Universities run the psilocybin trials, and the American government buys the drug

The four medicines are Marinol and Cesamet in 1985, Syndros in 2016 and Epidiolex in 2018. The plant they came from is still a Schedule I substance, and the rescheduling docket reached notice of hearing on 28 April 2026, which is a stage and not an outcome.

Four molecules completed the route. The plant did not.

One published series says what the licensed route eventually took, and two things have to sit in front of it. Cannabis had a retail channel from its first day, where every psilocybin route on this map is clinical or supervised, so the mechanism that moved the cannabis number does not exist here. And the unlicensed half of the series is a modelled national-accounts estimate, which makes every share on it partly a model. With both of those attached: Statistics Canada put the licensed route at 12.8 percent of household spending on non-medical cannabis in the quarter legal sales began, 24.3 percent across 2019 and 70.8 percent across 2025, on figures read on 7 September 2026.

Nothing on the psilocybin map has a decade of data.

Universities and hospitals sponsor 233 of the 328 trials registered on 5 September 2026, and 175 of the 328 name nothing that points at a funder. The United States government both funds the research and buys the drug.

The whole split is 233 public or academic sponsors, 38 companies, 43 unclassified and 14 filed by a named individual. The company figure is a floor. An unclassified row may well be a company nobody has typed yet. Of the 328, 153 carry some signal and 35 a government one. A signal names no payment, and no registry publishes an amount.

Company-typed starts are the one series here that moves, and the most recent of its eight buckets is an incomplete year. With that in front: they peaked at 11 in 2022 and were 6 in 2025, while total starts went from 30 to 58 between those two years.

Where the public money is stated it is concentrated in two places. NIDA administers 64 of the 150 NIH project-year records under the narrow predicate on the 7 September 2026 read, and NIMH 41, which is a count of records and not a share of dollars. On the same 5 September 2026 read, 12 trials of the 328 join a grant record by grant number, and 301 name no grant identifier at all. The grant field is registry housekeeping that sponsors fill in at their discretion.

The United States award file states 72 awards naming psilocybin on the 7 September 2026 read, being 63 grants, four procurement contracts and five direct payments. That 72 is not comparable with the 150 NIH project-year records above it: an award is not a project-year. The two purchases above sit in that file: the NIH’s from Benuvia Operations, and the Department of Veterans Affairs’ from Usona Institute. The other two contracts are SAMHSA awards to Oregon Health & Science University at US$249,999.00 each, for harm-reduction data. Four contracts is a floor, and it is a real one: the same query run against an approved comparator returns a full purchasing history, so the filter works and four is what it finds.

One of those sellers files a tax return, and two years of one organization’s contributions read as a collapse unless the reason sits in front of them: the earlier figure is read from the prior-year column of the later return, and the difference may be nothing but the accounting calendar of multi-year gifts. With that attached, Usona Institute states contributions of US$5,020,285 for tax year 2024 against US$53,766,608 for 2023.

On that same read, Usona supplies the study drug in 11 trials and sponsors four, and Filament Health Corp. supplies 16 and sponsors one. A sponsor index cannot see either of them, which is most of why nobody can tell you who supplies this field.

Eli Lilly agreed to buy AtaiBeckley on 15 July 2026, stated in AtaiBeckley’s own quarterly filing. Seven listed companies held a place on this shelf when the capital pack read it on 7 September 2026, every one of them a parent, and none of them sponsors a trial in its own name. The research is run by someone else.

What the seven have in common is thinner than the label suggests. Four state a psilocybin molecule of their own. One is a licensed producer, in the words of its Form F-1/A of 13 March 2026. One reaches psilocybin only through a wholly owned subsidiary, and one carries a programme that is not psilocybin at all.

The molecule test undercounts the capital in the room. An issuer that is not on this shelf names psilocybin twice in its own annual report, both times about somebody else’s candidate.

AtaiBeckley completed its acquisition of Beckley Psytech Limited on 5 November 2025, stated in its own annual report. A joint venture between Jaguar Health’s Magdalena Biosciences and Filament Health Corp. is stated in an 8-K exhibit of 21 April 2026. Across the whole axis, as it was re-derived on 9 September 2026, there are ten organization-to-organization rows over nine distinct relationships, and to the register a joint venture and a supply agreement are the same kind of thing.

No regulator has approved a psilocybin medicine, in anything read for this work. Three FDA pages about its own priority-voucher programme, read on 7 September 2026, mention no psilocybin product, and that is a fact about those three pages, not a reading of the agency’s files.

Its guidance on psychedelic clinical investigations is final at 91 FR 43101, with the notice of availability published on 14 July 2026, and a guidance states what the regulator expects to see in a trial while opening no route to an approval. A complete response letter for one psychedelic application, NDA 215455, midomafetamine, sits on the FDA’s own download host with the date 8 August 2024, naming functional unblinding and asking for a new trial. That document concerns a different compound.

Oregon charges a manufacturer, service centre or laboratory US$10,000 a year under a fee rule effective 1 January 2026, and a fee is what a licence costs its holder and never what the state collects. Who paid to put these measures on the ballot in the first place is a separate question, with two state registers behind it: who paid to change the law.

Eighteen market-research firms sell a figure for this field, and on the pages anyone can read, seventeen of them do not say what they counted. What governments and funders have counted instead is five different things on five incompatible clocks, none of them a revenue figure. That is its own piece: how big is the psychedelic drugs market.