UnitedHealth Group

IThe business

UnitedHealth Group is two companies sharing a name. UnitedHealthcare is the insurer. It sells health plans to employers, to people over 65 through Medicare Advantage, and to states for Medicaid, and it served 49.8 million people at the end of 2025 and 48.3 million at the end of September 2026.²³ Optum is everything around the insurer: doctors and clinics, a data and technology unit, and one of the three largest pharmacy benefit managers.¹⁴ In 2025 UnitedHealthcare had revenue of $344.9B and Optum $270.6B, and $168.0B of that was counted twice because one arm pays the other, so the group total was $447.6B.²

Two numbers explain most of what happens to the profit. The first is the medical care ratio: medical costs divided by premium revenue. When it moves one point, earnings from operations move about $3.5B at 2025 premiums C (calculated).² The second is Medicare. Payments from the federal Medicare program were 44% of group revenue in 2025, most of it Medicare Advantage.² The government sets those payments once a year; UnitedHealth cannot raise the price in the middle of the year if costs run ahead.

UnitedHealthcare: where the cost trend lives

UnitedHealthcare has three parts. Employer & Individual had $79.2B of revenue and served 29.7 million people at the end of 2025, 21.5 million of them in plans where the employer, not UnitedHealth, carries the claims cost.² That is a fee business and it is stable. Community & State had $94.4B and served 7.4 million, mostly Medicaid, a business where states set the rates and the people who stay enrolled are sicker.² Medicare & Retirement is the large one: $171.3B of revenue, up 23% in 2025, with 8.4 million people in Medicare Advantage at the end of the year.² Part of that growth was a rule change: under the Inflation Reduction Act the insurer now bears more Part D drug cost, which raises both revenue and medical cost.²

The ratio of medical cost to premium tells the story. It was 79.1% in 2020 and 83.2% in 2023. It was 85.5% in 2024 and 89.1% in 2025 C (calculated).²⁸ The 10-K gives the reasons in a list: Medicare funding reductions, elevated medical cost trend, the profile of patients newly added to value-based care, an early charge for expected 2026 losses at Optum Health, less favorable reserve development, the Inflation Reduction Act, and sicker people on the individual exchanges.² Plain terms: older patients used more care at higher prices than the premiums assumed, and the government paid less than the cost of that care.

What changed in 2026 is that UnitedHealth repriced and cut benefits, and accepted losing members. People served in Medicare Advantage fell from 8.4 million at the end of 2025 to 7.6 million at the end of September 2026, which is 840,000 fewer, or 935,000 once Medicaid-linked senior plans are counted. Total UnitedHealthcare enrollment fell 1.5 million.³ The medical care ratio for the third quarter of 2026 was 88.4%, against 89.9% a year before, and the company expects 88.1% plus or minus a quarter point for the full year.³¹⁵ The 10-Q says the cost trend is still rising: unit prices, more care per visit, more services bundled into each visit, and, in commercial plans, the arbitration process under the No Surprises Act and more aggressive billing.¹²

One detail matters more than it looks. The second and third quarters of 2026 each include a release of reserves set aside earlier: $860M in the second and $610M in the third, most of it for 2026 dates of service.¹⁵³ Without the third-quarter release the ratio would have been about 89.1% C (calculated), about the same as a year ago. That is not a sign of trouble. Reserves are an estimate, and UnitedHealth was cautious in the spring. But a fall of 1.5 points year on year in a quarter is partly accounting for what did not happen, and I do not treat it as a full cost improvement. In 2025 the same line was only $140M favorable for the whole year, down from $700M in 2024 and $840M in 2023.²

Medicare Advantage in more detail

Medicare Advantage lets people over 65 get their Medicare through a private plan, which receives a fixed monthly amount from the government for each member. That amount is adjusted for how sick the member is, using the diagnoses doctors report. Higher recorded illness means more money. This is why coding matters so much to both sides. The 10-K lists the regulatory pressure plainly: rate notices for several years that sat well below the expected rise in medical cost, a 2027 advance notice that was far below it, and large changes to the risk-adjustment model that reduce funding for people with the greatest health and social needs.² By the final 2027 notice, the 10-Q says, the rate had moved toward the industry cost trend and was still below it.¹² I could not open the CMS announcement and do not give the percentage.

The company works the levers it has: benefits, premiums, networks and counties. It has dropped plans in some counties, and its Stars ratings, which decide quality bonuses, are holding: at least 69% of members will be in plans rated four stars or better for 2027.¹²³ KFF counted 35.2 million people in Medicare Advantage in 2026, 55% of eligible beneficiaries. UnitedHealth had 26% of them, down from 29% a year earlier, while Humana rose from 17% to 20%.⁶ KFF’s count for UnitedHealth, 9.3 million, is larger than the company’s own 7.6 million. I believe the gap is Medicaid-linked senior plans that the company reports in another line, but I have not confirmed it.⁶³ So the company is shrinking in the business that made its profit, on purpose, and its closest rival is growing in it.

Optum: three businesses

Optum Health employs or contracts with doctors and takes the cost risk for patients in value-based contracts, where it receives a monthly payment and pays the claims. In 2025 it had $102.0B of revenue and lost $0.3B of operating earnings after earning $7.8B the year before; adjusted for the charge it still earned only $2.3B.²⁴ The cause was the same as the insurer’s: Medicare funding cuts and costlier patients on a fixed payment. In the third quarter of 2026 revenue was $22.9B, 10% lower, because Optum chose to serve fewer patients in markets where it lacked scale. Earnings were $735M, a 3.2% margin, against $41M a year earlier.³ It is a repair job. It is one of the places where I could be most wrong; the Medicare Advantage cost trend is the other.

Optum Insight sells software, analytics and outsourced services to hospitals, insurers, states and drug makers. It is the highest-margin unit: $19.4B of revenue and 13.5% margin in 2025, 23.4% in the third quarter of 2026 on a basis that now includes Optum Financial, the payments business and Optum Bank.²³ Its backlog of $31.1B includes $12.9B from affiliated companies, which means about 41% of its future revenue comes from UnitedHealthcare and Optum themselves C (calculated).² Insight also holds Change Healthcare, the claims processor hit by the 2024 cyberattack. The company booked its final direct cyberattack costs of $799M in the fourth quarter of 2025, and in 2024 it lent $9.0B to care providers to bridge the outage.⁴² Repayment is slow: $282M came back in the first nine months of 2026, and $10.8B of loan receivables of all kinds stood on the balance sheet at the end of June.³¹²

Optum Rx is the pharmacy benefit manager. It negotiates drug prices and rebates with manufacturers, builds the list of covered drugs for employers and plans, and runs mail-order and specialty pharmacies through about 64,000 retail pharmacies. In 2025 it managed $188B of drug spend, including $87B of specialty drugs, filled 1,659 million adjusted scripts, and earned $7.2B on $154.7B of revenue, a 4.6% margin.² It has the best earnings growth of the three, up 23% in 2025, and in the third quarter of 2026 scripts fell to 393 million from 414 million because of membership losses at UnitedHealthcare and some outside clients.²³ I come back to what reform does to this unit in the valuation.

Pharmacy benefit reform

The profit in a pharmacy benefit manager comes from three places: the fees clients pay, the share of manufacturer rebates it keeps, and the spread between what it pays pharmacies and what it charges plans. Reform attacks the second and third. In September 2024 the Federal Trade Commission sued the three largest managers, Caremark, Express Scripts and Optum Rx, over insulin prices.¹⁴ Express Scripts settled in February 2026 under a consent order that stops it favoring high-list-price versions of a drug, bases members’ costs on net price, and moves its retail pharmacy payments to cost plus a dispensing fee.¹⁷ Caremark settled in July 2026 with point-of-sale rebates, a separation of manufacturer fees from list price, and insulin caps; the FTC estimates up to $8.5B of consumer savings over ten years from the order.¹³ The FTC page for the Caremark order says the case against Optum Rx has been withdrawn from adjudication so the Commission can consider a proposed consent agreement. It gives no date and no terms.¹³

Optum Rx has moved ahead of the order. It launched what it calls a fully transparent, fee-based model that decouples its pricing from list prices and prescription volume, committed to pass 100% of manufacturer rebates through to clients by 1 January 2028, ended 33% of drug reauthorizations for about 270 chronic-condition medicines, and raised payments to about 2,300 independent pharmacies.³ Those moves cost something. Rebate retention and spread are the profit the old model earned, and the new one replaces them with fees the company has not yet disclosed. I could not confirm that a federal pharmacy benefit law has been signed in 2026, and I do not assume one. The FTC settlements are the reform I can document.

The Justice Department

There are two matters, and they differ in kind. The older is a civil False Claims Act case that a whistleblower filed under seal in 2011 and the Department chose to pursue in February 2017. It alleges improper risk-adjustment submissions, meaning that the company recorded diagnoses that raised its Medicare Advantage payments without support. In March 2025 a court-appointed Special Master recommended judgment for the company on all remaining claims, and in April 2025 the Department asked the court to reject that report. The company says it cannot reasonably estimate an outcome.²¹² The newer matter is an investigation of its participation in Medicare. After media reports, the company contacted the Department itself and said on 24 July 2025 that it had begun complying with formal criminal and civil requests and had full confidence in its practices.⁵ The 10-K and the second-quarter 10-Q add that some businesses have been or are under review for compliance with coding under the risk-adjustment model, and that CMS and the Inspector General can audit plans and claw back payments.²¹² The third-quarter release still lists the Department’s legal actions about Medicare participation among its risks.³ I found no filing since July 2025 that gives the scope, the people involved or a possible amount. The third-quarter 10-Q is not yet filed. That is what is public, and it is not enough to size the matter.

IIThe moat

An advantage lasts when a rival with money and talent cannot copy it in a reasonable time. UnitedHealth has several, and they are different in kind. Some are real and hard to build. Some are habits of customers that regulation can change. And one, its own relationship with doctors and patients, is the part most under attack.

The idea is simple. An insurer with 48 million members can buy care at a lower price than a small one. A data company that sees a large share of the claims can tell a hospital where it is out of line. A pharmacy benefit manager with $188B of spend can bargain with the drug makers. Put those under one roof and each helps the others: Optum can lower UnitedHealthcare’s cost per person and offer the same tools to rivals. That is the argument in the 10-K for how the company makes health care more affordable.²

The loop has a cost. Each part also invites a regulator to ask whether the group is paying itself, and the answers are now changing the economics. The test of an advantage is whether it has widened or narrowed. Here is the evidence both ways.

Evidence it is widening

• Operating margin in 2025, 4.2% (4.85% adjusted), was higher than at Elevance (3.6%), Cigna (3.3%), Humana (2.1%) and CVS (1.2%), and Centene lost money.²⁴⁷ C (calculated)
• It earned more than 4% in a year when its cost ratio rose 3.6 points, which no other insurer in my table did C (calculated).²⁷
• The Q3 2026 adjusted margin was 5.6% and Optum’s 5.3%, with Optum Health, the weakest unit, at 2.6% adjusted against 0.2% a year earlier.³
• About half of the 2025 earnings came from the two Optum units that do not carry insurance risk: Optum Insight and Optum Rx earned $9.8B of $19.0B, or $9.9B of $21.7B on the company’s adjusted figures, where Rx is $6.1B because the reported $7.2B includes a gain C (calculated).²⁴
• The Stars result for 2027 held at 69% of members in 4-star-plus plans, which the company says matched its expectations.³

Evidence it is narrowing

• Operating margin is 4.5 points lower than in 2023, and the medical care ratio rose 5.9 points in two years C (calculated).²
• Medicare Advantage enrollment is shrinking, 840,000 fewer since the end of 2025 (935,000 with the Medicaid-linked senior plans), and UnitedHealth’s share of KFF’s count fell 3 points in a year.³⁶
• The company has had to promise 100% pass-through of drug rebates to clients by 1 January 2028 and launched a fee-based pharmacy model. These weaken the margin on volume.³
• It committed $1B to the United Health Foundation, rebated 2026 profit on individual exchange plans, and has said it will end 30% of prior approvals by the end of 2026, all of which cost something.³
• The Optum Health value-based business needed an early loss-contract charge and exits from markets. The unit’s margin was 7.4% in 2024 and 3.2% in the latest quarter.²³

On balance the advantage is wide and intact in scale, and weaker in what it earns from it. The business can still win. The question that sets the value is how much of the old margin comes back once the government, hospitals and the Justice Department have each taken their share.

IIIThe competitive advantage

The honest comparison is the other big insurers, each of which has its own pharmacy and care arm. Elevance owns Carelon, Cigna owns Express Scripts and Evernorth, CVS owns Aetna, Caremark and the pharmacies, and Humana is the Medicare Advantage specialist. Centene is the Medicaid and exchange specialist. 2025 was a hard year for every one of them, which is itself the main fact: the cost trend was an industry problem, and UnitedHealth kept the highest margin of the group.

Three things stand out. First, UnitedHealth had the highest operating margin of the six, 4.2% and 4.85% adjusted. Elevance earned 3.6%, down from 5.0% in 2023, and Humana earned 2.1% on a Medicare Advantage book.²⁷ C (calculated) That is still 3.9 points below UnitedHealth’s own 2024.² Second, a lot of its profit does not come from insurance at all: Optum Insight and Optum Rx earned $9.8B of the $19.0B, and $9.9B of the $21.7B adjusted C (calculated).²⁴ Third, the market pays for that: 30 times 2025 earnings is more than Elevance or Cigna, though 2025 EPS of $13.23 was the lowest since 2018.⁷ At 2026 adjusted guidance it is 19.5 times, and at the 2027 consensus of $23.38 it is 17.1 times.³¹⁸

Where it wins

• Employer plans, where it is the largest carrier and its fee-based business gives a steady income.²
• Data, where Optum Insight earned a 23.4% margin in the third quarter of 2026.³
• Pharmacy benefit scale: $188B of drug spend gives it more leverage with manufacturers than a smaller manager has.²
• Capital: it pays a dividend at an annual rate of $9.28 a share, 2.3% of the price, and still bought back $5.0B of stock in the first nine months of 2026 C (calculated).¹²³

Where it loses

• Share in Medicare Advantage, where Humana rose from 17% to 20% of enrollment while UnitedHealth fell from 29% to 26%.⁶
• Trust. The 10-K lists reviews or audits by CMS, the Inspector General, state insurance departments, the FTC, the Justice Department and others, and the company’s own response includes ending 30% of prior approvals by the end of 2026.²³
• Value-based care. Optum Health was meant to be the way the group lowered the cost of Medicare patients. In 2025 it lost money doing it, and it is now serving fewer of them.²³

IVManagement

The senior team was replaced in about eighteen months. Timothy Noel became CEO of UnitedHealthcare in January 2025, succeeding Brian Thompson.¹¹ Patrick Conway became CEO of Optum in May 2025, replacing Heather Cianfrocco, who moved to run governance, compliance and information security.¹⁹ On 12 May 2025 Andrew Witty stepped down as CEO and Stephen Hemsley, who had run the company from 2006 to 2017 and been its non-executive chair since 2019, took the job back and kept the chair.⁹² Wayne DeVeydt, who was finance chief of Elevance from 2007 to 2016 and then a Bain Capital managing director, became CFO in September.²⁰ The board made F. William McNabb III, a former CEO of Vanguard, its Lead Independent Director in August 2025 and added Scott Gottlieb, a former FDA commissioner, that November.¹¹²¹ The company says it has refreshed nearly half of its top 100 leadership roles.¹⁶

It is a bench of insiders at the top of the two businesses and outsiders where the company most needed new eyes: a CFO from a rival insurer and private equity, and a new Lead Independent Director and a new independent director. Hemsley is 73. The proxy letter says the board does not intend to combine the jobs of CEO and chair indefinitely, and that succession planning covers the top 750 leaders, with plans for over 90% of them.¹¹ A shareholder proposal for an independent chair got 20.2% of the votes cast at the June 2026 meeting.²²

Pay, and what it pays them to do

Hemsley’s pay is unusual, and it tells you what the board wants him to do. His salary is $1.0M. He gets no annual cash bonus. In place of everything else he received one grant of $60M of stock options on 14 May 2025, struck at $308.01 a share, that vest after three years and cannot be exercised for profit before then; in February 2026 the board added a rule that he must keep the shares he gets from exercising them until 14 May 2030.⁹¹¹²³ Reported 2025 pay was $60.9M, almost all of it that option.¹¹ At $400.44 the option is worth about $56M above its strike C (calculated).¹¹¹

So follow the money. The option pays only if the share price is above $308.01 in May 2028, and more as it rises. It pays nothing for the medical care ratio, cash flow, a legal result or a fair outcome for patients. If the stock stays below $308.01 it is worth nothing, so his downside on the grant is the grant itself, not his own cash. That is a clean link to the share price. It is also a link to the multiple the market pays, which a CEO can move by guidance, buybacks and the tone of a call as much as by the business. The other executives are paid differently. Half their annual stock grant is performance shares judged on three-year cumulative adjusted earnings per share and three-year average return on equity, a quarter is options and a quarter restricted stock.¹¹ In December 2025 the committee added one-time Performance Improvement Awards for executives other than the CEO. They vest in 2028 only if adjusted earnings compound above a threshold, and they pay on margin recovery (40%), cyber scores (20%), enterprise cost cuts (20%) and earnings from artificial intelligence products (20%).¹¹ Noel received $9.0M of these awards and DeVeydt $5.0M R (reported).¹¹ Those are targets I can track, and I put them in the signposts.

Witty’s reported pay for 2025 was $24.6M, including a $23.0M stock grant; his performance shares from February 2025 were cancelled when he left.¹¹ At the June 2026 meeting 82.7% of the votes cast backed the pay plan and about 17% voted against it.²²

Their own money

The clearest sign of belief is what the new management did with its own cash. In the week after the CEO change, on 14 to 16 May 2025, Hemsley bought 86,700 shares on the open market at $288.57, about $25M; Rex, then the finance chief, bought 17,175 at $291.12, about $5.0M; and directors Gil, Flynn and Noseworthy bought smaller amounts.¹⁰²⁴ Hemsley’s purchase is worth about $35M at $400.44 C (calculated).¹ In total he reported 1,335,742 shares in April 2026, worth about $535M. But 280,500 of those are in charitable foundations (about $112M) and 692,255 in family trusts (about $277M), so what is plainly his own is a smaller figure.¹¹ All directors and executive officers together own 1,708,073 shares, about $684M and 0.19% of the company C (calculated).¹¹ The largest holders are index managers: Vanguard 10.1% and BlackRock 8.0% on their last filings.¹¹ The only open-market sales by officers I found since May 2025 were small ones by Conway, of 589, 800, 500 and 1,169 shares at $305 to $410.²⁴

What they did with the money, over a full cycle

I take 2016 to 2025 as the cycle. It covers a tax cut, a pandemic, an inflation burst, the Change Healthcare acquisition and cyberattack, and the Medicare repricing of 2024 and 2025. Over those ten years UnitedHealth generated $201.3B of operating cash. It spent $25.7B on property and software, $79.7B on acquisitions, $50.4B on dividends and $51.6B on buybacks C (calculated).⁸² Debt rose from $32.8B to $77.7B and goodwill from $47.6B to $110.5B.⁸²

Three things stand out. First, the buybacks were not made at low prices when it counted. UnitedHealth paid an average of $353 a share over ten years, but about $500, $529 and $462 in 2023, 2024 and 2025, against $400 now and a low of $238 in August 2025 C (calculated).⁸¹ It then cut its 2025 buyback to $5.5B, and did no repurchases in the fourth quarter of 2025.² It restarted in 2026, with at least $5B planned for the year.¹⁶¹⁵ Second, it was an acquirer on a large scale: $80B, much of it for businesses that now sit in Optum Health and Optum Insight, the two units that earned least in 2025 C (calculated).²⁸ The 2022 purchase of Change Healthcare was followed by the 2024 cyberattack. Third, it paid a rising dividend through all of it, $50B in total, while debt more than doubled.⁸

The return on that capital was good and then fell. On my measure, operating profit after tax over equity plus debt less cash, it was 17% in 2016 and 10% in 2025, with a range of 10% to 19% C (calculated).⁸ The company’s own return on equity went from 27.0% in 2023 to 12.8% in 2025.² A business that spent $80B on deals and ended with a lower return on capital than it began with has not proved that the deals paid. The gain in earnings per share, from $7.25 in 2016 to $23.86 in 2023, came from operating growth, buybacks and acquisitions together. Since 2023 EPS has fallen.

VFinancials

UnitedHealth’s accounts are large and mostly plain, but 2024 to 2026 are noisy, and I lean on the company’s adjusted figures for trend only when I say so. In 2024 there was an $8.3B loss on the sale of South American businesses that is not in operating earnings but is in GAAP EPS of $15.51; adjusted EPS was $27.66.²⁴ In 2025 the company took a $2.8B charge before tax for the final cyberattack costs, business exits and restructuring, including loss-contract reserves; GAAP EPS was $13.23 and adjusted $16.35.⁴ For 2026 the adjusted figure adds back about $1.55 a share of intangible amortization from past acquisitions, which is a real cost of buying the businesses even if it is not cash this year.³ I show both and use GAAP as the safer base.

Ten years

Margins

The operating margin was 7.0% in 2016 and 8.7% in 2020. It stayed between 8.1% and 8.8% from 2019 to 2024 and then fell to 4.2% C (calculated).²⁸ I want to be careful with that history. It was earned before the cost shock of 2024 and 2025, before the Inflation Reduction Act changed Part D, before the risk-adjustment model changed, and when Optum Health was earning $6.6B to $7.8B a year. All four have changed. The 7.0% to 8.8% range is a record of one regime. It is not a promise about the next one. Revenue also grew 10.3% a year on about $80B of acquisitions, and I do not assume that growth or that spending again.

The segment chart shows where the earnings were lost. UnitedHealthcare fell from $16.4B in 2023 to $9.4B in 2025, and Optum Health from $6.6B to a loss of $0.3B. Optum Rx grew from $5.1B to $7.2B, and Optum Insight fell from $4.3B to $2.6B C (calculated).² The comparison to watch is which parts have come back. The first nine months of 2026 are on a new basis, because Optum Financial moved to Insight on 1 January 2026, so I keep them in a separate table.

Three facts stand out. UnitedHealthcare’s earnings rose 37% in nine months on 0.3% more revenue, so almost all of the gain is margin from repricing and cutting benefits C (calculated).³ Optum Health’s earnings rose 63% on 6% less revenue, which fits a business shrinking to its profitable core.³ And the consolidated nine-month margin of 7.0% is not the full-year margin, because the first quarter, with a medical care ratio of 83.9%, carries most of the profit and the ratio rises through the year. If the July guide of 88.1% for the year still holds, the fourth quarter is near 93% C (calculated), against 92.4% a year earlier.¹⁵³

Guidance, and how often it moved

Three raises in a row is a good sign, and it is also what a cautious first guide looks like. Management set the bar low in January, at $17.75, after a year in which it had withdrawn its guidance. At $20.50 the 2026 floor is 15% above the first guide C (calculated).⁴³ Analysts expect $23.38 for 2027 and $27.36 for 2028, 14% and 17% higher.¹⁸ My base for 2027 is about $22.5, a little below the consensus E (estimate). The market already pays for a recovery to those numbers.

Cash and capital

Reported operating cash flow of $19.7B in 2025 was not a sign of strength. The release says it was helped by payments received in 2025 that had been expected in 2026.⁴ In the nine months to September 2026 it was $25.1B, 1.5 times net earnings, helped by $4.3B of working-capital change.³ The company’s July guide was about $24B of operating cash for 2026, a figure the first nine months have already passed.¹⁵³ Capital spending is small, about $3.3B a year or 0.8% of revenue, which is why the business converts earnings to cash easily C (calculated).²

The cash that matters for an owner is what can leave the insurance subsidiaries. At the end of 2025 only about $1.1B of $24.4B of cash was available for general corporate use, and the June 2026 10-Q repeats $1.1B of $28.6B.²¹² The rest sits in regulated subsidiaries, or supports $39.3B of medical costs payable.² Those subsidiaries held $43.1B of statutory capital against $23.2B required at the end of 2025, and the 10-K says the requirement is expected to rise because of the higher cost ratio.² In 2024 the subsidiaries sent a net $9.2B up to the parent; in 2025 the parent put in a net $0.5B.² The company also uses reinsurance deposit arrangements to reduce the capital it must hold, and these are not accounted for as reinsurance.² How much of the $19.9B of surplus can ever be released is something I can’t judge, so I count none of it.

Debt

Debt, including the current part, was $77.7B at the end of 2025 and $74.3B at 30 September 2026.²³ Interest expense was $4.0B in 2025, against $1.1B in 2016. Earnings from operations covered it 4.7 times in 2025 and 5.4 times on the company’s adjusted figure C (calculated).²⁴ Credit ratings were A2 (Moody’s), A+ (S&P) and A (Fitch) at the end of 2025, each with a negative outlook.² The company wants debt to capital at about 40% by the end of 2026; it was 41.7% at the end of September.³ I do not see a balance-sheet problem. But debt was cheap to add and costs $4B a year to carry, and for a business whose profit swings by $13B in two years it is the largest fixed claim ahead of the owners.

Earnings quality

Four points on what to trust. First, the 2026 recovery is real in the cash and in the cost ratio, but $1.47B of favorable prior-period reserve development in the second and third quarters helped it, and that can reverse.¹⁵³ Second, adjusted earnings exclude intangible amortization and portfolio items, so I use GAAP guidance of at least $19.50 for the safe base. Third, the loss-contract reserve set up in 2025 for expected 2026 Optum Health losses is being released, $282M in nine months, and the company excludes that from adjusted earnings.³ Fourth, several things are not in earnings at all: the IRS has sent notices of proposed adjustment for 2017 to 2020 on transfer pricing with a foreign subsidiary, and the company says its reserves are adequate.¹² I can’t size that, and I say so below.

Stock pay was $0.97B in 2025, 0.2% of revenue and about 5% of earnings from operations, so UnitedHealth is not a business that flatters earnings by paying staff in shares C (calculated).²

VIValuation

Start with what would go wrong

Before any numbers, I ask what would most likely make this a bad investment from here, and work back from that. For UnitedHealth the business is probably sound. What can fail is the margin it earns, and three things can take it.

1. The cost trend keeps running ahead of what the government pays. My base has the operating margin rising from 5.8% to 7.5%. If it stalls at 6.2% the value is $298; at 7.0% it is $347 E (estimate). Half a point of margin from 2029 on is worth about $29 a share, and a full point about $59 E (estimate). This is the likeliest way to lose money, because the 2024 and 2025 results showed the error is large: the medical care ratio rose 5.9 points in two years while the government raised Medicare Advantage payments by less than costs rose.²
2. The Justice Department matter changes how Medicare Advantage pay is earned, not only what is paid for the past. The fine is the smaller risk. A $10B payment costs about $9 a share E (estimate). A rule that cuts the revenue the company records for the same patients costs more, in the way the margin above does. I can’t judge which it will be, and the filings do not say.⁵¹²
3. Pharmacy benefit reform takes the profit out of Optum Rx. Optum Rx earned $7.2B of the group’s $19.0B in 2025, 38% of it, or $6.1B on the company’s adjusted figure C (calculated).²⁴ If a third of that went, about $2B after the cost of doing business differently, it would cost about $27 a share in my model E (estimate). The company has already promised 100% pass-through of rebates by 1 January 2028.³

The price adds a fourth risk: the multiple. The stock fell 62% from $625 to $238 and has since recovered 68%.¹ A buyer at $400.44 is paying 19.5 times the 2026 floor for adjusted earnings and expects those earnings to grow 14% and 17% in the next two years.³¹⁸ If the recovery arrives but the market pays 12 times earnings in 2036, the return from here is 7.5% a year E (estimate).

What an owner can really take out

Reported earnings are not the number I value. For 2026 the company guides to adjusted earnings of at least $20.50 a share, or about $18.6B.³ That figure is $0.9B above GAAP net earnings to shareholders of at least $17.65B, mostly intangible amortization that is added back.³ From there I take out what an owner needs to stand still: capital spending of about $3.3B and an allowance for small acquisitions of $2.0B, and I add back non-cash depreciation and amortization of $4.2B E (estimate).³ Stock pay of about $1.1B is already deducted. The result, about $16.5B, is 11% below adjusted earnings E (estimate). If the regulators force UnitedHealth to keep a further $1.5B in the insurance subsidiaries to match the higher cost ratio, it is $15.0B. With no allowance for deals it is $18.5B. At a market value of $361B that is a yield of 4.2% to 5.1%, about 4.6% in the middle, against a ten-year Treasury at 4.91% E (estimate).²⁵ The yield is no better than a bond. The case for owning it is that the cash grows, which the next sections test.

What the price implies

The consensus already sees a strong recovery: 2027 earnings of $23.38 is 14% above the 2026 floor and 2028 earnings of $27.36 is 33% above it C (calculated).¹⁸ My base for 2027 is about $22.5 E (estimate), a little under. For 2028 it is lower still. So the market and I agree on direction and differ on how far and how fast.

Over ten years the price rose 2.91 times, from $137.83 in November 2016, while diluted earnings per share went from $7.25 in 2016 to $13.23 in 2025.¹² The multiple on the calendar-year earnings in progress ran 17.6 to 36.4 C (calculated). Today’s 20.9 on the GAAP floor is not out of line with the 17.6 to 21.4 of 2017 to 2020. What is different is that those years had a margin of 8% to 9%, and this one has 5.8%.

Reverse DCF: what the price needs

Held at my base margins, the price needs revenue to grow about 4.7% a year for ten years, to $712B in 2036 C (calculated). My base is 4.0%. Revenue grew 10.3% a year in the last nine years, but with $80B of acquisitions and no repeat of a Medicare Advantage boom I would not use that rate. With my bull margin of 8.7% it needs only about 3.4% C (calculated). Or keep growth at my base and ask for margin: the price needs the operating margin to reach about 8.0% from 2031 onward, against my 7.5% and 8.5% in 2019 to 2024 C (calculated).

That is the clearest way I can put it. At $400.44 the market is paying for the margin to return to about 8%, most of the way back to the old level. I think that is possible. My base does not get there. Both views depend on the same unknown: how much of the 2019 to 2024 margin was the product of rates and risk scoring that have changed.

There are two other ways to read the same price. One is that it equals my base cash flows at a discount rate of about 8.2% C (calculated). That is the return the price offers if my base is right: 3.3 points above the Treasury and a little under the 8.5% I require. The other is to ask what a buyer in 2036 pays. If my base earnings come through and the market pays 17 times them in 2036, the return from $400.44 is 10.2% a year E (estimate). The multiple that makes the base earn my 8.5% is about 14 times.

The discount rate

I discount at 8.5%. The capital asset pricing model with the 4.91% ten-year Treasury and Damodaran’s 4.41% implied equity risk premium gives 8.0% on UnitedHealth’s beta against the market over 118 months, 7.6% over 60 and 8.9% over 36 C (calculated).²⁵²⁶²⁷ The longest and shortest windows are 0.69 and 0.89, so beta moved a lot as the stock became a story about one company’s problems, not the market’s. I take 8.5%, above the middle of that range, for two reasons. The cash flows depend on government payments, which are set by rules that can change in a year. And a legal outcome I can’t size is not diversifiable risk in any sense that a beta captures.

At 8.0% my base is $422, above the price. At 9.0% it is $337, below it E (estimate). So the price sits inside the range my discount-rate uncertainty alone produces. It tells me less than the margin does.

Three cases

My base case is a recovery that stops short of the old margin. Revenue grows 1.5% in 2027, because membership is still falling, then 4% to 5% as repricing finishes and health spending grows, easing to 3.5% by 2036, an average of 4.0% E (estimate). The operating margin goes from 5.8% in 2026 to 6.4% in 2027, which is operating income of $29.1B, and then to 7.4% in 2030 and 7.5% from 2031 E (estimate). Operating income is $50.0B in 2036, 6.7% a year from today. The cash flows also carry a legal payment of $3B in 2028, statutory capital kept for growth, and small deals, which a pure earnings view would skip.

The bull case has revenue growth near 6% for four years, a margin that returns to 8.7%, the 2022 to 2023 level, and a legal cost of $0.5B. Operating income is $64B in 2036, and the value is $547 E (estimate). The bear case has Medicare and commercial costs keep outrunning payments: the margin stays near 5.5%, revenue shrinks 1% in 2027 and grows 2.5% after 2028, and a $12B legal payment lands in 2028. It is worth $168 E (estimate). Even the bear is not a failing business. It is a company that earns a 5.5% margin on $550B of revenue and pays a dividend. It is only a poor purchase at $400.44.

What I can’t judge

Some inputs I can’t know, and I would rather say so than pretend. I can’t judge the Justice Department matter. The only primary text I found says the company is complying with formal criminal and civil requests about its participation in Medicare, from a filing in July 2025, and that in the older 2011 case a Special Master recommended summary judgment for the company in March 2025 while the Department asked the court to reject that recommendation.⁵²¹² The 10-K and the latest 10-Q, which is for the second quarter, do not describe the newer matter and say they cannot estimate an outcome. The third-quarter 10-Q is not yet filed. I can’t judge the IRS notices for 2017 to 2020 either, which the company disputes.¹² I do not know how much of the Medicare Advantage margin of 2019 to 2024 came from coding that the risk model now pays less for. I do not know how much of the $19.9B of surplus statutory capital can be released, or whether the 2025 reserve changes were enough. And I don’t know how much Optum Rx earns after rebate pass-through, because the company has not said. My answer to each is a range, not a point.

Multiply it out

My base needs several things at once. The medical care ratio back near 85% by 2029: I would give that 55%. Optum Rx keeping its profit on fees after rebate pass-through, without losing clients: 70%. The Justice Department outcome costing under $10B and leaving risk-adjusted pay intact: 65%. Revenue growth of 4% a year or better without big deals: 70%. And no new cost shock like 2024 and 2025: 60%. Multiplied, that is about 11% E (estimate). They overlap, since a good year for the cost ratio is also a good year for revenue, so the true figure is higher. Still, a case that needs five things is less likely than any one of them, and that is the reason I keep the margin below the old level.

Quality and price are two questions

How good is the business? It is a large, well-placed one. It earned a higher operating margin than any large rival in a bad year, it has a data and pharmacy business that makes 4% to 23% on revenue, a dividend, and a Medicare Advantage book that is shrinking to what it can price. How good is the price? Fair to a little high. At $400.44 the price is 7% above my base value and 27% below my bull. A good business at a fair price is a very different thing from a good business at a low price, and on this evidence it looks like the former.

It helps to compare it with the next best use of the same dollar. A ten-year Treasury pays 4.91% with no risk to the dollars promised.²⁵ UnitedHealth returns about 8.2% a year on my base, three points more, with every risk of a share. On this site, Booking Holdings returns 13.5% a year on its base case at its price, and 9.3% on its bear, while it sits 34% below its base value; its own page lists reasons the market might be right.²⁸ Insulet is 26% below its base and returns 10.7% on it. UnitedHealth is 7% above its own. These are different businesses, with different discount rates, so I use them only as a gauge of how much I am asking for in return for the extra uncertainty.

The margin of safety

My margin-of-safety price is $244, 35% below my base value. I use 35%, not the 30% I use for steadier businesses, because the bear is 55% below the base and the inputs I can’t judge are large. At $244 my bear case returns 7.1% a year, near but under the 8.5% I require, and the base returns 10.7% C (calculated). The price is 64% above it. The stock last closed below $244 in August 2025, at its low.¹

Where this could be in 5 and 10 years

My model values the cash UnitedHealth leaves for its owners, so I can ask what it says the shares are worth later. The table gives value per share at the end of September 2031 and 2036: the present value, at that date, of the owner cash that comes after it, on today’s share count and in that year’s dollars. Owner cash is free cash flow to the firm after tax, capital spending, deals and statutory capital, less 8.5% a year on today’s $73.2B of net debt, less the 3.5% minority claim.¹²³ The 8.5% charge is more than the company pays on its debt, so owner cash here is a little conservative.

For the base to be right at five years, UnitedHealth needs revenue of about $547B in 2031, growing 4.1% a year from 2026, an operating margin of 7.5% and operating income of $41.0B, against $26.2B this year. For ten years it needs $0.67T of revenue and a 7.5% margin, which leaves $33.8B a year of free cash flow to the firm.

The bear and bull differ on three numbers: revenue growth (2.0% against 5.1% a year), the operating margin (5.5% against 8.7%) and the legal cost. The discount rate matters too: at 7.5% the base is worth $699 in 2036, at 9.5% $468 E (estimate).

The single most likely way the base is wrong is the margin. If it stays at 6.2%, the value is $380 in 2031 and $450 in 2036, and the yearly return from $400.44 is 4.9% a year E (estimate). Acquisitions are the second. The 10-year average is 2.7% of revenue; at 2.0% a year my base falls to $323, and at 2.7% to $300 E (estimate). I assume the growth in my base comes from price, mix and membership, not purchases, and a reader who thinks UnitedHealth must keep buying to grow can use the lower figure.

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

VIIThe case for the price

What most people miss

First, the balance sheet has a cushion the headlines skip: $43.1B of statutory capital in the insurance subsidiaries against $23.2B required at the end of 2025, and I count none of it in my value.² Second, reserves are not being run down: days claims payable were 47.0 at September 2026 against 46.2 a year earlier, and a higher count means more of the year’s claims are still unpaid and reserved for.³ Third, the operating cost ratio fell to 12.8% from 13.5% in the quarter while the company kept spending on technology and people.³

The bull case

The bull case, as I understand it: 2024 and 2025 came from a cost shock and from growing too fast in Medicare Advantage and value-based care, not from a broken model. Management has repriced, cut benefits, left markets and sold weak units, and the numbers show it: a medical care ratio of 83.9% in the first quarter and 88.4% in the third, UnitedHealthcare’s nine-month margin at 4.8% from 3.5%, and an outlook raised three times.³ Stephen Hemsley said the third quarter reflected discipline and execution, and the company points to a Stars result of at least 69% of members in 4-star-plus plans for 2027.³ The people running it put in money: Hemsley bought 86,700 shares at $288.57 on 16 May 2025, about $25M, and Rex $5.0M, at a price 28% below today’s.¹⁰ Of thirty investors whose 13F filings I read, Tiger Global held 297,452 shares at 30 June 2026, down from 438,800 a year earlier, and Baron held 10,059. Berkshire Hathaway’s filings show 5,039,564 shares at the end of each quarter from June to December 2025 and none at 31 March and 30 June 2026. Appaloosa went from 2,450,000 shares to none between June 2025 and June 2026, and Viking’s 1,197,273 at the end of 2025 were gone by 31 March.²⁹ A holding is not a thesis, and most September 2026 filings are not yet due.

Against reality

Much of the bull case is true: the cost ratio is lower, the margin is higher and guidance has risen three times. Two things temper it. The second and third quarters include $860M and $610M of favorable reserve development, and without the third-quarter release the ratio would have been about 89.1%, level with a year ago C (calculated).¹⁵³ And the January guide was a low bar, so three raises show recovery from a low base, not certainty. So I tested what happens if UnitedHealth does what the consensus expects and keeps going. Revenue grows 1.5% in 2027 and 4.5% a year from 2028 to 2032, then fades to 3.5%. The margin is 6.6% in 2027 and 7.3% in 2028, the levels the consensus EPS of $23.38 and $27.36 implies on my reading, then 8.0% by 2031, with a legal cost of $3B E (estimate).¹⁸ That is worth about $404 a share and returns 8.5% a year at the price E (estimate). So if UnitedHealth hits the analysts’ margin and keeps climbing to about 8%, the price is reached. My base stops at 7.5% and is worth $376. From today’s 5.8% it is 2.2 points, about $10B of extra operating earnings, and Medicare Advantage rates, Medicaid rates and the Justice Department matter are each still open C (calculated).¹²

What a 10x would take. Ten times $400.44 is $4,004 a share, about $3.6T of market value on today’s share count C (calculated). At 17 times earnings that needs about $212B of net income, 12 times the 2026 GAAP floor of $17.65B. At a 6% net margin, above the 2.7% of 2025, it needs about $3.5T of revenue, 8 times this year’s C (calculated).³ That is about two-thirds of the $5.3T Americans spent on all health care in 2024.³⁰ In ten years it takes revenue growth of about 23% a year with no rise in the multiple, against 10.3% in the last nine, which included $80B of acquisitions C (calculated). The shares did rise 2.9 times in the last ten years, and 4.5 times to their 2024 peak, from a company worth about $130B.¹ From $361B, tenfold is close to impossible, and I would not build a case on it.

Where I come down

The cost ratio moved me: it fell three quarters running against the year before while membership shrank on purpose. The price did not, because it already pays for a margin near 8%, and neither did the reserve releases. What would raise my values most is a 2027 medical care ratio at or below 87% with normal reserve development, the Justice Department matter closing without a change to risk-adjusted pay, and Optum Rx holding $6B of earnings through 2028.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• Medicare Advantage rates and risk scoring. The 10-K says the base rate notices for several years were below cost trend, and the 10-Q says the final 2027 notice moved toward trend and remained below it. CMS and the Inspector General audit risk adjustment on a sample of plans, and the company says the audits can lead to refunds.²¹²
• Medicaid. There is a timing gap between how sick the people served are and the rates states pay, driven by behavioral, pharmacy and home health use. The company has left one state and enrollment fell 85,000 in the third quarter.¹²³
• Commercial cost trend. Hospital prices, bundled services, billing and the No Surprises Act arbitration process each push cost up. The company has pledged to rebate 2026 profit on individual exchange plans.¹²³
• Reserves. Medical costs payable were $39.4B at 30 September 2026, with days claims payable at 47.0, level with June.³ A shortfall in the estimate would show up as an unfavorable development line. It was favorable by $140M in 2025, $700M in 2024 and $840M in 2023.²
• Cyber and data. The Change Healthcare attack cost $799M in the final direct costs in 2025 and $9.0B of loans to providers in 2024. A committee of the board has engaged an outside cybersecurity firm.⁴²
• Tax. The IRS notices for 2017 to 2020 seek to raise taxable income significantly on transfer pricing with a foreign subsidiary, and the agency could pursue later years too.¹²
• Leadership. Hemsley is 73 and holds both chair and CEO roles, which the board says it will not keep indefinitely.¹¹ The next generation of leaders, Conway, Noel and DeVeydt, has been in role for under two years.²
• Ratings and funding. Each of Moody’s, S&P and Fitch had a negative outlook at the end of 2025.²

What would raise my base value: a 2027 outlook that lifts the margin to 6.4% or more with the ratio at 87% or below, a closed Justice Department matter, and Optum Rx holding its earnings through the 2028 change. What would lower it: a second year of cost ratios above 89%, a payment that changes how risk-adjusted pay is earned, or a large acquisition at a high price.³¹²⁵