Insulet

IThe business

Think of a razor, turned inside out. A razor is cheap and the blades are dear. With a tubeless pump the part you wear is the product, and you throw it away. An Omnipod pod holds the insulin, the pump, a small cannula that sits under the skin, and, in Omnipod 5, the program that decides how much insulin to give. It lasts up to three days. ³ A person using Omnipod buys about ten of them a month C (calculated).

Almost all the money is the pod you throw away every three days.

Revenue over the twelve months to June 2026 was $3,053M C (calculated). US Omnipod brought in $2,125M, or 70%. International Omnipod brought in $911M, or 30%. Drug Delivery, which is pods built for Amgen to deliver its chemotherapy-support drug, brought in $18M, or 0.6%, and has been shrinking for years. ³⁴ The company counted more than 600,000 active users at the end of 2025. ³ Divide 2025 Omnipod revenue of $2,674M by that year-end count and each user is worth about $4,460 a year C (calculated). The true figure is higher, because the user count rose during the year.

How the customer really pays

In the US the buyer is an insurer or a government program, and the sale runs through the pharmacy, as with a prescription drug. That is the point. No one has to find thousands of dollars up front, and the insurer can stop paying if the user stops. ³ The cost is that three wholesalers took 27%, 26% and 25% of all revenue in 2025, 78% together, and 86% of Omnipod sales worldwide went through intermediaries. ³

There is a second toll in the middle. Deductions for rebates, chargebacks and wholesaler fees were $848M in 2025, against $588M in 2024 C (calculated). ³ That is up 44% in a year when revenue rose 31%. It is the price of staying on the formulary, and it is the line I most want to see in each quarter, because the company reports revenue after it. Management said in August that US pricing has been stable to slightly up. ¹¹

Where the growth comes from

Omnipod 5 reached the US in 2022. It is now sold in 20 countries, and Spain became the 26th country for Omnipod overall in July. ⁵⁴ Outside the US, revenue grew 35.5% in the second quarter, 32.9% without currency, and it is now 30% of the total, up from 20% in 2016 C (calculated). ⁵³

The larger bet is type 2 diabetes. The FDA cleared Omnipod 5 for adults with type 2 in August 2024. ³ The company counts about 6 million people with type 1 in the countries it serves, another 6 million with insulin-treated type 2, and 3 million more in the US who take basal insulin only. ³ These are its own estimates. It also says about 40% of US type 1 patients already use a pump. On those counts 600,000 users is about 4% of 15 million people C (calculated). In the second quarter more than 40% of US new users had type 2, and more than 85% of US new users came from daily injections, not from another pump. ¹¹ That is why the company says its main competitors are the suppliers of multiple daily injections, including smart pens. ³

Where it is made

Pods come from two plants, one in Acton, Massachusetts and one in Johor, Malaysia, which began producing in 2024. A third is being built in Costa Rica, and a contract manufacturer in China also runs some lines. ³ Inventory is held in a small number of places, one of them run by a third party in China. In April 2026 the company agreed to guarantee up to $97M of the loan the seller took to build the Costa Rica plant. That is not on the balance sheet as debt, and I do not add it to net debt. ⁴ The company employed about 5,400 people at the end of 2025, 38% more than a year before. ³

IIThe moat

The patents are not the main thing. Insulet reports more than 1,000 of them, with expiry dates from 2026 to 2047, and more than 700 applications pending. ³ Some are already running out. What holds a customer is the shape of the product and the habits built around it.

What keeps a customer is a body-worn form and a pharmacy shelf, more than any single patent.

I count four things. First, the form. Tubeless is a different product from a pump with tubing, and for many users that is the reason they chose it. It is waterproof to 25 feet for an hour, and in the US it can be controlled from a phone. ³ Second, the algorithm lives in the pod, so each new pod carries the latest version. Third, the channel. Pharmacy reimbursement is a contract business, and Insulet sells almost all of its US volume that way, while pharmacy is 10% of Tandem’s US sales. Fourth, scale. It earns a 72% gross margin on a disposable that it makes by the million. ³ Each Omnipod 5 pod is bought on a three-day cycle, so a lost user shows up in revenue within months, once wholesaler stock is used up.

That last point cuts both ways. A subscription-like business has no installed base to defend, because the user can leave at any refill. I did not find a retention rate in the filings or on the call. Management said only that retention among type 2 users was worse than it had planned when it cut the US outlook in August. ¹¹

The part that is being tested

Insulet’s second layer is legal. It sued EOFlow, a maker of a patch pump sold in Korea and the EU, for taking its trade secrets. A jury found for Insulet in December 2024 and awarded $452M. In April 2025 the court upheld the verdict, cut the damages to $59.4M so they would not double up with an injunction, and barred EOFlow from selling products built on the secrets. ⁴ On 28 May 2026 the Federal Circuit reversed, holding that the claims were filed too late under the three-year time limit. On 22 June it widened a partial stay of the injunction to cover Korea, the EU, Qatar, Saudi Arabia and the UAE, and on 29 July Insulet asked for a rehearing. ⁴ Insulet had not booked the damages, so the reversal cost nothing in the accounts. The harm is to the belief that a copier can be stopped.

Does the product earn its place?

The strongest evidence is simply that people stay and that more of them arrive. More than 600,000 users at the end of 2025, 38% more staff, and revenue up 31% last year are what an earning product looks like. The weakest evidence is retention in the group the company is targeting next. In type 2 the user may be on several drugs, and the 10-K names the rising use of GLP-1 drugs as a risk. The company says the cause is its own execution and not the market, and that retention levels off after the first 90 days. ¹¹ I can’t test that. What I can see is that the US guide for the year was cut to 17% to 19%, from 20% to 22% in May. ⁵

IIIThe competitive advantage

Insulet has had the tubeless market to itself for years. That is changing, but slowly.

Three things are worth saying. First, what Insulet has that the others lack is a body-worn pump with no tubing. Of the three I read, only Tandem has filed to match it, and nothing has been cleared. Second, Tandem and Beta Bionics are copying the part Insulet built first: selling through the pharmacy. Beta Bionics took a pharmacy share of 38% of net sales in a year, and pharmacy is 10% of Tandem’s US sales. That chips at a channel advantage, not a product one. Third, MiniMed is the large one. Its quarter was bigger than Insulet’s, $843M to $802M, though the periods differ. With the extra week taken out its growth was about 11% to 13%, below Insulet’s 23.5% C (calculated), and it launched a new pump into the same market in June. I would not call it a threat yet. I would call it a reason the retention question matters. ⁴³⁴¹⁴²

Management said the 2027 outlook already assumes new competitors, and that the type 2 problem is not about price, competition or GLP-1s. ¹¹ I take that as the company’s belief. If the cause is Insulet’s own execution, it can be fixed. If it is the product, it cannot.

What is working

International is doing what the plan said. Revenue outside the US grew 35.5% in the second quarter and management raised its full-year outlook for it to 30% to 32% in constant currency, from 26% to 28% in May. ⁵⁶ Adjusted operating margin was 19.3%, up 1.4 points. That leaves out $25M of correction costs, and the GAAP margin was 16.2%, down 2.5 points. Adjusted gross margin was 72.9% on the call, up 3.2 points. ⁵¹¹ And the company beat the top end of its own revenue guide in each of the last three quarters. ⁷⁶⁵

What is not

US growth. In the first quarter US Omnipod revenue grew 28.3%. In the second it grew 20.1%, and the company now guides 14% to 16% for the third. ⁶⁵ About two thirds of the cut to the year, the company says, comes from type 2 users staying and using less than expected, and a third from fewer new starts and a slightly lower price. ¹¹ Net price is one place to look: gross-to-net deductions are rising faster than revenue. ³ The other is the forward range. The finance chief gave 2027 US growth of 9% to 14% on the call, with 12% at the midpoint, and expected total growth to leave 2026 in the mid-teens. ¹¹ US Omnipod grew 27.2% in 2025. ³

IVManagement

Ashley McEvoy became chief executive in April 2025 and joined the board the same month. She spent more than 25 years at Johnson & Johnson, the last five as worldwide chairman of its MedTech group, which ran surgery, orthopaedics and eye health, and earlier led its Ethicon suture business. ¹³ She replaced James Hollingshead, chief executive since 2022, who left on 28 April 2025 with the standard severance of twice salary and twice target bonus. ¹³ Flavia Pease became finance chief on 30 September 2025 after serving on the board and as chief financial officer of Charles River Laboratories. She replaced Ana Chadwick, who left the company in February 2026. ²⁴¹³

Add the rest. In December 2025 the audit committee, after a competitive process, replaced Grant Thornton with PricewaterhouseCoopers from the 2026 audit. The filing says there were no disagreements, and mentions a weakness in controls over IT systems outside North America, reported in the 2023 10-K and fixed in 2024. ²³ In September 2026 two directors stepped down: Timothy Scannell, a director for 12 years and chair for seven, for health reasons, and Michael Minogue to run for governor of Massachusetts. ²⁵ Jonathan Mazelsky, chief executive of IDEXX until May, joined the board on 1 July. ²⁶ In eighteen months Insulet has a new chief executive, a new finance chief and a new auditor, and its chair of seven years has left. None of that is a sign of trouble in itself. It is a lot of change for a company that has just had two product corrections.

How they are really paid

Follow that through. The bonus pays on revenue growth without currency, on operating profit with the company’s non-GAAP adjustments, and on how many new users start. ¹³ In 2025 adjusted revenue came in at $2,663M against a target of $2,480M, and adjusted operating profit at $466M against $409M, and the payout was 195.8% of target, close to the cap. ¹³ The long-term stock for 2025 to 2027 pays 70% on cumulative adjusted revenue and 30% on adjusted operating profit. ¹³

Nothing in the bonus pays on keeping a user, on cash, or on the capital used.

That helps explain how the retention problem could grow unseen. A new start counts, and a user who leaves a year later does not cost the executive anything. And because the 2025 definition of adjusted operating profit takes out items the company reports as non-GAAP, and the correction costs are among them, the $60M to $70M of the 2026 corrections reduces profit but not that bonus measure. ⁴¹³⁵ In December 2025 the board amended the plan to raise the maximum award, widen the cases in which a retiring executive is paid, and tighten the recoupment language. The filing does not give the 2026 measures, so I cannot say whether they changed. ²⁹ I take the 2025 design as the best guide.

How much of their own money is in

Less than the pay suggests. McEvoy’s beneficial holding at 23 March 2026 was 13,597 shares, including 6,037 options she could exercise within 60 days, and Pease held 543. All directors and executive officers together held 250,199 shares, under 0.4% of the company. ¹³ At $149.68 McEvoy’s holding is worth at most about $2.0M C (calculated), against a guideline of six times salary, or about $6.9M, which the proxy says is phased in. ¹³

What she has done on her own account is better. She bought 4,300 shares on 20 February 2026 at $239.35, about $1.03M, and 1,100 more on 21 August at $147.47, about $0.16M C (calculated). ¹⁴ Three directors also bought: Minogue 2,030 shares at $246.23 in February, and Stonesifer and Weatherman 2,790 and 3,450 shares at about $144 in June C (calculated). In all, insiders put in about $2.9M of their own money between December 2025 and August 2026 C (calculated). ¹⁴ The only open-market sale in 2026 was a director’s 418 shares for $60,000. In 2025, when the price was $275 to $349, thirteen sales by officers and directors came to $12.4M C (calculated). ¹⁴ The buying is small against the pay, but it has the right shape: little selling at the lows, and some purchases made with the price already well under the year’s high. Large owners went the other way: FMR (Fidelity) fell from 9.0% at the end of 2025 to 1.4% at 30 June, Capital Research from 11.1% in 2024 to 2.4% in March, and BlackRock held 7.7% at 30 September. ^16151713 Those are positions, not views.

What they did with capital over a full cycle

The record is mixed. Insulet bought no stock in the five years to 2024. Since 2016 it has expensed $413M of stock pay C (calculated), and the diluted share count went from 57.3M in 2016 to 73.9M in 2024, 71.9M in 2025 and 69.4M in the second quarter of 2026 C (calculated). ³⁷³⁴ That is 21% more shares than in 2016, and the buying back came late.

Two decisions stand out. In the first half of 2025 it paid $541.5M in cash to retire $419.9M of convertible notes, booked a $123.9M loss, and replaced them with $450M of 6.5% notes due 2033, getting $75.7M back from unwinding the related hedges. ⁴³ Then in February 2026 it bought $300M of its own stock through an accelerated plan, at an average of about $240 a share, 1.251M shares in all C (calculated). ⁴ At $149.68 that stock has lost $113M on paper C (calculated). The board extended its $125M authorization to the end of 2027 and added $350M, so it can do more. ⁴ Other uses were reasonable. The Malaysia plant opened in 2024 and the Costa Rica plant is under way, and the company refinanced its term loan at a lower margin in September. ³²⁷

One more thing to state without comment. For most of 2025, a member of the board was married to an executive of one of Insulet’s distributors. Sales to that distributor were $511.6M in 2025, 19% of revenue, and $587.8M in 2024, 28% C (calculated). The company says the terms were arm’s length and that it stopped counting the distributor as a related party from 1 October 2025. ³⁴

VFinancials

The record is one of fast growth that cost a great deal and only recently began to pay. Revenue went from $367M in 2016 to $2,708M in 2025, 24.9% a year, and was $3,053M in the twelve months to June C (calculated). Every year grew at least 18%. ^36333134 The operating margin went from −2.9% to 17.5%, and 2023 was the first year net income passed $100M.

Two of those years need a note. Net income in 2024 was $418M because $118M of it was a tax benefit from releasing a valuation allowance on deferred tax assets, not from selling pods. Net income in 2025 was $247M after a $124M loss for retiring convertible notes early. ³ Neither is a measure of the business. That is why I work from operating income and cash. The table is built from the ten annual reports and the filings that restate them. ^{36353433323130337} In 2021, operating cash flow was −$68M because inventory rose $154M, for demand and a third manufacturing line, and receivables rose $71M as the pharmacy channel, which pays more slowly, grew. ³³

That chart matters more than any other on the page. Gross margin rose from 57.5% to 71.1% as pods got cheaper to make. Selling and administrative cost stayed at 40% to 45% of revenue the whole way, and was 42.9% in the twelve months to June. ³⁴ The 2025 and 2026 gain is better than the 2022 low, but the company has not yet shown that selling cost shrinks as the business grows. My base needs it to. It is the single assumption on this page that I would most like to see tested.

What an owner really takes out

Three earnings numbers are in circulation, and they are close together. Net income for the twelve months was $375M. The company’s adjusted net income, which adds back the correction costs and one-off items but not stock pay, was $412M. ⁷⁶⁵ Neither says how much cash an owner could take out.

My number is about $405M. An owner could take that out each year after stock pay and the capital needed to stand still.

I get there like this E (estimate). Take operating income of $516M and add back $37M of correction costs, which will not recur. That makes $552M, or 18.1% of revenue, and it is already after stock pay. Tax it at 21%, close to the 20% the company paid in the half year. Take off after-tax interest on the debt, $32M on the $40M the company expects for 2026. Treat depreciation as the cost of standing still, since it was $99M against capital spending of $237M in the twelve months. That leaves $405M, or $5.83 a share, a yield of 3.9% on the market value.

The gap between $405M and $196M is the cost of growing, and it is large. Capital spending and software were $237M, $138M more than depreciation, mostly the Costa Rica plant and new lines. Working capital took $106M, mostly receivables from distributors’ timing and a planned build of inventory. ⁴³ Note how close my $405M is to the company’s adjusted net income of $412M. Insulet does not add stock pay back, so its adjusted figure is a fair one. The distance that matters is between that and the $196M. I do not count all of that as a permanent drain. I do count it once, which is why my model charges 35 cents of reinvestment for every added dollar of revenue. Over 2022 to 2025 the actual figure was 30 cents, and over 2021 to 2025, with the inventory build, 45 cents C (calculated). ^3331303

The difference between the company’s free cash flow and mine is stock pay. At $78M a year, 2.6% of revenue, it is not small, and I deduct it because an owner pays for it in shares. The company also paid $16M in the first half of 2026 in taxes on vesting stock, which is a further real cost of the program that does not appear in the stock-pay line. ⁴

The balance sheet

At 30 June 2026 Insulet had $535M of cash and $948M of debt, so net debt was $414M C (calculated). Debt is a $475M term loan due 2031, repriced in September at 1.75 points over SOFR, 0.25 lower than before, and $450M of 6.5% unsecured notes due 2033. It has $460M of the term loan fixed through swaps at 3.47%. A $500M revolving credit line, due 2030, is undrawn. ⁴²⁷ Net debt is about 0.6 times adjusted EBITDA of $736M for the twelve months C (calculated). ⁷⁴ I treat that as a small burden. The $97M Costa Rica loan guarantee sits outside it, and I mention it for completeness. Net interest is expected to run at about $40M in 2026. ⁴

Warranty is the other liability. The balance was $17M at the end of 2025 and $33M at 30 June 2026, and warranty expense for the corrections was $37M in the half year. The company expects the corrections to cost $60M to $70M in all, with the rest in 2027 from extra manual inspection until automated systems are in place. ³⁴ That is small against a business with $3B of revenue. It is the third and fourth corrections in four years, after two in late 2022 that carried a net charge of $57.9M. ³¹

VIValuation

What would ruin it

Before any numbers, I ask how this goes badly wrong. I see three ways, and I take them in the order I fear them.

One: users leave faster than new ones arrive. Two: the net price gets competed away. Three: something breaks in the factory or at the border.

First, users. A pod is replaced every three days, so the business has no installed base to hide behind. If a user leaves, revenue stops within months. The company says type 2 users are staying and using less than planned, and it cut its US outlook from 20% to 22% growth, to 17% to 19%, for that. ¹¹⁵ US growth went from 28.3% to 20.1% in one quarter and is guided to 14% to 16% for the next. ⁶⁵ If retention does not improve, the market is bigger on paper than in practice. My bear case is this world: US growth of 7% falling to 3%, international growth of 17% falling to 3%, and a margin after stock pay that slides to 15%. It is still a profitable business. It is worth $91 a share E (estimate).

Second, price. The company reports revenue after rebates, and rebates, chargebacks and wholesaler fees rose 44% last year against revenue up 31%. ³ Each 1% off the net price, with no change in cost, costs $30.5M a year before tax and $24.1M after, or 5.5% of the $436M an owner earns to stand still C (calculated). A 5% give-back would cost 28% of it C (calculated). Tandem and Beta Bionics are building pharmacy sales, and Beta Bionics’ pharmacy share is 38% from 23%. ⁴¹⁴² A rival does not have to be better. It only has to make payors ask for a lower price.

Third, something breaks. The first correction was for up to $40M of cost; the FDA’s website cited 476 serious-injury reports on it, and the company confirmed 29 serious adverse events. ²⁰²¹ The second correction covered about 7 million pods. ²² A third event, an FDA action, or a loss of the tariff exemption that the 10-Q says would have a material effect, would turn bad luck into a pattern. ⁴ Production sits in Acton and Malaysia, with a contract plant in China and a third plant in Costa Rica not yet open. ³

I leave the price off this list, though it is the usual third. At $149.68 the stock is already 57% below its November 2025 high and 47% below where 2025 ended C (calculated). ¹ A fall is not the same as a low price. The question is whether the business is worth more than the price.

What the price implies

Market value at $149.68 is $10.40B on 69.46M diluted shares C (calculated). ¹⁴ Add net debt of $414M and the business is priced at $10.81B, which is 24.8 times what it earned to stand still after tax C (calculated).

Here is the plain way to read it. Suppose an owner keeps 12 cents of each revenue dollar as free cash, after stock pay and all capital. That is about where 2025 landed, at 10.9% C (calculated). Then at 9.0% the price needs revenue growth to start at 14.4% and fade in a straight line to 3.5% by year ten, a ten-year growth rate of 8.9% a year E (estimate). At 10 cents it needs 18.9% to start. At 16 cents it needs only 7.4%.

The price asks for about the growth I expect, if the margin settles about two points below today’s. I expect it to rise instead.

The discount rate

I use 9.0%. The capital asset pricing model gives 8.6% on a beta of 0.88 from 120 months of returns against SPY, with the 10-year Treasury at 4.94% and Damodaran’s 4.20% implied premium C (calculated). ³⁹³⁸⁴⁰ The 60-month beta of 1.05 gives 9.3% C (calculated). Rounded to the nearest half point as elsewhere on this site, those are 8.5% and 9.5%. I take the middle, 9.0%. This is one product, with an unreported quarter, a guide cut and new management, so I lean toward the higher figure. It is above the 8.5% I use for ASML and level with Visa and Mastercard at 9.0%, and below the 10.0% I use for Intuitive Surgical. ^48464745 Debt is about 8% of capital at market value, so a cost of capital that includes it would be a little lower, about 8.6%. I do not use that.

The price sits inside the grid. At 10.5% and 3.0% growth the base is $150, about the price; the base equals the price at a rate of 10.7% C (calculated). Someone who requires 10.5% will not see a gap.

What I cannot judge

Four things, and I would rather say so than give a false figure.

Whether type 2 users stay. The company gives no retention rate. Each point of growth a year in both regions is worth about $20 a share on my base E (estimate). The 2027 number barely matters: the management range for US growth is 9% to 14%, and moving my 2027 US growth between those costs or adds about $3 a share E (estimate). ¹¹ What matters is whether the rate holds in 2030. I can’t tell you that.

What the net price does. The company reports revenue after rebates, and the rebate line has grown faster than revenue. The filings let me see it once a year, in a schedule at the back of the 10-K. ³ The call said US pricing is stable to slightly up. I have no way of testing that between reports.

Whether selling cost falls. My base takes operating margin after stock pay from 18.1%, the normalized figure, to 21%. Each point of long-run margin is worth about $11 a share E (estimate). Selling cost has stayed at 40% to 45% of revenue for ten years, as the chart in the financials shows. I think leverage will show once the large launch spending is done, but I have no proof of it. I test the path below.

What the new products are worth. Omnipod 6 is due in 2027 and a fully closed-loop system for type 2, with a filing in 2027 and a launch in 2028. ¹¹⁴ I put no separate value on either. They are why I do not use a lower growth rate in the base, and they could also be a reason the bear is too bleak.

Testing the margin

The base is above the $177 an earlier rough screen gave, and margin is the reason. First the history. Operating margin was 13.0% in 2023, 14.9% in 2024 and 17.5% in 2025, gains of 1.9 and 2.6 points. ³³⁰³¹ Adjusted for the corrections it was 17.6% in 2025, and the company guides about 100 basis points more for 2026. ⁷⁵ My base adds 1.0, 0.8 and 0.6 points to reach 21% in 2029. That is slower than the last two years and in line with the guide, so the path is not the stretch.

The level is. Gross margin is 71% to 72% and research is 11% of revenue, so 21% needs selling cost near 40% of revenue, down from 42.9% C (calculated). It has not been below 40% in ten years. ⁴⁵ If margin stops near 18.6%, my model gives about $177, the screen’s figure. The base needs the last two points, and I cannot show that they come.

Three cases

I value fifteen years of cash, 2027 to 2041, at 1 January 2027, then a growing perpetuity. Each year builds up from US and international revenue. I take operating margin after stock pay, tax at 21%, and reinvestment of a set number of cents for each added dollar of revenue. Interest is left out because I subtract net debt at its face value. I use net debt at 30 June and give no credit for cash made in the second half, which is conservative by about $2 a share E (estimate).

The bear is the world where type 2 never works and tubeless rivals take share. US growth is 7% in 2027 and settles at 3% by 2030, international slows from 17%, and margin after stock pay slides to 15%. After 2041 it grows 3.0% for ever, the rate its own growth has settled at since 2030. The base takes management’s framing for 2027, US growth of 11% inside its 9% to 14% range and international growth of 24%, down from the 30% to 32% it guides for 2026, then fades. Margin rises from today’s level to 21% by 2029 and stays there. The bull has US growth of 14% in 2027 and double digits to 2031, international growth of 28%, and margin reaching 28%, as SG&A falls to about 35% of revenue. It is still growing 4% in 2041 and steps down to 3.5% after that, as the base does.

What closes the gap

The base is $203 and the price is $149.68, a gap of $53. Here is what it takes, one change at a time, with everything else left as in the base.

No plausible single change closes the gap by itself except a bear half. The price is about where one of my two large worries comes true: growth about three points a year under what I expect for a decade, or a margin that settles about two points below today’s. If both do, the value is $91. If neither does, it is $203. Which of these the market holds I cannot see.

How the odds shrink

The bull case needs several things to go right at once. These are my judgments.

That comes to about 1.6% E (estimate). The four are linked, since a strong business helps all of them, so the true figure is higher. If I raise each to 75%, it is 32% C (calculated). The bear needs fewer links: type 2 fails to stick, a tubeless rival takes share, and margin stalls. At 50%, 40% and 40% that is 8% E (estimate). Neither is a forecast. The base is a central path, not a chain, which is why I put more weight on it than on either end.

The next best use of the dollar

The closest business on this site is Intuitive Surgical: a medical device company whose customers keep paying for the consumable. It is the better business. Its margin is above 30%, it has no debt, and it is growing. But its price needs owner cash to grow 14.5% a year for fifteen years, and at its price my base earns 7.0%. ⁴⁵ At Insulet’s price my base earns 10.7%, which is 3.7 points more. Even my Insulet bear earns 6.8% at this price, more than a Treasury, if its cash flows arrive and growth does not fall below 3%. That is just under the 7.0% my Intuitive Surgical base earns at its own price. It is a statement about the price, not about quality. MercadoLibre at 12.8% earns more than Insulet, with risks its own page sets out. ⁴⁹ I did not recompute the other pages on this site for this comparison.

Quality and price

These are two questions. On quality, I think this is a good business with a hard edge. The product is real and the form is hard to copy. Gross margin is 71%, revenue grew 23.5% last quarter, and international is on its plan. But it makes one product, 78% of revenue goes through three wholesalers, the corrections have recurred, there is new management and a new auditor, and the US is slowing as it tries something new in type 2. On price, the stock is below my base and above my bear. I would call it a good business at a price that already assumes part of what could go wrong. It is not a great business at a bargain.

The margin of safety

I take 30% off base, which gives $142. I use 30% because retention can’t be seen, the margin has to rise to make the base work, the third quarter is unreported, and the corrections are new. The price is 5% above $142 C (calculated). The stock closed below $142 on 34 of the 62 trading days since the August report, and on 14 of 22 days in October, last on 21 October. ¹ So the market has recently offered the price I would call a margin of safety, though not at the latest close. If the market is right, it is because users keep leaving, or because selling cost does not shrink and the 21% margin never arrives. Someone who accepts 9% a year and expects my base would see the price as below value. I want to say that plainly, and I want to say that it is a view about a business whose key inputs I cannot see.

Where this could be in 5 and 10 years

The value at a date is what the owner cash after it is worth then, at 9.0%, less today’s net debt of $414M, divided by today’s 69.46M shares. Owner cash is free cash flow after stock pay and all capital, before interest, and I assume it is paid out. Dates are year ends, a little over two months after this page. The yearly return starts at $149.68, adds the cash, and assumes the price equals value at the end date.

For the base to hold at five years, revenue has to reach $5.6B in 2031, with US growth near 7%, international near 12% and margin at 21%. At ten years it needs $7.8B and $1.15B of owner cash a year. The rise from $203 to $265 is the 9% a year I require, less the $44 a share paid out on the way.

The bear and the bull differ on three numbers. Growth: 2031 revenue of $4.4B in the bear and $6.7B in the bull. Margin: 15% against 28% from 2033. Capital: 45 cents of reinvestment for each added dollar of revenue against 28. The bear’s value rises only because it is nearer and pays cash out, and its yearly return of 3.4% is below the 9.0% I require.

The most likely way the base is wrong is margin. It needs selling cost near 40% of revenue, which ten years have not shown. If margin stops at 19%, the value today is $181, and $237 and $292 at the two dates, about 11% lower. E (estimate)

These are estimates of value, not forecasts of the share price. The price can sit above or below value for years.

VIIThe case for the price

What most people miss

Three facts cut in the company’s favour. International Omnipod grew 32.9% in constant currency in the second quarter, the full-year outlook for it was raised to 30% to 32%, and Omnipod 5 is sold in 20 of the 26 countries where Omnipod is. ⁵ Insulet beat the top end of its own revenue guidance in each of the last three quarters, including the one in which it cut the US year. ⁷⁶⁵ And the corrections are small: $60M to $70M in all, about 0.6% of enterprise value C (calculated), mostly in 2026. ⁴

The bull case

Management’s case is that the market is barely used. On the August call, as a third-party transcript has it, the company said only about 5% of roughly 2.5M US adults on multiple daily injections have moved to automated delivery, with about 3M more on basal insulin alone. Chief executive Ashley McEvoy called the type 2 shortfall a matter of execution, not of the market, price, competition or GLP-1 drugs. ¹¹ Omnipod 6 is planned for 2027, and a closed-loop type 2 system for a 2027 filing. ¹¹⁴ She has bought stock twice, $1.03M at $239.35 in February and $0.16M at $147.47 in August. ¹⁴

Outside views are thin and secondhand. MarketBeat lists 26 analysts with an average target of $186.68, a low of $144 and a high of $375. ⁴⁴ Baillie Gifford owned 2.04M shares at 30 June. ¹⁸ FMR (Fidelity) and Capital Research had cut to 1.4% and 2.4% by mid-2026. ¹⁶¹⁵ Those are positions, not statements of view. ARK owned none ¹⁹, and I found no named investor’s statement on Insulet to quote.

Against reality

In my model, if US growth is 11% in 2027 and international 24%, and margin after stock pay climbs from 18% to 21% by 2029, a share is worth $203. The price needs less: growth about three points a year lower for a decade, or a margin that settles two points below today’s. Today’s numbers are ahead of the path on growth and behind it on proof. The Q3 guide of 17.5% to 19.5% is above the mid-teens exit rate I assume. But US growth has fallen from 28% to a guided 15%, retention has no published figure, and selling cost has not yet fallen.

What a 10x would take. Ten times the price is $1,497 a share, a market value of $104B C (calculated). At 25 times owner earnings after tax, that needs $4.2B of them, and at 30 times, $3.5B. At my base margin of 21%, taxed at 21%, that is $25B of revenue at 25 times and $21B at 30 times. At my bull margin of 28%, it is $19B and $16B, against $3.05B today C (calculated). By 2036 that is 18% to 23% a year for ten years, against 9% in my base and 12% in my bull. The stock itself did not: its November 2025 high was 9.3 times its October 2016 close, in a decade when revenue rose 7.4 times C (calculated). ¹³⁶³ I have not counted other $3B companies, so I give no rate. For Insulet, a 10x from here is not plausible on anything I can see.

Where I come down

Three things moved me: international is working, margin is rising, and insiders bought at $144 to $147. Two did not: US growth is falling faster than I expected, and a 2025 bonus that paid 195.8% of target with no retention measure shows how little the pay tracks what I fear. Evidence that type 2 users stay would raise my values: US growth near 14% into 2027, adjusted margin up a point a year, and no third correction. My bear, base and bull stay at $94, $203 and $370.

VIIIRisks, and what would change my view

The business could be hurt by seven things, beyond those in the valuation.

One product. Omnipod brings in 99% of revenue and the company says it depends on it. ³ A fault, a recall or a better rival hits everything at once.

Channel and payors. Three wholesalers took 78% of 2025 revenue, and rebates rose 44% while revenue rose 31%. Coverage can change, and a pharmacy benefit manager can prefer another pump. ³

Quality and the FDA. Two corrections this year, two in 2022, 476 serious-injury reports to the FDA on the first, and a class action over statements about product safety. ²¹⁴ The company is also exposed to tariffs. The 10-Q says that if the exemption for certain medical devices ended, the effect would be material. ⁴

Competition. MiniMed launched a new pump in June, Tandem has filed for a tubeless pump, and Beta Bionics grew 30% in its latest quarter from a small base. GLP-1 drugs could delay or reduce the need for insulin in some type 2 patients, and the 10-K says so. ^4341423

The courts. The EOFlow reversal is under a petition for rehearing. If it stands, copies of the pod are easier to make and sell. The securities class action has no accrual. ⁴

People and controls. New chief executive, finance chief and auditor in 18 months, and a chair of seven years gone. A material weakness in IT controls outside North America was reported in the 2023 10-K and fixed in 2024. ²³²⁵ The 2025 10-K says internal control was effective, as audited by Grant Thornton, and I found no restatement. ³ That can change quickly in a company growing 20% a year with 5,400 people.

The balance sheet and the world. Debt is $948M against cash of $535M, most of it fixed or hedged. The 10-Q lists the war with Iran among its macro risks, and the company runs plants and a contract manufacturer abroad. ⁴

The thresholds here follow the same base path as the table in section VII.