Eli Lilly

In our Novo Nordisk report, published a week ago, we gave a verdict titled "Accumulate, Eyes on Lilly." We left half the story deliberately untold. Here is the other half, and it is the more remarkable one.

Eli Lilly is a 150-year-old Indiana drugmaker that, in July 2026, became the first healthcare company in the history of the world to be worth a trillion dollars. It got there on the back of a single molecule — tirzepatide, sold as Mounjaro for diabetes and Zepbound for obesity — which does something no pill or injection has previously managed at scale: it makes people substantially, durably thinner. Roughly 65% of Lilly's revenue now comes from two brand names of that one molecule. Revenue has tripled in a decade, from $21 billion to $65 billion, and nearly half of that increase arrived in the last two years alone.

And here is the sentence that this entire report exists to explain. In the first quarter of 2026, Lilly's revenue grew 56% — because volumes grew 65%, while the prices it actually realised fell 13%. That is not a forecast or a worry; it is in the results. Lilly is winning a war in which the prize is being repriced downward as it wins. Understanding whether that is a temporary toll or the permanent new shape of the business is the whole investment question, and it is genuinely unresolved.

"You are better off in that business by buying a group… I can't pick winners in the pharma industry." — the substance of Buffett's long-held objection to pharmaceuticals (paraphrase, not verbatim)

Buffett has spent sixty years explaining why he mostly avoids drug companies, and the objection is not that they are bad businesses. It is that the durability of the earnings is unknowable — it rests on patents with fixed expiry dates and on scientific outcomes that no amount of business analysis can forecast. Lilly is the hardest possible test of that objection, because it partially defeats it: the moat here is unusually visible. This is not "we hope the pipeline works." It is "we have won four head-to-head trials against the only serious competitor on Earth, our patent runs to 2036, and we are pouring fifty billion dollars of concrete that no start-up can match." That is about as knowable as pharmaceutical economics ever get. Whether it is knowable enough to justify forty-two times earnings is the question we will spend eleven parts on.

Lilly's history is the best possible education in what a pharmaceutical business actually is: a company that periodically owns the most valuable molecule in the world, and then loses it on a date known years in advance.

The 2011–2014 episode is the single most useful thing an investor can know about this company. Lilly was, briefly, a cautionary tale — an old pharma house whose best molecules had expired and whose pipeline looked thin. It rebuilt not by acquisition but by research, and the reason the 2036 patent cliff on tirzepatide should worry you less than the arithmetic suggests is that this management has already walked off that cliff once and climbed back. The reason it should worry you anyway is that they got lucky with a molecule nobody could have forecast. Both things are true.

The machine, in five steps:

Here is the distinction that matters, and it is subtle. Lilly's business is more understandable than most people assume — obesity and diabetes are enormous, growing, visible markets; the product plainly works; the competitive position is measurable in published trial data rather than guesswork. On the pure business test, Lilly passes comfortably. But the price demands something the business cannot give you. At forty-two times trailing earnings, you are not buying 2026 — you are underwriting a chain of estimates that runs to 2030 and beyond, in an industry where a single Phase III readout, one formulary decision by a pharmacy middleman, or one act of Congress can move a third of the value. That gap between a knowable business and an unknowable decade is precisely where Buffett has always drawn his line on pharmaceuticals, and it is honest to say that Lilly narrows that gap without closing it.

Lilly's revenue is extraordinarily concentrated — and the concentration is the bull case and the bear case simultaneously.

Why the concentration cuts both ways. Owning two-thirds of your revenue in one molecule is, in the ordinary course, a red flag — it is the profile of a company one bad headline away from disaster. But the ordinary rules bend here, because this is not a niche drug with a fragile market. Obesity affects something like a billion people; the constraint on Lilly's sales has never been demand, it has been how fast they can physically make the stuff. That is why the capital expenditure line (Part VIII) matters so much: for three years Lilly's binding constraint was manufacturing capacity, not customers. The concentration risk is therefore not "will people want it" but the narrower and more answerable pair of questions this report keeps returning to — what will Lilly be paid per dose, and what happens on 5 January 2036 when the compound patent expires. Everything else is detail.

Lilly's moat is built of four materials, and it is more inspectable than almost any pharmaceutical moat in history. It also has one proven, humiliating breach.

1 · Randomised evidence. Moats are usually argued; this one is measured. In SURMOUNT-5, the direct comparison against Novo's semaglutide, tirzepatide delivered −20.2% body weight versus −13.7% at 72 weeks — and, remarkably, with fewer patients quitting for gastrointestinal side effects (2.7% vs 5.6%). Winning on efficacy and tolerability simultaneously almost never happens. Doctors read these numbers; so do formulary committees.

2 · Manufacturing scale — the real, non-patent moat. Peptide drugs are savagely difficult to produce: industry figures put the waste at roughly 13,000 kg per kilogram of finished product, and a single peptide plant can cost north of $500 million. Lilly has committed more than $50 billion to new US manufacturing. That is a barrier no start-up crosses and few incumbents can match, and — crucially — it is a moat that survives patent expiry in a way a patent does not.

3 · The patent thicket. The tirzepatide compound patent runs to 5 January 2036 — ten years of runway. Formulation and method-of-use patents arguably extend protection further, but that thicket is politically exposed and we would not underwrite it. Treat 2036 as the honest planning date.

4 · The pill. Orforglipron, Lilly's oral GLP-1, is a small molecule made by ordinary chemical synthesis rather than peptide manufacturing — which means it can be produced at a scale and cost injections never will. It is the weapon that decides the endgame, and note the uncomfortable corollary: the second thing orforglipron disrupts is Lilly's own injectable franchise.

And now the breach. On 1 July 2025, CVS Caremark — a pharmacy benefit manager, a middleman — simply removed Zepbound from its main formularies, standardising on Novo's Wegovy and affecting roughly 200,000 patients. The clinically superior drug, backed by a won head-to-head trial, was deleted by a company that makes no medicines at all. It was reinstated only effective 1 October 2026 — fifteen months in which winning the science was worth nothing commercially. That episode is the single best evidence that Lilly's moat is narrower than its trial data implies: it protects the company from other drugmakers, but not from the people who decide which drugs get paid for. I score the moat an 8 — exceptionally visible and genuinely deep against competitors, marked down because it demonstrably does not extend to the parties that set the price.

Lilly has beaten Novo Nordisk in every direct comparison that matters. The most devastating of them is worth stating precisely, because it is the kind of evidence that almost never exists in investing. On 23 February 2026, Novo reported the results of REDEFINE-4 — a Phase III trial that Novo itself designed and ran, pitting its next-generation combination CagriSema against Lilly's already-marketed Zepbound. Novo chose the battlefield. Novo lost: 23.0% weight loss versus Zepbound's 25.5%, missing its primary endpoint of non-inferiority. Novo's shares fell about 15% after hours. Lilly now holds roughly 60% of the US market and has overtaken Novo outside America too; the two companies' market values are ~$1.05 trillion and ~$225 billion respectively.

And yet in the very quarter Lilly was winning all this, the prices it realised fell 13%. Here is why.

Now weigh it honestly, because this is the crux of the entire investment. The bull reading: Lilly traded price for volume in a market where volume is nearly unlimited, and the arithmetic is already working — a 13% price cut against a 65% volume increase produced 56% revenue growth. Trading margin for a vastly larger installed base is what you want a dominant company to do when the market is young. Medicare coverage beginning three weeks before we published is a genuine step-change in the addressable population.

The bear reading, and it is the more interesting one: Lilly has permanently altered what kind of business it is. A company that guarantees most-favoured-nation pricing on every product it will ever launch has surrendered the one thing Buffett prizes above all others — the ability to raise prices. As our research put it with uncomfortable precision: a business whose selling price is negotiated with the state is not a business with pricing power; it is a regulated utility with a research department. Buffett has happily owned regulated utilities and railroads for decades — but he bought them at utility multiples, around ten to fifteen times earnings, precisely because their returns are capped by regulators. Lilly trades at roughly thirty-three times next year's earnings.

Our read: the bulls are right about the next three years and the bears are right about the terminal value. Volume growth will very likely overwhelm price cuts through the late 2020s — the Medicare Bridge alone is a large, dated, verifiable catalyst. But the multiple embeds an assumption of pricing power that Lilly has, in writing, given away. You are being asked to pay a growth multiple for a business whose long-run price is set by someone else. That does not make it a bad company — it is a magnificent company. It makes $1,179 a demanding price for it.

The competitive picture divides cleanly into two categories, and investors consistently worry about the wrong one. Among drugmakers, Lilly's position is close to commanding. Novo has been beaten in every direct comparison, most damningly in a trial it designed itself. Amgen's MariTide underwhelmed on efficacy and is left competing on dosing convenience. The genuinely interesting challenger is Roche/Zealand's petrelintide, entering Phase III in the second half of 2026 and pitched explicitly on tolerability — attacking the one axis where these drugs are weakest, since a meaningful share of patients stop taking them because of gastrointestinal side effects. That is a smart line of attack and worth watching, but it is years from market and Lilly has its own answers coming (retatrutide reached up to −30% body weight at 104 weeks in a dose-escalated subgroup).

The competitor that has actually hurt Lilly, meanwhile, makes no drugs at all. The CVS Caremark episode is the template: a pharmacy benefit manager, negotiating rebates it does not disclose, removed the clinically superior product from its formularies for fifteen months. No trial result prevented it and no patent was infringed. When we assess the durability of Lilly's earnings, the relevant question is not "can Amgen catch up" — on current evidence, no — but "what share of the value gets captured by the intermediaries and the government standing between Lilly and the patient." That is the competition that matters, and Lilly's record against it is decidedly mixed.

If you read one paragraph of this report for its investment lesson, read this one. In 2023 and 2024, Lilly poured roughly $7.4 billion and $8.4 billion of capital expenditure into new plant — sums so large that in 2023 the company generated negative free cash flow despite being wildly profitable on paper. A less disciplined management would have looked at a soaring share price, borrowed cheaply, and bought back stock to flatter earnings per share. Lilly did the opposite: it left roughly $12.4 billion of its own buyback authorisation unused and spent the money on factories instead. Management looked at a fifty-times-earnings stock and decided concrete was the better buy. They were right, and the 2025 numbers prove it — free cash flow snapped back to $8.97 billion as the plants came online and the volume arrived.

On ownership, the anchor is the Lilly Endowment, a philanthropic foundation holding roughly 9.8% of the company — a stable, long-horizon shareholder of the kind Buffett has always favoured, alongside Vanguard (~9.1%) and BlackRock (~7.4%). The dividend is a token by design: eleven consecutive years of increases, raised 15% in each of the last two Decembers to $1.73 a quarter, but a yield of just 0.55% on a 22% payout. This is not an income stock and does not pretend to be. I score management a 9 — an outstanding operating record, genuinely rare capital-allocation restraint, and an explicit, correct priority order. The only mark against is that the MFN bargain, whatever its merits, permanently capped the pricing of every product they will ever launch.

Two things in this table deserve more than a glance. The first is the gap between earnings and cash: EPS of $28.26 against free cash flow per share of just $15.18 — cash is barely half of reported profit. This is the same pattern we flagged at Microsoft, Alphabet, Amazon, Meta and Novo, and the cause is identical: an enormous capital-expenditure programme (~9.6% of revenue) that accounting spreads over decades but cash pays for today. It is not a quality-of-earnings problem — Lilly's earnings are real. But it means that anyone valuing this company on price-to-free-cash-flow will get a frightening number (~82×) that reflects a construction boom, not the underlying economics. The truth lies between the two lenses, and it will migrate toward the earnings lens as the plants finish.

The second is the balance sheet's quiet shift: total debt has risen from $25.2 billion in 2023 to $42.5 billion in 2025, against equity of $26.5 billion, to fund that same build-out. Lilly remains financially strong by any sensible measure (Altman-Z of 9.87, net debt around 1.1× EBITDA, an 83.8% gross margin throwing off cash), so this is not a solvency question. It is simply worth naming plainly: Lilly borrowed to build. That was the right call, and the 2025 free-cash-flow recovery to $8.97 billion vindicates it — but a company carrying more debt than equity has less room for a bad surprise than the headline profitability suggests.

Lilly presents one of the most genuinely difficult valuation puzzles on our board, because every backward-looking measure screams expensive and every forward-looking measure whispers reasonable. Take them in order and resist the urge to pick the one that suits your prejudice.

Start with the fact that most flatters Lilly, because it is true and genuinely unusual: the stock rose and got cheaper. Over the past year, as Lilly became the first trillion-dollar healthcare company, its trailing multiple fell from roughly 49× to ~42×, because earnings grew faster than the share price. That is what a real earnings inflection looks like, and it is the strongest argument the bulls have. Extend it and the case gets better still: on consensus estimates, you are paying about 22.6× the 2028 number and ~18.2× the 2030 number. For context, that makes Lilly cheaper on 2030 earnings than Costco is (~33×) — despite Lilly's trailing multiple being almost identical to Costco's.

Now the discipline. Those 2028 and 2030 figures rest on the estimates of nine and three analysts respectively — a vanishingly thin evidentiary base on which to hang a trillion-dollar valuation, forecasting a business whose realised prices just fell 13% and whose principal patent expires in 2036. The DCF, meanwhile, produces $495.67, some 58% below the market price — the second-most-bearish reading on our entire board after Costco's. We have learned to distrust the DCF for companies in a capital-expenditure boom (it did the same to Alphabet, Amazon and Novo), and it is unquestionably too harsh here. But when a model that mechanically extrapolates today's cash flows says a stock is worth 42 cents on the dollar, the honest translation is not "the model is broken" — it is "virtually all of this valuation lives in years that have not happened yet."

Two final signals. The stock sits at an all-time high, having nearly doubled from its $623.78 fifty-two-week low — you are buying after the move, not before it. And Wall Street's lowest price target, $1,135, sits below today's price, with the average at $1,315 offering only ~11.6% upside; even a Street that rates the stock a Buy (33 buy, 9 hold, 3 sell) sees limited room. There is no margin of safety at $1,179. That is not a prediction that the shares fall; magnificent businesses often stay expensive for years. It is a statement that at this price you are being paid nothing to be wrong, in an industry where being wrong is a Phase III readout away. I score valuation a 4, and set the buy zone toward $950 — about 21× the 2027 estimate — a level a drug company reaches on one disappointing trial or one pricing headline, and one this very stock traded far below within the last year.