The Compounding Boom, and How It Ended

In the first week of February 2026, the telehealth company Hims & Hers announced a compounded pill version of semaglutide. Within days the Food and Drug Administration said it would restrict the active pharmaceutical ingredients intended for non-approved compounded GLP-1 drugs, warned that no company could present such products as generic versions of approved medicines or as the same as them, and noted that its tools included seizure and injunction 1. The pill was withdrawn. On 9 February the manufacturer of the approved drug sued the company for patent infringement 8.

It was the last act of a story that should, on paper, have ended a year earlier. An earlier piece on this site told how American compounding law came to be written after a 2012 outbreak, and how its shortage provision opened a window for copies of the GLP-1 drugs in 2022 and closed it in 2025. This is the sequel: the business that grew inside that window, why it kept operating after the window shut, and the enforcement year — 2026 — that brought it to an end in the United States, Britain and South Africa within a few weeks of each other.

A business built on a list

The word compounding conjures a pharmacist with a mortar. What grew between 2022 and 2025 looked nothing like that. The typical arrangement had three parties: a telehealth platform that marketed the product and ran a short online consultation, a prescriber working with or for the platform, and a pharmacy or registered outsourcing facility that prepared the medicine and shipped it to the patient's door. The patient paid the platform, usually by monthly subscription, and very often never learned which pharmacy had made what arrived.

Every element of that model depended on the shortage listing. The listing did not create the demand, which came from trial results and from list prices that most insurers would not cover for obesity. What it did was suspend the rule against making essentially a copy of an approved drug, which allowed a platform to offer the same molecule, prepared from bulk active ingredient bought on the open market, at a price the manufacturer's own channel was not offering. A statutory safety valve had become the foundation of a consumer business.

The branding was the part regulators came to object to most. When the agency later reviewed these companies' websites, the violations it cited fell into two families: claims or implications that a compounded product was the same as, or a generic of, the approved one; and products advertised under the telehealth firm's own name or trademark without qualification, so that the platform appeared to be the maker 2. Neither was a technicality. Both hid from the person paying the thing the approval system exists to make visible — who made this, and to what standard.

The window closed; the business did not

The regulator declared the tirzepatide shortage resolved in December 2024 and the semaglutide shortage in February 2025, and the last grace period, for registered outsourcing facilities, ran out on 22 May 2025. On paper, the mass copying of these molecules was over.

In practice it continued, and the clearest evidence came from the party with most to lose. In July 2025, when the manufacturer of semaglutide lowered its outlook for the year, its filing stated that despite the expiry of the grace period, its market research showed that what it called unsafe and unlawful mass compounding had continued — and that, as a result, its own product's penetration among cash-paying American patients had been lower than expected 4. A compounding sector was now visible in a multinational's profit warning.

The legal space that remained was the patient-specific route. A pharmacy may still prepare a medicine for an individual patient where a prescriber judges that a change from the approved product — a different strength, an added ingredient, a different form — makes a clinically significant difference for that person. Used as intended, that is exactly what compounding is for. Used at the scale of a subscription business, with the same modification applied across an entire customer base, it was hard to distinguish from the copying that the shortage provision had allowed and its ending had forbidden. Much of the enforcement that followed turned on that difficulty.

February 2026

The pill announcement mattered because of what it implied. Oral semaglutide is an unusually demanding formulation: the molecule crosses the gut wall poorly, and the approved tablet depends on an absorption enhancer and years of development. A compounded version, prepared outside that development and outside any shortage, was a direct test of how far the patient-specific route could be stretched. The regulator's answer was to go upstream. Rather than pursue preparations one pharmacy at a time, it said it would restrict the active pharmaceutical ingredients themselves where they were intended for non-approved compounded drugs 1.

That was a change of method rather than of law. Bulk active ingredient is where every compounded preparation begins, and restricting it reaches every downstream product at once. The manufacturer's lawsuit, filed three days after the agency's statement, alleged that the company's compounded products infringed one of its American patents and that its promotion suggested equivalence with the approved medicine 8.

On 3 March the agency sent 30 warning letters to telehealth companies over their marketing of compounded GLP-1 products, citing once again the implication of sameness and the branding that obscured the real compounder 2. On 30 April it proposed to exclude semaglutide, tirzepatide and liraglutide from the list of bulk substances that registered outsourcing facilities may compound from, stating that it had not identified sufficient clinical need for any of them 3. The sequence that shortage had set in motion in 2022 was being run backwards: first the shortage ended, then the ingredient was restricted, then the route itself was proposed for closure.

The same month, elsewhere

Britain never had an American-style compounding sector for these molecules. Its medicines law has its own route for unlicensed preparations made to meet an individual patient's special clinical needs, but nothing resembling the American shortage exemption, and no comparable boom grew around it. What it had instead was manufacture with no pretence of pharmacy at all. On 28 May 2026 officers from the medicines regulator's criminal enforcement unit raided a country estate near Northampton and recovered around 12,000 doses of unlicensed weight-loss medicines, including preparations labelled as retatrutide and tirzepatide alongside other peptide products. Two men were arrested on suspicion of offences under the Human Medicines Regulations 2012. The agency called it its largest ever seizure of such products 5.

Five days earlier, South Africa's medicines regulator and its pharmacy council had announced intensified joint enforcement against the unlawful manufacture and distribution of unregistered GLP-1 and GIP medicines containing semaglutide, tirzepatide or both. An inspection of one pharmacy operation on 11 May had found, the regulator said, medicines being produced and supplied under the pretext of compounding but outside the legal framework, with illegally imported active ingredients and no analytical testing to confirm identity, potency or purity. The injectable products found on site were seized 6.

Three legal systems, three vocabularies — compounding, unlicensed manufacture, unregistered medicines — and one underlying pattern. Demand for a class of medicine far exceeded what the approved channel supplied at a price people could pay, and a secondary supply grew in whatever legal form each country's law made nearest to hand. Within a few weeks of one another in 2026, all three regulators decided that form had been stretched past what it could bear.

What actually ended it

Enforcement was necessary; it was probably not sufficient. The compounding boom lived in the gap between what the approved product cost and what people were willing or able to pay, and in late 2025 that gap began to close from the other side. In November the semaglutide manufacturer cut the monthly cash price of its injectable products for existing self-paying American patients from $499 to $349, and introduced a lower introductory price for new ones 9. An approved oral version for obesity followed in January. A subscription to a compounded copy is a much harder sell when the approved medicine costs not very much more.

The regulator, meanwhile, kept a running count of a different kind of cost. By the end of May 2026 it had received 990 adverse-event reports associated with compounded semaglutide and more than 730 associated with compounded tirzepatide, and it stressed that events of this kind are under-reported 7. Reports like these do not establish that a product caused any particular harm. They are, however, the only systematic record that exists for preparations that never went through the approval process, and they were accumulating.

Looking back, the whole episode has the shape of a natural experiment in what a shortage provision does when the thing in short supply is also the most commercially attractive medicine of its decade. The provision worked as drafted when it opened. It worked as drafted when it closed. What nobody had drafted for was a consumer business that had grown so large inside the opening that it could not simply stop, and that had to be closed three times — by the end of the listing, by the restriction of its raw material and by the proposed removal of its route — before it did.

Where the story stops

The boom's end leaves a question that law cannot answer and chemistry can. When a compounded preparation and an approved product carry the same molecule name, what, measurably, differs between them? The regulator has pointed to some of the candidates: salt forms of semaglutide, such as the sodium and acetate salts, for which it says it lacks information on whether they share the chemical and pharmacological properties of the approved active ingredient; impurity profiles that depend on the synthesis route; and sterility and endotoxin control, which belong to a facility rather than to a molecule 7. Those are analytical questions with analytical answers, and they are where a technical article would begin.