The GLP-1 economy in five minutes: verified drug prices, and the business behind them.
The cliff that is not a cliff.
Starbucks stops covering weight-loss GLP-1s in October. We read four writeups of that decision and three never say what the drug would cost you. Here is the number, where it comes from, and the expiry date sitting ten weeks behind the change.
RETRACTION: we published something false about what a drug costs, and this is the first thing in the edition because that is where it belongs.
On Saturday, August 22, 2026, we published a card and caption on LinkedIn stating that Lilly’s published Zepbound prices of $299 to $449 require commercial insurance that does not cover the drug, and that a cash payer cannot get them. The caption went further, arguing the two cheapest prices sit at opposite ends, one requiring coverage and one requiring you to give up on it, with insured-but-denied patients stranded between them.
Both claims are false. Lilly runs two programs at identical prices. The first, the Zepbound KwikPen Savings Card, states you “must have commercial drug insurance that does not cover Zepbound single-patient-use KwikPen.” The second, the Zepbound KwikPen Self-Pay Savings Card Program, states the card “is for self-paying (cash) patients only” and “operates outside of any health insurance program.” A cash payer can get $299 to $449.
The post ran on LinkedIn only, for approximately 22 hours, and has been deleted.
THE SKIM
- Starbucks stops covering GLP-1s for weight loss in October. The medicines may still be covered when prescribed for other conditions.
- Where coverage does give a number, it is a list-price band. The cash prices a dropped employee actually pays are $299 to $449.
- Every major cash price we track expires on December 31, 2026, ten weeks after the change takes effect.
Starbucks is dropping a health benefit in October, and most of the coverage telling you about it never says what the drug costs. We read four writeups. Three give no patient price at all. The one that does quotes a range topping out above the drug’s own stated price.
The floor
THE FLOOR
United States
$149 USD/mo
Wegovy pill, NovoCare
Canada
C$88 CAD/mo
Apo-Semaglutide, Costco
Set a free alert and we email you the day one of these moves, or the day a drug that has never had a price gets one. Set an alert →
Cheapest verified price we hold per market, promotional and introductory offers excluded.
Every figure above, and every other price we hold, is on the price board with the date we checked it and the method we used.
The cliff that is not a cliff
THE LEAD: an employer stops paying, and the question is what the person actually pays instead.
Starbucks will stop covering GLP-1 medicines, the glucagon-like peptide-1 drugs sold as Wegovy and Zepbound, when they are prescribed for weight loss. The change begins in October 2026. The change affects benefits-eligible employees. The medicines may remain covered when prescribed for other medical conditions, which is a distinction most of the coverage has flattened.
The Obesity Action Coalition, a patient advocacy group, called on the company to reverse it on August 21, 2026. Its president and chief executive, Joe Nadglowski, put the objection in thirteen words.
An employee should not lose access to a treatment simply because the diagnosis is obesity.
A dropped employee does not pay a list price. They pay a cash price, and those are published. Zepbound is $299 a month at the 2.5 milligram starting dose, $399 at 5 milligrams, and $449 at 7.5 milligrams and above, which is where a patient on a maintenance dose actually sits. Both of Lilly’s programs, the insurance-based card and the cash-only card, land on the same numbers.
Now compare that with how the change has been described. The Society for Human Resource Management, writing on August 13, 2026, told benefits managers the drugs “typically cost between $1,000 and $1,500 a month for consumers.” HR Executive, on August 17, 2026, gave no patient price at all, and neither did two other writeups we read. The story has largely been priced for the employer rather than for the employee.
Both halves are a problem. The silence leaves a reader with no idea what they face. And the range that does appear runs to $1,500, above Zepbound’s stated original of $1,087 on TrumpRx, the federal site where that figure is published. Against the $449 a maintenance patient pays, that original is 2.4 times higher. Against the $299 starting dose, 3.6 times.
Zepbound self-pay, both Lilly programs, read at zepbound.lilly.com on August 23, 2026
| dose | a month | against the $1,087 list |
|---|---|---|
| 2.5 mg, starting dose | $299 | 3.6x |
| 5 mg | $399 | 2.7x |
| 7.5 mg and above | $449 | 2.4x |
An employer dropping coverage is not people leaving treatment. It is people moving into a cash market that already has prices, and those prices are set by the manufacturers rather than by a plan. The benefit did not disappear. It changed hands, from a plan sponsor to a company running a discount program on its own product.
That is a different story from the one being told, and for a patient it is a worse one, in a specific way.
The documents
THE PART NOBODY HAS REPORTED: read the terms, and the cash market a dropped employee walks into has a closing date.
We read Lilly’s program terms rather than a summary of them. Three sentences in that document are doing more work than the prices.
The first: “Card expires and savings end on 12/31/2026.” Starbucks stops covering these medicines in October. The cash prices a dropped employee moves onto are scheduled to end roughly ten weeks later. Every major cash route we track carries the same date or one near it.
The second is the definition of a month. For this program, “month is defined as 28-days and up to 1 single-patient-use KwikPen.”
The third is the annual cap. The card “may be used for a maximum of up to 11 prescription fills per calendar year.”
Put those two together and the program does not cover a year. At a 28-day month a calendar year needs thirteen fills. The card allows eleven.
There is also a 45-day rule. To keep the price, a refill has to be purchased “within 45 days of the delivery/received date” of the previous one. Miss the window and, in the document’s words, “purchase offer will not apply.”
None of this is hidden. It is on the page, under the number, and it is the difference between a price and a price you can actually hold for a year.
One survey, and why only one
Three organisations published employer coverage figures in 2026, and they do not agree. We are using one of them and naming it, because the alternative is to line up three incompatible numbers and let a reader think they are watching a trend.
The one we are using: Mercer, in a survey of 604 US organisations published on June 11, 2026, found 6% of large employers dropped coverage for these drugs in 2026, with another 5% planning to drop it in 2027 or actively considering it.
The ones we are not using, and the reason: two other bodies published 2026 coverage shares from different samples, different definitions of a large employer, and different field dates. Their numbers are not wrong. They are answers to different questions, and setting them beside Mercer’s would manufacture a pattern out of a definitional gap.
One figure from elsewhere is worth carrying because it measures something else entirely. The International Foundation of Employee Benefit Plans, a benefits research body, reported on July 7, 2026 that these medicines rose from 6.9% of employer annual claims in 2023 to 11.4% in 2026. That is not a coverage share, so it does not collide with Mercer’s. It is the reason an employer does this.
Watching for
what moves next.
- October 2026, and whether anything moves before it. The Starbucks change takes effect in October. The Obesity Action Coalition asked for a reversal on August 21, 2026. Whether the company answers, and what it says about coverage for other conditions, is the thing to watch.
- December 31, 2026. Lilly’s card program expires that day. So does Novo’s introductory pricing. A cash market with a common expiry date is not a market, it is a promotion, and January is when we find out which it was.
- Judge Quraishi, still silent. Novo Nordisk’s injunction motion against Eli Lilly has been fully briefed and sitting with the court, submitted on the papers with no hearing. It can land any morning.
- The twins. Plosbrio and Poviztra, Novo’s own cheaper versions of its own drugs, approved in Canada and still never sold. Seventeen product listings, checked every six hours.
That is the brief. Same time Wednesday.