Cannabis policy and earnings

$MSOS $TLRY $CGC $ACB: Inside the 274 Pages That Closed the Cannabis Rescheduling Record

On August 17 every party in the DEA rescheduling case filed its closing brief. The Government asked for Schedule III. All five of the other filings asked that marijuana stay in Schedule I. Here is what the documents actually say, what the sector reported while the lawyers wrote, and which dates are real.

What happened on August 17

The evidentiary phase of the federal marijuana rescheduling case ended on a Monday, quietly, in a filing inbox. Chief Administrative Law Judge Derek C. Julius had set August 17 as the deadline for optional transcript corrections and optional post-hearing briefs, capped at fifty double-spaced pages each. Every designated party used the option. Together with the Government's own submission, the closing record runs to 274 pages.

The arithmetic of the filings is the fastest way to understand the proceeding. One brief, 52 pages, argues that marijuana should move to Schedule III. Five briefs, 222 pages, argue that it should not move at all, and they carry the signatures of all seven designated parties, because the states filed jointly with Smart Approaches to Marijuana and DUID Victim Voices filed jointly with Kenneth Finn. That imbalance is not an accident of drafting: it is the structural feature of this hearing, and it has been visible since June, when the DEA designated the seven outside participants and not one of them was a company or trade body in favour of rescheduling.

The file has two tracks, and they are constantly confused

Almost every misunderstanding about cannabis policy in 2026 comes from collapsing two separate proceedings into one headline. They started on the same day and then went in opposite directions.

Track one: a rule that is already law

On April 28, 2026 a final rule took effect the day it was published. It moved a defined subset into Schedule III: marijuana products approved by the FDA, and, per the DEA's own description of the action, medical marijuana products regulated under state licence. That rule is in force now. It is also the rule that Smart Approaches to Marijuana and the National Alcohol and Drug Screening Association challenged in early May, with petitions later consolidated alongside one filed by Nebraska, Indiana and Louisiana. The challenge is about authority, not pharmacology: whether the Attorney General could reschedule a slice of marijuana defined by the kind of licence the seller holds.

Track two: the case that is still open

The broader question, moving marijuana as a class, is still the proposed rule from May 2024, revived by the hearing notice published on the same April 28. The hearing opened on June 29 before Judge Julius, ran seventeen hearing days, and closed with final arguments on July 15. The August 17 briefs are the last word from the parties in that proceeding. Nothing has been decided.

Inside the briefs: one test, two versions, and a fight about who decides

Strip away the volume and the case turns on a single statutory phrase. To sit in Schedule I a substance must have a high potential for abuse, no currently accepted medical use in treatment in the United States, and a lack of accepted safety for use under medical supervision. Only the middle clause is genuinely contested here, and the argument is not about whether patients use cannabis. It is about which test measures whether that use counts.

The Government's case

The Government brief, signed by James J. Schwartz and Jarrett T. Lonich and dated August 17, asks the tribunal to recommend transfer to Schedule III. Its spine is an opinion from the Office of Legal Counsel that binds the DEA. That opinion concluded that the traditional five-part test for accepted medical use is insufficient for marijuana, because it takes no account of the therapeutic use happening inside state-sanctioned programmes, and that the two-part test used by the Department of Health and Human Services is legally sufficient to establish it. On the facts, the Government leans on the HHS review: more than 30,000 practitioners recommending marijuana to over six million patients across 43 jurisdictions, a landscape it reads as incompatible with a finding that there is no accepted safety under medical supervision. On abuse potential, the argument is comparative rather than absolute — that the profile aligns with Schedule III substances rather than with Schedule I or II — and it asks for significant deference to the HHS assessment.

The case against

The joint brief from Smart Approaches to Marijuana and the opposed states, signed by John M. McNichols and Patrick F. Philbin, is the longest single filing at 73 pages, and it is framed as proposed findings of fact and conclusions of law rather than as rhetoric. Its central claim is that the five-part test is the correct one, that federal agencies applied it at least nine times over half a century and reached the same answer every time, and that a two-part test introduced for this proceeding cannot carry the weight of a scheduling change. The Tennessee Bureau of Investigation, in 26 pages, presses the burden of proof: under 21 C.F.R. 1316.56 the proponent of rescheduling must prove its case, and in its reading the Government has not. The same brief raises the jurisdictional argument that runs underneath the whole file — that the Attorney General had no authority to issue the notice of proposed rulemaking, which would make any resulting rule contrary to law under the Administrative Procedure Act. NDASA, DUID Victim Voices with Kenneth Finn, and the pharmacist Phillip Drum fill out the record with impairment, workplace testing, paediatric exposure and product-potency arguments.

What happens next, and why no one can give you a date

The procedural map from here is short and entirely open-ended. Judge Julius issues a recommended decision, which is not binding and for which no date has been set. Parties then have twenty days from receipt to file exceptions. The complete administrative record passes to the DEA Administrator, Terrance Cole, who decides. No provision of law imposes a deadline on that decision. If a final rule follows, it is published in the Federal Register and is itself appealable.

That is four sequential steps, three of which have no clock attached. A December decision is possible. So is nothing at all before 2027. Any specific date circulating in cannabis social media is a guess wearing the clothes of a fact, and the difference matters because positioning built on a date that does not exist is positioning with a hidden expiry.

The three channels that move money while the courtroom is quiet

The scheduling headline is the loudest part of the file and, for most listed operators, not the part with the fastest cash consequence. Three other channels are live right now.

Tax, which is the one that reaches the bank account

Treasury and the IRS have said guidance is coming on how Section 280E applies after the April order, including apportionment for businesses that hold both medical and adult-use licences. For a US operator still filing under 280E, that guidance decides how much of its own gross profit it is allowed to keep. It is worth more, in cash terms, than any wording about schedules, and it does not require the DEA to do anything.

The D.C. Circuit, which could reach backwards

The consolidated challenge to the April order is still at the threshold stage. A ruling there would not be a verdict on whether cannabis is dangerous; it would be a verdict on whether the Attorney General had the power to reschedule a subset of marijuana defined by licence category. An adverse decision would unsettle a piece of the file that the market has already filed under settled.

The health bureaucracy, which has already moved federal money

In April the Centers for Medicare and Medicaid Services launched an initiative covering up to $500 a year of hemp-derived products for eligible Medicare patients, focused on CBD but allowing up to 3 milligrams of total THC per serving. A lawsuit brought by legalization opponents to stop it was dismissed and is now on appeal. Separately, the FAA has said it is weighing new marijuana standards for pilots and air traffic controllers. Neither story trends. Both are the federal government behaving as though the April order is real.

The hemp deadline moved, and the trade-off is not the obvious one

On August 2 Senate Appropriations leaders released the text of a continuing resolution funding federal agencies through December 11. Inside it are provisions delaying the redefinition of hemp — the rule that from November 12 would have limited legal hemp products to 0.4 milligrams of total THC per container — to that same December 11 date. The delay carries a carve-out: synthetic cannabinoids that a cannabis plant cannot naturally produce would still be recriminalized on November 12, as originally planned.

Two qualifications travel with it. The House has already passed its own version of the resolution with no hemp language, so the provision must survive another vote in both chambers before it is law. And the effect on licensed marijuana operators runs the opposite way to the intuition: survey work published in late July found that state hemp bans push consumers toward licensed marijuana retailers. A delay to the federal ban therefore postpones a competitive tailwind for the licensed trade rather than removing a threat to it.

What the companies reported while the lawyers were writing

Between late July and mid-August the sector produced its own set of facts, and they are more useful than the policy noise because they are audited, dated and comparable to the same quarter a year earlier. Currencies differ by issuer and are stated as reported.

Three readings come out of that table, and none of them is the one the sector usually tells itself.

Profitability arrived, cash did not follow. Four of the five names printed positive adjusted EBITDA. Four of the five also burned cash in the same period. Adjusted EBITDA has become the number the sector reports and the free cash flow line has become the number that describes it. Canopy is the clearest illustration: an adjusted EBITDA loss that narrowed by 59% year on year, sitting next to a free cash outflow that more than doubled.

Growth is coming from outside the home market. Aurora's total revenue fell 9% while its international medical business grew 17% to C$43.3m, and Canadian medical fell 25%. Cronos grew 58% at group level. The Canadian adult-use market is not what is producing the improvement; international medical and non-Canadian channels are.

Net income is not the number to read. Organigram posted C$105.5m of net income on C$105.8m of revenue, which sounds like a transformation and is not one. Operating income for the quarter was C$5.3m. The distance between those two lines is fair-value accounting: the interim statements show roughly C$100.9m of positive fair-value movement on derivatives, preferred shares and other financial assets, plus C$2.3m on contingent consideration, with pre-tax income of C$103.9m and a C$1.6m tax recovery closing the gap. The bottom line is instruments being revalued, not cash arriving.

And one live corporate situation

On August 11 Curaleaf announced an intention to launch an unsolicited take-over bid for Aurora, at 0.3463 of a Curaleaf subordinate voting share plus US$0.75 in cash per Aurora share, an indicated value of US$4.00 with a US$5.00 cap. No formal offer has commenced. Aurora formed a special committee and told shareholders to take no action. That is a proposal, not a transaction, and the distinction is the whole story for anyone modelling it.

$MSOS: the instrument most people use to trade this file, and what is actually inside it

When retail flow wants exposure to a US federal cannabis headline, it usually buys the AdvisorShares Pure US Cannabis ETF rather than a single operator, because most US multi-state operators do not trade on a major US exchange. That makes $MSOS the de facto ticker of the policy trade, and the contents of the vehicle decide what that trade actually is.

As of August 18, 2026 the fund reported net assets of about US$914.1m, a NAV of US$4.55 and 200.88 million shares outstanding. The gross expense ratio is 0.81% and the net ratio 0.78%, against a contractual cap of 0.74%. The official holdings file dated the same day shows the concentration: Trulieve at 28.74%, Curaleaf at 27.23%, Green Thumb Industries at 19.04% through a swap, Glass House Brands at 7.55% and Verano at 6.05% through a swap.

Three of those positions are roughly three quarters of the fund. Several of the largest are held through total return swaps rather than directly, because the underlying shares are not readily accessible to a US-listed fund, and the holdings file also carries a large negative cash line that is the mirror image of that swap financing. None of this is hidden — it is published — but it changes what the ticker means. A buyer of $MSOS is not buying a diversified index of the US cannabis industry. They are buying a concentrated position in three operators, some of it synthetically, wrapped in a fund that costs about 0.78% a year to hold.

$CGC: a vote, a quorum problem and an auditor who walked

Canopy Growth filed its definitive proxy on August 7 and mailed the meeting materials on August 17, the same day the DEA briefs landed. The annual general and special meeting is on Friday, September 25 at 1:00 p.m. Eastern, virtual only, with a record date of July 31 and proxies due by 1:00 p.m. Eastern on September 23.

The company is unusually direct about why it is pushing for early votes. Nasdaq rules require a quorum of one third of outstanding shares, and the proxy states that reaching it depends on strong participation given the shareholder base, with an adjournment and additional cost if it is missed. That is a retail-heavy register talking about itself.

What is on the ballot

Seven items. The election of five directors. The appointment of MNP LLP as auditor for fiscal 2027. A share consolidation proposal, at a ratio the board picks within a one-for-five to one-for-fifteen range, exercisable at its discretion within a year. An advance notice by-law. Renewal of the omnibus incentive plan. Say-on-pay, and say-on-frequency.

The consolidation is preventive rather than remedial. The stated objective is to be in a position to raise the per-share price to meet continued listing requirements in the event the price declines. No deficiency notice has been received. Shareholders approved a substantially identical resolution in 2025 and the board never used it. The share price context is not comfortable — the stock closed fiscal 2026 at US$0.95 on March 31 — but the authorisation and the action are two different things, and the 2025 precedent is the evidence for that.

The item that was not in the preliminary proxy

On August 7, PKF O'Connor Davies resigned as auditor with immediate effect. The stated reason is not about Canopy: the firm is stepping back from the cannabis sector. The Form 8-K records no disagreements on accounting principles, practices or disclosure, and the reportable events it lists are ones already on file — the adverse opinion on internal control over financial reporting at March 31, 2026, and the restatement of fiscal 2024 and 2025 disclosed in May. The audit committee engaged MNP LLP the same day for the year ending March 31, 2027.

Put in sequence, that is a company with an unremediated material weakness in the accounting for equity-linked instruments, changing auditor mid-cycle, and asking shareholders to ratify the new firm at a meeting whose quorum is not assured. Each element is explainable on its own. Together they describe where the reporting credibility of this issuer currently sits.

What Schedule III would actually change, and what it would not

The word rescheduling carries more freight than the mechanism can bear, so it is worth separating the parts of the story that a Schedule III outcome would touch from the parts it would leave exactly where they are.

It would change the tax base of US operators

Section 280E of the tax code denies ordinary business deductions to any trade trafficking in a Schedule I or II controlled substance. Moving marijuana to Schedule III removes that denial prospectively. For a multi-state operator, this is not a marginal improvement in the effective tax rate: it is the difference between paying tax on gross profit and paying tax on actual profit. It is the single largest cash item in the whole debate, and it is also the item that Treasury and the IRS can shape through guidance without any further action from the DEA.

It would make research meaningfully easier

Schedule I research requires registrations, secure storage and protocol review that most academic groups will not attempt. Schedule III lowers that wall. The effect is real, slow and invisible in any quarter: it shows up in trial registrations over years, not in a print.

It would not create interstate commerce

A federal schedule change does not merge fifty state markets into one. Product would still not cross state lines, licences would still be state-issued, and the cost structures that come from building a supply chain per state would remain exactly as they are. Anyone modelling national scale on the back of a scheduling decision is modelling a different reform.

It would not fix banking, and it would not uplist anyone by itself

Access to banking is governed by anti-money-laundering supervision and by the risk appetite of institutions, not by the schedule alone. Exchange listing is governed by exchange rules. Both would become easier conversations, and neither becomes automatic. This is the gap between the headline the sector wants and the mechanism it would get.

Why the record is one-sided, and why that is a two-edged fact

Seven designated participants, all of them opposed, is an unusual configuration for a proceeding of this size. It happened because participation was granted through a designation process that ran in April and May, and the parties that pursued designation hardest were the ones with an institutional interest in the status quo: state law-enforcement bodies, drug-screening and impaired-driving organisations, prohibitionist advocacy, and individual clinicians. Industry, by and large, did not appear as a party.

That produces two effects that pull against each other. In the short term it means the evidentiary record now sitting in front of Judge Julius was built almost entirely by the side that wants nothing to change, and the Government is the only voice in the file arguing for the move. In the longer term it means the opposing parties have spent seventeen hearing days and 222 pages building the record they would need on appeal, which is what a party does when it expects to lose the first round.

Both readings are visible in the briefs themselves. The jurisdictional argument, that the Attorney General had no authority to issue the notice of proposed rulemaking, is not a scheduling argument at all. It is an argument aimed at a court, filed into an administrative record so that it exists there when the case moves. Reading the closing briefs only as an attempt to persuade the judge misses half of what they are for.

How to read the next cannabis headline without getting caught

The file is now in a phase where the loudest stories will be the least decisive, because nothing is scheduled and everything is anticipated. Four filters keep the reading honest.

Ask which track the headline belongs to

If a story concerns FDA-approved products or state-licensed medical marijuana, it belongs to a rule already in force and its news value is about implementation or litigation. If it concerns marijuana as a class, it belongs to a case with no decision and no date. The two get merged in headlines constantly, and the merge is where mispricing starts.

Ask whether the document exists

The August 17 briefs are real and quotable, and they are still not on the DEA docket. A recommended decision, when it comes, will be a document with a date on it. Until a document exists, a report about a decision is a report about expectations. The distinction sounds pedantic right up to the moment it costs money.

Ask who is exposed to it

A Schedule III outcome for marijuana as a class would mostly change the tax position of US operators under 280E. It does not automatically create interstate commerce, it does not resolve banking, and it does not uplist anyone to a major exchange by itself. A Canadian LP with international medical revenue and a large cash pile has a different exposure to the same headline than a US operator carrying a 280E tax bill, and $MSOS, being three quarters concentrated in three US operators, is not a hedge against either.

Ask what the balance sheet allows

The one thing this proceeding has made certain is that it will take longer than anyone wants. In that world, the relevant question about a company is not what it thinks of Schedule III but how many quarters it can fund while waiting, and whether the cash statement agrees with the adjusted EBITDA line. On the numbers above, that question separates the sector more sharply than the policy does.

Dates that are real

The bottom line

August 17 closed the record in the biggest federal cannabis proceeding in fifty years, and it closed it with the argument almost entirely one-sided in volume and entirely one-sided in direction: five of the six filings ask that nothing change. The Government's case rests on a legal opinion that binds the agency and on an HHS finding about how medicine is actually practised in 43 jurisdictions. The opposing case rests on the older test, on the burden of proof, and on a jurisdictional argument that, if it lands, makes the substance of the fight irrelevant.

What the sector does while that is decided is now measurable. Tilray closed a record fiscal year with almost no net debt. Cronos is sitting on more cash than several of its peers are worth. Aurora is growing where the margin is and shrinking where it is not, while a competitor circles it. Canopy is improving its profit and loss faster than its cash flow, changing auditor mid-cycle, and asking a dispersed shareholder base to show up to a meeting. None of that depends on Judge Julius, and all of it will still be true when he writes.