The migration to other molecules.

The prediction, when the FDA closed the semaglutide shortage on February 21, 2025, was that the compounding industry would contract. It did not contract. It moved, and it kept moving, and the direction it has taken over the last eighteen months is more interesting than anything the molecules themselves are doing.

Here is the thesis. What the sector did after losing the right to copy GLP-1s was not diversification into new therapeutic categories. It was the relocation of a single business model, regulatory arbitrage on the definition of what may lawfully be compounded, toward whatever molecule that definition leaves open. First sideways, into modified versions of the same drugs. Then outward, into a different class of peptides entirely.

And then, on July 23 and 24, 2026, the model changed shape in a way that has been reported almost entirely as a peptide story. It is not a peptide story. An FDA advisory committee voted to recommend six of seven peptides for compounding eligibility after the agency's own scientists had recommended against all seven in writing. I think those vote tallies are the most consequential thing to happen to compounding policy since the shortage closed, and the margins matter more than the outcome.

The minimum background. After the FDA resolved the tirzepatide shortage on October 2, 2024 and reaffirmed it on December 19, 2024, and resolved semaglutide on February 21, 2025, the legal basis for compounding essentially-copies of those drugs expired on a staggered calendar running from February 18 to May 22, 2025. We covered the enforcement mechanics in the compounded GLP-1 crackdown. This essay is about where the capacity went afterwards.

The first migration went sideways, and the tell was vitamin B12

The initial move was not to a new molecule at all. It was to a modified version of the old one.

Federal law bars compounding a drug that is essentially a copy of a commercially available product. So compounders differentiated. The most common addition was vitamin B12, followed by B6, niacinamide, glycine and carnitine, none of which has demonstrated benefit for weight management or glycemic control. Alongside that came dosing presented as individualized rather than standard, which does additional work in the same legal direction.

I want to be precise about why this is the tell. Nobody added B12 to a GLP-1 for a clinical reason. The additive answers a legal question, not a physiological one. This was never a therapeutic strategy. It was a compliance strategy in a therapeutic costume.

It also turned out not to be inert. A 2026 analysis found that when tirzepatide is compounded with B12, the two can chemically bond, producing a molecule not present in the approved product. Whatever that is, it is not a copy, and nobody characterised it before it was being dispensed at scale.

Regulators moved on the marketing before they moved on the chemistry. The FDA issued more than 55 warning letters to online sellers of compounded GLP-1s on September 16, 2025, and a further tranche of roughly 30 to telehealth companies over their promotional claims. Updated guidance in June 2026 drew the line harder, barring platforms from implying clinical equivalence to the branded products or any FDA approval of their custom mixtures.

The financial evidence that this worked is on the public tape. Hims & Hers reported $586 million in Q1 2025 revenue, up 111% year over year, at the peak of the compounded book. After pivoting toward branded product, a subsequent quarter showed revenue of roughly $608 million and a net loss of about $92 million. On February 6, 2026, HHS General Counsel Mike Stuart announced the company had been referred to the Department of Justice. Revenue held. Margin did not. That gap is the price of the model.

The second migration went to molecules that were never available

The more interesting movement was outward, and it ran into a wall that predates the GLP-1 era entirely.

Between October 2023 and early 2024, the FDA moved seventeen peptides into Category 2 of its 503A bulk drug substances review, the bucket for substances judged to raise significant safety concerns. Category 2 is stricter than a shortage rule in one crucial way: it does not depend on supply. A Category 2 substance is off the table whether or not anything is scarce.

So the demand did not get served through pharmacies. It got served outside them. Cryptocurrency inflows to unregulated peptide sellers have grown for six consecutive quarters and are tracking toward a $100 million annual run rate in 2026, against roughly $1 million per quarter in 2024. A hundredfold move in about two years.

Ozemback does not name, rank or link to any of those sellers, and will not. The figure is cited for one reason: it is the cleanest measure of what happens to demand when a legal channel closes and an illegal one does not. Demand did not disappear. It deregulated.

April 15, 2026 looked like a door opening. It was a hinge

Then the wall moved, and most coverage misread the movement.

On April 15, 2026, the FDA announced it would remove twelve peptides from Category 2. The stated reason was procedural: the nominations had been withdrawn by the parties who submitted them. Safety questions were not resolved. Nothing was affirmatively cleared. A substance left a restrictive list because the paperwork that put it there was pulled.

This was widely read as authorization. It was not. Removal from Category 2 does not permit compounding: a substance still has to clear Pharmacy Compounding Advisory Committee review and then be placed on the 503A bulks list through notice-and-comment rulemaking. April removed an obstacle. It granted nothing.

On July 9, 2026, the FDA posted its scientific briefing documents ahead of the committee meeting and recommended that the panel add none of the first seven peptides under review. The stated grounds were incomplete molecular characterization, little or no human efficacy data, and unresolved immunogenicity concerns. That is the agency's own scientific staff, on the record, against all seven.

Then the committee overruled the agency

On July 23 and 24, 2026, the fourteen-member Pharmacy Compounding Advisory Committee voted the other way on six of them.

Day one: BPC-157, TB-500 and KPV each passed 8 to 6 with one abstention, and MOTS-c passed 7 to 5 with two. Day two: Semax passed 8 to 5 and Epitalon 7 to 5, each with one abstention, while emideltide, also called DSIP, was voted down 6 to 7 with one abstention. Six recommendations, one rejection, every one of them against the written position of the agency's own scientists. A further five peptides are scheduled for a committee meeting before the end of February 2027.

The most underrated fact here is not which peptides passed. It is that none passed comfortably. An 8 to 6 result is not scientific consensus. It is a split room, and one already told by staff scientists that the characterization data is incomplete. Read the tallies rather than the headline and you are looking at a governance outcome, not an evidentiary one.

None of it is binding. A PCAC recommendation is advice, and final placement on the 503A bulks list requires a separate rulemaking that the FDA controls and can decline to initiate.

Why this changes the business model, not the catalogue

Here is why I think the votes matter more than the molecules.

Shortage-based compounding is structurally temporary. It is lawful because of a supply failure, and supply failures end, on a schedule the compounder cannot influence. Every business built on a shortage is built on a countdown. That is what the 2024 to 2025 deadlines demonstrated.

A bulks-list placement has the opposite property. It does not expire when a fill-finish plant comes online, and does not depend on anyone else's manufacturing failure. It is a permanent statutory home.

So the sector's incentive migrated too, from monitoring shortage lists to influencing an advisory process. Watching a database is cheap and reactive. Working a rulemaking is expensive, slow, and durable if it lands. The industry spent eighteen months learning that the first strategy always ends, and has now demonstrated, by a margin of two votes, that the second can work.

Let me be explicit about scope. This is market, regulatory and field analysis. Ozemback does not recommend, endorse, rank or refer anyone to any pharmacy, compounder, platform, seller, clinic or prescriber, and nothing here describes any substance as suitable for use by anyone. Several compounds named above are, by the FDA's own July 2026 assessment, incompletely characterized with little or no human efficacy data. That is a fact about the regulatory record, not a comment on what anyone should do.

The strongest case that I am wrong

Three objections, and the second is the one with real force.

One: PCAC recommendations are advisory and frequently go nowhere. The agency's scientists are already on record against, rulemaking takes years, and reading a non-binding committee vote as a structural shift is reading an input as an outcome.

Two: the steelman for the compounders is better than critics allow. The objection that a molecule lacks human efficacy data is partly circular when the trials that would generate it cost hundreds of millions and nobody will fund them for a compound that cannot be protected by patent. Meanwhile the Category 2 nominations were themselves thin, which is precisely why twelve of them evaporated the moment the nominators walked away.

Three: the economics are trivial. Seven peptides at cash-pay volumes are a rounding error against a GLP-1 market measured in tens of billions.

My read: the first is correct on the facts, which is exactly why I framed this as a signal rather than a result. The second is the strongest and I concede much of it, because the evidentiary standard really is applied to unpatentable molecules in a way that is close to unmeetable by design. The third is the weakest, because it measures the wrong thing. The significance is not the revenue attached to seven compounds. It is the precedent.

What actually migrated

The compounding sector did not diversify after 2025. It carried one capability across a series of legal boundaries: the ability to find where the definition of a lawfully compoundable drug is soft, and build volume there before it hardens. B12 additives were that. The gray market absorbed what the pharmacies could not. And the July 2026 votes are the sector's first serious attempt to move a boundary rather than chase one.

What ended in 2025 was compounding's dependence on scarcity. What began in July 2026 is its bid for permanence. Those are not the same business, and the second is far more durable.

The molecules will keep changing, because they always have. The question worth watching through February 2027 is not which peptide is next. It is whether the boundary moves with them, and who decides that it does.

Ozemback — August 2026