Somewhere on Instagram right now, a man with excellent teeth is eating 500 milligrams of THC before running a marathon. The comments are full of fire emojis. The engagement is spectacular. And in ten years, this clip will be watched the way we now watch Four Loko commercials: as an artifact from the brief, loud window when a substance was legal enough to sell and new enough that nobody had figured out what it was for.

Here is what it's for. Pain, sleep, nausea, anxiety, appetite, the long grinding symptom load of chronic illness and cancer treatment. Here is the prediction, with dates attached: by 2036, edibles are primarily a medical category, the stunt economy around them is dead, and smoking cannabis occupies the same cultural position cigarettes occupy today. Declining, aging, faintly embarrassing.

This site obviously has a position in that future, so discount accordingly. Then look at the mechanics.

The stunt economy was always a phase

Every newly legal intoxicant goes through a carnival period. Post-Prohibition alcohol had one. Early vaping had one (remember cloud competitions, a genuine competitive scene built around exhaling). Cannabis edibles are in theirs now, and the tell is that the content economy around edibles rewards dose as spectacle. Nine gummies. 500 milligrams. "I ate an entire bag and went to Costco." The product is not the gummy. The product is the audience watching a man metabolize poor judgment.

Novelty economies burn out on their own, but this one has a regulator problem stacked on top. The intoxicating hemp market that supplied most of the stunt inventory, the mail-order 100mg-per-piece stuff with a COA from a lab nobody can call, starts shutting down federally this November, with the remainder winding down in December. What survives is the state-licensed dispensary channel, which is capped, tested, tracked, and deeply allergic to being the backdrop of a viral overdose video. Canada already shows the end state: legal edibles capped at 10 milligrams of THC per package, sold in packaging designed by someone who was told, at gunpoint, never to make it fun.

The party branding dies first because it was always the most fragile part. Nobody builds a durable business on giddiness.

The alcohol precedent, read correctly

If you want to know how fast an intoxicant's brand can rot, alcohol is running the demonstration live. In 2023, 62 percent of American adults drank. Today it's 54 percent, the lowest figure since Gallup started asking in 1939, before anyone reading this was born. A majority of Americans now says even moderate drinking is bad for your health, a belief held by 28 percent as recently as 2015. The surgeon general put a cancer warning on the category in January 2025, and average weekly consumption has fallen to levels last seen in the mid-nineties.

Read that carefully before drawing the lazy conclusion. Gallup's own analysis says the decline is not people switching to weed; cannabis use has been flat for four years while drinking fell off a cliff. Edibles did not beat alcohol. Health perception beat alcohol, and it beat the single best-constructed brand in the history of drugs. Centuries of accumulated meaning, wine as sophistication, beer as belonging, whiskey as masculinity, an entire architecture of identity, and it lost eight points in three years to the sentence "this gives you cancer."

That is the lesson that transfers. An intoxicant's commercial moat is its cultural position, and cultural positions collapse at the speed of a health narrative, decades of brand equity be damned. The one corner where actual substitution shows up makes the same point from the other side: THC seltzers earned their liquor-store shelf space by borrowing alcohol's occasions, the barbecue, the after-work wind-down, while shedding the calories, the hangover, and the warning label. They sell the ritual without the indictment. That is a brand-position play, executed against a wounded incumbent.

Now ask which cannabis format is standing where alcohol stood.

The boring structural argument for medicine

The medical turn is not a mood. It's already the revealed preference of every serious institution that has touched the plant.

The FDA has approved cannabinoid medicine twice: dronabinol, synthetic THC, in 1985 for chemotherapy nausea, and Epidiolex, purified CBD, in 2018 for seizure disorders. Nabiximols, the THC:CBD spray behind most of the cancer pain trial literature we cite on our cancer pain page, is approved in over two dozen countries. The DEA is deep into a formal proceeding to move cannabis to Schedule III, with post-hearing briefs filed this month. Schedule III is not legalization. It is something more consequential for this argument: it unlocks research at pharmaceutical scale, because Schedule I status is the main reason cannabinoid trials have been small, short, and underpowered for fifty years.

Follow that thread forward. Real trials produce real indications. Real indications produce FDA-approved formulations with defined doses. Approved formulations eventually produce the thing no dispensary product has ever had, which is insurance reimbursement. The moment an oncologist can prescribe a covered 2.5mg THC:CBD formulation for treatment-related insomnia instead of sending a patient to squint at a Wyld label, the medical demand routes through medicine. The demand was always medical anyway; a large share of "recreational" purchasing has been self-treatment for pain and sleep wearing a leisure costume because that was the only door open.

Recreation persists through all of this. It just stops being the story. It becomes what beer is becoming: a commodity in managed decline, tightly ruled, dully packaged, no longer pretending to be a culture. The interesting money, the research money, and the regulatory momentum all point the other way.

The timeline

2026 to 2028. The hemp wind-down consolidates the market into licensed operators. Schedule III lands. The stunt brands die or pivot; the first casualties are the ones whose entire identity was potency-as-content. State regulators, watching the hemp era's excesses, tighten potency caps and marketing rules in a wave.

2028 to 2031. The research dam breaks. Properly powered trials launch on the indications where the existing evidence is suggestive and thin: cancer symptom clusters, sleep maintenance, neuropathic pain, anxiety. Some fail publicly, which is healthy. Pharmaceutical companies that spent a decade pretending not to watch stop pretending.

2031 to 2034. The first FDA approvals beyond Epidiolex arrive, most plausibly in palliative and oncology settings where the risk calculus is most forgiving. The word "edible" starts to split in two: a regulated oral cannabinoid formulation on one side, candy on the other. Dispensary shelves bifurcate the same way, medical-grade low-dose products with real COAs versus commodity recreational.

2034 to 2036. Reimbursement begins for approved formulations. Combustion falls to a minority format in legal markets. The category's center of gravity is a pharmacy, and the marathon clip is a museum piece.

Legislation is slow, agencies are slower, and this schedule assumes no catastrophic political reversal, which is the one genuine wildcard. Everything else in that sequence is already in motion.

What happens to smoking

Cigarette smoking in America went from 42 percent of adults in 1965 to roughly 11 percent today, and it managed that decline without a substitute sitting on the same shelf. Cannabis combustion declines faster, for the alcohol reason and the pharmacy reason at once. The alcohol reason: smoke's health narrative is indefensible in a way wine's never was. Wine had antioxidant studies and French grandmothers to hide behind for twenty years; nobody has ever attempted a "combustion in moderation" campaign, because there is nothing to say. When the health-perception wave that just took eight points off alcohol reaches cannabis formats, and it will, it lands on the joint first and hardest. The pharmacy reason: no trial will ever be run on combustion, no physician will ever prescribe it, no insurer will ever cover it. Each medical expansion is, automatically, a non-smoking expansion.

Flower doesn't hit zero by 2036. Tobacco hasn't hit zero either. What smoking loses is the center: the default format, the cultural shorthand, the thing the word "cannabis" makes you picture. It becomes the legacy format, kept alive by ritual and price and an aging cohort, while the growth, the margin, and the legitimacy all live in oral dosing. We wrote about the individual version of this switch in edibles versus smoking; this is the market running the same calculation at scale, with alcohol's collapse as the pace car.

The closing thought is about the man with the teeth. He is not the enemy of the medical future. He is its unpaid advance man, proof of concept that the compound is survivable at absurd doses, performed nightly for an audience of millions. The industry that replaces him will never thank him, and he will never understand that the joke, held at the right angle, was the clinical trial nobody ran. The 2.5 milligram tablet that a palliative care physician prescribes in 2034 will owe something to a decade of men eating 500 for engagement. History does not distribute dignity evenly.