Eli Lilly LLY

> ## ⚠ EARNINGS BANNER — LILLY REPORTS TOMORROW

>

> Eli Lilly reports second-quarter 2026 results on 2026-08-05 — ONE DAY after this dive was written.

>

> | | |

> |---|---|

> | Report date | 2026-08-05 (calendar entry last updated 2026-08-04) |

> | Consensus EPS | $6.06 |

> | Consensus revenue | $20.685B |

> | Year-ago quarter (Q2 2025) | EPS $6.31, revenue $15.558B |

> | What consensus therefore implies | Revenue +33.0%, EPS −4.0% |

> | Beat history — five straight, all large | 2026-04-30 $8.55 vs $6.97 · 2026-02-04 $7.54 vs $6.91 · 2025-10-30 $7.02 vs $5.69 · 2025-08-07 $6.31 vs $5.60 · 2025-05-01 $3.34 vs $3.26 |

>

> Three things you must hold in mind before reading anything below.

>

> 1. The data in this file stops at 2026-03-31. The most recent quarterly income statement ends 2026-03-31 and the most recent 10-Q was filed 2026-04-30. The second quarter of 2026 is not in this file — it is tomorrow's print. Every trailing multiple, margin and cash-flow figure in this dive is a quarter stale, and we say so wherever it matters.

>

> 2. A five-beat streak cuts both ways. Four consecutive beats of 23%, 9%, 23% and 13% mean the company is out-executing the models — and that the market now expects it to. The asymmetry around a sixth beat is worse than around the first.

>

> 3. Consensus for tomorrow is a DECLINE in earnings on a 33% increase in revenue, and that is not an error. Lilly includes acquired in-process research and development charges in its reported earnings — it is why the first quarter of 2025 printed $3.34 after a $1.72-per-share charge on the Scorpion transaction. The 10-Q filed 2026-04-30 discloses pending acquisitions with "potential amounts payable at closing … up to approximately $12 billion," up from "less than $3 billion" three months earlier in the 10-K, and the 2026-05-20 8-K confirms $9.0 billion of notes issued partly to fund the Centessa acquisition. A headline "miss" tomorrow may be deal accounting rather than deterioration, and a headline "beat" may be a rebate true-up rather than demand. Do not trade the headline.

>

> This timing is an explicit input to the verdict and is dealt with directly in Section 11.

Reference table

1. What the business actually is

The 10-K is unambiguous about structure: "We discover, develop, manufacture, and market products in a single business segment—human pharmaceutical products." It reiterates this in the segment note: "We operate as a single reportable segment … Our determination that we operate as a single segment is consistent with the nature of our operations and the financial information regularly reviewed by the chief executive officer, in his capacity as the chief operating decision maker."

This matters for how you read the vendor file. The vendor's seg_prod block for LLY is not a product breakdown at all — it contains only two lines:

The two lines sum exactly to reported revenue, so the table is arithmetically clean — but it tells you nothing about which products. The filing's own product-revenue table exists but is stripped from the text extract (it is rendered as a table object), so we take the product economics from the MD&A narrative instead, which is more informative than the vendor block by an order of magnitude. We use the filing.

What the filings actually say about the products:

• 10-K (FY2025): "Mounjaro and Zepbound accounted for 56 percent of our total revenues in 2025."
• 10-Q (Q1 2026): "Mounjaro and Zepbound accounted for 65 percent of our total revenue for the three months ended March 31, 2026."
• 10-K on Mounjaro: "Revenue of Mounjaro increased 53 percent in the U.S. … Revenue outside of the U.S. was $9.3 billion in 2025 compared to $2.6 billion in 2024."
• 10-K on Zepbound: "Revenue of Zepbound increased 174 percent in the U.S., driven by increased demand, partially offset by lower realized prices."
• 10-Q on Mounjaro: "Revenue of Mounjaro increased 59 percent in the U.S. … Revenue outside the U.S. … was $4.4 billion compared to $1.2 billion."
• 10-Q on Zepbound: "Revenue of Zepbound increased 79 percent in the U.S., primarily driven by strong demand, partially offset by lower realized prices, including previously announced reductions in cash pay prices."
• 10-K on the aggregate: "Revenue increased in 2025 driven primarily by increased volume, partially offset by lower realized prices. The increased volume and lower realized prices in 2025 were primarily driven by Mounjaro and Zepbound."

Every single growth sentence in both filings carries the same clause: more volume, lower realised price. That is the defining economic fact of this business right now and it recurs in Sections 3, 6a and 11.

Geography (seg_geo, cross-checked and clean — the five lines sum exactly to $65.179B):

The U.S. share of revenue actually FELL, from 67.4% to 66.7%, because Europe grew 67.0% against the U.S. at 43.1%. That is a healthier picture than the headline concentration implies, and the 10-Q confirms Mounjaro has now "launched … in all major markets." Note also that the 10-Q flags China's National Reimbursed Drug List addition as a source of lower realised prices — volume for price, explicitly.

Distribution and customer concentration, from the 10-K: "three wholesale distributors in the U.S.—McKesson Corporation, Cencora, Inc., and Cardinal Health, Inc.—each accounted for a significant percentage of our consolidated revenue." Lilly is also building a direct channel: "we utilize LillyDirect, our direct-to-patient digital health care platform … Sales through LillyDirect represented a growing portion of our business in 2025."

People: "At the end of 2025, we employed approximately 50,000 people, including approximately 27,000 employees outside the U.S. Our employees include approximately 12,000 people engaged in research and development activities." Twenty-four percent of the workforce is in research — the vendor profile's flat "50,000" hides the most important composition fact on the page.

2. The single most important analytical issue: what the last beat was actually made of

This section exists because the verdict turns on it. Lilly has beaten consensus five consecutive times and reports again tomorrow. The question is not whether the beats happened — they did, and they were large — but whether they were made of the same material each time. The 2026-04-30 10-Q says they were not.

The most recent quarter reported $8.55 against a $6.97 estimate, a 22.7% beat. Reconstructing it from the filing and the vendor income statement:

The honest conclusion is mixed, and we state it as mixed. The quarter contained a real, very large, volume-driven revenue result — and it also contained a rebate true-up, a tax benefit and an easy prior-year comparison. A 22.7% beat is not evidence that the next one will be 22.7%. The 10-Q's own forward language is notably guarded: "incretin volume fluctuations due to channel dynamics or demand can have a disproportionate impact on our results of operations in any given period."

Now apply it to tomorrow. Consensus is $6.06 on $20.685B. That is revenue +33.0% and EPS −4.0% against the year-ago quarter's $6.31. Reconciling those two requires a large charge below the revenue line. The most likely source is disclosed in the same 10-Q: "As part of our business development activities in 2026, we have entered into acquisition agreements, subject to closing conditions. Potential amounts payable at closing for these pending acquisitions would be up to approximately $12 billion." Three months earlier, the 10-K had said the equivalent figure "would be less than $3 billion." The pending-deal book quadrupled in one quarter, and the 2026-05-20 8-K confirms $9.0 billion of notes were raised partly to fund the Centessa transaction, with several tranches subject to a special mandatory redemption at 101 if that deal does not close.

Therefore: tomorrow's headline earnings number is likely to be a poor instrument for measuring tomorrow's business. (That inference is ours, labelled as ours; the filings disclose the deal book and the charge mechanism, not the specific second-quarter accounting.) The numbers to read instead, in order: revenue against $20.685B; the incretin share of revenue against 65%; gross margin against 81.9%; and any language change on realised price.

One further check on management's own arithmetic, half-weighted as guidance always is. The knowledge base records that on 2026-04-30 Lilly "Raised full-year 2026 revenue guidance by $2B to $82.0-$85.0 billion" and "Raised full-year 2026 non-GAAP EPS guidance to $35.50-$37.00." Take the first quarter's $8.55 and the consensus $6.06 for the second: the first half implies about $14.61, which means the second half must deliver $20.89-$22.39, or 59-61% of the full year. That is a very steep back-half. It is consistent with an orforglipron launch ramp and front-loaded deal charges — and it is also exactly the shape that produces a guidance cut if anything slips.

3. Growth, margins and the price problem

Annual:

\*Operating margin as computed by the vendor. The 10-K's own operating-expense structure differs and we flag it in Section 8 (item 3): the vendor appears to place acquired in-process R&D charges below the operating line, which flatters operating margin relative to the filing's presentation. The vendor's pretax income of $25.730B and tax of $5.092B give an effective rate of 19.79%, which matches the 10-K's stated 19.8 percent exactly — so the top and bottom of the statement are reliable even where the middle is rearranged.

Quarterly — and this is where the price problem becomes visible:

Read the last two rows together. Revenue grew 55.5% year-over-year and gross margin fell from 85.1% sequentially and from 82.5% year-over-year to 81.9%. The 10-Q gives the reason in one sentence: "Gross margin as a percent of revenue decreased 0.6 percentage points for the three months ended March 31, 2026, primarily driven by lower realized prices." The vendor's quarterly margins reproduce the filing's stated change exactly (82.5% → 81.9% is precisely −0.6 points), which is a good cross-check on the quarterly block.

Contrast the annual direction. The 10-K says "Gross margin as a percent of revenue in 2025 increased 1.7 percentage points compared with 2024, primarily driven by favorable product mix and improved cost of production, partially offset by lower realized prices." The vendor's annual figures imply +2.5 points (81.3% → 83.8%), not +1.7. That is a filing-versus-vendor contradiction and it is logged in Section 8. We use the filing's +1.7 points.

Operating expenses, from the filings:

• 10-K: "Research and development expenses increased 21 percent in 2025." Vendor: $13.338B / $10.991B = +21.4%. ✔ Consistent.
• 10-K: "Marketing, selling, and administrative expenses increased 29 percent in 2025." Vendor: $11.094B / $8.132B = +36.4%. ✘ Contradiction — logged in Section 8; we use the filing's 29%.
• 10-Q: "Research and development expenses increased 28 percent … Marketing, selling, and administrative expenses increased 19 percent." Vendor Q1 R&D: $3.510B / $2.734B = +28.4%. ✔ Consistent.

The composition is exactly what you want to see in a launch year: R&D up 21%, selling expense up 29% "primarily driven by promotional efforts supporting ongoing and planned launches." R&D intensity has actually fallen — from 24.4% of revenue in 2024 to 20.5% in 2025 — not because Lilly cut research but because revenue outran it. In absolute terms R&D rose $2.347B.

Consensus forward path (analyst counts in brackets; EPS from epsAvg only — every EBITDA/EBIT/net-income estimate row in this file is corrupt, see Section 8):

†FY2025 non-GAAP EPS reconstructed from the four earn_cal actuals ($3.34 + $6.31 + $7.02 + $7.54 = $24.21) against a FY2025 consensus epsAvg of $23.590 — a beat, confirming that the estimate series is on the non-GAAP basis while inc_a/inc_q are GAAP. See Section 8, item 2.

Two observations on the consensus set itself, both of which reduce our confidence in it. First, FY2028 rests on six analysts and FY2029 on four; nothing in this dive depends on either. Second, and more oddly: consensus revenue of $85.276B sits ABOVE the top of management's own guided range of $82.0-$85.0B, while consensus EPS of $35.039 sits BELOW the bottom of the guided $35.50-$37.00. The street models more sales and less profit than the company does. That is a coherent position — it prices in the deal charges and the price deflation — but it means the FY2026 EPS row is not a clean anchor, and it is one reason our fair-value work runs off FY2027.

4. Balance sheet and capital returns — where the vendor file is genuinely out of date

The vendor's net-debt arithmetic is essentially fine. Its vintage is not.

Recomputation: $7.163B + $0.105B − $42.503B = −$35.235B. The vendor reports $35.340B, a $105 million difference — exactly the short-term investment balance, which the vendor ignores as usual. At 0.3% of the figure this is immaterial and we accept the vendor number for FY2025. Enterprise value: market cap $1,051.087B + debt $42.503B − cash and short-term investments $7.268B = $1,086.322B, against the vendor's $1,089.175B — a +0.3% residual. Both are usable.

The lease double-count pattern we test for on every file is ABSENT here, and we checked specifically because Lilly is building manufacturing capacity aggressively. capitalLeaseObligationsCurrent and capitalLeaseObligationsNonCurrent are both 0.00, and longTermDebt ($40.868B) is not equal to any lease line. The 10-K corroborates: Note 10 says "We primarily have leases for corporate offices, research and development facilities, vehicles, and equipment" and discloses operating leases only. There is no finance-lease overhang and no double count. Lilly owns its plants outright — the capital shows up as property and equipment and as debt, not as leases. Property, plant and equipment net rose from $17.102B to $24.675B, +44.3%.

Now the part the vendor file cannot see, and it is material.

FILING OVERRIDES VENDOR. The vendor's balance-sheet block stops at 2025-12-31 and carries $42.503B of total debt. The 10-Q shows $43.4B a quarter later and the 8-K adds $9.0B of principal issued on 2026-05-20 — so gross debt is now approximately $52 billion, roughly 23% above the vendor's figure, and net debt correspondingly higher. Every leverage ratio in ratios_ttm (debtToEquityRatioTTM 1.390, netDebtToEBITDATTM 1.138, debtToAssetsRatioTTM 0.372) is computed on the stale base and understates current leverage.

A structural feature of that May issuance is worth naming. The 8-K states that if the Centessa acquisition is not consummated by roughly 2027-03-31, Lilly "will be required to redeem" five of the eight tranches "at a redemption price equal to 101% of the principal amount." That is standard special-mandatory-redemption language, but it tells you the acquisition programme and the balance sheet are now coupled: a failed deal is a forced refinancing.

Other balance-sheet facts (FY2025): total assets $112.476B (from $78.715B, +42.9%); inventory $13.744B (from $7.589B, +81.1%) of which the 10-K discloses $1.5 billion is capitalised pre-launch inventory, "primarily related to orforglipron" — an asset that existed only because the company expected approval, and which the April 2026 approval has now validated; net receivables $20.155B (from $13.275B); goodwill $5.898B and intangibles $6.521B (a modest 11.6% of assets). Book value per share $34.88, so the stock trades at 32.0x book; tangible book $19.78. Current ratio 1.50, quick ratio 1.10.

Working-capital intensity is extreme and is the single most under-discussed number in this file: days sales outstanding 107.0, days inventory outstanding 445.0, days payable outstanding 154.0, cash conversion cycle 398 days. Lilly is carrying more than fourteen months of inventory. That is partly deliberate (launch stockpiling, capacity-constrained products) and partly the reason free cash flow is so far below net income.

Returns on capital are exceptional and should be stated as such: return on equity 101.3%, return on invested capital 32.1%, return on capital employed 41.5%, return on assets 21.7%. Interest coverage 37.5x. (Note that the 101% ROE is flattered by a small equity base — $26.535B of equity against $112.476B of assets, financial leverage 3.74x — and should not be read as five times better than a 20% ROE business.)

Capital returns. The 10-K: "In 2025, we repurchased $4.1 billion of shares under our $15.0 billion share repurchase program that our board authorized in December 2024. As of December 31, 2025, we had $10.9 billion remaining." The 10-Q updates it: "During the three months ended March 31, 2026, we repurchased $2.3 billion of shares … As of March 31, 2026, we had $8.6 billion remaining." The buyback pace more than doubled in the first quarter of 2026 ($2.3B in one quarter against $4.1B in all of 2025). Dividends: "Dividends of $6.00 per share and $5.20 per share were paid in 2025 and 2024 … The quarterly dividend was increased to $1.73 per share effective for the dividend to be paid in the first quarter of 2026, resulting in an indicated annual rate for 2026 of $6.92 per share."

Note the vendor's lastDiv of $6.46 is a trailing twelve-month figure, not the current rate. The indicated forward rate is $6.92, a 0.62% yield at $1,116.11. Payout ratio 22.1%.

Share count does not reconcile, and we treat it as an open question — see Section 8, item 6. Diluted weighted-average shares were 895.9M in Q1 2026 and 898.0M for FY2025, and dividends paid of $5.384B at $6.00 per share independently imply ~897.3M shares. But the 2026-05-07 8-K reports 944,818,881 shares "outstanding and entitled to vote" at the 2026-02-25 record date, and the vendor's $1,051.087B market cap at $1,116.11 implies 941.7M shares. These cannot both be right. All valuation in this dive is done per share (price ÷ EPS), which is immune to the question; but if the income statement is correct, the market capitalisation is nearer $1,002B and Lilly is not yet a trillion-dollar company.

5. Cash flow — the gap between profit and cash

Three readings, in order of importance.

First: free cash flow is less than half of reported net income, and has been for three years. Net income was $20.638B in 2025; free cash flow was $8.972B. incomeQualityTTM is 0.81 on operating cash flow alone, and far worse after capital expenditure. The mechanism is entirely working capital: the cash-flow statement shows −$8.093B of change in working capital, driven by inventory (+$6.155B on the balance sheet) and receivables (+$6.880B). This is what a capacity-constrained product being un-constrained looks like from the inside, and it should reverse — but it has not yet.

Second: shareholder returns exceeded free cash flow, and were funded with debt. Dividends of $5.384B plus buybacks of $4.108B is $9.492B against $8.972B of free cash flow. Net debt issuance of $8.051B covered the difference and the acquisitions. That is a defensible choice for a business earning 32% on invested capital, but it should be named for what it is: Lilly is currently a net borrower to pay its shareholders, and the 2026-05-20 issuance of a further $9.0B extends the pattern.

Third: capital expenditure is going higher and the company says so. The 10-K: "Capital expenditures were $7.8 billion during 2025, compared to $5.1 billion in 2024. We are making investments in global facilities to manufacture existing and future products. These investments … will result in meaningfully higher capital expenditures in the near term." The 10-Q repeats the sentence verbatim three months later. Capex was 12.0% of revenue in 2025 and capexToDepreciationTTM is 3.38x — Lilly is spending more than three times its depreciation, which is the signature of a build-out rather than a maintenance programme.

Two off-balance-sheet commitments the vendor file does not carry, both from the 10-K:

> "Executed agreements related to our medicines in development could, under certain circumstances, require us to pay up to approximately $10 billion if we do not purchase specified amounts of goods or services over the durations of the agreements, which are generally up to 8 years."

> "As of December 31, 2025, we had approximately $902 million of unfunded commitments to invest in venture capital funds." (The 10-Q updates this to ~$850 million at 2026-03-31.)

The $10 billion contract-manufacturing commitment is the important one. It is contingent, not a liability, and it appears in no leverage ratio anywhere. It is also, in a real sense, the mirror image of the growth thesis: Lilly has pre-committed to buy manufacturing capacity for eight years on the assumption that incretin demand keeps rising. If demand plateaus, that commitment converts from an enabler into a cost.

6. Valuation — priced in or room?

At $1,116.11:

The trailing four-quarter diluted EPS sum is $28.15 (quarters ending 2026-03-31, 2025-12-31, 2025-09-30, 2025-06-30), a 39.6x trailing multiple on the GAAP basis. On the non-GAAP basis the same four quarters sum to $29.42 ($8.55 + $7.54 + $7.02 + $6.31), a 37.9x multiple. We quote both and use the non-GAAP series for forward comparison because that is the basis every estimate in the file is built on.

Vendor scoring model. Rating B, overall score 3 out of 5, with return-on-equity and return-on-assets sub-scores of 5 (the maximum) and debt-to-equity 1, price-to-earnings 2, price-to-book 1. The model's verdict is that this is an outstanding operating business at a demanding price on a leveraged balance sheet — which is, in three numbers, the correct summary of this dive.

Street position. Consensus target $1,315.42 (+17.9%), median $1,300, high $1,500, low $1,135. The low target is above the current price. Ratings: 0 strong buy, 33 buy, 9 hold, 3 sell. Two things stand out. Three outright sell ratings on a consensus-Buy mega-cap is unusual and worth respecting — someone with a published model thinks this goes down. And a target range whose floor is above spot means the published sell side has no articulated downside case at all, which historically is a condition that precedes rather than prevents drawdowns.

Peer context. The vendor peer list is the correct one for once — AbbVie, Amgen, AstraZeneca, Gilead, J&J, Merck, Novo Nordisk, Novartis, Pfizer, Sanofi — a genuine large-pharma set including the only direct incretin competitor. But no peer multiples are supplied in the file, so no peer-multiple comparison is drawn. We note only the price data present: Novo Nordisk at $44.28 with a $196.77B market capitalisation against Lilly's $1,116.11 and ~$1.05T — a five-to-one gap between the two companies that share this market.

6a. What today's price assumes (the inversion)

At $1,116.11 — 24.7x FY2027 consensus EPS of $45.178 and 31.9x FY2026E — the price embeds roughly the following falsifiable claims. Each is dated and sourced.

• Revenue reaches $85.276B in 2026 (+30.8%) and $99.316B in 2027 (+16.5%). (Consensus-derived: 18 analysts on each row.) Testable at every print, beginning tomorrow, where the implied quarterly figure is $20.685B. Supporting evidence is strong: Q1 2026 already ran at $19.799B, so the full year needs an average of about $21.8B in the remaining three quarters — a 10% sequential step from Q1.
• EPS reaches $45.178 by 2027, i.e. +86.7% over two years from the $24.21 non-GAAP earned in 2025. (Consensus-derived, 13 analysts.) Note this is below management's own trajectory, which guides to $35.50-$37.00 for 2026 alone against consensus of $35.039.
• Realised price declines at a rate the volume growth comfortably outruns. (Our number, derived from the filings and one half-weighted management claim.) Both filings attach "lower realized prices" to every growth sentence; the chief executive has guided to "single-digit (~5-10%) annual price deflation in the obesity category over time; net price already cut from ~$1000 to $499." At 5-10% annual price erosion, volume must grow roughly 40-45% to deliver 30% revenue growth. This is the most fragile assumption in the price and it is fragile in a specific way: it is not a demand assumption, it is a net-price assumption, and net price is set by payers, PBMs, the U.S. government's Bridge Program and China's reimbursement list — not by Lilly.
• The incretin concentration does not become a de-rating event. (Our reading of the 10-K's 56% and the 10-Q's 65%.) At 65% of revenue in two molecules, the market is currently applying a diversified-pharma multiple to what is arithmetically a single-product company with a very good pipeline. That is a judgement, not a fact, and it can reverse quickly.
• The government does not select an incretin for administered pricing inside the valuation window. (Filing-derived: the 10-K discloses Jardiance selected effective 2026 and Trulicity and Verzenio selected in January 2026 effective 2028, and states "Given our product portfolio, we expect other significant products will be selected in future years." ) Mounjaro was approved in 2022; the nine-year small-molecule clock in the 10-K's description would put it in scope around 2031 — outside the explicit window, but well inside the horizon of anyone underwriting 2028-2030 estimates.
• The manufacturing build-out earns its cost of capital. (Our number, from the cash-flow statement and the 10-K.) $7.841B of capex in 2025, "meaningfully higher" ahead, plus roughly $10 billion of contingent contract-manufacturing purchase commitments over up to eight years.

6b. The return bridge (why the multiple moves)

Expected return over 12-24 months decomposes as EPS growth (+44.7% in 2026, +28.9% in 2027 on consensus) + multiple drift (we assume COMPRESSION, from 24.7x to ~27x on the following year — which is compression on a rolling basis) + shareholder yield (~1.0%: 0.62% indicated dividend plus roughly 0.4% net buyback).

This is the opposite structure to most names in this batch, and it is worth being precise about. Our base case of $1,220 is 27.0x FY2027E, higher than today's 24.7x FY2027E — but the same $1,220 is only 23.4x FY2028E, against 21.4x today. The multiple is not expanding; the earnings are. Essentially the entire base-case return is earnings growth, and we are assuming the market pays a slightly lower forward multiple next year than it does today, because the growth rate decelerates from +30.8% to +16.5% to +12.1% and because concentration risk rises.

Why compression rather than expansion. Three reasons, all evidenced above. The revenue growth rate roughly halves between 2026 and 2027 on consensus. Free cash flow conversion is 43.5% of net income and shareholder returns are debt-funded. And the incretin share of revenue is rising, not falling — a market that pays 25x for diversified growth pays less for concentrated growth, and the concentration is going the wrong way.

Sanity check on the destination. If EPS simply reaches the FY2027 consensus of $45.178 and the multiple holds at today's 24.7x, the price is $1,116 — precisely today. So the entire base-case gain requires the market to pay 27x instead of 24.7x on that year's earnings. That is not a heroic assumption for a company compounding earnings at 29% — but it is the assumption, and we name it. Our base sits 7.3% BELOW the street's $1,315.42, which requires 29.1x.

6c. Variant perception (where we differ, what would surprise)

• We differ from the street on the destination, and in the cautious direction. Our base of $1,220 is 7.3% below the consensus $1,315.42 and 6.2% below the median $1,300. The difference is entirely the multiple we are willing to pay on 2027 (27.0x versus the street's implied 29.1x), and the reason is concentration plus net-price direction. A dive that lands below the street on a consensus-Buy mega-cap is a real, if modest, variant perception. Watchable number: the incretin share of revenue at each print — 56% (FY2025), 65% (Q1 2026), next reading tomorrow.
• We differ from the vendor file, and from most screens, on how much leverage this company carries. Screens show total debt of $42.503B from the 2025 year-end balance sheet. The 10-Q shows $43.4B at 2026-03-31 and the 2026-05-20 8-K adds $9.0B of new notes — roughly $52B, about 23% higher, against total equity of $26.535B. Every published debt-to-equity and net-debt-to-EBITDA figure for LLY is currently understated. Watchable number: total debt as disclosed in the second-quarter 10-Q.
• We differ on the quality of the beat streak. The consensus read is that Lilly beats by 10-25% routinely. The 10-Q shows the most recent beat contained a "favorable one-time adjustment to estimates for rebates and discounts," a 380-basis-point drop in the effective tax rate, and an easy comparison against a $1.72-per-share prior-year charge. Revenue growth of 55.5% is the real number in that quarter; $8.55 is not a clean run-rate. Watchable number: gross margin against 81.9%, and any recurrence of rebate-estimate language.
• We do NOT have a variant perception on the science, and we should say so plainly. The knowledge base is overwhelmingly bullish, and we largely agree with it — but agreeing with a lane that is 88% bullish and 68% supplied by the seller is not edge. Our edge on this name is in the balance sheet, the earnings composition and the entry discipline; it is not in the biology. That should constrain position size.
• Positive surprise that would force us higher: an orforglipron launch trajectory that shows the oral franchise is additive rather than cannibalising, visible as total incretin revenue accelerating while the injectable lines hold; or a first sequential increase in gross margin alongside revenue growth, which would mean price erosion has been outrun by manufacturing scale; or free cash flow conversion moving above 60% of net income as the inventory build stops. Watchable numbers: gross margin against 81.9%; the working-capital line in the cash-flow statement against −$8.093B.
• Negative surprise that would break the thesis: a full-year guidance cut from the $82.0-$85.0B / $35.50-$37.00 range; gross margin below ~80%; the incretin share of revenue above 70% with no offsetting pipeline revenue; or an announcement that an incretin has been selected for government-administered pricing. Watchable events: tomorrow's release and the annual HHS selection cycle.

Synthos fair values

All three anchors are multiples of FY2027 consensus EPS of $45.178 (13 analysts, non-GAAP basis), cross-checked against FY2026E of $35.039 (13) and FY2028E of $52.151 (6, and treated as indicative only).

• Bear ~$875 — 19.4x FY2027E. Cross-check: 25.0x FY2026E; 16.8x FY2028E. The scenario: net price erodes faster than the guided 5-10% as payers, the Bridge Program and international reimbursement lists compound; a competitor's next-generation asset takes share; the pending $12 billion of acquisitions produces large charges without near-term revenue; and the market re-rates a 65%-concentrated business toward a large-pharma multiple. $875 is below the 200-DMA of $1,027.79 and would represent a 29.2% drawdown from the 52-week closing high. −21.6%.
• Base ~$1,220 — 27.0x FY2027E. Cross-check: 34.8x FY2026E; 23.4x FY2028E; 41.5x trailing non-GAAP. Sits 7.3% below the street consensus of $1,315.42 and 6.2% below the median. The scenario: revenue lands in the consensus range, orforglipron launches into the capacity that has been built, price erosion runs at the guided single-digit rate, and the multiple compresses modestly on a rolling forward basis as growth decelerates from +30.8% to +16.5%. +9.3%, plus roughly 1.0% of shareholder yield.
• Bull ~$1,490 — 33.0x FY2027E. Cross-check: 42.5x FY2026E; 28.6x FY2028E; essentially the street high of $1,500. The scenario: the oral franchise proves genuinely additive and opens the ex-U.S. markets at scale; indication expansion (sleep apnea, cardiovascular, inflammation, addiction) converts the incretin into a platform rather than a product; the manufacturing build stops consuming working capital and free cash flow converts above 70%; and the market pays a premium multiple for the only listed asset with this combination of growth and gross margin. +33.5%.

Base is 9.3% above spot; asymmetry roughly 1.55:1 to the upside (21.6% down, 33.5% up). That is a positive but not a commanding skew, and it is the arithmetic reason the position should be built in pieces rather than taken in one — quite apart from the print tomorrow.

7. Knowledge base — 180 claims, and why depth is not the same as evidence

Tracked knowledge-base claims naming Eli Lilly: 199 raw hits, of which 180 are entity-level, drawn from all 51,928 records. Patterns searched: LLY (case-sensitive word boundary), Eli Lilly, Lilly, tirzepatide, Mounjaro, Zepbound, orforglipron, retatrutide. Match distribution: Eli Lilly 103, LLY 68, tirzepatide 8, Lilly 1.

This is nominally the deepest single-name lane in the batch. It is also the most compromised, and the honest accounting matters more here than anywhere else in this dive.

Discards, in order.

Stance mix across all 180: 159 bullish, 18 neutral, 3 bearish. That is 88.3% bullish and 1.7% bearish.

The three bearish claims are, in full: the chief executive on counterfeit and mass-compounded incretins; the chief executive on tariffs being the wrong mechanism; and a physician on the harm of cycling on and off the drug. There is not one bearish claim in this entire file from an independent investor about the equity. Compare the same batch's technology names, where the bear lane runs 25-30%. A file that is 88% bullish, 68% supplied by the seller, and contains zero independent scepticism is a description of narrative saturation. It is a reason to lower conviction, not raise it.

Date range 2025-01-09 to 2026-07-25. Distinct sources 11.

The independent lane — what the 37 fully-weighted claims actually say

The strongest and most specific bull case in the file, from the highest-skill independent voice (skill 2.0):

> 2026-07-11 · jordi_visser · bullish · conviction 90 · skill 2.0

> "Huge belief in Eli Lilly and the compounding side of curing diseases; wouldn't get out even if it three-bagged from here; Lilly already runs its own sovereign AI (~1,000 [top-end accelerators]) with biotech."

> 2026-06-06 · jordi_visser · bullish · conviction 85 · skill 2.0

> "'Mark this down': Eli Lilly has a chance to be the largest company in the world and #1 AI company within 5 years — specialized model, own 1000-[accelerator] data center, 150 years of proprietary data, [protein-structure prediction and drug-design research partners] + [its chip supplier] + Tune Lab."

> 2026-06-06 · jordi_visser · bullish · conviction 80 · skill 2.0

> "Peptides are the 'API key for the human body' and the next massive trend; Eli Lilly is reaching root-cause metabolic treatment with a $200-300B obesity TAM barely accessed — watch its stock after 18 months of consolidation."

> 2026-05-30 · jordi_visser · bullish · conviction 80 · skill 2.0

> "Eli Lilly is an AI/longevity trade — revenue up 55% YoY at trillion-dollar scale, only healthcare name in his 100-name portfolio, AI partnerships ([two drug-discovery research partners and its chip supplier]), driving GLP-1 cost down via Jevons paradox."

Two observations. First, the 55% year-over-year revenue figure in the 2026-05-30 claim is independently verified by the filing — Q1 2026 revenue of $19.799B against Q1 2025 of $12.729B is +55.5%. When a tracked claim contains a checkable number and the number checks, that is worth recording. Second, the 2026-06-06 claim explicitly frames the stock as having spent "18 months of consolidation," which the price series supports: six-month return +6.9% against SPY's +11.1%, and a price 9.7% below the 52-week closing high. That claim was made two months ago and the consolidation has continued.

Note also that the same source frames the position, in four separate claims, as "an AI trade." (2026-07-25, 2026-07-18, 2026-07-04, and the quarantined duplicates.) We record this and we do not adopt it. Nothing in the 10-K or either 10-Q supports valuing Lilly on artificial-intelligence grounds; the filings describe a pharmaceutical manufacturer with 12,000 research employees and $13.338B of research spending. A framing that converts a drug company into a technology multiple is exactly the kind of narrative extension that should raise scepticism, not conviction — however good the underlying business is.

The specialist biotechnology lane (19 claims) is the most useful part of the file because it is the only place where anything critical appears:

> 2025-08-28 · biotech_hangout · neutral · conviction 60

> "Two straight disappointing Lily quarters (down 20-25%) took bloom off the obesity rose short-term, but entrenched leaders Novo/Lily still see enormous category success."

That is the most operationally relevant claim in the entire file for a dive written the day before a print. It is direct evidence that this stock has been punished 20-25% on earnings before. It is a year old, and the four quarters since have all been beats — but it establishes that the distribution of outcomes around a Lilly print has a long left tail.

> 2025-08-26 · biotech_hangout · bullish · conviction 70

> "Obesity shifting from Lilly/Novo duopoly to a biotech game; Amgen, Metsera, Kailera and oral peptides offer differentiated tolerability/dosing upside."

> 2026-01-23 · biotech_hangout · bullish · conviction 85

> "Trump-Lilly-Novo deal ratified a new 'large markets, medium prices' benchmark that reshapes pharma economics for a decade."

> 2026-01-16 · biotech_hangout · bullish · conviction 55

> "Lilly at ~30x is not expensive when Costco and Walmart trade at ~40x — pharma multiples are cheap relative to broad-market staples."

> 2025-12-15 · biotech_hangout · neutral · conviction 55

> "Lilly's pirtobrutinib ASH data was mixed: relapsed/refractory failed to beat, frontline looked good but very early with marginal safety edge over covalent inhibitors."

> 2026-01-16 · biotech_hangout · bullish · conviction 68

> "Generalists are rotating into pharma/biotech because it looks cheap versus an expensive AI-driven S&P; interest is broadening beyond mega-caps."

The 2026-01-23 claim is the one that best matches the filings. "Large markets, medium prices" is precisely what the 10-K and 10-Q describe: volume up, realised price down, Medicare access via the Bridge Program, China's reimbursement list. The independent lane and the filings agree on the direction of the business model, and it is a lower-price, higher-volume model.

> 2025-11-26 · forward_guidance · bullish · conviction 80

> "Healthcare is where the AI bubble gets biggest but revenues are infinite; Eli Lilly breaking out, aging solved within five years."

We log that one and discount it heavily. "Revenues are infinite" and "aging solved within five years" are the vocabulary of a top, not of an analysis.

The management lane — half-weighted, and one claim that is worth more than all the rest

123 claims (68.3%) come from the chief executive or the investor-relations lane. All are half-weighted and none is used as a valuation input. They are consistent, articulate and almost entirely promotional: the incretin as a metabolic master switch, roughly 100 studies in progress, a triple agonist coming in 2027, oral small molecules scaling to a billion patients, a development cycle 40% faster than peers, $50B of U.S. manufacturing, a supercomputer in Indianapolis.

But four of them are genuinely informative precisely because they cut against the company's interest, and those are the ones we cite:

> 2025-09-29 · eli_lilly_ceo (half-weighted) · neutral · conviction 90

> "Pharma has no enduring franchise—every drug goes to zero at patent expiry; Mounjaro reverts to baseline ~2030s, forcing reinvestment or return of cash."

This is the single most important claim in the file and it comes from the chief executive. It is a statement of the terminal-value problem in a business where the terminal value is legally scheduled. It is also, notably, filed in the non-entity block of the knowledge base — a claim about Mounjaro rather than about LLY — which means a naive entity-level read of this lane would miss it entirely. We elevate it.

> 2025-09-29 · eli_lilly_ceo (half-weighted) · neutral · conviction 70

> "Guided the street to expect single-digit (~5-10%) annual price deflation in the obesity category over time; net price already cut from ~$1000 to $499."

This corroborates, from the seller's own mouth, the "lower realized prices" clause that appears in every growth sentence of both filings. It is the quantification the filings decline to give.

> 2025-07-23 · eli_lilly_ceo (half-weighted) · neutral · conviction 65

> "Up to 50% of weight-management users discontinue within 12-14 months due to side effects/cost — a real durability and outcomes headwind."

Half of the installed base leaves within about a year. That is the demand-durability question stated by the party least motivated to state it.

> 2025-10-28 · eli_lilly_ceo (half-weighted) · neutral · conviction 70

> "GLP-1s now represent a little more than 50% of Lilly's overall revenue."

Cross-check this against the filings and it is already obsolete: the 10-K says 56% for FY2025 and the 10-Q says 65% for Q1 2026 — for two products, not the whole class. The claim was accurate when made and is now stale, which is a useful demonstration of why the filing outranks the lane on any measurable fact.

Two further half-weighted claims are used only as dated markers of the guidance: "Raised full-year 2026 revenue guidance by $2B to $82.0-$85.0 billion" and "Raised full-year 2026 non-GAAP EPS guidance to $35.50-$37.00" (both 2026-04-30, tagged skill 0.5). These are inputs to Section 2's back-half arithmetic and to nothing else.

The counterparty lane — ten claims, discounted

Ten claims come from a hardware vendor whose company announced a co-innovation laboratory with Lilly and supplies the accelerators for its biology supercomputer — "[The supplier] announced a co-innovation lab with Lilly; only [its] full stack and digital-biology expertise can stand up a pharma AI factory." A supplier describing the value of what it sold is a sales claim, not independent evidence, and we treat it as such. It is also the mechanism by which the "Lilly is an AI trade" framing entered the file in the first place.

Why Moderate-High rather than High

The lane is deep (180 entity-level claims), recent (through 2026-07-25), and it contains a genuinely high-conviction position from a skill-2.0 independent voice with a verifiable number in it. But 68.3% of it is the seller, 5.6% is a supplier, 4.4% is drug-level rather than equity-level, and there is not one independent bearish claim in 199 hits. Breadth is 11 sources, of which 5 carry full weight. Depth without dissent is not conviction; it is consensus. We rate it positive, not positive-high, and we size accordingly.

8. Data integrity — what we rejected from the vendor file and why

Nine items. Four are direct filing-versus-vendor contradictions, and in every one the filing wins.

1. est EBITDA, EBIT and net-income rows are CORRUPT in seven of eight years — ALL REJECTED. We use epsAvg only. The pre-computed integrity block flags six rows in which netIncomeAvg exceeds ebitdaAvg, which is arithmetically impossible for a taxpaying, interest-paying company:

The FY2030 row is the clearest demonstration: revenue of $131.496B with EBITDA of $42.216B is a 32.1% EBITDA margin, while net income of $58.152B on the same revenue is a 44.2% net margin. A company cannot have a net margin twelve points above its EBITDA margin. Every EBITDA, EBIT and net-income estimate row in this file is rejected. All forward valuation in this dive runs on epsAvg exclusively, and no EBITDA-based forward multiple is quoted anywhere. (The trailing evToEBITDATTM of 32.5x comes from km_ttm, not from est, and is reported once for context only.)

2. FILING-CONFIRMED BASIS MISMATCH — the income statement is GAAP, the estimates and the earnings calendar are NOT, and the vendor never labels either. inc_q reports Q1 2026 diluted EPS of $8.26; earn_cal reports the same quarter's actual as $8.55. The gap runs through the whole series:

The reconciliation that settles it: the four 2025 earn_cal actuals sum to $24.21 against a FY2025 consensus epsAvg of $23.590 — a beat, consistent with every other datapoint. The GAAP FY2025 diluted EPS was $22.95, below the consensus, which would imply a miss that no source records. Therefore the estimate series is on the company's non-GAAP basis. We quote the trailing multiple on both bases (39.6x GAAP, 37.9x non-GAAP) and run all forward work on the non-GAAP basis, because that is what the estimates are.

3. FILING OVERRIDES VENDOR — the vendor's annual gross-margin change is wrong. The 10-K states: "Gross margin as a percent of revenue in 2025 increased 1.7 percentage points compared with 2024." The vendor's figures give 81.3% → 83.8%, an increase of 2.5 points. We use the filing's +1.7 points. (The quarterly series does not carry the defect: the vendor's 82.5% → 81.9% reproduces the 10-Q's stated −0.6 points exactly, so the quarterly gross-margin block is accepted.)

4. FILING OVERRIDES VENDOR — the vendor's selling-and-administrative growth is wrong. The 10-K states: "Marketing, selling, and administrative expenses increased 29 percent in 2025." The vendor's sga of $11.094B against $8.132B implies +36.4%. We use the filing's 29%. Related and probably the same cause: the vendor's operating income of $29.696B (45.6% margin) reconciles to gross profit less R&D less SG&A with a $487M residual, while $3.171B of expense sits between operating income and pretax income — consistent with the vendor placing acquired in-process R&D charges BELOW the operating line where the filing places them within operating expenses. The vendor's 45.6% operating margin therefore flatters the filing's presentation and every operating-margin figure in this dive is labelled as vendor-computed. The pretax and tax lines are reliable: $5.092B on $25.730B is 19.79%, matching the 10-K's stated 19.8 percent to the basis point.

5. FILING OVERRIDES VENDOR — the debt balance is a quarter and a bond deal out of date. The vendor's newest balance sheet is 2025-12-31 with total debt of $42.503B. The 10-Q filed 2026-04-30 states "As of March 31, 2026, total debt was $43.4 billion." The 8-K filed 2026-05-20 discloses $9.0 billion of aggregate principal issued that day across eight tranches, with net proceeds of "approximately $8.94 billion." Gross debt is therefore approximately $52 billion today, roughly 23% above the vendor's figure, and every leverage ratio in ratios_ttm is computed on the stale base. We use the filings.

6. THE SHARE COUNT DOES NOT RECONCILE, AND TWO SEC FILINGS DISAGREE WITH EACH OTHER. Three independent figures:

• 895.9M — diluted weighted-average shares, Q1 2026 inc_q; 898.0M for FY2025.
• ~897.3M — implied by the cash-flow statement: $5.384B of dividends paid at the 10-K's stated $6.00 per share.
• 944,818,881 — the 2026-05-07 8-K: "the 944,818,881 shares that were outstanding and entitled to vote as of February 25, 2026" (internally consistent, since 847,254,010 votes cast is 89.7% of it, matching the filing's "approximately 90%").
• 941.7M — implied by dividing the vendor's $1,051.087B market capitalisation by the $1,116.11 close.

The income statement and the cash-flow statement agree with each other at ~897M and disagree with the proxy-vote count at ~945M by 5.3%. We cannot resolve it from the material available and we do not pretend to. Consequence: every per-share figure in this dive (all P/E ratios, EPS, book value per share) is unaffected, and we use per-share arithmetic throughout for exactly that reason. Every aggregate figure — market capitalisation, enterprise value, price-to-sales — carries a potential 5% overstatement. If the income statement is right, Lilly's market capitalisation is approximately $1,002B and it has not yet crossed a trillion dollars. Flagged, not silently smoothed.

7. netDebt and enterprise value are, unusually, close to correct — and we say so. Vendor netDebt $35.340B against a recomputed $35.235B (a $105M gap, exactly the discarded short-term investments) and a corrected EV of $1,086.322B against the vendor's $1,089.175B (+0.3%). Both are accepted for FY2025. (A refinement: the 10-K notes Lilly "held total investments of $2.9 billion" beyond cash, which if netted would put FY2025 net debt at $32.433B. We mention it but do not use it, since long-term investments are not cash equivalents.) The caveat from item 5 applies: correct as of 2025-12-31, obsolete as of today.

8. The lease double-count pattern was tested for specifically and is ABSENT. capitalLeaseObligationsCurrent and capitalLeaseObligationsNonCurrent are both 0.00 and neither equals longTermDebt. The 10-K's Note 10 discloses operating leases only, "primarily for corporate offices, research and development facilities, vehicles, and equipment." There is no finance-lease liability and no not-yet-commenced lease disclosure of the kind that dominates the technology names in this batch. Lilly's capacity build is owned, not leased, and it appears as $24.675B of property and equipment (up 44.3%) and as bonds. A clean result, and worth stating because the a priori expectation for an aggressive builder was the opposite.

9. Minor internal inconsistencies, flagged and resolved. (a) quote.price is $1,116.11 and tech.last is $1,115.68 — a $0.43 (0.04%) mismatch. We use $1,116.11 throughout, and note that the tech block's derived percentages (−9.70% from high, +78.32% from low) are computed off $1,115.68; at four hundredths of a percent this changes nothing. (b) quote.yearHigh $1,249.45 / yearLow $623.78 are intraday extremes; tech.hi52 $1,235.56 / tech.lo52 $625.65 are closing extremes. We use the closing series. (c) quote.sma50 1,147.32 and sma200 1,024.79 differ from tech.sma50 1,150.85 and tech.sma200 1,027.79 by about 0.3%; we use the tech block for consistency with the return series. (d) lastDiv of $6.46 is the trailing twelve-month dividend; the 10-K gives an indicated 2026 rate of $6.92 and we use that for the forward yield.

Not defects, correctly reported. The 445-day inventory figure, the 398-day cash conversion cycle, the 1.29% free-cash-flow yield and the 32.0x price-to-book all look wrong for a company of this quality and are correct — they reflect the genuine working-capital build and small equity base documented in Sections 4 and 5. The 101.3% return on equity is likewise arithmetically correct and should be read alongside 3.74x financial leverage.

Non-equity tripwire: checked and clear, with one observation. LLY is NYSE-listed common stock, average volume 3.0M shares, a 52-week closing band of $625.65-$1,235.56 (a 97% range — the opposite of the narrow band that characterises a preferred or baby bond), and a rising discretionary dividend. The one item that momentarily resembles the pattern is the beta of 0.506, which is low for a stock that has nearly doubled in a year. That is a genuine characteristic of large-pharmaceutical revenue rather than a signal of a fixed-income-like instrument, and given the 97% price range it does not trip the tripwire. Common equity, confirmed.

9. Technicals — a strong twelve-month name fading into a print

• Price $1,116.11, down 0.47% on the day (a $5.25 decline from a $1,121.36 previous close), on volume of 3.1M against a 3.0M average — an ordinary session. The day's range was $1,109.16 to $1,135.83; the stock closed near the low of its daily range on the eve of earnings.
• −9.70% from the 52-week closing high of $1,235.56; +78.32% above the 52-week closing low of $625.65. Maximum drawdown from peak over the trailing year is −9.70% on closing prices — a shallow correction, not a break.
• Below the 50-day average and above the 200-day. Spot is 3.06% BELOW the 50-DMA of $1,150.85 and 8.59% above the 200-DMA of $1,027.79. This is the constructive half of a pullback: the intermediate trend has rolled over, the primary trend has not.
• RSI 42.1 — neutral, mildly weak, nowhere near oversold or overbought. MACD −1.08 — negative, i.e. short-term momentum is against the position. By the standards of this batch, where several names carry RSI readings in the 70s and 80s, this is one of the better entry conditions available.
• Relative performance, and the shape that defines the moment:

Read the three rows in order and the pattern is unusually clean. Over twelve months LLY beat the index by 22 points — an exceptional result. Over six months it lagged by four points. Over three months it beat by eight. This is a leader that led, stalled for a quarter, and has been recovering — but has not yet reclaimed its high or its 50-day average. The knowledge base independently describes this as "18 months of consolidation" in a claim dated 2026-06-06.

• Practical read, and it is directly relevant to the verdict. Unlike most names in this batch, the chart is not the obstacle here — the calendar is. An RSI of 42, a price below the 50-DMA and a 9.7% drawdown from the high are the conditions under which one would ordinarily want to buy a compounding franchise. The 200-DMA at $1,027.79 sits 7.9% below and would be an excellent second entry; the 52-week low at $625.65 is not a realistic reference. The reason to stage rather than complete is the print in under 24 hours, not the technical position. A gap-down to the 200-DMA on a disappointing release would be an opportunity rather than a thesis break, and a gap-up through $1,150.85 would confirm the intermediate trend has turned — both outcomes argue for holding capital back tonight.

10. Insiders — routine compensation mechanics only

This table contains no signal and we will not manufacture one. Every acquisition is a zero-price or award-priced grant; the four-, nine-, eleven- and fourteen-share director awards on 2026-07-20 are dividend-equivalent accruals on existing holdings, not decisions. The Seymour entries are a single restricted-stock-unit vesting with the customary in-kind share withholding to cover tax. Not one insider bought or sold a share on the open market in the period covered.

Two things are worth one sentence each. First, no officer or director sold discretionarily during a 9.7% drawdown from the high — weak evidence, and we weight it as weak, but heavy selling into a fade would have been meaningful and did not occur. Second, the most recent filing is dated 2026-08-01, four days before the earnings release, which is consistent with a closed trading window and tells you nothing about anyone's view of the print.

11. Verdict, kill-criteria and flip conditions

Buy — Tactical.

The business deserves ownership. The entry deserves patience of a very specific and short-dated kind.

What the filings establish. Revenue of $65.179B in 2025, up 44.7%, and $19.799B in the first quarter of 2026, up 55.5% year-over-year, at an 83.8% gross margin, driven by volume: Mounjaro U.S. +53% in 2025 and +59% in the first quarter of 2026, Mounjaro ex-U.S. from $2.6B to $9.3B, Zepbound U.S. +174% then +79%. Europe grew 67.0% against the U.S. at 43.1%, so the growth is broadening geographically even as it concentrates by product. The FDA approved Foundayo (orforglipron) for obesity in April 2026, validating the $1.5 billion of pre-launch inventory the 10-K had capitalised and removing the injectable-capacity constraint. U.S. government pricing agreements were finalised in the first quarter, converting a policy overhang into Medicare access. Return on invested capital is 32.1%, interest coverage 37.5x, research spending $13.338B across approximately 12,000 research employees.

What you are accepting, stated without softening. Mounjaro and Zepbound were 65% of revenue in the first quarter of 2026, against 56% for full-year 2025 — concentration rising nine points in a year. Every growth sentence in both filings ends with "lower realized prices", and management guides to 5-10% annual category price deflation. Free cash flow was $8.972B against $20.638B of net income (43.5% conversion), with $8.093B absorbed by working capital and 445 days of inventory on the balance sheet. Dividends and buybacks of $9.492B exceeded free cash flow and were funded with debt. Gross debt has gone from $33.6B to roughly $52B in eighteen months against $26.535B of equity, and roughly $10 billion of contingent contract-manufacturing purchase commitments sit off the balance sheet. HHS has selected three Lilly products for administered pricing and the 10-K says "we expect other significant products will be selected in future years." Two federal product-liability MDLs cover the incretins. And the chief executive himself has said that "every drug goes to zero at patent expiry; Mounjaro reverts to baseline ~2030s."

On the earnings print — the decision this dive was written to make. Lilly reports tomorrow, 2026-08-05, on consensus EPS of $6.06 and revenue of $20.685B, after five consecutive beats. We are striking a Buy recommendation into that, and here is the justification, stated as a specific decision rather than a preference.

Why we are not waiting entirely. The valuation does not depend on the quarter: at 24.7x FY2027 consensus for a business growing revenue 30% this year, a single quarter moves the fair value by low single digits. The technical set-up is favourable rather than stretched — RSI 42, below the 50-DMA, 9.7% off the high — which is the opposite of the conditions that argue for waiting. The entire published analyst target range, floor included, sits above the current price. And the structural catalysts (orforglipron launch, Medicare access, ex-U.S. ramp) are already in the filings and do not require tomorrow to confirm them.

Why we are not buying the full position. Three reasons, in order of weight. First, the expectation is now in the price: beats of 22.7%, 9.1%, 23.4% and 12.7% mean the market prices a beat, so the payoff around a sixth is asymmetric against the buyer. Second, tomorrow's number is close to uninterpretable in the first hour: consensus implies an EPS decline of 4.0% on a revenue increase of 33.0%, which reconciles only through large charges — most plausibly the acquired in-process research from the up-to-$12 billion of pending deals the 10-Q discloses, up from under $3 billion three months earlier. Third, the last beat's composition was mixed: the 10-Q attributes part of it to "a favorable one-time adjustment to estimates for rebates and discounts," the tax rate fell 380 basis points, and the prior-year comparison carried a $1.72-per-share charge. We would be underwriting a run-rate that the filing itself tells us was not a clean run-rate.

The operational instruction, therefore: establish roughly one third of the target weight today at $1,116.11; hold the remainder for the session after the release, sized according to what the release actually says about realised price and incretin concentration rather than about the headline. A staged entry is Buy — Tactical by definition.

Pre-registered KILL criteria — what would break the thesis:

• A cut to the full-year 2026 guidance range of $82.0-$85.0B revenue or $35.50-$37.00 non-GAAP EPS. The implied back-half of $20.89-$22.39 (59-61% of the year) is steep enough that a cut is a live possibility, and it would falsify the growth path directly.
• Gross margin below approximately 80%. It has gone 82.5% → 84.3% → 82.9% → 85.1% → 81.9%, and the 10-Q names lower realised prices as the cause of the last decline. Sub-80% would mean price erosion has outrun manufacturing scale.
• The incretin share of revenue above 70% without offsetting pipeline revenue. 56% → 65% in one year is already the fastest-moving risk on this page.
• An incretin selected for U.S. government-administered pricing. The 10-K explicitly anticipates further selections.
• Free cash flow conversion staying below 50% of net income for a second full year while capital expenditure continues to rise. One year is a capacity ramp; two is a business model.
• A material adverse development in either incretin product-liability MDL, given the 10-K's disclosure that Lilly is "predominantly self-insured for litigation liability losses."
• Failure of the Centessa acquisition, which would trigger the special mandatory redemption at 101% on five of the eight tranches issued 2026-05-20.

Pre-registered FLIP TO BUY — CORE (upsize) criteria:

• A clean second-quarter release in which revenue meets or exceeds $20.685B, gross margin holds at or above 81.9%, and no new one-time rebate adjustment appears. This is the single most likely path to full size and it resolves within 24 hours.
• A retracement toward the 200-DMA at $1,027.79 (7.9% below) with the fundamentals intact — the second entry point, and a materially better one.
• Evidence that the oral franchise is additive rather than substitutive: total incretin revenue accelerating while the injectable lines hold.
• Working capital reversing — inventory days falling from 445 and free cash flow conversion rising above 60% of net income.
• A first sequential gross-margin increase alongside revenue growth, which would mean scale has finally outrun price erosion.

Where LLY fits in the Synthos Framework Portfolio. The biotechnology / diseases-of-aging sleeve — the house anchor, at a 3-4% target weight built in two or three tranches: approximately one third today, the balance after the 2026-08-05 release and on any move toward the 200-DMA. This is the highest-conviction sector placement in the batch and one of the more measured position recommendations, and that tension is deliberate. The house standing rule — pick forward next-exponentials, not trailing compounders — is satisfied here in a way it is not by most mega-caps: Lilly's forward consensus growth (+30.8% revenue in 2026) exceeds its trailing five-year average, the room-to-run is quantified by the company's own penetration numbers, and the acceleration is visible in the quarterly series rather than inferred. What restrains the sizing is not the business but the combination of a 9.3% base-case gain, a 65% single-franchise concentration, and a print tomorrow. Logged as a tracked Synthos call (Buy — Tactical) as of 2026-08-04 at $1,116.11.

On overlap within the batch. LLY, AbbVie and Johnson & Johnson (all covered 2026-08-04) occupy the same sector but not the same role. Lilly is the growth position — 44.7% revenue growth, 24.7x FY2027E, a 0.62% yield and a leveraged balance sheet. It should not be sized as though it were a defensive pharmaceutical holding, because its beta of 0.506 flatters a security whose earnings depend on two molecules and an administratively set price. If the biotechnology sleeve holds only one name, it should be this one — but it should be the smaller half of a sleeve, not the whole of it.

Single biggest risk: realised price. Not demand, not the science, not the competition — price. Both filings pair every volume gain with "lower realized prices." The first quarter of 2026 flattered itself with a one-time rebate true-up. China's reimbursement listing was explicitly a price-for-volume trade. The U.S. government has finalised a discounted-access agreement and has already selected three other Lilly products for administered pricing. Management guides to 5-10% annual category deflation and has already cut net price "from ~$1000 to $499." At 65% single-franchise concentration, a 500-basis-point acceleration in net-price erosion does more damage to this equity than any clinical setback in the pipeline.

Most fragile assumption in the price: that volume growth continues to outrun price erosion by a wide enough margin to deliver 30% revenue growth in 2026 and 16.5% in 2027. At the guided 5-10% annual net-price decline, unit volume must compound at roughly 40-45% to hit the 2026 consensus. The first quarter delivered it. The question every subsequent quarter asks — beginning tomorrow — is whether it can keep delivering it as the installed base grows, as roughly half of weight-management users discontinue within 12-14 months on the company's own account, and as the payer, the government and the reimbursement list each take their turn on the price.

Provenance & disclosures

• Traceability: 180 entity-level knowledge-base claims name Eli Lilly (199 raw hits; 19 non-entity thesis/category mentions discarded; breadth 11 distinct sources; net conviction positive). Stance mix 159 bullish / 18 neutral / 3 bearish, spanning 2025-01-09 to 2026-07-25. Discards and weightings, in full: 8 entity-level claims are drug-molecule-level only (ents = tirzepatide / semaglutide / Mounjaro without the company or ticker) and are discarded from the equity read — six from a physician source on dosing and weight cycling, two from the chief executive on mechanism; this is the drug-versus-equity contamination the molecule-name search patterns were always going to produce. 3 claims tagged to a quarantined-misattribution source are excluded outright under the standing attribution policy. 123 claims (68.3% of the file) come from the chief-executive lane (118) or the investor-relations lane (5) and are HALF-WEIGHTED at skill 0.5 — note that we override the source data's skill-1.0 tag on the chief-executive lane, because a chief executive describing his own franchise is management framing however the record is tagged; none is used as a valuation input. 10 claims come from a commercial counterparty that supplies Lilly's computing hardware and are discounted as sales claims. 1 claim from a promoter-class voice is low-weighted under the standing house rule. That leaves 37 claims from 5 sources carrying full independent weight. Search patterns: LLY, Eli Lilly, Lilly, tirzepatide, Mounjaro, Zepbound, orforglipron, retatrutide, across all 51,928 records; match distribution Eli Lilly 103, LLY 68, tirzepatide 8, Lilly 1. All quoted claims are verbatim from the stored claim text, with bracketed substitutions only where a house naming restriction applies. There is not one independent bearish claim about this equity in 199 hits, and that fact is treated as a caution rather than as confirmation.
• Data as-of: fundamentals — the most recent income statement ends 2026-03-31 and the most recent 10-Q was filed 2026-04-30, so THIS FILE IS ONE QUARTER STALE; the second quarter of 2026 reports 2026-08-05, one day after this dive · estimates 2026-08-04 · prices 2026-08-04 ($1,116.11; 50-DMA $1,150.85; 200-DMA $1,027.79; RSI 42.1; MACD −1.08) · knowledge-base claims 2026-08-04. Every figure comes from the Synthos vendor data file for LLY or from the SEC filings in filings/LLY/; no figure comes from memory, recall or external retrieval.
• Filings consulted: 10-K filed 2026-02-12 (FY2025); 10-Q filed 2026-04-30 (Q1 2026 — the most recent financial statements available); 10-Q filed 2025-10-30 (Q3 2025); 8-K filed 2026-04-30 (Q1 earnings furnishing); 8-K filed 2026-05-07 (annual meeting results, 2026-05-04); 8-K filed 2026-05-20 (the $9.0 billion note offering). Targeted greps run across the 10-K and both 10-Qs covered: product and brand revenue, incretin concentration, gross margin and operating-expense drivers, research and development, the borrowings note, the lease note, commercial paper and credit facilities, capital expenditure and manufacturing commitments, pre-launch inventory, share repurchases and dividends, the IRA and Medicare selections, compounding and counterfeit exposure, contingencies and litigation, tax rate, segment structure, headcount, and the executive-overview risk language in both periods.
• Filing-versus-vendor overrides (four, all resolved in favour of the filing): (1) Debt is a quarter and a bond deal out of date — the vendor carries $42.503B at 2025-12-31; the 10-Q shows $43.4B at 2026-03-31; the 2026-05-20 8-K adds $9.0B of principal issued with net proceeds of ~$8.94B; gross debt is now approximately $52B, ~23% above the vendor figure, and every vendor leverage ratio is computed on the stale base. (2) Annual gross-margin change — the 10-K says gross margin "increased 1.7 percentage points" in 2025; the vendor's numbers imply +2.5 points; we use +1.7. (3) Selling-and-administrative growth — the 10-K says marketing, selling and administrative expenses "increased 29 percent"; the vendor implies +36.4%; we use 29%, and note the related finding that the vendor appears to place acquired in-process R&D charges below the operating line, flattering its 45.6% operating margin relative to the filing's presentation (the vendor's pretax and tax lines are reliable — 19.79% against the filing's stated 19.8%). (4) Product-level revenue — the vendor's seg_prod contains only "Product" and "Collaboration and other revenue" and is useless for analysing a single-segment pharmaceutical company; the filings' MD&A supplies the brand-level economics and the incretin concentration (56% for FY2025 per the 10-K, 65% for Q1 2026 per the 10-Q), and we use the filings throughout. A fifth item sits between the two sources rather than being resolved by either: the share count, where the income statement and cash-flow statement imply ~897M shares while the 2026-05-07 8-K reports 944,818,881 outstanding — a 5.3% gap we could not resolve, handled by doing all valuation per share.
• Other data rejections (detailed in Section 8): all est EBITDA, EBIT and net-income rows REJECTED — netIncomeAvg exceeds ebitdaAvg in every year from FY2025 to FY2030 (FY2030: $58.152B against $42.216B, a 44.2% net margin against a 32.1% EBITDA margin) and FY2023 shows EBITDA of $8.830B below EBIT of $9.205B; forward valuation runs on epsAvg exclusively and no forward EBITDA multiple is quoted anywhere. The GAAP-versus-non-GAAP basis mismatch between inc_q ($28.15 trailing) and earn_cal ($29.42 trailing) is identified and both bases are quoted. netDebt ($35.340B vs a recomputed $35.235B) and enterprise value (+0.3% residual) are accepted for FY2025 as within tolerance. The lease double-count pattern was tested for and is absent — both capital-lease fields are zero and the 10-K discloses operating leases only. Closing-price 52-week extremes are preferred over intraday; the tech moving averages are preferred over the quote block; quote.price $1,116.11 is used over tech.last $1,115.68 (a 0.04% difference); the indicated 2026 dividend of $6.92 from the 10-K is used over the vendor's trailing lastDiv of $6.46.
• Estimate coverage: FY2026 13 analysts on EPS, 18 on revenue; FY2027 13 and 18 — the primary forward anchor; FY2028 6 and 22; FY2029 4 and 15; FY2030 2 and 11. No conclusion in this dive rests on the FY2029 or FY2030 rows, and the FY2028 EPS row (six analysts) is used only as a cross-check. Note additionally that consensus FY2026 revenue of $85.276B sits above the top of management's guided $82.0-$85.0B while consensus FY2026 EPS of $35.039 sits below the bottom of the guided $35.50-$37.00 — an internal inconsistency in the consensus set, and a further reason the FY2027 row rather than the FY2026 row is used as the valuation anchor.
• Peer note: the vendor peer set is appropriate for once — AbbVie ($243.80 / $430.74B), Amgen ($390.02 / $210.50B), AstraZeneca ($155.62 / $241.21B), Gilead ($135.27 / $167.95B), Johnson & Johnson ($254.93 / $614.36B), Merck ($128.04 / $316.24B), Novo Nordisk ($44.28 / $196.77B — the only direct incretin competitor), Novartis ($153.27 / $291.32B), Pfizer ($25.41 / $144.82B), Sanofi ($42.68 / $102.48B). No peer multiples are supplied in the file, so no peer-multiple comparison is drawn.
• Fair-value caveat: the $875 / $1,220 / $1,490 anchors are multiples of FY2027 consensus EPS of $45.178 on the company's non-GAAP basis (19.4x / 27.0x / 33.0x), cross-checked against FY2026E of $35.039 and FY2028E of $52.151. Stated arithmetic, not a discounted cash flow. The base case requires the market to pay 27.0x on 2027 earnings against 24.7x today; at an unchanged 24.7x, FY2027 consensus earnings imply approximately today's price of $1,116, and that dependence is flagged in Section 6b as the fragile leg. Our base sits 7.3% below the street consensus of $1,315.42 — this dive is the more cautious of the two views.
• Timing: Eli Lilly reports second-quarter 2026 results on 2026-08-05 — ONE DAY after this dive. Vendor consensus EPS $6.06 on revenue $20.685B, calendar entry last updated 2026-08-04. The earnings banner at the top of this document is mandatory under the batch specification and the timing is an explicit input to the verdict, dealt with in Section 11. The most recent financial statements available at the time of writing end 2026-03-31 and were filed 2026-04-30; the second quarter is not in this file.
• Accessibility note: all tables are labelled with header rows, all emphasis is carried by text weight and wording rather than colour, every figure is stated in full rather than encoded in a chart, and no conclusion depends on a visual element.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.