Novartis Paid $8.7 Billion to Cure a Disease Once. The Hard Part Is Doing It Twice.

There is a kind of business problem most companies would kill for and a few biotech executives lie awake over: what happens when your product works so well that the patient never needs it again. Novartis paid $8.7 billion in 2018 for exactly that problem.¹ The asset was a single dose, infused once into an infant's bloodstream, that rewrites a broken gene and stops a fatal muscle-wasting disease before it can run its course. It became the best-selling gene therapy on earth. And then, faster than almost anyone modeled, it started to disappear.

The official story is that Novartis bought a blockbuster. It did — but a blockbuster of an unusual species, the kind that consumes its own market by curing it. Zolgensma was never going to behave like a daily pill. The strategic question Novartis was actually buying wasn't 'can we sell this drug?' It was 'can we build a platform that does this again, on a different disease, before this one runs dry?'

The trap hidden inside a one-shot cure

A traditional drug bills you forever. A gene therapy bills you once and then says goodbye. Zolgensma treats spinal muscular atrophy, and its first FDA approval in May 2019 was narrow on purpose: pediatric patients under two years of age, the window before the disease does irreversible damage.² That narrowness is the whole economic story. There is a finite, slowly-replenishing number of infants born with SMA each year, and a backlog of already-diagnosed children waiting for the first treatment to exist. When approval landed, Novartis drained that backlog fast — and a backlog, once drained, does not refill. Sales went from $361 million at launch to $920 million in 2020 to a peak of $1.35 billion in 2021.⁴ Then the curve bent the other way. The first annual decline came in 2023, less than five years after launch, and by the second quarter of 2025 quarterly sales had fallen to roughly $297 million — a 17% drop year-over-year.⁹ Nothing went wrong with the product. The market simply ran out of patients to cure.

This is the structural fact that makes gene therapy a different kind of bet than a normal acquisition. The product priced near $2 million a dose and earned more than $6.4 billion cumulatively through 2024 — a genuinely successful launch by any measure.⁴ But success here is self-extinguishing. The only way to keep a gene-therapy franchise alive is to keep finding new genes, new diseases, new patient pools. The asset isn't the molecule. It's the machine that makes molecules like it.

The scandal the FDA decided not to punish

There is a darker thread in this story, and a serious analysis has to pull it. Before Zolgensma was approved, AveXis personnel became aware of potential data manipulation in a mouse potency assay — as early as March 2019 by Novartis' own investigation timeline. Novartis did not disclose this to the FDA until June 28, 2019, more than a month after the agency had already approved the drug on May 24.³ Two senior AveXis executives were terminated. The crucial mitigating fact is that the manipulated data concerned an early-stage manufacturing potency assay, not patient clinical data — the FDA confirmed the product remained safe and effective. And the agency, controversially, chose to impose neither civil nor criminal penalties.³

Why this matters strategically, not just morally: a platform's value rests entirely on regulators trusting the data that comes out of it. Novartis got to keep its product and escaped formal punishment — but it spent a measure of the one currency a gene-therapy company cannot easily rebuild, which is the assumption of good faith. Every future filing now travels with a footnote. That's a real cost, even when the fine is zero.

Buying the second act before the first one runs out

Watch what Novartis did as Zolgensma's curve started to bend, and the strategy comes into focus. In November 2024 it acquired Kate Therapeutics in a deal worth up to $1.1 billion — an undisclosed upfront payment plus milestones — bringing in preclinical AAV gene therapies for Duchenne muscular dystrophy, facioscapulohumeral dystrophy, and myotonic dystrophy type 1.⁶ These are exactly the kind of fresh genetic targets a draining franchise needs. A year earlier, Novartis had finished remaking itself: with the October 2023 spin-off of Sandoz it became a pure-play innovative-medicines company,¹⁰ and it now classifies gene and cell therapy as one of three 'emerging platforms,' alongside radioligand therapy and xRNA.⁷ It is hedging across modalities — in October 2025 it moved to acquire Avidity Biosciences to push its xRNA approach in neuromuscular disease.⁸ Gene therapy is a leg of the stool, not the whole stool.

But the most telling move is the cleverest, and the nearest at hand. In November 2025 the FDA approved Itvisma — an intrathecal reformulation with the same active substance as Zolgensma, authorized for SMA patients aged two and older.⁵ Read that carefully. It is the same drug, redelivered, aimed squarely at the one population the original approval excluded by design. Where Zolgensma had drained its market, Itvisma opens a new one out of the very same science, priced at $2.59 million a dose with multibillion-dollar peak ambitions.⁵ This is the platform thesis being tested in the most direct way possible: can you wring a second franchise out of an asset whose first franchise was already curing itself out of existence?

Isn't this just an expensive one-hit wonder?

The fair objection is blunt: Novartis paid $8.7 billion, got one declining product and a pre-approval scandal, and the rest of the pipeline is largely preclinical — a long, expensive way from anything that bills. That is all true, and it should temper the cheerleading. Kate's programs are years from a market; xRNA and radioligand are separate bets entirely; and 'emerging platform' is corporate language for 'not yet proven.' But the steelman cuts the other way too. The thing Novartis actually bought in 2018 was the institutional capability to commercialize a gene therapy at scale — the manufacturing, the $2-million reimbursement playbook, the regulatory muscle — and that capability is what made Itvisma possible in 2025 — built on the same active substance and the same platform logic, even if it required its own intrathecal reformulation and separate trial program. The first product was the tuition. Whether it was worth it depends entirely on the second act, and Itvisma is the first chapter of that act arriving exactly when the franchise needed it. That timing is not luck; it is the strategy showing its work.

Novartis made the most credible big-pharma bet on gene and cell therapy not because Zolgensma was a triumph — it was, and then it wasn't — but because it understood early that in this business the launch is the easy part and the encore is everything. A cure is a product that fires once. A platform is a finger that can pull the trigger again. The $8.7 billion was never really for the drug that empties its own market. It was for the chance to keep reloading.