Analysis of the Form 10-K | Fiscal year ended 31 December 2025
What does Lilly sell, to whom, and how does the money reach the shareholder?
Lilly discovers, makes and sells prescription medicines and reports a single segment (Note 17). The chief executive reviews one consolidated net income figure, so the only disaggregation the filing offers is revenue by product and by geography. Products are sold in about 90 countries and made in the U.S., Puerto Rico, Ireland and other sites. At the end of 2025 the company employed about 50,000 people, up from about 47,000, of whom about 12,000 work in research. The certifying officers are unchanged: David Ricks, chair and chief executive since 2017, and Lucas Montarce, chief financial officer since 2024.
| Revenue, US$ mn | 2025 | Share | 2024 | 2023 | Δ 25/24 |
|---|---|---|---|---|---|
| Cardiometabolic health | 48,221 | 74.0% | 29,521 | 19,668 | +63.3% |
| — of which Mounjaro + Zepbound (tirzepatide) | 36,507 | 56.0% | 16,466 | 5,339 | +121.7% |
| Oncology | 9,376 | 14.4% | 8,753 | 6,658 | +7.1% |
| Immunology | 5,247 | 8.1% | 4,393 | 3,798 | +19.4% |
| Neuroscience | 1,391 | 2.1% | 1,474 | 2,879 | −5.6% |
| Other | 943 | 1.4% | 903 | 1,122 | +4.4% |
| Total | 65,179 | 100% | 45,043 | 34,124 | +44.7% |
| U.S. | 43,481 | 66.7% | 30,375 | 21,791 | +43.2% |
| Europe | 11,558 | 17.7% | 6,921 | 6,175 | +67.0% |
| Japan | 2,132 | 3.3% | 1,815 | 1,673 | +17.5% |
| China | 1,951 | 3.0% | 1,660 | 1,540 | +17.5% |
| Rest of world | 6,057 | 9.3% | 4,271 | 2,946 | +41.8% |
| Total | 65,179 | 100% | 45,043 | 34,124 | +44.7% |
One molecule now carries the company. Tirzepatide, sold as Mounjaro for diabetes and as Zepbound for obesity, produced 56% of revenue. The six products above US$3 bn each produced 82%, against 75% a year earlier. Europe was the fastest-growing region at +67.0%, almost all of it Mounjaro, whose revenue outside the U.S. rose from US$2.6 bn to US$9.3 bn.
A U.S. sale runs as follows. Lilly invoices a wholesaler at list price; three wholesalers each account for between 16% and 24% of revenue. At the moment of sale Lilly estimates what it will later pay back to managed-care plans, Medicare, Medicaid, chargeback and patient-assistance programmes, and books revenue net of it. In 2025 those reductions on the most significant U.S. programmes were US$62.1 bn, against U.S. revenue of US$43.5 bn: for every dollar kept, about 1.43 dollars of list value went back to the payment chain. Of what remains, cost of sales takes 17%, and royalties flow out to partners on some products (to Roche on Ebglyss). To sell at all, Lilly must own plants and carry inventory whose combined book value equals 58.9% of a year's revenue (section 5). A further step is specific to this company: much of the production runs through Ireland, and Irish tax on intercompany profit is paid before the product is sold to a third party. The prepaid taxes on the balance sheet are that timing gap (section 4).
The evidence on advantage is mixed and must be read in the filing's own terms. Price did not hold: the company attributes the fall in realized U.S. prices primarily to Mounjaro and Zepbound, the same products that carried volume (section 3). Research spending kept rising in absolute terms while falling as a share of revenue. Compound patents on tirzepatide run to 2036 in the U.S. and 2037 in major European countries, but U.S. data protection ends in 2027. Trulicity loses U.S. compound and biologics protection in 2027. Barriers such as manufacturing know-how and LillyDirect, the direct-to-patient platform described as "a growing portion of our business", are company assertions; the filing gives LillyDirect no revenue figure. The filing also describes what erodes the position: mass-compounded and counterfeit incretins, which the company says it cannot ensure are adequately regulated, payers that require employers to opt in to obesity coverage, and competitors from China licensing products into major markets. None of these is quantified. Whether the advantage is widening or narrowing cannot be established from this document.
A newcomer would read 45% growth as pricing power. It is the opposite: growth came from volume, the U.S. price fell, and the number that decides the case is how much of each list dollar Lilly keeps after the rebate chain.
What type of company is this, and where is the case decided?
Lilly is a rapid-growth company. Revenue grew 44.7% in 2025 after 32.0% in 2024, and capital expenditure ran at 3.93 times depreciation and amortization. It is not a "solid" company because neither growth nor margins are stable across the years presented: the operating margin we construct moved from 18.9% to 40.4% in two years. It is not cyclical: the filing ties revenue to launches and access, not to a commodity price or a replacement cycle.
The case reduces to one engine, tirzepatide, which explains US$20.0 bn of the US$20.1 bn revenue increase. The governing pair is volume of tirzepatide and net price per unit of tirzepatide. The filing does not publish units or per-product price, so the pair cannot be built in units. It publishes the company-wide split of growth into volume, price and currency, and a U.S. rebate roll-forward from which the share of list value given back can be computed. Those two items stand in for the pair, and the substitution is stated where it is used. Every other product took one line in the bridge: Verzenio added US$416 mn and the remaining portfolio lost US$321 mn net, within which Trulicity fell US$978 mn.
Three questions decide the case.
The first is whether tirzepatide growth is volume that has been bought with price. In 2025 volume rose far faster than price fell; the same mix sustained, with Medicare access discounted from July 2026 under the November 2025 government agreements, would be a different business. The document answers this only in part: it gives the split for the whole company, not for the molecule, and no units.
The second is why net income became much less operating cash and still less free cash flow, and which of the absorbing items will reverse. Receivables, inventory and prepaid expenses, mostly Irish taxes, took US$18.3 bn of cash in one year. Whether that is the cost of growing or a sign that earnings run ahead of cash is the question the balance sheet must answer.
The third is whether the capital being added earns, and who pays for it. Debt rose while dividends and buybacks exceeded free cash flow, and a third of gross plant is still under construction. This too is partly answerable: incremental returns can be computed, but part of the new capital does not yet produce revenue.
Some measures are mute here. Segment margins do not exist. The organic and acquired split is not material, because the acquisitions of the period brought pipeline, not revenue. Book-value multiples say little when the company expenses acquired research (section 8) and retires repurchased shares. Unit economics per patient are not disclosed.
The report goes deep in sections 3, 4 and 6.
Which lines moved the result, and what moved them?
Management gives three explanations. First, revenue rose "driven primarily by increased volume, partially offset by lower realized prices", both "primarily driven by Mounjaro and Zepbound". Second, the gross margin rose 1.7 points through "favorable product mix and improved cost of production, partially offset by lower realized prices". Third, net income rose "primarily due to higher gross margin, partially offset by increased marketing, selling, and administrative expenses and research and development expenses". All three hold against the statements, and the third understates one item: acquired research charges fell by US$370 mn, which lifted pre-tax income without any operating improvement.
| US$ mn | 2025 | 2024 | 2023 | Δ 25/24 |
|---|---|---|---|---|
| Revenue | 65,179 | 45,043 | 34,124 | +44.7% |
| Cost of sales | 11,052 | 8,418 | 7,082 | +31.3% |
| Research and development | 13,337 | 10,991 | 9,313 | +21.3% |
| Marketing, selling, and administrative | 11,094 | 8,594 | 7,404 | +29.1% |
| Acquired in-process R&D | 2,910 | 3,280 | 3,800 | −11.3% |
| Asset impairment, restructuring, special | 484 | 861 | 68 | −43.8% |
| Other—net, (income) expense | 571 | 219 | (97) | n.m. |
| Income before income taxes | 25,731 | 12,680 | 6,554 | +102.9% |
| Income taxes | 5,091 | 2,090 | 1,314 | +143.6% |
| Net income | 20,640 | 10,590 | 5,240 | +94.9% |
The complete statement, every line and every year with its source, is in Financials → Income statement on this company's page.
The scalability is visible line by line. Revenue grew 44.7% and cost of sales 31.3%. Selling costs grew 29.1% and research 21.3%. The statement has no operating-income line. Adding back "other—net" to pre-tax income gives a constructed operating income of US$26.3 bn, up 103.9%. The margin on that basis was 40.4%, against 28.6% in 2024. The 11.7-point gain came from three cost ratios falling at once. Research fell from 24.4% to 20.5% of revenue. Selling fell from 19.1% to 17.0%. Cost of sales fell from 18.7% to 17.0%, the gross-margin gain the company attributes to mix and production cost net of lower prices. Of the three, only the cost-of-sales ratio describes efficiency in making the product; the other two describe fixed spending spread over a larger revenue base, and would reverse if revenue stalled while the programmes continued. Advertising doubled to US$2.9 bn and is now 26% of the selling line, which is where the promotion of launches the company describes shows up. Inside research, early-stage spending grew 24.6% and late-stage 19.5%. The mix is moving toward programmes that are years from revenue, which is consistent with the pipeline table but means a larger share of today's research spend has no product attached yet.
Below the operating line, two items worked against the result. "Other—net" swung to a US$571 mn expense. Interest on borrowings rose to US$895 mn. Undesignated currency forwards booked a US$489 mn loss in the same line, but the company uses them to offset gains and losses on the underlying exposures, which the filing does not quantify. The FY2025 filing no longer breaks the line down, so the net effect of currency and the remainder are not explained. The effective tax rate rose from 16.5% to 19.8%, which the company attributes to the jurisdictional mix and to the U.S. tax law enacted in July 2025.
The pair: tirzepatide volume and net price. Volume here means the quantity of product shipped, and price means what Lilly retains per unit after the rebates, discounts and returns it books at the time of sale. Revenue is the product of the two, and the filing never states either one for the molecule. The volume half comes from the revenue bridge in Item 7; the price half comes from the same bridge and from the U.S. rebate roll-forward in the critical accounting estimates.
For the company as a whole, revenue grew 45%. Volume added 50% and price took away 6%, while currency added 1%. In the U.S. the split was volume +53% and price −10%; outside the U.S., volume +46% and price flat. In 2024 the same bridge showed volume +27% and price +5%, so the year turned price from a contributor into a drag. By region, the U.S. supplied US$13.1 bn of the revenue increase and the rest of the world US$7.0 bn, so the price decline sits in the larger half. The filing names Mounjaro and Zepbound as the source of both the U.S. volume and the U.S. price decline. Mounjaro grew 52.5% in the U.S., where "strong demand" was "partially offset by lower realized prices". Zepbound grew 173.7% in the U.S. on "increased demand", with the same offset.
The rebate roll-forward provides a second reading of the price half. The reductions booked against U.S. sales, set out in section 1, rose 49.9% in 2025. Dividing them by U.S. revenue plus the reductions gives the share of list value handed back: 58.8% in 2025, 57.7% in 2024 and 63.5% in 2023. This is our construction, not a figure the company reports, and it has two limits: the roll-forward covers only the "most significant" programmes, and U.S. revenue includes collaboration income. On this reading the share given back rose by about one point in 2025 after falling by almost six in 2024.
The mechanism the filing describes for the fall in price is a change in channel and payer mix rather than a list-price cut. Zepbound is an obesity medicine that U.S. self-insured employers "must generally opt in" to cover, Medicare access to discounted obesity medicines is expected only from July 2026, and CVS Caremark removed Zepbound from preferred status on some plans in July 2025. Lilly also sold to self-paying patients through LillyDirect and, in November 2025, reached preliminary agreements with the U.S. government that lower Medicaid prices and give Medicare beneficiaries discounted access by July 2026. The filing does not price these channels, but it describes both the government arrangements and the self-pay market as dependent on lower prices. A shift of volume toward them is consistent with a 53% volume gain and a 10% price decline. Outside the U.S. the mechanism is launches: Mounjaro was launched "in all major markets", and outside-U.S. Mounjaro revenue rose 3.6 times.
The company does not publish units shipped, net price per dose, the split between Mounjaro and Zepbound of the volume and price effects, LillyDirect revenue or the share of tirzepatide sold under each payer programme. Without them, the pair can be described in direction but not measured.
Growth was organic for practical purposes. The three business combinations of the period, POINT in December 2023, the Wisconsin plant in 2024 and Verve in 2025, brought research assets and a factory, not revenue the filing reports. Collaboration revenue includes one-time Jardiance amendment benefits in both years (section 7), and the 2023 base includes about US$2.0 bn of revenue primarily from the sale of the olanzapine and Baqsimi rights. Neither changes the reading of 2025.
The year's result was built on volume growth at a falling net price, absorbed by costs that grew far more slowly than revenue. What sustains the margin is therefore volume, and the line that will show whether volume keeps outrunning price is the price component of next year's revenue bridge.
Is the profit cash, and does it belong to the company?
Operating cash flow was US$16.8 bn against net income of US$20.6 bn, a ratio of 0.81. The ratio was 0.83 in 2024, so the gap is not new. Non-cash items added US$4.3 bn net. The largest were acquired research of US$2.9 bn, expensed in the income statement but paid in investing, and depreciation and amortization of US$2.0 bn, less a deferred tax benefit of US$1.7 bn. Working capital then absorbed US$8.1 bn. Receivables took US$7.0 bn, inventory US$4.7 bn and prepaid expenses US$6.6 bn. Higher payables and accruals released US$10.2 bn against them. Free cash flow, operating cash less capital expenditure, was US$9.0 bn after US$3.8 bn in 2024.
Owner earnings, defined once for this report as net income plus non-cash items less capital expenditure and less working capital absorbed, were US$5.4 bn. Without the working-capital deduction they were US$13.5 bn. The judgment uses the figure with working capital, because a company that must carry more receivables and inventory to sell more is investing as surely as if it bought a plant. Owner earnings fall below free cash flow by US$3.5 bn because our definition keeps stock compensation and acquired research as costs, while the cash flow statement adds both back to operating cash.
Three questions follow.
The first is how much of operating cash is other people's money held temporarily. The rebate accrual rose US$5.8 bn and trade payables US$2.2 bn. Together, US$8.0 bn equals 89% of free cash flow. The rebate accrual is cash owed to payers on sales already booked; it is paid with a lag and grows with sales, so it funds Lilly while volume rises and reverses when volume stops. Payables days rose by more than a third (section 5), a change the filing does not explain.
The second is how far apart tax expense and tax paid are. Lilly paid US$10.8 bn in cash taxes against a tax charge of US$5.1 bn, a gap of US$5.7 bn. The filing explains it: "taxes paid on intercompany profit not yet recognized, primarily related to Ireland", as Irish plants produced for inventory that had not yet been sold, plus a prior-year payment. Prepaid taxes rose from US$7.1 bn to US$12.9 bn. This is the reverse of the usual pattern. The company pays tax ahead of the income statement, and the asset will unwind into tax expense as the inventory is sold to third parties. Cash taxes were also helped in the U.S. by immediate deduction of research costs and accelerated depreciation under the July 2025 law.
The third is whether capital expenditure captures all investment, and it does not entirely. Contract-manufacturing and supply agreements could require up to about US$10 bn if minimum quantities are not bought, over periods of up to eight years. Pre-launch inventory, capitalized before approval because the company judges approval probable, rose to US$1.5 bn from US$548.1 mn, primarily orforglipron. Interest is reported net of capitalized interest, and the capitalized amount is not disclosed while construction in progress rose 57.8%. Long-lived asset spending including software was US$8.7 bn, against US$7.8 bn of property purchases.
Accounting choices move earnings in identifiable places. Acquired research is expensed immediately when the purchase is classified as an asset acquisition; a business combination would have put it on the balance sheet. That classification is a management judgment, and in 2025 it charged US$2.9 bn to the income statement. The non-GAAP measure the company reports excludes amortization, special charges, equity-investment results and a U.S. tax-law effect, which added US$1.26 per share to reported EPS of US$22.95, per the proxy. The effective rate sits below the statutory 21%: non-deductible acquired research adds 1.7 points, while Irish rates, the foreign-derived income deduction and tax credits subtract more.
The early-deterioration comparisons, 2025 against 2024:
| Comparison | Item growth | Reference growth | Gap |
|---|---|---|---|
| Inventory vs revenue | +81.1% | +44.7% | +36.4 pts |
| Receivables vs revenue | +61.4% | +44.7% | +16.7 pts |
| Allowance vs gross receivables | not available | ||
| Payables vs cost of sales | +66.6% | +31.3% | +35.3 pts |
| Pre-launch inventory vs R&D expense | +173.7% | +21.3% | +152.4 pts |
| Operating cash vs revenue | +90.7% | +44.7% | +46.0 pts |
Five gaps exceed five points, and they point in different directions. Among operating balances, inventory is the largest unfavourable gap; the pre-launch comparison is wider but concerns a smaller balance. Work in process more than doubled to US$8.2 bn, consistent with the capacity build the company describes, but days of inventory rose by more than a third (section 5), and part of the stock is pre-launch inventory whose value depends on approvals. Receivables days also rose; the filing states that the allowance is "not material" and does not give it, so a loosening of terms cannot be excluded from the document. Payables grew twice as fast as cost of sales, which flatters cash. Operating cash grew faster than revenue because net income nearly doubled; working capital absorbed more, not less, than in 2024.
The liability side of working capital released US$7.6 bn more than in 2024, a figure larger than the US$5.2 bn growth in free cash flow; the asset side absorbed US$10.3 bn more. Net, working capital cost free cash flow US$2.7 bn more than a year earlier, so the growth in free cash flow did not come from working capital. The gross liability flow does, however, exceed a third of that growth, and it reverses if volume stops. It is carried to the weaknesses in section 11.
The profit is real in the sense that it is the company's and is collected: receivables are concentrated in three large wholesalers and the history of revenue adjustments is small. It is not yet cash in proportion, because a growing company with Irish production pays tax and builds inventory before it sells.
What stands ahead of the shareholder, and what is it worth?
Total assets rose to US$112.5 bn from US$78.7 bn. Nine asset lines exceed 5% of the total. Property and equipment of US$24.7 bn, 21.9% of assets, includes US$13.0 bn of construction in progress. That is 35% of gross cost, plant already paid for that does not yet produce. Receivables are US$17.8 bn and inventory US$13.7 bn. Prepaid expenses are US$14.3 bn, almost all prepaid tax (section 4). Deferred tax assets of US$10.0 bn rest mainly on research costs capitalized for tax (US$4.5 bn) and on rebate accruals (US$2.8 bn). Correlative adjustments, the mirror of tax disputes between jurisdictions, add US$2.4 bn. A valuation allowance of US$1.2 bn covers carryforwards the company does not expect to use. Other noncurrent assets of US$7.0 bn include pension and retiree-health surpluses of US$3.5 bn and lease right-of-use assets of US$1.3 bn; the remainder is not broken down. Cash is US$7.3 bn.
Goodwill is US$5.9 bn after US$127 mn from Verve; no impairment was recorded in the three years. The filing does not describe the reporting units or give any headroom. Intangibles of US$6.5 bn comprise marketed-product rights of US$5.0 bn, with a remaining weighted life of about eleven years, and acquired research of US$1.6 bn, which is not amortized. The company states it is "likely" that some acquired research assets "will become impaired in the future". The capitalization policy is narrow: research is expensed, milestones are capitalized only at or after approval, and acquired research is capitalized only in business combinations or when it has an alternative use.
On the liability side, the rebate accrual is 15.5% of assets and the subject of the auditor's critical audit matter (section 7). Other current liabilities of US$8.5 bn are not broken down beyond forward-contract, lease and pension current portions. Long-term income taxes payable are US$5.9 bn, of which US$1.1 bn is due in 2027 and US$4.8 bn has no estimable timing.
Total debt is US$42.5 bn, all notes, after commercial paper of US$4.3 bn was repaid. Operating leases of US$1.4 bn are outside this figure. Maturities are long: US$12.6 bn, or 29.4%, falls due by 2030 and US$14.1 bn after 2050. Fair value was US$39.8 bn, below carrying value. Covenants are not described. Unused committed facilities are US$10.1 bn. Off-balance-sheet, the supply commitments (section 4) and unfunded venture commitments of about US$902 mn stand ahead of the shareholder. Net debt is US$35.2 bn, treating all cash as available because the filing does not separate restricted cash.
Working capital is measured in days on year-end balances against revenue or cost of sales of the same year. Receivables stood at 100 days, against 89 in 2024 and 97 in 2023. Inventory stood at 454 days, against 329 and 298. Payables stood at 178 days, against 140 and 134. Equity is US$26.5 bn; non-controlling interests are not material, and the employee benefit trust holds 50 million shares at a cost of US$3.0 bn.
Three readings are required. Return on equity is 101.2% on average reported equity. Adding back the trust gives 88.1%; repurchased shares are retired against retained earnings, so the full history of buybacks cannot be added back from this document. The figure measures a small, buyback-reduced equity base more than a return. Retained earnings rose US$10.9 bn to US$24.5 bn, and they must not be read alone, since a company that distributed everything would show them flat without being worse.
Solidity, told rather than scored: interest on borrowings takes 3.4% of constructed operating income, and long-term debt equals 1.98 years of current net income.
About 24.4% of assets are cash and receivables carried at contractual or invoice amounts, 34.2% are plant and inventory at cost, and 19.9% (goodwill, intangibles, deferred tax assets) depend on management's estimates of future income. Prepaid taxes, a further 11.5%, depend on the inventory being sold through the Irish chain.
If growth stopped, tangible equity would be US$14.1 bn, or US$15.70 per diluted share. Net working capital is US$20.4 bn, a current ratio of 1.58. Cash is US$8.08 per share, and cash net of gross debt is minus US$39.18 per share. If revenue stopped growing and margins returned to the 2023–2025 average, receivables and cash would protect the shareholder, while pre-launch inventory, construction in progress and prepaid taxes would protect the shareholder only to the extent the products they were built for are sold.
Does the capital earn, and does the shareholder receive it?
Per diluted share, the company moved as follows. Three years make two observations, not a trend.
| Per diluted share, US$ | 2025 | 2024 | 2023 |
|---|---|---|---|
| Revenue | 72.48 | 49.82 | 37.78 |
| Net income (EPS) | 22.95 | 11.71 | 5.80 |
| Free cash flow | 9.98 | 4.16 | 0.88 |
| Equity | 29.51 | 15.79 | 12.03 |
| Diluted shares, mn | 899.3 | 904.1 | 903.3 |
Revenue per share grew 45.5%, slightly faster than revenue, because the diluted count fell 0.5%. The growth was not paid for with shares. Free cash flow per share rose 2.4 times but remains less than half of earnings per share.
Return on invested capital, which we construct because the company reports neither operating income nor invested capital, was 39.7% in 2025 and 27.7% in 2024. Capital is equity plus debt less cash, averaged over the year; the return is constructed operating income after the effective tax rate of each year. The incremental return, added after-tax operating income over added capital, was 60.3% in 2025 and 49.3% in 2024. An incremental return above the average return means the new capital is earning more than the old, the opposite of a company living on a past franchise. The return can be split into margin and turnover. After-tax operating margin rose from 23.9% to 32.4%. Revenue per unit of average capital rose only from 1.16 to 1.22. The improvement in return came from margin, not from using capital more intensively. The limit is the denominator: added capital includes US$4.8 bn more construction in progress, US$5.8 bn more prepaid taxes and US$6.2 bn more inventory, balances that had not yet produced revenue at year end. The incremental return is computed partly on capital that is not yet working, so the return on the capital already producing is higher than the figure shown.
The reinvestment rate, added capital over after-tax operating income, was 81.2% in 2025 and 105.6% in 2024. Multiplied by the return on capital, it implies growth of 32.2% in 2025. Realized growth in constructed operating income was 103.9%. The difference did not come from leverage, which is below the operating line, or from acquisitions, which added no revenue. It came from operating leverage on capacity built earlier: operating costs before acquired research rose far less than revenue (section 3). Lower acquired-research charges added a further amount (section 3).
Return and distribution meet here. The dividend paid per share rose 15.4% to US$6.00. The indicated 2026 rate is US$6.92, a further 15.3%. A fixed dividend that grows faster than free cash flow per share would commit a growing share of cash before any decision on plant or pipeline is taken; in 2025 free cash flow per share grew faster, so the commitment was covered. Lilly paid out 46% of net income in dividends and buybacks, which on its own would leave most of that return compounding. The retained part was not funded by free cash flow: distributions of US$9.5 bn were 105.8% of free cash flow, so reinvestment came from new debt.
Self-funding is the main finding of the section. Over 2023–2025, operating cash was US$29.9 bn and capital expenditure US$16.3 bn, leaving free cash flow of US$13.5 bn. Dividends and buybacks took US$21.5 bn, 159% of that free cash flow, and acquired research and acquisitions a further US$13.0 bn. The difference was borrowed: net debt raised was US$25.6 bn. For three years Lilly has paid its shareholders more than its free cash flow and borrowed to build its plants and buy its pipeline. The company is not self-financing in this period; it depends on the debt market, at an average cost of 2.35% on the reported interest line.
Capital intensity is rising. Capital expenditure was 3.93 times depreciation and amortization in 2025, against 2.86 and 2.26 in the two prior years. Over the three years it absorbed 44.8% of cumulative net income. The filing does not split maintenance from expansion; using depreciation and amortization of US$2.0 bn as a stated proxy for maintenance, about US$5.8 bn of 2025 capex was expansion. If expansion stopped, free cash flow would have been about US$14.8 bn on 2025 figures, before any change in working capital. Growth spending is not only in plant. Research and selling together take 37.5% of revenue (section 3), against capex at 12.0%. Acquired research adds another 4.5%, so plant is the smaller part of what Lilly spends on growth.
The multiples use a closing price of US$1,074.68 at the last trading day of 2025 and 943,357,420 shares at 9 February 2026. Both were supplied with the request and not taken from the filing. The price date matches the balance-sheet date so that numerator and denominator describe the same moment. The supplied share count matches issued shares (944.8 million at year end), which include the 50 million trust shares that the company excludes from earnings per share. Market capitalization on that count is US$1,013.8 bn, and enterprise value, adding debt and deducting cash, is US$1,049.0 bn.
At that price the shareholder pays 46.8 times diluted earnings per share, or 49.1 times net income on the supplied share count. The same price is 107.7 times free cash flow per share and 36.4 times book value per share. A buyer of the whole company before its financing would receive constructed operating income equal to 2.51% of enterprise value, the one multiple not distorted by tax timing. Enterprise value is 37.1 times EBITDA. The declared dividend of US$6.23 yields 0.58%.
The arithmetic of the price is that it pays for income that does not yet exist. At that operating yield on 2025 profit, the price assumes the capacity under construction and the pipeline will produce returns well beyond today's, which is the question the capital-intensity and self-funding figures leave open.
Which notes change the reading, and by how much?
The statements say how much. The notes say how much of each figure is a management estimate.
The rebate accrual is the only critical audit matter. It stands at US$17.4 bn, and its counterpart is a deduction from revenue. The auditor calls it challenging because historical rebate data "may not be predictive", so management "supplements its historical data analysis with qualitative adjustments", particularly for the products that contribute most to revenue. The company gives a sensitivity: a 5% change in the liability moves revenue by about US$897 mn, or 3.5% of pre-tax income. A shift of volume toward higher-rebate channels, such as the Medicaid and Medicare arrangements for obesity, would raise it; a gap between estimated and actual payments would be booked against later revenue. Adjustments for prior periods have been below 1% of U.S. revenue in 2025 and below 3% in 2024. The accrual grew 50.6%, faster than revenue.
Income taxes contain the other large judgment. Gross unrecognized tax benefits are US$5.1 bn, up from US$4.0 bn, and US$3.2 bn would lower the effective rate if recognized. The pricing of intercompany transactions for 2016–2018 is in competent-authority procedure, and the U.S. examination of 2019–2021 continues. Prepaid taxes (section 4) are themselves taxes on intercompany profit. The company states that a 5% change in uncertain positions moves net income by US$160 mn. The share of pre-tax income earned in the U.S. and Puerto Rico rose from 20% to 54% in one year, so the tax base is moving while the dispute concerns the old allocation.
Pre-launch inventory and purchase commitments are a judgment on approval. Lilly capitalizes inventory before approval "when we believe that future commercialization is probable"; the balance (section 4) is primarily orforglipron, submitted for approval in the U.S., EU and Japan. The supply commitments relate to "medicines in development". A rejection or a delay would turn both into charges, and the filing gives no sensitivity.
The litigation note contains a charge without a figure. After the Seventh Circuit affirmed a jury verdict on Average Manufacturer Price calculations, Lilly "recognized a charge related to the matter" in 2025. Note 5 attributes the US$484 mn of special charges primarily to "a litigation charge" and Verve costs, and gives neither amount. The incretin product-liability claims are in two federal multidistrict litigations, and the company is "predominantly self-insured". No ranges of loss are given.
The Boehringer Ingelheim collaboration contributed US$3.4 bn of revenue, 5.3% of the total, and contains a recurring "one-time" item. Amendments that reassigned Jardiance markets produced one-time benefits of US$370 mn in 2025 and US$300 mn in 2024, and a US$200 mn sales milestone was also recognized in 2025. Up to US$910 mn of further milestones are possible in 2026, while Jardiance faces a government-set Medicare price from 2026. Revenue from this line depends partly on contract events rather than prescriptions; it is recognized as collaboration revenue but shown in the product table alongside product sales.
Pension plans are not a pressure point: assets exceed obligations by US$558 mn, and the retiree health plan is overfunded by US$1.5 bn. Three-quarters of pension assets are valued at fund-reported net asset values.
The notes shift the reading toward caution on revenue and tax rather than on assets. The two largest estimates sit on the revenue line and on cash already paid to Ireland.
What did management do with the money, and do they say so when it goes badly?
David Ricks is chair and chief executive; the board has a lead independent director and all-independent committees. Lilly Endowment holds 9.8% of the shares. Vanguard holds 8.6%, BlackRock 6.8% and PNC 5.4%. Directors and executive officers as a group own 1,362,076 shares, under 1%, worth about US$1.46 bn at the supplied price. Mr Ricks holds 815,535 of them. The proxy does not say whether these were bought or received through awards. Pledging and hedging are prohibited. Ernst & Young has audited the company since 1940, for fees of US$19.6 mn. No related-person transactions were reportable in 2025. The board is classified, and the 2026 proxy proposes to declassify it and remove supermajority voting provisions.
Capital allocation over the three years is set out in section 6: plant, acquired research and distributions to shareholders, with debt filling the gap. Of the acquired research, the DICE oral IL-17 molecule (US$1.9 bn in 2023) was discontinued in 2024 "in favor of another molecule in development". The Vitrakvi intangible was impaired in 2024. Verve added US$608 mn of acquired research in 2025, with goodwill described in section 5.
Lilly bought back 4.8 million shares in 2025 for US$4.1 bn, an average of about US$855 per share. In the fourth quarter it paid an average of US$1,001.60. Both are below the supplied year-end price (section 6). Measured against free cash flow per share, the company bought itself at about 86 times in 2025, 200 times in 2024 and 371 times in 2023. The multiple fell because free cash flow rose faster than the price paid. US$10.9 bn of the US$15.0 bn programme remains.
Net dilution was negative: 1.7 million shares were issued under employee plans against 4.8 million retired. The buyback did more than offset stock compensation, which cost US$626 mn, about 1.0% of revenue.
Adjusted measures exclude the same categories every year (amortization, special charges, equity-investment results), with a U.S. tax-law item added in 2025. Over three years the non-GAAP exclusions add US$3.06 per share to reported EPS. The bonus goes further: it also excludes acquired research, which over 2023–2025 totalled US$9.99 bn, or US$10.70 per share, plus the remaining tax-law effect, currency differences from plan and divestiture revenue. For bonus purposes 2025 EPS was US$27.06. Acquired research is the company's main way of buying its future pipeline, and the annual bonus does not charge managers for its price; the long-term awards, paid on share price, are the only part of pay that reflects it.
The 2025 decisions read together say one thing. Lilly issued US$13.3 bn of notes in two offerings, repaid commercial paper and raised spending on plant by 55%. It bought two research programmes for US$1.9 bn and Verve for US$0.5 bn net. It increased the buyback and, in early 2026, signed further acquisitions with closing payments under US$3 bn. Each decision has its own source note; together they show a company that has chosen to finance capacity, pipeline and distributions at the same time, with debt filling the gap. The consequence for next year's balance sheet is more plant, more debt and, if the capacity runs, much larger depreciation.
Compensation rewards three things: revenue of seven named products, EPS adjusted as above, and pipeline progress, each a third of the bonus; the pipeline multiple was 230%. Long-term awards pay on share price against a 9.48% cost of capital and on relative shareholder return. Chief executive compensation was US$36.7 mn, up from US$29.2 mn.
On candour, the filing contains admissions in factual form: the DICE molecule was discontinued in favour of another, the Vitrakvi impairment was "driven by expected commercial projections", and the CVS Caremark decision "has negatively impacted access". The management discussion does not comment on the price decline beyond attributing it to two products, and the litigation charge appears without its amount.
What would have to happen for capital to be lost permanently here?
The risks, in order of weight for this case:
Three of these have already materialized in the 2025 figures: the U.S. price decline, the CVS Caremark exclusion and the Average Manufacturer Price charge. The Jardiance price is fixed and takes effect in 2026. The others remain potential, and capacity is the one with the largest balances already behind it.
Comparison with the FY2024 risk factors shows what entered. The November 2025 agreements with the U.S. government are new, as are the CVS Caremark exclusion, the risk on "patient participation in cash-pay markets" through LillyDirect, and the explicit naming of "intensifying competition... including from China". Orforglipron is now named as a "highly anticipated" submission. The EU agreement to shorten maximum data protection is new. Concentration was raised: the six-product share rose (section 1), and the tirzepatide reference now replaces a three-product group that included Trulicity. Outside the risk factors, the supply-commitment figure fell from about US$14 bn to about US$10 bn, with a reworded scope. The headings are largely unchanged; the changes are inside them.
The pre-mortem. Three years from now this investment has produced a permanent loss of capital; three mechanisms in the document could have done it.
The first is price outrunning volume. Under the government agreements and payer pressure, the price component of the bridge turns more negative than −6% while volume growth slows. Today's measure is the share of U.S. list value given back, 58.8% by our construction. The signal would be the price line of the revenue bridge and the reductions in the rebate roll-forward growing faster than U.S. revenue.
The second is capacity that is not needed. If orforglipron is delayed or competing products take volume, the plants and inventory built for it would be written down. Today the exposure is the plant under construction, the inventory and the commitments listed in risk 4; inventory days, at 454, are the measure that moves first. The signal would be inventory days rising further and an inventory or impairment charge in the special-charges line.
The third is financing. Lilly pays out more than its free cash flow and has borrowed heavily in three years (section 6). If cash flow stalls while distributions are maintained, debt rises further. Today net debt is US$35.2 bn, 1.25 times EBITDA. The signal would be the difference between dividends plus buybacks and free cash flow widening.
What was promised, and what became of it?
The catalysts the filing itself names are these:
None of these controls the price at which the new volume will be sold.
This is the first year of coverage, so the check runs against what the FY2024 filing said. That filing gave no revenue or earnings guidance; its expectations were about capital spending, financing, distributions, supply and the pipeline. On the pipeline, tirzepatide for heart failure with preserved ejection fraction, then submitted, is now approved in the EU; insulin efsitora alfa moved from Phase 3 to submission; orforglipron moved from Phase 3 to submission for obesity. One molecule bought in the period was dropped: the DICE oral IL-17 molecule, charged in 2023, was discontinued in 2024 in favour of another.
| Stated in FY2024 10-K | Realized in FY2025 | Gap |
|---|---|---|
| "Meaningfully higher capital expenditures over the next several years" | Capex US$7.8 bn vs US$5.1 bn | In line |
| Feb 2025 notes (US$6.5 bn) to fund business development and repay commercial paper | Commercial paper from US$4.3 bn to zero; acquired research and acquisitions US$3.7 bn | In line |
| Indicated 2025 dividend US$6.00 per share | Paid US$6.00 | In line |
| US$15.0 bn buyback authorized, nothing yet bought | US$4.1 bn bought, US$10.9 bn left | Executed in part |
| Tirzepatide "supply currently exceeds demand in the U.S." | U.S. volume +53%, price −10% | Volume delivered at a lower price |
| Pre-launch inventory US$548.1 mn, primarily orforglipron | US$1.5 bn; orforglipron submitted | Exposure nearly tripled before approval |
| Supply commitments up to about US$14 bn | Up to about US$10 bn, scope reworded | Not comparable |
The capital and distribution commitments were kept, and most stated pipeline steps advanced by one stage. What the earlier filing did not foresee was the fall in realized price that accompanied the volume: in February 2025 it wrote that supply exceeded demand, and the year then brought both more volume and a lower U.S. price. The commitment figure is the one line where the change cannot be read, because its description moved from incretin medicines to medicines in development.
The three questions of section 2 can now be answered with the figures already shown.
On the first question, growth in 2025 was volume bought partly with price: volume rose far faster than price fell, and the share of U.S. list value given back rose about one point (section 3). That trade was profitable in 2025 because costs grew far more slowly than revenue. The reading holds as long as the volume line of the bridge stays well above the price line; it fails if the Medicare and Medicaid channels, which the company says may not offset their concessions, push the price component down faster than volume grows.
On the second question, the gap between profit and cash is explained and mostly temporary in nature: prepaid Irish taxes, work-in-process inventory and receivables from three wholesalers. It is also a gap that can only close through sales. The reading holds if prepaid taxes and inventory days fall as capacity enters production; it fails if they keep rising faster than cost of sales.
On the third question, the capital added in 2025 earned an incremental return above the average return (section 6), but the period's distributions exceeded free cash flow by a wide margin and debt paid for the rest. The reading holds if free cash flow covers dividends and buybacks once capex stops rising; it fails if debt keeps growing to sustain them. At the supplied price, the operating yield on enterprise value means the market price depends on those outcomes, not on 2025 profit.
Each strength rests on a figure demonstrated in the body.
Each weakness rests on a figure demonstrated in the body.
These are things the document does not say and that would change the judgement.
The judgement is that 2025 shows a company converting volume into profit at high incremental returns, financed partly by debt and by other people's money, at a price that already assumes the conversion continues.
What next year's filing will settle
- Price component of the revenue bridge: −6% consolidated, −10% U.S. in 2025. It answers the first question: whether volume keeps outrunning price.
- U.S. rebate reductions and accrual: US$62.1 bn of reductions, US$17.4 bn accrued. They show whether the share of list value given back keeps rising.
- Inventories and prepaid taxes: US$13.7 bn and US$12.9 bn. They answer the second question: whether the cash gap begins to close.
- Construction in progress: US$13.0 bn. Its transfer into service, and the depreciation that follows, answer the third question.
- Total debt against dividends plus buybacks less free cash flow: US$42.5 bn of debt; distributions 105.8% of 2025 free cash flow. They show whether distributions are still borrowed.
This report is based on Eli Lilly and Company's Form 10-K for the fiscal year ended 31 December 2025, with auditor's report and management certification dated 12 February 2026. The FY2024 Form 10-K is used for comparison and the 2026 Proxy Statement for ownership, compensation and governance. Multiples use the closing price of US$1,074.68 at the close of FY2025 and the share count supplied at 9 February 2026, both provided with the request. This report is for analysis purposes only and does not constitute investment advice.
The full data table behind this report — every line item with its source in the filing and the arithmetic checks — is published separately: Eli Lilly and Company FY2025 data table.