). Real limitations were everywhere. Presentation generators were not good enough. Code copilots were messy to the point where extracting value from them was itself a job. The trend was clear — people with AI could arguably do things better than people without it — but it felt incremental. An upgrade, not a transformation.

Then, in January 2026, I tried Claude’s tools, based on the Opus 4.6 model, and some agentic features. And after playing with it for a week or so, I realized things genuinely changed. And how fast, if you think about it!

It was no longer about chatting with AI. It was a full-scale workflow — files, deliverables, projects — and on top of that, you could create agents to put it on autopilot. They could work with your browser, your file system, your email, Notion, external tools, etc., etc. And the hallucination rate, in my experience, had dropped visibly, to the point where the work was actually getting done in a productive manner.

Even , a vocal critic of LLMs and things like ChatGPT, published an unusually optimistic piece on the neurosymbolic nature of Claude systems and how it changes everything: The biggest advance in AI since the LLM

I kept discussing this with peers — software developers, content creators, media professionals, and finance people. Everyone is excited and surprised, but above the excitement, people feel concerns about how good AI automation has suddenly become. In the early ChatGPT era, people talked about opportunities to make their jobs easier. Now the tone has clearly shifted to “how do I not lose my job in the first place?”

This essay is about what happens when that shift plays out across the economy.

How to Run a Drug Discovery Project for ‘the Price of Lunch’

I recently came across a preprint out of Stanford, from Dr. James Zou's lab, and it's one of those papers that makes you stop and stare.

The team ran a complete drug target evaluation with eleven AI agents organized to act autonomously like an entire biotech company — virtual Chief Scientific Officer (CSO) at the top, specialist divisions below, over a hundred connectors to real biomedical databases, and a reviewer agent that checks reasoning. Agents coordinate all work between themselves, automatically and to a degree — recursively.

The CSO agent receives a query, decomposes it into sub-tasks, and delegates to specialist agents. The specialist agents use their tools, produce outputs, and pass them back. The reviewer agent checks reasoning. The CSO synthesizes and communicates the result.

The kind of analysis that, inside a pharmaceutical company, would typically involve multiple specialist teams coordinating over weeks or months.

One of the cases: the system evaluated B7-H3 as a therapeutic target in lung cancer. It found — on its own — that B7-H3 upregulation was concentrated in cancer-associated fibroblasts, not the cancer cells themselves, and that these fibroblasts were running an immunosuppressive program. It validated this with spatial transcriptomics. It recommended an antibody-drug conjugate.

For context, its training data stopped in January 2025. In August 2025, the FDA granted Breakthrough Therapy Designation to exactly that kind of drug, with a 48% response rate. The system arrived at the same logic independently.

For $46 in Anthropic API credits, in less than a day.

In another case, it autopsied a failed clinical trial and concluded the biology was probably right but the wrong patients were enrolled. That cost $54.

In the largest demonstration, the system spun up over 37,000 agents in parallel to curate outcome data from nearly 56,000 clinical trials — and found that drugs targeting cell-type-specific genes were 40% more likely to advance from Phase I to Phase II, with roughly a third fewer adverse events.

These are not chatbot conversations. This is the actual analytical work, including hypothesis generation, evidence synthesis, and cross-disciplinary reasoning, that scientists spend years learning to do.

But here is the bigger thing. The $46 did not simply replace one scientist or a specific function. It replaced the need for the whole arrangement — the teams, the handoffs, the reviews, the months of coordination. The work got done, and the organizational structure that existed to support the work became unnecessary in this particular case.

Yes, sure, it is just a proof-of-concept case study, a lab experiment, real-world deployment would be way more expensive and complex, and perhaps it will still require humans-in-the-loop. But this is a hint of where things are going, quite rapidly.

The coordination tax

Until recently, the mainstream framing was that AI replaces tasks. That is true of course, but with the rapid emergence of agentic AI, what it actually impacts is something I think of as the coordination tax — and it operates at every level of knowledge work, not just at the management level.

You see, every knowledge worker is a coordination node, if you will. The drug discovery scientist synthesizes biological and chemical data, clinical literature, and cell biology into a target assessment. The marketing analyst pulls customer insights, campaign metrics, and competitive data into a recommendation. The compliance associate reads contracts against regulatory checklists. A software developer is building a module in the enterprise software, against the tech requirements and specific context of a bigger system. Even though they are not “managers” per se, all of them are involved in pulling disparate information together, translating it across contexts, and producing something coherent. That is coordination. And the overhead of maintaining a network of humans doing it — the meetings, the reporting, the institutional memory, the sheer effort of keeping everything connected — is the coordination tax, or an operations overhead, or whatever you call it. It is often huge. Administrative expenses account for nearly 25% of health care spending in the US, for example.

For decades, the coordination tax was unavoidable. You could not get specialized knowledge synthesized without humans synthesizing it. Agentic AI threatens, for the first time in the history of human labor, to change that dynamic. It goes far beyond just helping workers do their specific functions faster. After all, we have had productivity tools since the spreadsheet. But now agents promise to not only do functions, but also to do coordination between functions automatically, at every node, without a human in the loop. Or, rather, with substantially fewer humans in the loop.

Goldman Sachs grasped this early. They embedded Anthropic engineers for six months to build agents for trade accounting and client onboarding. The CIO Marco Argenti described the agents as "digital co-workers" for complex, process-intensive professions.

Note the word: professions.

Not just functions. The actual professions — the people who do compliance, who do accounting, and client relations. Meanwhile, the CEO David Solomon announced a multiyear plan to reorganize the bank around generative AI, with a stated goal of “constrain headcount growth.”

But here is where the story gets more interesting than the usual “AI takes jobs” headline.

Mastercard launched what they call a Virtual C-Suite — an AI agent that acts as a digital CFO for small businesses, analyzing cash flow, flagging risks, and recommending actions based on 175 billion annual transactions. Most small businesses never had anyone doing this work. They couldn't afford a CFO, so financial strategy was whatever the owner could figure out between everything else. Now it is a subscription. For millions of small businesses, agents could be delivering expertise that simply didn't exist before — not replacing a human, but filling a gap no human ever occupied.

In healthcare, both sides of the tension are visible simultaneously.

A multi-agent system for rare disease diagnosis published in Nature, DeepRare, was reported to outperform experienced physicians across 6,401 cases. For the 300 million people enduring diagnostic odysseys averaging five years, that is transformative. In a broader context, administrative costs eat up a quarter of US healthcare spending.

At the same time, only 3% of healthcare organizations have deployed agents in live clinical workflows, according to one study.

But whatever it is, the AI capability is accelerating, no question about it. At Shanghai Jiao Tong University, for instance, ASI-Evolve — a system that reads papers, forms hypotheses, designs experiments, runs them, and uses the results to improve its own architecture — produced 105 AI architectures that beat the best human-designed AI.

Yes, you heard it right, AI making AI better. My brain pictures scenes from the “I, Robot“ movie, which I watched for the first time like a decade ago, and now it seems like at least part of it is becoming reality (progress in robotics/drones + AI + agentic systems).

Anyway, then they pointed ASI-Evolve at drug-target prediction, and it arguably outperformed human experts (this claim should be treated with caution, and in the context, though, I would treat it as a theoretical proof of concept, not a deployment-grade thing. But the trend is there…).

“Looking ahead, the scope of AI self-acceleration extends beyond individual models to the full AI development stack—architecture, data, algorithms, and infrastructure yet to be explored. As agentic systems take on more of the 14 References SII-GAIR implementation and iteration work, human scientists can shift from being the executors of solutions to the definers of problems—concentrating their expertise on the questions that matter most and leaving the expansive search through hypothesis spaces to AI. We expect this paradigm to drive not only the self-improvement of individual models, but the self-evolution of the entire AI field,” the authors conclude.

The recursive loop is gradually closing.

The labor displacement could already be real

In healthcare, the agent is already arriving with a name and a face. Hippocratic AI built “Ana” — an autonomous AI agent that calls patients, prepares them for appointments, answers medical questions, and operates around the clock in multiple languages. The company initially marketed it at $9 an hour, compared with $40 for a registered nurse. Hundreds of US hospitals now use AI systems that go beyond monitoring vitals — they trigger step-by-step care action plans, directing what nurses should do before the nurse has evaluated the patient.

Michelle Mahon of National Nurses United put it this way: “The entire ecosystem is designed to automate, de-skill, and ultimately replace caregivers.”

Meanwhile, only 3% of healthcare organizations have deployed agents in live clinical workflows. The capability is racing ahead of the institutional readiness to use it safely — and the people whose judgment built the profession’s trust are being told the machine knows better…

While I am not a big fan of alarmist predictions, pretty much every report I have reviewed during the literature search for this article tells one thing: the job market is changing rapidly.

However, the displacement picture is more nuanced than either the hype or the panic suggests. BCG's microeconomic modeling estimates that over the next two to three years, 50% to 55% of US jobs will be reshaped by AI — but reshaping is not replacing.

Only 10% to 15% of jobs face actual elimination in a five-year horizon.

The more interesting finding is where the impact concentrates. Jobs with 40% or more automatable tasks, about 43% of all US roles, are the ones that trigger organizational redesign.

And the pattern is not uniform: call center representatives get substituted because their demand is bounded, while software engineers get amplified because lower costs unlock more demand for what they produce. The most structurally dangerous category is what BCG calls "divergent" roles (about 12% of jobs), where entry-level positions get automated while senior roles persist or grow.

Insurance sales assistants disappear, while senior advisory roles expand. IT support technicians handling routine tickets disappear while systems oversight roles grow… etc.

The career ladder doesn't just lose rungs at the bottom, it loses the bottom entirely. And the skills required for the remaining roles are higher, not lower — more judgment, more oversight, more cognitive intensity.

The people who thrive will be those who can grade the agents' work, not those who compete with it.

Now, here is what I find genuinely alarming about job market prospects: CEOs are not waiting for the agents and infrastructure to be ready. They are not waiting for the governance to be built. They are not even waiting for agents to be reliable. The displacement does not require AI and agents to be perfect — all it needs is that decision-makers believe they are good enough. I think with models like Opus 4.6 and beyond, Claude Cowork/Claude Code, and all sorts of other tools, that bar has been effectively cleared.

A game theory paper from UPenn and Boston University formalizes what this means at scale. AI-driven automation, they show, is a Prisoner’s Dilemma operating across the entire economy. Every company that automates gains a cost advantage. Every company that does not — gets outrun by competitors that did. But every displaced worker is also a consumer who stops buying. The rational choice for each individual CEO is to automate, but it becomes collectively suicidal when everyone makes it simultaneously.

The researchers tested UBI and profit taxes — neither resolves the coordination failure in their model. The only mechanism that could work is a Pigouvian automation tax that prices the downstream damage into the automation decision itself.

Furthermore, I think the dilemma does not stop at the level of firms, it probably works between nations, too.

In Europe, the EU AI Act is the most ambitious regulatory attempt in the world, and as a result, the continent’s AI companies continue to accept risks of falling behind American and Asian competitors (especially China), with talent migrating to less restrictive jurisdictions. China is running the opposite experiment: Tencent is wiring agents into WeChat for hundreds of millions of users, government-backed infrastructure at a scale that is hard to match even by Western democracies, especially considering the governance model in China arguably leans toward state capacity over individual worker protection. In the US, activists stalled $98 billion in data center projects in a single quarter while the administration pushes AI as geopolitical necessity. In just one example of many, Muscogee Nation citizens blocked a hyperscale data center on tribal land purchased for agriculture, a representative saying: “This feels like a modern-day land run.”

Every jurisdiction is making the same choice under different constraints:

1) Accelerate and absorb the social cost.

2) Regulate and absorb the competitive cost.

3) Or find a third path that nobody has demonstrated works.

The Luddite fallacy — that technology always creates more jobs than it destroys — was true in the long run, but the long run could be 60 to 80 years, via painful disruptions. Also, previous technologies augmented workers at their coordination nodes, where various functions still needed an operator. This time, with AI agents, things are different: agents are the autonomous operators. Will the Luddite fallacy hold?

Flying blind

But out of all the contradictions that we’ve just reviewed, the most ironic thing is that agents, with all the mind-blowing capabilities, are still half-baked for massive public and corporate use at scale...

A practitioner, Mark Kropf, founder & CEO of Rensei (enterprise OS for AI agent fleets), running autonomous coding agents in production, described 19 releases in 14 days, none of which made agents “smarter”. All made the system less likely to catch fire.

His real incidents: agents in infinite retry loops burning $200 an hour, workers crashing mid-task with no reassignment, agents editing each other’s files in a monorepo, one agent’s API calls rate-limiting the entire fleet. His prediction: the AI agent hype is going to hit a wall in 2026, because nobody solves infrastructure issues.

Neha Deodhar from Google wrote a great technical summary of why agents are still breaking in production, unlike “classical“ software.

Challenges with agents are as deep as their capabilities are impressive.

For instance, 82% of business leaders in a recent Celonis survey believe AI will fail to deliver ROI if it does not understand how the business actually runs. The context that makes the work legible disappears with the people who carried it, being substituted by agentic-driven automation.

The knowledge base agents consume is itself corruptible.

Researchers recently invented a fake disease, bixonimania, published preprints with fictional authors, thanked “Professor Sideshow Bob” in the acknowledgements, and wrote “this entire paper is made up” in the methods.

Four major AI platforms recommended ophthalmology visits for a condition that does not exist. Then real researchers cited the fake papers in a real journal. The AI treats formatting as credibility. The humans never read what they cited. Both systems, algorithmic and institutional, seem to reward volume over verification…

And when AI is used not just to consume knowledge but to substitute for it, the results are worse.

The FDA recently issued a warning letter to a pharmaceutical manufacturer that had used AI to create its drug specifications, manufacturing procedures, and production records.

When investigators asked why the company had not performed required process validation, officials replied that they “were not aware” of the legal requirement because AI had not informed them.

The facility had insects, dirt, and an open dock door exposing the manufacturing area to the outside. A company making real drugs, sold to real people, that outsourced its understanding of its own regulatory obligations to a chatbot. Nobody checked whether the outputs were accurate or even legal. An attorney who reviewed the case put it simply: no one at the company knew enough to recognize what AI was telling them was wrong.

Next, agentic AI cybersecurity is the wild west…

McKinsey’s Lilli AI platform was breached by ethical hackers who gained full access to 100,000 documents used by 30,000 consultants. Over 1,800 OpenClaw instances were found leaking API keys and credentials.

Then there is the pricing question.

It is known that the current wave of AI and agent adoption runs on subsidized compute — AI labs pricing tokens below cost to capture market share.

Users are already reporting that a single message on some platforms consumes a meaningful percentage of their plan allocation. As the honeymoon ends and consumption-based pricing takes hold, the startups built on cheap inference may discover their unit economics don’t survive.

The enterprises that restructured around agents may find the cost merely shifted from salaries to compute bills. Who bears that cost, and who profits from it, is a power question the industry has barely begun to ask…

Finally, there is one limitation of modern AI that no infrastructure or pricing model can fix, at least now.

ran what he calls a “Time Machine” experiment — sending today’s best AI models back to 2012 to make one of the biggest strategic calls in modern oncology. Two companies were racing to bring cancer immunotherapy to market. Bristol-Myers Squibb had a two-year head start. Both faced the same fork: run a broad trial enrolling all patients, or restrict enrollment to the subset most likely to respond based on a promising but unproven biomarker. BMS chose broad. Merck chose narrow. The broad trial failed. The narrow one succeeded so overwhelmingly it was stopped early. By 2024: Merck's drug generating $29.5 billion a year. BMS's: $9.3 billion.

Chang assembled AI agent teams and gave them only the information available before June 2012. Both teams recommended the broad path, unanimously. The path that lost. The competitive intelligence agent identified the exact scenario that materialized — and still voted broad.

I should be honest about a critique I have heard from several people in the industry: the models were not truly time-locked. They were trained on post-2012 text discussing how the story unfolded. But that actually makes the point stronger. Even with hindsight baked into its training data, the AI converged on the defensible, consensus choice. It could not make the contrarian bet that won.

“AI can give you the best possible analysis. It can’t give you the courage to go against it,” says Chang.

Organizations might be handing strategic authority to consensus machines while letting go of the people whose judgment — imperfect, intuitive, sometimes irrational — was the only thing that could override consensus when consensus was wrong. That is not a bug to be patched, it is the nature of the AI systems.

What’s ahead?

's Superagency argues that AI is the ultimate human amplifier — not a replacer but an extension of human capability, a superpower distributed to millions simultaneously.

It is a compelling and optimistic thesis, I’d like to believe it.

However, the question this essay raises is whether the institutions, the incentive structures, and the career pipelines that turn amplified individuals into a functioning economy can survive the transition.

After all, the "amplifier" framing assumes the benefits distribute. The game theory we discussed above, however, shows they concentrate.

Hoffman himself acknowledges this — he cites Ted Chiang's critique that "the economic value created by the personal computer and the internet has mostly served to increase the wealth of the top one percent of the top one percent" and says the case for AI rests on breaking that pattern.

My own opinion about the promise of an agentic future is still evolving. However, what concerns me is that the emerging agent economy seems to be running on a resource it is actively destroying.

Every institution deploying agents depends on human judgment to direct them, evaluate their output, and override them when they are wrong. That judgment was built over years inside the coordination layer — the junior roles, apprenticeship, the institutional memory, the domain knowledge accumulated through experience. The same coordination layer that agents are arguably starting to eliminate at scale.

We are simultaneously increasing the demand for human judgment and dismantling the system that develops it, which I think is the central paradox of the agent economy.

And it belongs to everyone — the CEO deciding how many people to keep, the government choosing how to regulate, the founder deciding what to build, and the twenty-year-old student wondering whether the career she is studying for will even exist by the time she is ready for it…

The twentieth century built its prosperity on a sort of bargain: invest in human talent and coordination, and the growth compounds. The agent economy breaks that bargain, at a scale unknown to history.

What it builds in its place might be the defining question of the next decade…

Thanks for reading!

— Andrii

You have likely read about Rosie the dog by now. If you have not, a Sydney engineer used ChatGPT and AlphaFold to design a personalized mRNA cancer vaccine for his dying dog, and the tumor shrank. The story went hyper popular, and it generally divided commentators into two camps. One says a guy with AI just did what pharma spends billions on. The other says this is one dog, no controls, no peer review, meaningless noise that does not deserve the attention of this magnitude.

I want to suggest that the real matter is different.

The question is not whether Rosie’s vaccine works — clinically, an uncontrolled N-of-1 in a dog tells us almost nothing.

And it is not whether biohackers with AI can “replace” pharma researcb — they obviously can not.

The question is what happens when the design of a personalized therapy, the part that used to require institutional teams and years and millions of dollars, becomes something a technically skilled and well-connected outsider can do with publicly available AI tools, while everything after the design remains locked behind the same institutional gates as before.

That gap between what one can now figure out and what one is allowed to do about it is the story. And it is bigger than one dog’s fate…

What Actually Happened

Let me lay out the facts, because the viral versions are overstating things.

In 2024, Paul Conyngham’s rescue dog Rosie was diagnosed with aggressive mast cell cancer. Surgery and chemotherapy failed, and vets basically gave her months. Conyngham has no background in biology, but the viral framing of a random entrepreneur and pet owner stumbling into science is misleading. He is a computing engineer with 17 years in machine learning, a co-founder of a data consultancy, and a former director of Australia’s Data Science and AI Association.

He paid about $3,000 for genomic sequencing of Rosie’s tumor at the University of New South Wales (UNSW), arguably used ChatGPT to navigate bioinformatics pipelines he had never worked with, ran the data through AlphaFold to model mutated proteins, and produced a half-page mRNA vaccine formula targeting Rosie’s specific mutations.

Martin Smith, the computational biologist who directs UNSW’s Ramaciotti Centre for Genomics, described himself as “gobsmacked” when he saw what Conyngham had done with the data. This kind of analysis pipeline would normally require an institutional team with a specialized biological background.

Then the wall.

Conyngham identified an existing immunotherapy matching Rosie’s cancer. The manufacturer refused to supply it for a dog. He pivoted: Pall Thordarson, director of UNSW’s RNA Institute, synthesized a bespoke mRNA vaccine from Conyngham’s formula in under two months. But regulatory approval to inject it took three more months, which was longer than designing the vaccine in the first place. Rosie got her first shot in December 2025.

The largest tumor shrank dramatically, by half to three-quarters, depending on the measurement. Importantly, and unlike what many media stories tell, she is NOT cured, the tumour did not go away completely. Moreover, another tumor has not responded.

That is the story. Accurate, interesting, and from a clinical standpoint, almost meaningless. Here is why.

mRNA cancer vaccines have been in development for over a decade, and the clinical record is sobering.

CureVac’s prostate cancer vaccine missed its survival endpoint in 2017. Argos Therapeutics’ Phase III trial in renal cell carcinoma was stopped for futility. BioNTech and Genentech’s personalized cancer vaccine missed efficacy signals in advanced solid tumors as recently as March 2025. Gritstone’s GRANITE missed its endpoint in colorectal cancer in 2024... and so on.

The consistent pattern: mRNA vaccines can provoke immune responses, but translating that into actual survival benefit in advanced cancer has been brutally hard. BioNTech’s chief medical officer, Özlem Türeci, acknowledged the core problem — the six to eight weeks needed to build a T-cell response is too slow to control rapidly growing tumors.

One dog’s tumor shrinking, with no controls and no peer review, does not change this record.

Professionals in oncology know that partial responses happen, spontaneous regressions happen, and the graveyard of promising N-of-1 results is vast.

The clinical skeptics are right in this case of a highly hyped anecdotal case.

Now, I should say that mRNA as a platform, the ability to write biological instructions as code and deliver them in lipid nanoparticles, is arguably the most important therapeutic technology to emerge in a generation.

COVID proved it could be manufactured at a population scale in an unprecedented time. The cancer vaccine applications are still early and struggling, but the platform itself is reshaping how we think about programmable medicine, from rare genetic diseases to personalized oncology to pandemic preparedness.

That story about who controls mRNA manufacturing, how the platform is being regulated, and where the next generation of applications will come from is its own essay, and I am already drafting it. I will publish in the coming weeks.

Subscribe now

For now, the point is narrower: the specific application that went viral this week sits on the weakest branch of a very large tree.

What makes the story structurally important is something different: Conyngham compressed a bioinformatics pipeline that previously constituted one of pharma’s core competitive advantages into a solo project using tools available to anyone with an internet connection, his connections, and drive. Basically, a personal biohacking project.

It points to a possibility that the design layer of drug development is starting to leak out of the institution. But when he tried to act on what he’d designed, he hit a wall that had nothing to do with the quality of his science.

The Wall Is Older Than the Modern AI Trend in Bio

More than a decade before Conyngham opened ChatGPT, a medical student named David Fajgenbaum was dying at the University of Pennsylvania.

In 2010, during his third year, he was diagnosed with a rare subtype of Castleman disease — a disorder that triggered a catastrophic immune reaction, causing his liver, kidneys, and bone marrow to fail simultaneously. Over the next several years, he nearly died five times, cycling through rounds of chemotherapy that kept him alive but couldn’t stop the relapses. His doctors, some of the best in the country, told him they had nothing left.

What Fajgenbaum did next is worth sitting with, because it contains the structural seed of everything happening now. Between hospitalizations, he started running experiments on his own blood samples. He combed PubMed, reading hundreds of papers by hand. No AI, no AlphaFold, no generative models at the time were available to him. Just a desperately sick researcher with access to a medical library and enough training to understand what he was reading.

He found sirolimus, an immunosuppressant routinely prescribed to prevent kidney transplant rejection. It had been on the market for years, and it was cheap. It also had extensive safety data. And no doctor treating Castleman disease would have thought to prescribe it, because it was categorized as a transplant drug, not an oncology drug.

Long story short, Fajgenbaum took it, his disease went into remission, and it has now been in remission for over a decade.

This is the fact that reframes the entire Rosie the dog conversation: the wall between what could save a patient and what the system actually delivers to them existed long before AI touched drug development. Fajgenbaum’s drug already existed, it was already manufactured, and it was already proven safe. The only thing missing was someone willing to connect it to his disease, and the system, organized around disease-specific treatment algorithms, couldn’t do it. He had to do it himself while facing a sobering diagnosis.

He later co-founded a nonprofit called Every Cure, using machine learning to systematically screen thousands of approved drugs against thousands of diseases, industrializing the act of desperation that saved his life.

But the wall that made it necessary has not moved much. Drugs still get stuck inside regulatory categories, and patients still fall between the gaps in treatment algorithms. The system still optimizes for population-scale evidence, and individuals with rare or refractory diseases still get told: we’re out of options, good luck.

AI makes the design layer faster, and there is some evidence of that, but it does not make the wall meaningfully thinner for an average individual.

Share

What It Actually Costs to Break Through the Wall

If you want to understand what it looks like when someone throws everything they have at that wall, look at Sid Sijbrandij’s truly remarkable story of biohacking his way out of cancer.

Sijbrandij co-founded GitLab, a publicly traded software company with over 2,000 employees that pioneered fully remote work at enterprise scale. In November 2022, he was diagnosed with osteosarcoma — a six-centimeter tumor growing from his spine. He went through surgery, radiation, and chemotherapy so intense that it required four blood transfusions. The cancer went into remission. Then it came back in 2024. His medical team told him, essentially, that the standard of care was exhausted.

What happened next is the most resource-intensive case of a patient taking over their own treatment that I’m aware of.

Sijbrandij deployed every frontier diagnostic available, including single-cell sequencing, bulk genomic analysis, multiple types of minimal residual disease testing, organoid models grown from his own cancer cells. He hired a full-time director of his care and assembled clinical and scientific advisory boards. He filed five separate expanded access applications with the FDA to obtain experimental drugs with no available clinical trials — each was approved, arguably, within 48 hours.

The Century of Biology

Going Founder Mode On Cancer

Welcome to The Century of Biology! This newsletter explores data, companies, and ideas from the frontier of biology. You can subscribe for free to have the next post delivered to your inbox…

Read more

4 months ago · 300 likes · 35 comments · Elliot Hershberg

Then he flew to Germany for an experimental radioligand therapy targeting a protein called FAP that his sequencing had identified as overexpressed in his cancer cells. The treatment used Lutetium-177, the same radioactive payload in Novartis’s Pluvicto, delivered by a targeting ligand that had first been validated by diagnostic imaging. The tumor shrank enough to become operable again. Post-surgical analysis showed T-cell infiltration had surged from 19 to 89 percent. His cancer is now undetectable.

Here is the uncomfortable part, at least, it is what Sijbrandij describes based on his experience.

At every stage, Sijbrandij encountered institutional resistance that had nothing to do with the science. Hospitals wouldn’t release his own tissue samples without prolonged bureaucratic negotiation. Some of the most promising experimental drugs he found had been shelved by companies, not always because the science was wrong, but because the market was too small to justify development costs. Institutional review boards could block treatment on the basis of a single member’s concern.

Sijbrandij frames the core tension this way: getting a drug approved for a population costs around a billion dollars. Dosing a single person with a personalized therapy costs roughly a million. And that gap keeps widening because the cost of designing new medicines keeps falling while the cost of proving them at a population scale keeps rising.

He spent what appears to be millions of dollars on his treatment. He had a team of people working full-time on his care. He traveled internationally. He had the social capital to persuade companies and researchers to collaborate. He is alive, in remission, and building a new software company.

The question is what his story means for everyone who doesn’t have what he has…

The Real Shift — and Its Limits

The hype says AI is disrupting pharma. The reality is more specific and more consequential.

What AI is actually compressing is the discovery layer of drug development: identifying targets, screening compounds, predicting molecular interactions, designing candidates.

This is real.

At MIT, James Collins trained generative AI to design 36 million novel antibiotic compounds, synthesized 24, and found two that effectively kill drug-resistant MRSA and gonorrhoea through entirely new mechanisms. At Cambridge, Michele Vendruscolo can now screen billions of small molecules against Parkinson’s disease targets in days for a few thousand pounds, work that previously took six months and cost millions. In commercial realms of AI drug discovery, progress is even more real, as I described in an industry report last year. These are real advances that are published, validated, and progressing toward clinical testing.

But discovery is one layer.

What AI does not compress too much is everything that follows: GMP-grade manufacturing, formulation, safety testing, regulatory approval, clinical trials, and the logistics of getting a physical product into a patient’s body. Vendruscolo said it clearly: “AI is revolutionising drug discovery. But only in very specific ways.”

Now, I am an avid user of LinkedIn, and the hype about Rosie the dog’s vaccine “miracle” is overwhelming as much as it is misleading.

Say, there is this “Napster moment in pharma” analogy that circulated in the discourse; it is instructive precisely because it is wrong. Napster made distribution free for a product, a digital music file, that required zero manufacturing and clinical testing, risking human life. An MP3 is pure information; copy it and you have the product. A drug design is also information. But a drug is a physical object: synthesized molecules, lipid nanoparticles, cold chain, quality controls, and a trained professional to administer it. You cannot download a vaccine.

This distinction is where the real power story lives.

Where Power Migrates

When one layer of a value chain commoditizes, the layers adjacent to it don’t become less valuable, they become more valuable. This is what is happening in drug development, and almost nobody in the viral discourse on LinkedIn and elsewhere is talking about it.

Manufacturing is becoming the chokepoint.

Contract development and manufacturing organizations with the certifications to produce clinical-grade therapeutics, the CDMOs, are increasingly becoming the most strategically important players in the system. Conyngham needed UNSW’s RNA Institute. Sijbrandij needed a specialized German facility. During COVID, Moderna and BioNTech had the vaccine designs ready in a matter of hours or days, and still needed months to scale manufacturing despite virtually unlimited funding and political urgency. As more people and more AI systems can design therapies, the power shifts to whoever can make them. This is not a software problem. It is an atoms problem, and atoms don’t scale like code.

Leave a comment

Regulatory frameworks are the new dividing line.

The FDA’s expanded access pathway worked for Sijbrandij, but it’s designed for individuals, not populations.

A new “plausible mechanism pathway,” proposed after the first personalized CRISPR therapy was administered to an infant in 2025, represents the earliest attempt at a regulatory framework for individualized medicine at scale.

Meanwhile, China has adopted a different posture toward experimental treatments — creating a geopolitical dimension to what looks on the surface like a procedural question. How fast a country’s regulatory system adapts to personalized therapies may determine where the next generation of biomedical innovation actually happens.

Leave a comment

Proprietary data is the hidden gate.

The tools are opening up — AlphaFold is free, ChatGPT, Claude, Gemini, etc. are ubiquitous. But as Collins at MIT noted, the critical datasets on drug absorption, distribution, metabolism, and toxicity are held by pharmaceutical companies and are not publicly available. The front door to drug design is unlocking, but the back room where the real knowledge lives is not.

Capital is still the ultimate filter.

Conyngham needed $3,000 and a personal connection to a willing nanomedicine pioneer. Fajgenbaum was already inside the system as a Penn medical student. Sijbrandij needed millions, a full-time care team, and the ability to fly to Germany on short notice. The design layer is democratizing. The execution layer may be getting less equal because as design becomes abundant, the scarce resources (manufacturing, regulatory navigation, data, clinical access) command more leverage, not less.

And the AI tool providers are becoming something new entirely in the biomedical and healthcare fields.

This is the shift that is happening. When Conyngham used AlphaFold to model Rosie’s mutated proteins and ChatGPT to navigate bioinformatics pipelines, those tools weren’t passive infrastructure, like cloud servers or MS Excel used to be, the way a microscope or a sequencing machine is. They were making consequential analytical judgments: which protein structures to prioritize, which pipelines to recommend, which targets looked promising. The tool was doing work that used to be done by specialized teams inside pharmaceutical companies. A sequencing machine generates data, and AI interprets it. That’s a fundamentally different position in the value chain. DeepMind, OpenAI, Anthropic, Google, and the growing ecosystem of specialized biotech AI platforms — Insilico Medicine, Recursion, Isomorphic Labs — are not selling picks and shovels to the gold rush. They are becoming the analytical intelligence layer of drug development itself. Whoever controls the models controls the lens through which the next generation of therapies is conceived. Pharma’s old intellectual moat, which has been the institutional expertise to go from data to hypothesis, is leaking to pharma outsiders, like big tech or even individual biohackers with resources. It is a new phenomenon that is now enabled by more and more powerful AI tools.

Leave a comment

The Real Question

Here is the thing that makes me most uneasy about the Rosie discourse.

The viral narrative — “biohacker beats pharma” — is factually wrong but politically powerful. Millions of people will read this story and may start believing, based on a single dog’s anecdotal medical case, that pharmaceutical development costs are mostly artificial, that the regulatory apparatus is mostly obstruction, and that AI has made the whole system optional. None of this is true. But it does not need to be true to reshape perception and, potentially, even policy.

The notorious $2.6 billion average development cost is not primarily science — it is the cost of proving that a drug is safe and effective at a population scale. Clinical trials, adverse event monitoring, manufacturing consistency, and post-market surveillance. This infrastructure exists because without it, drugs could harm people at scale — thalidomide in the 1960s, which caused thousands of severe birth defects, is the reason most of modern drug regulation exists in the first place.

Rosie’s vaccine was personalized, designed for one dog, nobody is proposing it be given to millions. But the regulatory system that governed the red tape Conyngham complained about is the same system that governs every drug on the market. When public contempt for that system grows, fed by stories where “the red tape was harder than the vaccine,” the consequences extend far beyond one dog or one biohacker. They reach into the framework that protects everyone.

At the same time, the institutional defense — “N-of-1 means nothing, move along” — also misses the momentum. The design layer is commoditizing, it is an obvious trend. Outsiders can now do analytical work that was recently an institutional monopoly. And every time a Fajgenbaum or a Sijbrandij breaks through the wall and survives, the question of why that wall is so high for everyone else gets harder to dismiss.

The tension is not between innovation and regulation. It is between the speed at which individuals can now identify what might save them and the speed at which institutions can verify whether it actually will. AI is accelerating the first, but not the second — not yet. And the gap between those two speeds is going to produce a decade of collisions, in courts, in legislatures, in hospital ethics boards, and in the lives of patients who can, with help of AI, theoretically understand what might help them and can’t reach it in practical life.

Sid Sijbrandij calls himself “the Kool-Aid Man breaking through the wall.” He can afford to be. The harder question is what happens for everyone else. How do you make personalized medicine accessible to people who aren’t billionaire founders, without abandoning the safety standards that exist?

Nobody has a good answer yet. But the pressure to find one is building faster than most institutions realize. AI is increasingly a player in the game, or should I say, the new generation of companies that possess the keys to the most powerful AI systems on the planet…

Thanks for reading. Please share it with those who may find it useful!

— Andrii

In this newsletter, I’ll be sharing reflections on emerging ideas in science and technology: how societies change, how power accumulates, and how certain futures quietly take shape. Molecules will show up sometimes, but mostly as a prerequisite for talking about something bigger.

Some issues will be quite technical, others more generalist and strategic; some will be speculative and futuristic, others more down-to-earth. We’ll see how it goes.

Today, I would like to take on a topic that has been on my mind for almost a year: the “wild west“ of weight loss drugs, the rise of new pharmaceutical empires, a potentially much bigger future for GLP-1 receptor agonists as the first longevity drug category at scale, and also emerging and growing risks and what it all means for society.

Off we go. Hope you enjoy it!

— Andrii

Subscribe now

“It’s not just about the weight you lose, it’s about everything you gain” — Novo Nordisk

This was one of the ad campaign slogans promoting Wegovy, one of the drugs from a new revolutionary class of weight-loss medicines, based primarily on GLP-1 receptor agonists. These molecules mimic a natural hormone that regulates appetite in the brain.

Few people expected this, but weight-loss drugs are on track to become the biggest-selling drug class in medical history.

• Analysts from major banks (Goldman Sachs, Morgan Stanley, BofA) project the global GLP-1 market could reach $100–150 billion annually by the early 2030s.

• If the trend keeps its pace, GLP-1s would surpass sales of statins at their peak, all oncology blockbusters as a class, and HIV antivirals.

• In 2023–2024 alone, GLP-1s generated tens of billions in annual revenue, with growth rates still extremely high.

Sales of GLP-1 receptor agonists helped Denmark-based pharma giant Novo Nordisk become Europe’s most valuable public company and contributed an estimated 1–1.5% of Denmark’s GDP growth in a single year (2023).

In parallel, US-based drugmaker Eli Lilly rode the same drug class to a market capitalization of more than $1 trillion — the first and only pharma company in history to do so.

Beyond market leaders, there is a growing number of other players, big and small, trying to catch a piece of the cake… Amgen, Roche, AstraZeneca, Viking Therapeutics, Boehringer Ingelheim / Zealand Pharma, etc.

Within three years of Wegovy’s approval in 2021, GLP-1 drugs reached tens of millions of users globally, and the trend is rapidly growing. For instance, by 2025, nearly 12% of Americans had used a GLP-1 weight-loss drug at least once, according to surveys, while in the UK, an estimated 1.6 million adults reported using weight-loss medications in the previous year

Astonishingly, GLP-1s are already crossing species boundaries. San Francisco–based Okava Pharmaceuticals has launched a pilot study testing a six-month GLP-1 implant for obese cats, signaling potential expansion of GLP-1s beyond human medicine into veterinary metabolic health.

None of this was supposed to happen. I am pretty sure executives at Eli Lilly or Novo Nordisk had little idea a decade ago about the magnitude of what was about to unfold…

Why are weight loss drugs such a big deal?

First, let’s zoom out for a moment: what are GLP-1 receptor agonists?

Most notable names on the market include injectable drugs Ozempic and Wegovy (based on the chemical compound semaglutide, developed by Novo Nordisk) or Mounjaro and Zepbound (based on a different compound, tirzepatide, developed by Eli Lilly). Oral formulations are now moving through approval and launch pipelines to further expand reach.

This drug class acts on the incretin system, a hormone signaling network centered in the gut. The incretin system links food intake to insulin release, appetite regulation, gastric emptying, and energy balance by coordinating communication between the intestine, pancreas, and brain.

GLP-1 drugs were originally developed to manage type 2 diabetes. What they revealed instead was a lever on human metabolism so powerful that it reshaped appetite, body weight, cardiovascular risk, and disease progression all at once.

Clinical trials showed average weight loss exceeding 15%, up to a 94% reduction in progression from prediabetes to diabetes, and a 20% drop in major cardiovascular events—results strong enough to move GLP-1s from endocrinology into the center of modern medicine.

The effects were obvious almost immediately. People felt less hungry within days. Weight dropped within weeks. Blood markers improved within months. By 2024, an estimated 15 million Americans were using a weight loss drug, many without diabetes, often paying cash, often navigating shortages because global supply could not keep up with demand. Marketing was following a strong biology signal…

A few years ago, weight-loss drugs “lived” in clinics and pharmacy queues. Today, they show up in places that don’t look medical at all.

Due to intense hype, these drugs are [frequently marketed on social media](https://www.edweek.org/leadership/weight-loss-drugs-are-the-talk-of-social-media-and-teens-are-listening/2024/03#:~:text=Commentary may contribute to girls,self-esteem and mental health.) (e.g., TikTok, Instagram) as “quick-fix” solutions, rather than treatments for chronic disease, driving demand for “beach-body” ready results.

They appear in subscription flows that resemble streaming services… Telehealth companies, for instance, have adopted the user-friendly, recurring-payment structures typically seen in services like Netflix or Spotify, to manage high-demand medications like Wegovy and Zepbound. Just one example of many, a company Hims & Hers, offers weight-loss programs for the price of a mid-range gym subscription for compounded GLP-1 injections, bundled into what they advertised as a ‘seamless’ online experience that includes unlimited consultations.

Weight loss drugs start arriving through telehealth checkouts, sometimes without a long-standing relationship with a physician. They are framed as maintenance, prevention, or simply “something people are on.”

This shift from medicine to lifestyle phenomenon is visible in how the market now talks about access to weight loss drugs.

Executives increasingly compare GLP-1 medications to everyday consumer goods—something you sign up for, manage through an app, and fit between the gym and dinner. That framing would have been unthinkable when these drugs were positioned narrowly as treatments for type 2 diabetes, in earlier years!

Advertising reflects the same drift. Much of the messaging avoids disease language entirely. Instead of obesity or diabetes, it emphasizes energy, confidence, control, or simply “feeling like yourself again.” In the United States, this framing also allows intermediaries to market aggressively without triggering the stricter rules that apply to direct pharmaceutical advertising.

The result is a subtle inversion. GLP-1s are no longer introduced into life as an exceptional medical intervention. They increasingly appear as a background option—something available, discussed, and normalized long before a doctor’s office enters the picture.

To be fair, we have seen medicine move into everyday lifestyle before…

When sildenafil was approved in 1998, Viagra entered the market as a treatment for erectile dysfunction, yet its cultural impact quickly exceeded that diagnosis. It became one of the first prescription drugs marketed directly to consumers at scale in the United States, framed around confidence and quality of life rather than disease. But still, its use remained episodic: taken as needed, producing temporary effects, and leaving baseline physiology unchanged once discontinued.

The oral contraceptive pill marked a deeper shift. Approved in 1960, it brought continuous medication into the lives of healthy people. By the late 1960s and 1970s, tens of millions of women worldwide were taking hormones daily to regulate fertility and life planning rather than to treat illness. Contraceptives normalized long-term hormonal modulation when it served a clear, future-oriented purpose.

But GLP-1 receptor agonists are a different beast, representing a more foundational turn.

They are neither episodic like Viagra nor function-specific like contraceptives. They act continuously across appetite, metabolism, insulin signaling, inflammation, and cardiovascular risk—systems that define the baseline physiological state of the human organism. Their emerging role centers on maintaining a biological trajectory over time.

That difference matters. It marks a transition from medicine used occasionally or for specific functions to medicine that “stabilizes” the entire system —and, in doing so, begins to redefine what “normal” health means for many people.

The Race of Empires

Modern medicine is built around diseases and their diagnoses. Reimbursement, regulation, and clinical guidelines all assume discrete diseases with clear endpoints. GLP-1s do not quite fit that model, they act upstream, on the biological circuitry that influences many downstream outcomes at once.

Once the effects of GLP-1s became undeniable, the bottleneck moved upstream. Demand surged faster than any modern pharmaceutical supply chain was designed to handle.

By 2022, Novo Nordisk and Eli Lilly were no longer just competing on efficacy, the competition shifted to capacity. Both companies announced plans to spend tens of billions of dollars expanding manufacturing, a scale of capital investment rarely seen in drug development. Even so, supply lagged.

In the United States, the Food and Drug Administration placed multiple GLP-1 products on its official shortage list in 2022-2024, an extraordinary designation for drugs generating billions in annual revenue. These days, most of these shortages are resolved.

However, the shortage status changed the market “overnight”.

Federal rules at the time allowed compounding pharmacies to produce copies of approved drugs, while shortages persisted, opening a legal window that telehealth platforms moved through immediately.

Companies such as Noom, Ro, and Hims & Hers Health would offer compounded GLP-1s at prices hundreds of dollars below branded versions. Analysts estimated hundreds of thousands of users within months; executives claimed totals exceeding one million.

This was not a disruption in the classic startup sense. No new biology had been invented, the molecule already worked. But what changed was the access model. Cash-pay channels, online prescribing, and mail-order fulfillment removed friction faster than insurers or clinics could adapt. The result was a parallel distribution system operating alongside — and occasionally in tension with — the branded pharmaceutical market.

Drugmakers responded by tightening control.

Both Novo and Lilly pursued lawsuits over advertising practices and safety claims, while simultaneously building direct-to-consumer distribution channels and partnering selectively with telehealth platforms. The goal was clear: reclaim continuity before the market fragmented permanently.

At the same time, a second race unfolded inside research and development. Weekly injections had proven the point, but pills promised something bigger.

Oral formulations lowered psychological barriers, expanded global reach, and aligned with preventive use over decades rather than acute treatment. Whoever succeeded would not just capture market share, they would define how a population takes a drug meant to be used for years.

The first FDA-approved, daily oral GLP-1 receptor agonist specifically for chronic weight management in adults with obesity or overweight is the Wegovy pill (oral semaglutide), approved in December 2025 and launched in January 2026. It is the first pill version of the popular weekly injections, offering a daily, non-injectable option and what I think of as “an iPhone moment“ of the GLP-1 race.

Eli Lilly, in its turn, has orforglipron (LY3502970), a small-molecule, non-peptide oral GLP-1 receptor agonist in late-stage (Phase 3), designed for daily use without food restrictions. It is not yet on the market, but clinical trials showed significant weight loss up to 11.2% over 72 weeks.

The stakes extended beyond the companies themselves. National health systems, insurers, and employers watched a therapy that could plausibly reduce long-term cardiometabolic costs — and plausibly increase short-term spending at unprecedented scale. Coverage decisions became strategic bets on future disease curves rather than simple reimbursement choices.

By the time shortages began to ease, the market had already changed shape. Platforms had been built, habits had formed. Millions of users had started a therapy that evidence suggested would be difficult to stop without reversal.

What looked like a supply crisis was something deeper: the first sign that a drug acting on foundational metabolism would reorganize not just treatment pathways, but the institutions responsible for delivering care.

But recently, hints of an even bigger story started to emerge…

The Longevity Turn

During the Aging Research and Drug Discovery (ARDD) meeting in Copenhagen in 2025, I attended a presentation by Andrew Adams, Group Vice President of Molecule Discovery at Eli Lilly, where he presented a reframing that crystallized years of accumulating data. He argued that GLP-1 receptor agonists combine two properties rarely found together in medicine: mechanistic relevance to age-related disease and clinical scale sufficient to influence population-level healthspan.

The evidence supporting that claim was already familiar, though rarely assembled under a longevity lens. Trials showed reductions in progression from pre-diabetes to diabetes of up to 94%, significant declines in major adverse cardiovascular events, and sustained weight loss accompanied by improvements in inflammatory and metabolic markers.

• They lower blood glucose and insulin resistance.

• They suppress appetite and reduce caloric intake.

• They slow gastric emptying, changing satiety timing.

• They reduce inflammation and improve several cardiovascular risk markers.

Patients report improvements in blood pressure, lipid profiles, joint pain, sleep apnea symptoms, and energy levels.

Adams extended the argument further. He highlighted early signals linking GLP-1s to improvements in vascular function, cognitive trajectories, and psychiatric and addiction-related conditions, while emphasizing that these areas remain under active investigation.

The unifying theme was delay: delaying disease onset, delaying functional decline, and shifting risk curves later in life.

This framing placed GLP-1s within a broader transition already underway in healthcare. Preventive, proactive management gained priority over episodic intervention. Long-acting delivery formats entered the discussion recently, including RNA-based approaches and other technologies designed to sustain adherence across aging populations. The operational goal became continuity at scale.

A parallel message during the same ARDD2025 conference came from Lotte Bjerre Knudsen, Chief Scientific Advisor at Novo Nordisk and a central figure behind semaglutide. At the same conference, she opened her presentation with a direct title: “Semaglutide as a Proven Longevity Medicine.” The slide signaled how fully the internal scientific narrative had evolved alongside clinical results.

Together, these moments reframed the category. GLP-1s began to function as tools for managing long-term trajectories rather than addressing isolated conditions. Risk profiles over decades replaced short-term endpoints. Individual prescriptions scaled into population-level strategies tied to aging itself.

At this point, the implications settled into place. A therapy positioned to delay multiple age-related diseases simultaneously reshapes how healthcare systems plan, invest, and govern care across lifetimes. It turns prevention into infrastructure and metabolism into a managed domain.

Going from type 2 diabetes treatment to weight loss to systemic longevity management is an extraordinary shift… although many things are still to be proved in clinical trials.

Every medical miracle has a darker side…

As GLP-1 use moved from a clinical niche to mass adoption, the increasingly pressing question has become this:

By late January 2026, about 4,400 lawsuits had been filed in the U.S. over GLP-1 drugs such as Ozempic, Wegovy, and Mounjaro (filings dating back to 2023), according to reporting that cited USA Today. The allegations span severe gastrointestinal injury and other serious harms—claims manufacturers deny and say they will fight in court.

Regulatory systems are optimized for therapies with clear indications and bounded time horizons. GLP-1s are drifting into a different category: continuous biological management. That changes the risk calculus in at least three ways:

1) Prevention lowers tolerance for harm.

When a therapy is used to reduce future risk rather than treat acute illness, tolerance for serious side effects drops sharply. Vaccines illustrate this dynamic well: they are judged less by potential benefit and more by how safely they shift long-term population risk. GLP-1s increasingly fall into a somewhat similar logic.

2) Rare events stop being rare at the population scale.

Even low-probability side effects become governance-level problems when the denominator is in the millions. That’s why regulators are updating guidance and surveillance, not just labeling.

In the UK, for example, the MHRA recently updated guidance on GLP-1 medicines and highlighted the risk of severe acute pancreatitis, noting 1,296 Yellow Card reports of pancreatitis (2007–Oct 2025) associated with GLP-1 receptor agonists. (They still emphasize the drugs are safe and effective for most people, but “most people” is not the relevant unit when systems manage scale.)

3) Population benefit vs. personal harm becomes the central political tension.

At the population level, GLP-1s may reduce cardiovascular events and delay metabolic disease, and even be the first “longevity drug” on the market. At the individual level, however, a subset of patients alleges devastating harm, and courts are where that asymmetry gets priced.

That tension is now being formalized in the structure of litigation itself. In December 2025, a federal judicial panel created a separate mass litigation to centralize lawsuits alleging GLP-1 drugs caused NAION-related vision loss, distinct from the existing GI injury litigation. Reuters reported nearly 3,000 lawsuits in the gastrointestinal MDL at that time, alongside dozens of vision-loss cases being centralized.

This is what “side effects” look like once a drug class becomes a pillar of preventive medicine: not just medical risk, but institutional risk, regulatory, legal, and reputational, accumulating alongside adoption.

So, what’s next?

A drug class that regulates metabolism at scale does more than change bodies. It reshapes how societies think about responsibility, aging, and care over a lifetime.

Doctors are already starting to treat GLP-1s as long-term drugs. For patients with obesity or metabolic disease, not being on one increasingly looks like an untreated risk rather than a neutral choice.

The same shift is happening on the money side. In the United States, big insurers like CVS Health, UnitedHealthcare, and major Blue Cross Blue Shield plans are running the numbers on what GLP-1s cost over a lifetime and what they might save by preventing heart attacks, diabetes, and other expensive diseases.

Coverage decisions now come down to whether paying more today reduces bigger bills later.

Employers are doing the same math. In the U.S., most workers are covered by self-insured employers, and those companies are deciding whether to pay for GLP-1s, how long to cover them, and who qualifies. Some limit duration. Some require people to try cheaper options first.

Benefits consultants are now part of the picture. Firms like Mercer, Aon, and Willis Towers Watson publish guidance on how GLP-1 coverage affects employer health spending and what trade-offs companies should expect.

Outside the United States, the same questions appear under different institutional names.

In the UK, the National Health Service evaluates GLP-1s through cost-effectiveness thresholds rather than employer budgets. The debate centers on whether long-term use justifies upfront spending when benefits accrue over decades.

Across Europe, national health systems face similar trade-offs. In countries like Germany and France, reimbursement discussions increasingly show up in the language of prevention economics: delayed diabetes, fewer cardiovascular events, and lower long-term disability costs weighed against continuous drug spending for large populations.

In Asia, the pressure comes from demographics. Aging populations in Japan, South Korea, and China drive interest in therapies that reduce metabolic and cardiovascular disease over time. Public and private payers are cautious, but the logic is familiar: long-term metabolic control looks cheaper than managing complications later. Access expands slowly, through pilot programs and narrow indications, rather than mass rollout.

Different systems, same problem.

For patients, the shift feels personal everywhere. Many people who stop these drugs see their weight and appetite return quickly. Over time, the question changes. It’s no longer “should I start?” It becomes “can I afford to stop?”

The gatekeepers of this new biology modulation era, institutions, companies, regulators, platforms, and employers, all are faced with effects that span decades and populations, they will have to figure out and build systems to manage continuity, risk, and global scale.

Thanks for reading. Please share it with those who may find it useful!

— Andrii

Subscribe now

So, MIT Technology Review just released its 10 Breakthrough Technologies of 2026.

If you skim the list, there are a few usual suspects: hyperscale AI data centers, generative coding tools, new energy technologies, and commercial space stations. They reflect where capital, research effort, and industrial momentum already are.

But what really jumped out at me this year is something else: about a third of the list is life sciences–related, and those entries are some of the most socially loaded technologies on it…

Measuring, editing, and reintroducing biology

The biotech-related breakthroughs on the list are different on the surface, but they’re all variations on the same theme: how directly we can “hack” our biological destiny.

The first is embryo scoring.

Embryo screening itself isn’t new. Fertility clinics have long used genetic tests to identify severe inherited diseases or chromosomal abnormalities before implantation.

What’s changed is that advances in genome-wide sequencing and polygenic risk scoring now allow clinics to estimate probabilistic risks for a much wider range of conditions, and in some cases even non-medical traits.

From a technical standpoint, this is understandable: sequencing is cheaper, datasets are larger, and the statistical tools exist. In practice, though, it remains highly controversial. These predictions are probabilistic rather than deterministic, their accuracy varies widely across traits and populations, and the decisions they inform happen extremely early, with irreversible and far-reaching consequences. That combination of early intervention, uncertain prediction, and implicit value judgments, is what makes embryo scoring one of the most socially sensitive technologies on this year’s list.

Do you think the general scientific direction of “quality control” of future babies is the right thing to do, if we talk not just about specific, obvious disease risks, but about more vague traits, like intelligence or physical strength? It is easy to answer in theory (probably), it is harder when it is an actual decision.

Anyway, the second item is base-edited and personalized gene-editing treatments.

This breakthrough tech category on this year’s list encompasses therapies that go beyond generic approaches to tailor genetic intervention to individual patients’ DNA.

Base editing, for instance, is a refined form of CRISPR technology that can change single DNA letters without cutting the double strand, allowing highly precise changes that avoid some of the risks of older methods.

In late 2025, researchers reported the world-first base-edited gene therapy used to treat a previously untreatable blood cancer, demonstrating the real clinical potential of these tools. Even more striking, clinicians in the U.S. designed and administered a personalized CRISPR-based gene editing therapy for a baby with a rare metabolic disorder, developed and delivered within months, and showing early signs of benefit, a milestone in bespoke genetic medicine.

Beyond individual cases, multiple clinical trials are underway using base editors to target conditions from leukemias to sickle-cell disease and hypercholesterolemia, reflecting how fast the field is moving from laboratory proofs-of-concept into real clinical applications rather than speculative future treatments.

The third is gene “resurrection.”

Banks of genetic information from extinct species, and related techniques, are being used to restore traits, with implications for conservation, medicine, and climate-adjacent biology. This isn’t about bringing back dinosaurs, but it does raise practical questions about how biological traits move across time and ecosystems.

In a striking example, researchers recently sequenced the entire genome of a woolly rhinoceros from the preserved stomach contents of a 14,400-year-old wolf pup found in Siberian permafrost, a first for any Ice Age animal and a powerful demonstration of how high-quality ancient DNA can still be recovered.

Beyond that, a team in northern Saudi Arabia extracted preserved DNA from naturally mummified cheetahs, offering new insights into genetic diversity and potential rewilding strategies for big cats that once roamed the Arabian Peninsula. Meanwhile, biotech groups like Colossal Biosciences have created “woolly mice” — genetically edited rodents that express traits inferred from extinct woolly mammoths, as a proof of concept for trait restoration in living species.

Share

None of these technologies are flashy consumer products, like AI tools, but they operate at a much more foundational level, and their impact is more likely to accumulate slowly than announce itself all at once. That’s part of what makes them so socially and ethically controversial.

Studying LLMs like an alien life form

One non-biological item on the list that is quite notable, perhaps even counterintuitive, is the mechanistic interpretability of large language models.

What’s interesting here is not so much the technical work itself, but the framing. Researchers are increasingly trying to understand large language models by observing and probing them (like mapping internal behavior, testing responses, and identifying patterns), rather than treating and documenting them like good old software systems.

In practice, that means studying LLMs almost like complex natural systems. Less “we defined every rule, and it is in the description,” and more “we built something, and now we need to understand how it behaves, changes over time.”

Anyway, what stood out to you in this year’s MIT Technology Review’s list of 10 Breakthrough Technologies 2026? Was it the biology, the AI, or something else entirely? I’m curious how others read it.

In this newsletter, I’ll be sharing reflections on emerging ideas in science and technology — not as an end in themselves, but as a way to think about bigger things: how societies change, how power accumulates, and how certain futures quietly take shape. Molecules will show up sometimes, but mostly as a prerequisite for talking about something larger.

Some issues will be quite technical, others more generalist and strategic; some will be speculative and futuristic, others closer to the ground. We’ll see how it goes.

I’m still figuring out the exact shape of this newsletter, and I hope you’ll help shape it over time, through your feedback, replies, and conversations, which are very welcome!

For now, I thought I’d start with one idea that stuck with me after several recent conferences on aging research and drug discovery.

Anyway, off we go. Hope you enjoy it!

— Andrii

Subscribe now

We are not one organism.

Roughly 70% of the human immune system sits in the gut, constantly interacting with trillions of microbes we do not control and rarely think about.

Yet for most of modern history, aging has been understood as something that happens to a single biological self.

Time passes. DNA accumulates damage. Cells become senescent. Chronic inflammation rises. Systems wear down. Eventually, things stop working as well as they once did.

This view is intuitive. It fits neatly with metaphors we already understand: clocks, entropy.

It also aligns with how medicine has traditionally approached aging: treat the organs, fix the parts, manage the failures.

But this assumption does more than shape explanations. It determines what kinds of interventions get built. If aging is something that happens inside a bounded organism, then medicine looks inward: repair DNA, clear senescent cells, tweak metabolism.

If that assumption is incomplete, however, then entire classes of interventions may be systematically undervalued or missed altogether.

Aging, in other words, may not be failing parts inside a machine.

It may be the gradual destabilization of a biological ecosystem.

You Are Not One Organism

The average human body contains roughly 30 trillion human cells.

But it also contains between 38 and 40 trillion microbial cells if we talk about an average 70 kg person.

These microbes — mostly bacteria, but also viruses, fungi, and other microorganisms — live on our skin, in our mouth, in our lungs, and most densely in our gut. Together, gut microbiota weigh approximately 0,2 kilograms, roughly the mass of a mango.

Interestingly: The microbiome is often said to weigh ~2 kg, but revised estimates put it closer to ~200 grams. The original overestimate came from assuming the entire gut had colon-level bacterial density; in reality, most microbes are confined to ~0.4 L of colon content, only part of which is bacterial mass.

The gut microbiome alone contains over 1,000 species of bacteria. Collectively, their genetic material outnumbers the human genome by a factor of about 150 to 1. In terms of raw biochemical activity, the microbiome is not a side system — it is a dominant one.

When functioning properly, this microbial ecosystem performs essential tasks:

• Regulating immune responses

• Breaking down nutrients we cannot digest ourselves

• Producing signaling molecules and metabolites

• Supporting tissue repair and metabolic balance

For most of our lives, we barely think about our own bacterial communities, and that may be precisely why their role in aging has been underestimated.

Balance, Not Presence, Is the Key

The microbiome is not inherently good or bad. What matters is balance.

A healthy microbiome is diverse, stable, and cooperative with its human host. When this balance is disrupted, a state known as dysbiosis, problems begin to emerge.

Dysbiosis can be triggered by many factors:

• Chronic stress

• Poor or highly processed diets

• Repeated antibiotic use

• Persistent infections

• Environmental exposures

When microbial balance breaks down, harmful species can dominate, beneficial ones diminish, and the immune system is pushed into a constant state of low-grade activation.

This leads to chronic inflammation, one of the most consistent biological features of aging. But inflammation is only the beginning.

Share

Dysbiosis and the Biology of Aging

Modern aging research describes aging through a framework known as the hallmarks of aging. Originally proposed in 2013 and expanded in 2023, this framework identifies 12 core biological processes that drive aging across tissues and species.

They include:

• Genomic instability

• Telomere attrition

• Epigenetic alterations

• Mitochondrial dysfunction

• Loss of proteostasis

• Deregulated nutrient sensing

• Cellular senescence

• Stem cell exhaustion

• Altered intercellular communication

• Chronic inflammation

• Immune system decline (immunosenescence)

• Reduced tissue regeneration

It turns out, increasingly supported by evidence, that microbial imbalance influences nearly all of these processes. Chronic dysbiosis can:

• Sustain inflammatory signaling that damages DNA

• Accelerate immune exhaustion

• Impair stem cell function and tissue repair

• Disrupt nutrient-sensing pathways

• Alter epigenetic regulation through microbial metabolites

At a major aging research conference, the Aging Research and Drug Discovery Meeting (ARDD 2024), which I attended in 2024 in Copenhagen, a CEO of a Texas, US-based biotech company, Maxwell Bioscience, Scotch McClure, presented this idea in what I found to be deliberately provocative terms.

He argued that aging should be understood, at least in part, as a communicable process, not because we “catch” aging from others, but because persistent pathogens and microbial imbalance act as upstream drivers of the aging process itself.

In his words:

“The human microbiome expresses over 200 times 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.”

The claim, as I understand it, is not that microbes are the sole cause of aging, but that they may be a critical accelerator.

In this view, aging is not driven solely by the passage of time and the wear and tear of our bodies.

It is shaped by decades of biological interaction between human cells and the microbial environment with which they coexist.

Evidence From Cancer Research

This idea does not exist in isolation; for instance, in oncology, the role of the microbiome is already sound.

As I learned from a presentation by Rafik Fellague-Chebra, Global Medical Director at Novartis Oncology, during another event, the Drug Discovery Innovation Programme (DDIP) in Barcelona this year, clinical data show that antibiotic use can reduce the effectiveness of cancer immunotherapy by up to 50%.

In some cases, patients who do not respond to treatment regain responsiveness after receiving fecal microbiota transplants (FMTs) from patients who did respond.

This has led researchers to identify the oncobiome — the microbial ecosystem associated with tumors — as an active modulator of immune signaling, drug resistance, and cellular pathways such as mTOR and AKT.

The lesson is clear: microbes are not passive passengers, they shape biological outcomes at the highest level.

If they can determine whether advanced cancer therapies work, it is reasonable to ask whether they also influence how quickly bodies age.

Share Molecules & Empires

Rethinking Intervention

If aging were purely mechanical, the solution would be mechanical: repair the damage, replace the parts, slow the clock.

But ecosystems do not respond well to brute force.

This is one reason why broad-spectrum antibiotics, while lifesaving in acute settings, are a poor long-term strategy. They destroy harmful microbes — but also beneficial ones. They destabilize microbial communities. And they drive resistance.

An ecological problem requires an ecological solution. But medicine is structurally better at repairing “individual machines” than ecosystems of organisms.

For clarity, ecological in this context is not about climate. In a broader sense, ecological = arising from interactions within a complex, multi-actor system, rather than from failures of individual components.

Anyway, clinical trials and regulatory pathways were built around interventions with a single, isolatable active ingredient and a relatively stable target, e.g. “block this receptor” or “replace this hormone.”

Microbiome-facing interventions collide with the opposite reality: massive baseline variability across individuals, feedback loops, and context dependence that make “one size fits all” harder to prove in a standard randomized trial design.

The incentives structure reinforces the same bias.

In many health systems, fee-for-service reimbursement rewards discrete, billable acts (visits, procedures, prescriptions) more reliably than long-horizon maintenance of complex risk factors, especially when benefits arrive years later and accrue to a different payer. In other words, we pay for repair more readily than we pay for ecological stewardship.

One emerging approach focuses on reinforcing innate immunity rather than eliminating microbes indiscriminately.

A case study is earlier mentioned biotech Maxwell Biosciences, which is developing synthetic analogs of LL-37, a naturally occurring antimicrobial peptide produced by the human immune system.

LL-37 is a remarkable molecule, and we all have it in our bodies!

It has broad activity against bacteria, viruses, and fungi, but in its natural form, it degrades too quickly to be used therapeutically.

Leave a comment

Using a proprietary platform, the company created synthetic peptoids — chemically stable molecules that mimic LL-37’s function. These peptoids:

• Are stable at room temperature

• Do not require cold-chain storage

• Show no resistance development in pathogen exposure studies

• Do not disrupt healthy gut microbiota

• Disrupt microbial membranes and biofilms while preserving host tissue

The lead candidate is delivered as a nasal spray, targeting the nasal–brain axis, a pathway increasingly implicated in inflammation and neurodegenerative disease.

The broader ambition, as I understand, is to create a synthetic immune system — modular, tunable molecules that restore immune balance, clear chronic pathogens, and reduce inflammatory burden without destabilizing microbial ecosystems.

The company plans its first clinical trials in 2026, with expansion into healthspan and aging applications.

Whether this specific approach succeeds remains uncertain. But the underlying logic is difficult to ignore.

A Different Story About Aging

Seen through this lens, aging is not simply decay, rather, it is the long-term outcome of coexistence with trillions of microbes. With environments — physical, social, and microbial — that gradually shape biological trajectories over decades.

It kinda changes what kind of problem aging is.

If aging is partly “ecological”, then it cannot be addressed only at the level of isolated organs or individual choices. It implicates food systems, public health infrastructure, infection control, living conditions, and the incentives that determine what kinds of interventions get built in the first place.

Some societies will learn to manage these ecosystems better than others. And the differences may compound slowly, invisibly, and profoundly.

Longevity, in this framing, is less about focusing on the body organs and tissues, but more about maintaining balance, biological, microbial, and institutional, under constant pressure.

In other words, we are not just bodies moving through time, we are environments embedded in larger environments.

And how well those environments are understood, governed, and maintained may shape not only how long we live, but who gets to age well.

Leave a comment

To close this newsletter, here is a pretty entertaining (alarming, I should say?) article I think you may find interesting at any age:

“I Thought The Decline Would Be Much More Subtle”: Older Adults Reveal The Jaw-Dropping Truths About Aging That Nobody Prepares You For

Thanks for reading. Please share it with those who may find it useful!

— Andrii