For years the story in neuroscience was retreat. Failure rates were brutal, endpoints were fuzzy, and one big name after another shut down its CNS unit. Our radar shows the retreat reversing. 52 accelerating compounds touch the nervous system, and the shape of the cluster is the interesting part.
From the Knitify Pharma Demand Radar — our running read on where pharmaceutical demand is heading. We publish the observations; the method behind them stays in-house.
Two poles, both defined by precision
The accelerating CNS chemistry concentrates in two places. On one side, precision epilepsy — molecules for genetically-defined seizure disorders where the biology is now specific enough to drug. On the other, a new generation of rapid-acting antidepressants working through mechanisms the older SSRIs never touched.
What stands out is how little sits between the two poles. This is not a broad return to every hard brain disease at once; it is a concentrated, almost deliberate move into the corners of neuroscience where a clear molecular story already exists. Both clusters share the same underlying logic — a target you can name, a patient population you can define, and an effect you can measure — and that shared logic is the real observation, more than any single molecule. The field did not get braver about the brain in general. It got selective.
Both poles have already produced approvals in conditions that were effectively untreatable a few years ago: fenfluramine and ganaxolone for rare childhood epilepsies like Dravet syndrome and CDKL5 deficiency disorder, and esketamine as a rapid-acting therapy for treatment-resistant depression. [1]
The science: why the brain was so hard
The retreat had real causes. A central-nervous-system drug takes on the order of twelve years to develop — meaningfully longer than the industry average — and historically only around 6% of CNS candidates that enter clinical trials are ever approved, against roughly 13% in other areas. [2] Two things made it brutal: the blood-brain barrier keeps out the overwhelming majority of small molecules, and the diseases were defined by symptoms rather than biology, so trials measured fuzzy endpoints against a heavy placebo response. In Alzheimer's, the decade from 2002 to 2012 produced a failure rate above 99%.
The barrier itself is worth understanding, because it is the physical reason so much promising chemistry died in preclinical work. The blood-brain barrier is a tight sheet of specialized cells lining the brain's blood vessels, reinforced by active efflux pumps that recognize foreign molecules and shove them back out into the bloodstream. It evolved to protect the brain from toxins, and it does its job indiscriminately — it cannot tell a carefully designed drug from a poison. A molecule that works beautifully in a test tube can be a complete failure in a patient simply because it never reaches the tissue it was built for. Chemists are left threading a narrow needle: a compound has to be small and lipophilic enough to cross, yet not so lipophilic that it fails on solubility, safety, or selectivity everywhere else in the body.
On top of the delivery problem sat a measurement problem. The brain's output is behavior and experience, neither of which reports cleanly. Animal models translate poorly — a mouse cannot describe its mood, and rodent proxies for depression or neurodegeneration rarely reproduce the human disease, so a candidate can clear every preclinical hurdle and still miss in people. Human trials then lean on rating scales that drift with expectation, and in mood disorders the placebo response is famously large. Stack it up: an invisible target organ, a barrier that blocks the molecule, endpoints measured by proxy, and a strong placebo effect. That is the recipe for a decade of expensive failure.
The precision-genetics turn
What broke the pattern was not a better guess at the old, broad diagnoses — it was a shift toward diseases with a single, identifiable cause. Many of the rare childhood epilepsies now driving the epilepsy pole are monogenic: they trace to one gene. Dravet syndrome and CDKL5 deficiency disorder are diagnosed genetically, which changes the economics of a trial from top to bottom. You enroll a homogeneous population confirmed by testing rather than a grab-bag of patients who happen to share a symptom, you often start with a mechanistic hypothesis instead of a hunch, and regulators have well-worn pathways for rare, severe, high-unmet-need conditions.
The endpoints cooperate too. A seizure is a discrete, observable event — you can count it. Measuring how often seizures occur, and whether a drug reduces that count, sidesteps exactly the fuzzy rating-scale problem that made mood and cognition trials so treacherous. That combination — a defined population, an objective endpoint, and a clear regulatory route — is what de-risks these programs, and it is why severe pediatric epilepsies such as Dravet, CDKL5, and Lennox-Gastaut have become a template that developers keep returning to. The lesson generalizes: when you can define the patient and count the effect, the risk that killed the field for a decade comes down sharply.
The mechanisms: inhibition and excitation
The two poles also happen to sit on the two halves of the brain's signaling system, which is part of why they read as one coherent trend. Ganaxolone is a neurosteroid that tunes the GABA-A receptor, the brain's main inhibitory switch. Turning up inhibition calms the runaway electrical activity that drives seizures, and because neurosteroids act at a site distinct from the older benzodiazepines, they can retain activity where those drugs lose it over time. It is a mechanistically precise way to quiet a hyperexcitable, genetically-defined circuit.
Esketamine works the opposite side of the ledger, on the glutamate/NMDA system that carries the brain's main excitatory signaling. Its antidepressant effect arrives in hours rather than the weeks that monoamine-based drugs require — a genuinely different mechanism from the SSRIs that dominated depression treatment for a generation, and one of the few real departures from that playbook in decades. The through-line across both molecules is the same: a defined mechanism you can measure. When the biology is precise, the endpoint is precise, and a field that spent years chasing invisible targets suddenly has something concrete to aim at.
The demand: the capital has already turned
The clearest sign this is real isn't in the science — it's in the money. In 2025, neuroscience overtook oncology in total M&A deal value for the first time: Johnson & Johnson bought Intra-Cellular Therapies for roughly $15 billion, Novartis took Avidity Biosciences, a neuromuscular RNA specialist, for about $12 billion, and the pattern repeated through the year as large companies bought back into a field they had spent a decade abandoning. When the acquirers reverse course that decisively, demand for the underlying chemistry follows — and the patent record is where it shows up first.
The detail that gives this weight is that these are, in many cases, the same organizations that closed their CNS units in the retreat. Capital of that size does not move on sentiment; it moves when the risk-adjusted math changes, and the precision turn is what changed it. Neuroscience passing oncology in deal value is a threshold moment, and thresholds like that tend to pull the whole supply chain behind them, from discovery chemistry to manufacturing. The molecules accelerating in the record now are the leading edge of that pull.
Who this moves
A revival concentrated in specific chemistry lands on specific suppliers, not the industry evenly. CNS-focused API makers and CDMOs are the most direct beneficiaries: many of these molecules are complex to synthesize, and several fall into controlled-substance categories that demand licensed handling, secure facilities, and tight chain-of-custody. Manufacturers who can already work with scheduled compounds and intricate small-molecule chemistry sit in a narrower, more defensible lane than commodity producers, and that is precisely the kind of work the two poles generate.
Rare-disease developers are the second group. The genetically-defined epilepsies reward organizations built around orphan indications — companies fluent in genetic diagnosis, patient identification, and the regulatory mechanics of small, well-characterized populations. Their commercial footprint is modest by big-pharma standards, but the model is proven, and the precision turn plays directly to it. The third group is more upstream: blood-brain-barrier delivery specialists. If the barrier is the gate that blocked the field for a generation, then the technologies that help molecules cross it — shuttles, transporter-targeting approaches, and alternative routes into the brain — become broadly valuable the moment CNS programs multiply. Whoever helps open the gate participates in far more than one drug.
What could still go wrong
None of this repeals the arithmetic. A CNS drug still takes about twelve years, and the historical odds — roughly 6% approval — do not vanish because a handful of programs finally worked. Concentrated early wins can look like a durable trend before they are one, and the honest reading of the record is a return to specific, definable slices of neuroscience, not a solution to the brain. The broad, heterogeneous diagnoses that broke the field are still broadly broken.
The commercial size of rare disease is its own caveat. Monogenic epilepsies are, by definition, small populations; revenue depends on premium pricing and on how many patients are actually diagnosed and reached, and not every program clears that bar. There is also an operational tax the precision story tends to gloss over: molecules acting on glutamate/NMDA or neurosteroid pathways can carry abuse potential and scheduling considerations, which bring restricted distribution, supervised administration, and added regulatory and logistics weight. That burden is real, and it lands hardest on manufacturing and distribution — the very stakeholders the revival is pulling in. The signal is genuine; the discipline is in respecting how narrow, and how demanding, the openings still are.
Editorial commentary from the Knitify Pharma Demand Radar. Not investment, legal, regulatory, or medical advice.