By Abbas Kazimi, CEO of Nimbus Therapeutics, as a part of the From The Trenches function of LifeSciVC
In early 2009, when international markets had been bottoming out, you possibly can think about what it felt like strolling the halls of the JPM Healthcare Convention. Biotech felt fragile. Capital was scarce, pipelines had been skinny, and quite a lot of buyers had been quietly asking whether or not the enterprise mannequin in drug discovery even labored anymore.
That was additionally the second when a small group of companies, Atlas Enterprise included, had been sketching a distinct type of biotech on whiteboards with legal professionals and scientists. Not a completely built-in pharma. Not a lab-heavy roll up. However a centered, small-molecule firm constructed round chemistry, main computational instruments (Schrödinger in our case), and a digital working mannequin that would survive volatility and nonetheless produce actual medication.
Nimbus was born out of that period.
We didn’t got down to construct the largest labs. In actual fact, Nimbus has by no means had its personal labs. We got down to construct the quickest studying engine.
Sixteen years later, Nimbus has bought medication to Gilead (right here) and Takeda (right here), partnered packages with Genentech (Roche) (right here), Celgene (BMS) (right here), and most lately with Lilly (once more) (right here, right here), and constructed what we consider is among the most constant small-molecule engines in biotech. We did that by proudly owning the Design–Make–Test–Analyze loop, utilizing international companions to execute at scale with out ever giving up management of what mattered.
At present, DMTA at Nimbus is concrete and central to the enterprise. We design molecules on computer systems. We make them via CRO companions. We take a look at them in proprietary, bespoke organic screening cascades constructed for every goal. We analyze the information and feed it straight again into the subsequent design cycle utilizing built-in AI and physics-based approaches.
That lived expertise is why a latest essay on geopolitical threat in biotech sparked conversations about whether or not Western pharma could have inadvertently funded a competing biotech ecosystem by treating preclinical biology as a commodity.
The piece argues that outsourcing execution inevitably results in outsourcing studying, leaving Western biotech’s licensing property from international ecosystems that their very own capital, years earlier than, helped prepare.
Our expertise suggests one other path is feasible. Nimbus continues to make compounds which are absolutely ours: the molecules, the information, and the training.
The IP sits within the molecule, assigned cleanly to a chosen subsidiary. Extra importantly, the training loop lives inside the corporate, giving us repeatability and conviction.
It is usually value acknowledging how a lot the panorama has modified. Chinese language biotechs and TechBio corporations have made actual progress, notably in medicinal chemistry effectivity and the fast development of me-too medical candidates. Timelines for these “follow-on” packages have compressed, execution has improved, and competitors has elevated.
However that progress has been uneven. Generic medicinal chemistry and early medical execution are more and more commoditized. What stays troublesome, and much much less transferable, is preclinical biology and the novel chemical fairness to crack it open: figuring out high-value targets, constructing illness conviction, producing distinctive chemical instruments, and designing experiments that separate sign from noise earlier than sufferers and capital are put in danger. That is the place many of those fashions stay weakest right now.
Nimbus has all the time invested there. As execution turns into extra intensely aggressive, organic perception turns into the differentiator.
DMTA Is Not New. Compression Is.
One of the best small-molecule groups have all the time labored in loops. Kind a speculation. Make a molecule. Take a look at it. Analyze the information. Determine what to do subsequent. That loop has all the time existed inside sturdy core groups, from Merck in its small molecule heyday of the 1980-Nineteen Nineties to right now’s progressive “good chemistry” companies like Nimbus and others.
What AI and physics-based strategies have modified isn’t the existence of the loop, however its pace and determination. They allow you to discover extra chemical area, filter extra aggressively, and arrive at higher experiments sooner. Nimbus has been integrating generative design, free-energy perturbation (FEP), Absorption, Distribution, Metabolism, and Excretion (ADME) and pharmacokinetics (PK), prediction, and structure-based modeling into each day resolution making for years.
Whereas we by no means had a ‘platform’ the loop was all the time the engine – our engine. AI now compresses it.
Asset-Mild Does Not Imply Studying-Mild
There is a crucial distinction between outsourcing execution and outsourcing studying. The priority that corporations may be coaching their rivals is legitimate when the training loop itself strikes offshore.
Shared infrastructure doesn’t inherently weaken drug design capabilities. That is precisely what cloud computing did for software program. No startup constructed its personal information middle. But the winners nonetheless owned their code, their structure, and their product selections. Shared infrastructure didn’t weaken these corporations. It strengthened them by permitting expertise and judgment to pay attention the place they mattered most.
Nimbus matches cleanly into that logic.
Not each startup must construct a full-stack lab. What issues is sustaining management of the training loop. CROs and CDMOs are execution layers. Nimbus owns the hypotheses, the molecular designs, the prioritization logic, the information integration, and the go or no-go selections.
The training stays inside the corporate. That distinction issues.
Rational Capital Constructions, Intentional Technique
Asset-light fashions emerged partly from enterprise financing and governance realities. That’s true. However for Nimbus, the digital mannequin was not merely a response to capital constraint. It was an intentional technique that allowed us to outlive a number of capital cycles, protect optionality, compound studying over lengthy horizons, and companion repeatedly from a place of power.
As a result of we weren’t anchored to mounted infrastructure or a single area of experience, we might adapt organically to essentially the most compelling science, whether or not in immunology, oncology, cardiometabolic illness, or liver illness. The absence of in-house labs was not a limitation. It created freedom to pursue packages opportunistically and strategically, guided by information relatively than construction.
Sixteen years of outcomes counsel this could be a profitable method when executed with self-discipline.
Sovereignty Lives in Selections
There’s a actual and vital query about sovereignty and resilience on the business degree. However startups should not meant to be sovereign infrastructures. Constructing and sustaining home capability at scale for long-term industrial resilience is a nationwide welfare drawback, sometimes carried by governments, strategics, or shared platforms.
Nimbus’s mannequin matches that actuality. We don’t purpose to be a sovereign infrastructure. We purpose to be a sovereign resolution engine.
That’s precisely the function startups ought to play. Deal with judgment, studying, and conviction, whereas leveraging shared execution capability to maneuver sooner and keep versatile.
Coming Full Circle
Biology isn’t semiconductors. It’s messy, far more advanced, and deeply human. Worth accrues in layers. Execution and information era sit on the base. Speculation era, molecular design, and the mixing of weak indicators into coherent selections sit on the prime.
Nimbus has all the time invested on the prime of that stack.
As I replicate on the JPM Healthcare Convention now in 2026, practically sixteen years after these early conversations, what struck me most was how acquainted the dialogue nonetheless felt. At conferences, dinners, and facet conversations, the main target was not on infrastructure or geopolitics.
It was on the standard of the medication. The molecules. Whether or not they would matter for sufferers.
That’s the place our IP lives. That’s the place our sovereignty lives. Nimbus was constructed for that world lengthy earlier than it was modern to explain it that means.
Many thanks for the contributions and editorial recommendation from Cindy Fung (Nimbus head of company affairs), Peter Tummino (President, Analysis & Growth), and Bruce Sales space, (Associate, Atlas Enterprise).
