The GLP-1 economy in five minutes: verified drug prices, and the business behind them.
We read the documents.
A cease-and-desist the trade press gave one line, an ad quietly revised in the middle of the fight about it, and a pricing framework the internet quotes in the wrong unit. All of it was public. None of it was read.
On April 22, Novo Nordisk sent Lilly a letter telling it to take the ad down. Lilly never answered, changed the ad instead, and three months later Novo was in federal court.
The letter’s date, the three months of silence, and the quiet revision have not been reported anywhere we can find. The letter itself got one line in the trade press. It has been sitting in public court filings the whole time.
This week we stopped trusting summaries and pulled the primary documents. What fell out: a cease-and-desist the trade press gave one line, an ad quietly revised in the middle of the fight about it, a pricing framework the whole internet quotes in the wrong unit, a fourth generic manufacturer we found in a government database, and corrections to our own record, printed in daylight.
Asked in April, sued in July
Everything in this section comes from one document: Novo’s 47-page brief supporting its preliminary-injunction motion, filed July 24. We read it in full.
Two cautions before the facts. This is Novo’s account, written to persuade a judge. And Lilly’s answer, its opposition brief and seven declarations, is not free on the public docket, so we have not read Lilly’s version of any of it.
Here is what Novo says happened.
In March, the FDA approved Wegovy 7.2 mg. That is the dose that posted about 47 pounds of average weight loss, and its approval is what made the SURMOUNT-5 trial behind Lilly’s commercial an outdated comparison.
On April 22, Novo sent Lilly a letter. It demanded that Lilly suspend the advertising and stop making consumer-directed comparative claims based on that trial.
Lilly did not respond to the letter. In Novo’s words: “Lilly never responded and instead made only cosmetic revisions.”
One footnote the documents force on us: Novo has told this part two ways.
The brief says Lilly never responded. Novo’s own general counsel, in a July interview, said Lilly refused to pull or correct the ads, and a refusal is a response.
Both are Novo’s account, a month apart.
We print both because the difference is exactly the kind of thing this edition is about.
Those revisions are the part nobody has written about. Lilly changed three things and kept the one that matters.
The dose labels changed. The original ad called both products “MTD,” maximum tolerated dose, which implied the trial ran each drug at its best. The revised ad names the real numbers: Zepbound 10 mg and 15 mg against Wegovy injection 1.7 mg and 2.4 mg.
Small print was added, saying Wegovy 7.2 mg “was not evaluated in this study” and “has since been approved.”
And the headline stayed. Fifty pounds for Zepbound against 33 for Wegovy, which is the exact claim the lawsuit is about.
So which version has been running? On the brief’s own impressions clock, the revised one, since late April. The brief counts approximately 700 million impressions from late April, and elsewhere attributes that same 700 million to the revised commercial. It never says the words “we revised it on this date.”
This sharpens the picture we printed in edition 2. We reported the sprint: sued Tuesday, moved for an injunction Friday, three days. That is still exactly what happened.
What we did not know is that the three days were the end of three months of being ignored.
The full sequence now reads: April 22 letter, silence, a quiet revision, then in June an expansion to sponsored TikTok and Facebook ads calling the trial a “head-to-head study” and saying “those are just the facts.” Then July 21, the suit. July 23, Lilly’s counsel refuses. July 24, the motion.
Read the sequence the way a lawyer would. Lilly got a letter demanding the ad come down, said nothing, changed the labels, and kept the headline.
You do not quietly fix the part you can defend. You quietly fix the part you cannot, and keep the part you are betting the case on.
The revision reads less like a correction than a map of where Lilly thinks the line is: the dose labels were indefensible, the 50 versus 33 is the hill.
On Monday a judge starts deciding whether the hill holds.
That is Novo’s account of the sequence, and Lilly’s answer sits behind a paywall we have not read, which we say plainly because it is true.
The calendar closes tomorrow. Monday, August 17 is the day the fully briefed motion goes to Judge Zahid Quraishi, decided on the papers, no appearances required. Briefing closed August 10.
A ruling can land on any ordinary morning after that, with no hearing and no warning.
The framework everyone quotes wrong
Second document, second surprise, and this one costs us something.
The pan-Canadian Tiered Pricing Framework sets generic prices as percentages of a brand reference price, counted per brand-reference category. Not as dollar caps. That single distinction is the whole error, and we made it too.
the pan-Canadian tiered pricing framework, as written
| competitors in the category | price, as a share of the brand reference | note |
|---|---|---|
| 1 | 85% | or 75% falling to 55% after three months of funding, where a listing agreement exists or ever existed |
| 2 | 50% | applies to every dosage form |
| 3 or more | 25% for oral solids, 35% for everything else | the framework names liquids, patches, injectables and inhalers |
Read the bottom row again. Pens floor at 35 percent, never 25.
Which means every line you have read about semaglutide generics falling to a quarter of the brand price was written about pills.
The framework also settles the question we promised readers we would answer, by telling us we asked it slightly wrong. It counts products, not companies, inside a category keyed to the brand reference product. A competitor is any product holding a Notice of Compliance and a DIN that is either marketed, or approved with supply available in the previous 12 months.
That moves two Canadian stories into different lanes.
Sevmia, Apotex’s generic Wegovy, is not a third anything. No generic Wegovy is on the Canadian market, so Sevmia would be the first generic in its own category, at 85%, or at 75% falling to 55%. Not the steep tier.
Aspen is the candidate third in the Ozempic-reference category, where two generics are already selling.
And there is a live question underneath all of it, which is the better one.
The framework only governs where the brand reference product is reimbursable by a member drug plan. No provincial plan covers weight-management GLP-1s. So if Wegovy is reimbursed nowhere, the framework may not govern Sevmia’s price at all, and the public-tier story everyone has told about it, us included, would be beside the point.
We have asked the documents. The pCPA can be asked directly, and that is now the watch item.
The fourth manufacturer
Third document, and this one is a database.
On August 12, running the weekly check before edition 7 went out, we swept Health Canada’s Drug Product Database by active ingredient. It returned a semaglutide product we had never recorded.
Aspen Pharmacare Canada, approved July 17, two DINs, Ozempic-label multi-dose pens. Still not launched.
Three editions of this brief published in that window.
The tier math is why it matters. Two generic manufacturers are actually on the Canadian market today: Dr Reddy’s since May 6, Apotex since May 14. That is the 50% tier. Aspen reaching market would make three and take that category to 35%.
The sweep now runs on a standing schedule, every six hours, alerting only when something moves. Nothing hides that long again.
As of this morning, all 20 DINs we watch still read approved, none marketed, no market date on any of them.
One more thing the sweep answered, in the negative. There is no Sandoz semaglutide product in that database in any status, across 47 records. A Sandoz launch we had on file did not happen.
Here is the uncomfortable part.
Health Canada’s database is free, public, and searchable, and a fourth generic manufacturer sat in it for 26 days while everyone covering this beat, including us until last week, read summaries instead.
The moral is not that we are clever. It is that almost nobody on any beat reads the primary documents, which means almost every number you encounter is a copy of a copy.
We fixed our process; the weekly sweep now reads the whole ingredient group. The copies will keep circulating either way.
Pick your sources accordingly.
What we got wrong, and the rule that governs it
On August 12 we published two corrections in one day.
Edition 7 took one for the motion date and the pricing framework. Edition 4 took one for the same tier framing, plus a Sandoz launch that does not exist in the database at all. Both are stamped and dated in the editions where they appeared.
One tier claim survived the sweep. Edition 1’s “roughly 35% of brand” step was correct for pens the whole time.
Now the distinction that governs all of it, because this edition creates a case of the second kind.
Edition 2 told you the disputed commercial had been seen more than 700 million times since late April. The number holds against the primary document. The window holds. What that sentence missed is that there are two versions of the ad, that the disputed campaign was already running before April 22, and that the late-April clock belongs to the revised cut.
That is not a wrong fact. It is an incomplete picture, and the two get fixed differently.
Where a fact was wrong, we stamped it, dated, in the edition it appeared. Where the picture was incomplete, the fix is more reporting, and you are reading it.
So edition 2 keeps its record as printed. This section is the correction to the picture.
The rule that comes out of a week like this one is short. A primary source we hold beats a secondary source we do not, every time.
Watching for
- Monday, August 17: motion day in Novo v Lilly. Briefing is closed, so a ruling can arrive any morning from then, on the papers, without a hearing. If it is granted, the paywalled declarations become worth buying and we can date the ad revision precisely.
- The direct ask to the pan-Canadian alliance: whether the framework governs a generic whose brand no member plan reimburses. We said it here, so it is now owed.
- Aspen’s market status. The flip from approved to marketed is the 35% trigger in the Ozempic-reference category. August 19 is also the twins deadline we put in print, and all 20 watched DINs were still approved and unlaunched this morning.
That is the brief. Same time Wednesday.
Updated August 26, 2026: the briefing timeline is corrected, Novo filed a reply brief August 17 and briefing closed August 18, not August 10 as we wrote; the docket allows what we said it didn’t.