Illumina ILMN

Reference table

1. Data integrity — what we rejected before writing a line

REJECTED — the entire forecast EBITDA and EBIT series in est. Identical vendor defect to the one found in NTRA in this same batch: every forecast year carries a large negative EBITDA alongside a large positive net income. FY26E EBITDA −$511.4M against net income +$817.3M; FY30E EBITDA −$662.5M against net income +$1,379.8M. EBITDA cannot sit $2.04B below net income. The forecast EBIT line is similarly negative throughout (−$862.1M FY26E). Only the revenueAvg, netIncomeAvg and epsAvg lines from est are used in this dive. Those three pass an internal cross-check: FY26E net income of $817.3M over FY26E EPS of $5.313 implies ~153.8M shares, consistent with the 153M weighted average in Q1'26.

REJECTED — inc_q[0] interest expense. The most recent quarter (2026-06-28) reports interestExpense: 0. Every other quarter in the file runs $24M-$36M, and the company carries $2,553M of total debt. The zero is a vendor artifact; we use ~$25M per quarter where the figure matters. Note that this quarter is otherwise internally consistent: revenue $1,159M, gross profit $770M (66.4%), operating income $245M, net income $207M, EPS $1.37 on 153M diluted shares (207/153 = $1.35, close enough to $1.37 to be a rounding difference on the exact share count).

FLAGGED — earn_cal revenue near-duplication. The 2026-02-05 row reports revenueActual of $1,159,000,000, and the 2026-07-30 row reports the same $1,159,000,000. The corresponding inc_q figures are $1,160M (Q4'25) and $1,159M (Q2'26). A $1M discrepancy on the Q4'25 row. Immaterial, but it is the kind of duplication that signals a stale write; we use inc_q for all revenue figures.

FLAGGED — net debt is ambiguous across the file. bal_a FY25 gives cash and equivalents $1,418M, cash and short-term investments $1,633M, long-term investments $338M, total debt $2,553M, and states netDebt: 1135000000 (i.e. it nets only cash and equivalents). km_ttm reports an enterprise value of $31,468.6M against a $29,981.6M market capitalisation, implying net debt of $1,487M — a third figure. Our computation, using cash plus short-term investments against total debt: $2,553M − $1,633M = $920M of net debt, or $582M if long-term investments are included. We use $920M and enterprise value of ~$30.90B, and we label it as our computation. The spread across the three figures is $567M — 1.9% of market capitalisation, so no conclusion in this dive turns on it.

FLAGGED — GAAP versus non-GAAP EPS. inc_a FY25 reports EPS of $5.48, and the quarterly GAAP EPS sum ($0.82 + $1.50 + $0.98 + $2.18) reconciles exactly to it. earn_cal reports the non-GAAP figures the market trades on ($1.19, $1.34, $1.35 for FY25 quarters; $1.15 and $1.31 for FY26). Trailing GAAP EPS is $5.41 (Q3'25 through Q2'26); trailing non-GAAP is $5.15. The est EPS series is on the non-GAAP basis (FY26E $5.313 sits naturally above the $5.15 trailing). All forward multiples in this dive use the non-GAAP est series, and are labelled as such.

GAP — no consumables versus instruments split. seg_prod gives only "Sequencing" $3,985M and "Microarray" $358M for FY25. For a razor-and-blades business the consumables/instrument mix is the most important single disclosure, and it is absent. No claim in this dive rests on that mix, though its absence is precisely why we cannot distinguish "customers stopped buying machines" from "customers stopped running them" — a distinction that would change the verdict.

GAP — forecast depth collapses beyond FY28. FY29E rests on 8 revenue and 5 EPS contributors; FY30E on 7 and 2. A two-analyst FY30 EPS of $8.845 is not a consensus. The base case is built on FY27E, where coverage is 13 and 12.

CLEAN — everything else. Quarterly revenue sums to the annual line for FY25 ($1,041M + $1,057M + $1,083M + $1,160M = $4,341M against the stated $4,338M — a $3M rounding difference). Quarterly net income sums to $850M, matching exactly. Balance sheet ties. Cash flow ties. No sign flips, no zeroed share counts.

2. The four-year revenue decline — the fact the price ignores

Four consecutive annual declines. Cumulative change FY22 to FY25: −5.4%. Trailing revenue of $4,493M is still 1.99% below the FY22 peak of $4,584M, four and a half years later.

Set against that, the profit recovery is spectacular and should be given full credit: FY24 operating income of −$833M became +$864M in FY25 — a $1,697M swing on essentially flat revenue. That is entirely a cost and impairment story. The mechanics are visible:

• R&D: $1,321M (FY22) → $1,354M (FY23) → $1,169M (FY24) → $951M (FY25). Down 29.8% from peak, and from 30.1% of revenue (FY23) to 21.9% (FY25) — trailing 21.3%.
• SG&A: $1,612M (FY23) → $1,092M (FY24) → $1,078M (FY25). Down 33.1% from FY23.
• Gross margin: 60.9% (FY23) → 65.4% (FY24) → 66.7% (FY25) → 66.8% trailing.
• Goodwill: $2,545M (FY23) → $1,113M (FY24) → $1,113M (FY25). A $1,432M write-down in FY24, and total assets fell from $10,111M to $6,303M in the same year — the balance sheet was roughly a third smaller after the FY24 restructuring.

The honest read. Illumina removed roughly $920M of annual operating expense and wrote off roughly $1.4B of goodwill, and on flat revenue that produced a $1.7B earnings swing. It worked. But cost removal is a one-time act with a floor. R&D at 21.3% of revenue is already lean for an instrument company that must defend a technology position against lower-cost entrants; there is limited further reduction available without impairing the franchise. From here, EPS growth must come from revenue growth or buyback. Consensus assumes ~6% of the former and the company delivered $742M of the latter in FY25 (~2.5% of shares annually).

3. Geography — the China number is the most important line in the file

China revenue has fallen in every single one of the last four years, from $502M to $243M — a 51.6% decline. That $259M of lost revenue is, on its own, larger than the entire $246M cumulative revenue decline the company reported over the same period. Stated another way: absent China, Illumina's revenue would have grown modestly. China is the whole decline.

Two readings are available and the data cannot distinguish between them:

1. Geopolitical / procurement policy — a market Illumina cannot serve on the terms it once did, in which case the erosion is largely complete at ~$243M and the rest of the business is quietly growing.

2. Competitive displacement by lower-cost domestic sequencing platforms — in which case China is the leading indicator, and the same cost dynamic eventually reaches EMEA and Asia-Pacific pricing.

The distinction matters enormously and we cannot resolve it from this file. What we can observe: Americas revenue also declined for two consecutive years ($2,521M FY23 → $2,441M FY24 → $2,406M FY25), which is at least consistent with the second, more pessimistic reading. EMEA is the one bright spot, growing from $1,140M (FY23) to $1,264M (FY25), +10.9%.

Anyone underwriting Illumina at 33x forward earnings is implicitly betting on reading (1). We flag that as an assumption, not a finding.

4. Product mix and the razor-and-blades question

FY25: Sequencing $3,985M (91.7% of the $4,343M product total), Microarray $358M (8.3%). Microarray has declined in every year the file records ($417M FY21 → $413M FY22 → $392M FY23 → $358M FY25) — a legacy technology in managed run-off, worth roughly 8% of revenue and shrinking ~4% annually. Its disappearance costs about $15M of revenue a year, which is noise.

What we cannot see, and it matters: the split between instrument placements and consumables. In FY18 the file carried that detail (Consumables $2,156M, Instruments $569M, Services and Other $584M — i.e. consumables were 79% of product revenue), but the disclosure was discontinued. Without it we cannot answer the central operating question: is flat revenue the result of a paused instrument-purchase cycle (temporary, and followed by a consumables ramp) or of flat run-rates on the installed base (structural)?

The circumstantial evidence leans toward the latter. Days of inventory outstanding is 153.9 — over five months of stock on hand — and inventory has been broadly flat at $547M-$587M for three years. A company anticipating an instrument-cycle upswing typically builds; a company managing a plateau holds steady. But this is inference, not evidence, and we label it as such.

5. Financials — the receipts

• Revenue: FY25 $4,338M (−0.8%); TTM $4,493M. Quarterly run-rate now $1,091M-$1,160M.
• Profitability: FY25 operating income $864M (19.9% margin), net income $850M (19.6%), GAAP EPS $5.48. TTM operating margin 21.0%, net margin 18.4%, EBIT margin 24.3%.
• Gross margin: 66.8% trailing, up from 60.9% (FY23). Q2'26 66.4%.
• R&D: $951M (21.9% of FY25 revenue; 21.3% trailing) — down from $1,354M and 30.1% in FY23.
• Cash flow: FY25 operating cash flow $1,079M, capex $148M, free cash flow $931M. FY24 $837M/$128M/$709M. FY23 $478M/$195M/$282M. Cash generation has more than tripled in two years. Free cash flow yield 3.1%.
• Capital return: buybacks $742M (FY25) against $116M (FY24) and $0 (FY23). No dividend. Share count 159M (FY24 diluted) → 156M (FY25) → 153M (Q2'26) — roughly 3.8% retired in eighteen months.
• Balance sheet: cash and short-term investments $1,633M plus long-term investments $338M; total debt $2,553M ($1,490M long-term, $499M short-term); net debt ~$920M (our computation). Net debt/EBITDA ~1.1x. Interest coverage 12.6x. Current ratio 1.80x, quick ratio 1.40x.
• Returns: return on equity 31.1%, return on invested capital 13.5%, return on assets 12.4%. The ROE figure flatters — equity is only $2,723M because of the accumulated write-downs (retained earnings −$392M), so a modest profit produces a large ratio. Return on invested capital of 13.5% is the honest number, and it is respectable but not exceptional.
• Stock compensation: $275M (FY25), 6.3% of revenue — down from $380M (FY23). Modest by software standards, normal for instruments.

Overall: this is now a genuinely well-run, cash-generative, moderately levered business. Nothing in the financial statements argues against owning it. The argument is entirely about what it costs.

6. Valuation — what is priced at $198.16

At $198.16 (market cap $29.98B, net debt ~$0.92B, EV ~$30.90B):

The valuation problem stated as plainly as possible: 33.0x FY27E earnings for a business consensus expects to grow revenue 5.8% that year. The PEG on revenue growth is 5.7x. Even on the more flattering EPS growth (+13% annually FY26-28, driven by margin and buyback), the multiple is 2.5x growth.

Relative to the vendor's own peer set: Illumina at 6.9x trailing EV/revenue sits above Quest Diagnostics ($25.6B cap), Labcorp ($25.2B), Hologic ($17.0B) and ICON ($12.1B) — all of which are lower-growth service businesses, so the premium is arguably right in kind. It sits well below Natera (15.1x, covered separately in this batch) — which is growing 20%+ rather than 6%. Illumina is priced between a mature service business and a hypergrowth diagnostic, while growing like the former.

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

Reverse-engineering $198.16 into falsifiable claims. Consensus-derived arithmetic, labelled as such.

• Revenue growth resumes and holds at ~6% annually through FY28, ending the four-year decline permanently. Consensus (13 analysts) has $4,877.0M for FY27E and $5,214.1M for FY28E. That requires 2026, 2027 and 2028 each to grow — after four consecutive years of contraction. Falsifiable quarterly. Q2'26 supports it (revenue $1,159M, +9.7% versus Q2'25's $1,057M). Q3 consensus of $1,150.6M does not (it implies a sequential decline).
• Gross margin holds at or above 66% and operating margin advances toward ~22-24%. The EPS path from $5.313 (FY26E) to $6.932 (FY28E) on only 13% revenue growth requires roughly 300bp of further operating-margin expansion. Since R&D is already cut to 21.3% of revenue from 30.1%, most of the remaining leverage must come from gross margin or from revenue growing faster than the cost base — i.e. from volume. This is the most fragile assumption in the price.
• The buyback continues at or above the FY25 pace ($742M annually, ~2.5% of shares). Roughly 2.5 points of the ~13% annual EPS growth is share-count reduction. FY24's buyback was only $116M, so this is not an established run-rate — it is one year of behaviour extrapolated.
• China stabilises at or above $243M. Consensus does not break this out, but a continuation of the 2021-2025 trend (−17% annually) removes another ~$40M of revenue per year, roughly offsetting one-sixth of the modelled growth.
• The market keeps paying 30x+ forward for a mid-single-digit revenue grower. No de-rating despite the growth profile. This is not a company-controllable assumption at all — it is a bet on the multiple, and it is what today's price is mostly made of.

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

Expected return over a two-year horizon: EPS growth (+13% annually, consensus) + multiple drift (negative) + shareholder yield (+2.5% from buyback, no dividend).

Our base case assumes multiple COMPRESSION. We apply 29x to FY27E EPS against the 33.0x the market pays for FY27E today — a genuine de-rating on the same forward year, not the mechanical rolldown that a flat price produces. The justification: a company growing revenue 5.8% with a completed cost programme is a mid-to-high-teens or low-twenties P/E business in most market regimes; 29x is already generous and reflects the franchise quality, the 66.8% gross margin and the optionality on clinical volume inflection.

So the base case return is negative (−12%): roughly +13% of EPS growth and +2.5% of buyback yield, more than offset by a ~12% de-rate and a starting price 9.4% above the consensus target. The bull case requires the opposite — that the multiple holds at 33x while EPS grows into FY28 — which is only reasonable if revenue growth re-accelerates to double digits. That is the fragile leg, and it is the leg the last four years of revenue have not supported.

Note carefully: the drop from 37.3x FY26E to 28.6x FY28E at a constant price is mechanical rolldown, not re-rating. It is not evidence the stock gets cheaper; it is arithmetic. Only the multiple the market actually pays at each date counts, and today it pays 33x for FY27.

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

• We differ from the street on what the last twelve months of price action represents. The consensus framing is that a turnaround has been recognised. Our reading of the file is that the turnaround was a one-time $1.7B cost-and-impairment swing on flat revenue, and that the market has capitalised a non-repeatable event into a permanent multiple. The test is simple and quarterly: does revenue grow? Negative surprise, watchable 2026-10-29: revenue at or below the $1,150.6M consensus, confirming a sequential decline and no top-line inflection.
• We think China is being treated as an idiosyncratic geopolitical footnote when the data reads as a competitive leading indicator. Illumina's China revenue fell every year for four years while Americas fell for two. If lower-cost sequencing platforms are the cause, the erosion is not contained. Negative surprise, watchable annually: China below ~$200M in FY26, or Americas declining for a third consecutive year.
• We may be wrong on the buyback. If management steps up repurchases materially from the $742M FY25 pace — the balance sheet could support $1.5B+ annually against $931M of free cash flow and only ~1.1x net leverage — the EPS path accelerates meaningfully without any revenue help. Positive surprise, watchable in cash-flow statements: annualised buyback above ~$1.2B.
• The genuine bull catalyst we are not modelling is a clinical-volume inflection — population-scale screening or a step-change in oncology profiling that finally converts installed instruments into consumables volume. The file gives no evidence for it, so we assign it no value, but it is the reason not to be short. Positive surprise, watchable: any quarter above ~$1,250M of revenue, or a return to double-digit year-over-year growth.
• Where we do NOT differ: on the quality of the repaired P&L, the 66.8% gross margin, the cash generation, or the balance sheet. Those are all as good as the bulls say. Our disagreement is confined to price and to the durability of the growth assumption.

Synthos fair values

• Bear ~$130 — 22x FY27E EPS ($6.008). Assumes the four-year decline resumes, China erosion continues, gross margin plateaus, and the market re-rates toward a mature-instrument multiple. This is roughly the 200-day average ($140) broken and held below. −34%.
• Base ~$175 — 29x FY27E EPS ($6.008) = $174.2. Assumes ~6% revenue growth lands, margins hold, the buyback continues, and the multiple de-rates modestly from 33x toward a level consistent with mid-single-digit growth. Sits just below the $181.18 consensus target and just above the $170 median. −12%.
• Bull ~$250 — 36x FY28E EPS ($6.932) = $249.6. Assumes revenue re-accelerates to high single digits or better, China stabilises, and the market pays a genuine growth multiple. Above the $230 street high. +26%.

Base sits ~12% below spot. The distribution is roughly symmetric in magnitude (−34% / +26%) but centred below the current price — which is exactly what a Watch looks like.

7. Technicals — extended, not exhausted

• Price $198.16, down 3.4% on the day, −3.4% from the fifty-two-week high of $205.10.
• +12.2% above the 50-day average ($176.55) and +41.3% above the 200-day average ($140.22). Both rising. The trend is powerfully intact.
• RSI 55.98 — neutral. MACD +6.72, positive. No exhaustion signal in either direction, which means the chart offers no tactical edge and no protection.
• Relative strength: +92.9% over twelve months (S&P +19.9%, Nasdaq-100 +23.9%), +45.9% over three months (S&P +5.1%). Illumina has outperformed the index by 73 points in a year.
• max_dd_from_peak of −61.1% is a useful reminder of what this chart looked like on the way down — the fifty-two-week low is $91.00, and the stock has more than doubled off it.
• Read: the 200-day at $140 is 29% below spot. The 50-day at $176.55 is 11% below. There is a very large air gap between the current price and any technical support, which is the practical risk in a momentum name where the fundamental case rests on a cost programme that is finished.

8. Insider activity and governance

• Scott M. Davies (Chief Legal Officer) sold 615 shares at $200.00 on 2026-07-30 — the day of the Q2 earnings release — leaving 20,338 held. Small and almost certainly programmatic.
• Michael C. Sullivan (officer) filed a Form 3 on 2026-07-29 with zero holdings — a new appointment.
• Keith A. Meister (director) received 142 shares at $149.19 on 2026-06-30, holding 6,780. Meister is the principal of an activist investment firm, and his presence on the board is a material governance fact visible in this file. It is consistent with the aggressive cost programme documented in §2 and with the step-up in buybacks from $116M to $742M — activist-influenced boards typically prioritise exactly those two levers. It also implies that the easy activist work is largely done, which reinforces the "cost story has finished" read.

No insider buying appears anywhere in the file. With the stock up 92.9%, that absence is unsurprising and not itself a signal — but neither is there a director stepping up at these levels.

9. Knowledge base — three claims, all stale, and they do not net bullish

A search of 51,928 knowledge-base entries for "ILMN" and "Illumina" returns three claims:

Eli Lilly CEO (2021-03-04, conviction 50, bullish, horizon: thesis) — categories genomic sequencing / pathogen surveillance, entities Illumina, ILMN: "US genomic surveillance is badly underfunded (~0.3% vs needed 5% of cases); sustained variant tracking implies structural demand for sequencing capacity." Read: five and a half years old, made during a pandemic surveillance surge, at conviction 50 (low). It is a public-health observation that happens to name Illumina, not a company view. The demand it anticipated did not show up in the revenue line — FY21 revenue was $4,526M and FY25 was $4,338M.

Rhonda Patrick (2022-03-09, conviction 72, bullish, horizon: thesis) — categories liquid biopsy / multi-cancer early detection / diagnostics, entities Grail, Illumina: "Grail's liquid biopsy detects circulating tumor DNA for 50 cancers from a blood draw; life-changing for catching cancer at stage zero/one." Read: this claim is about an asset Illumina no longer owns. The file itself evidences the divestiture — goodwill fell from $2,545M (FY23) to $1,113M (FY24), total assets from $10,111M to $6,303M, and FY24 carried a −$1,223M net loss consistent with the write-down. The technology claim may well be correct; it simply is not an Illumina claim any more.

David Sinclair (2022-04-18, conviction 42, bullish on the competitor) — category DNA sequencing, entities Oxford Nanopore: "Oxford Nanopore's pore-based sequencing reads thousands to tens of thousands of bases per run — the more advanced, upcoming technology versus Illumina's older microscopy-based chips." Read: this is a mildly NEGATIVE claim for Illumina, tagged at low conviction (42), from a voice whose expertise is longevity biology rather than instrument markets. Discount it heavily as a market call — but note that its structural argument (that Illumina's platform is the incumbent rather than the frontier) is not contradicted by anything in the last four years of revenue.

kb_claim_count: 3. kb_breadth: 3. Net conviction: mixed and stale. One claim is five years old and about public policy; one is about a divested business; one favours a competitor. There is no live Synthos conviction lane on Illumina, and the tagged claims should not be read as panel support for the name. The dive stands on the fundamentals and the arithmetic.

10. Moat and competitive position

Illumina's moat is an installed base plus a consumables lock-in, reinforced by a validated-workflow switching cost: laboratories that have built protocols, LIMS integrations and regulatory submissions around a platform do not migrate casually. The 66.8% gross margin is the direct evidence that the moat still prices.

But three things in this file argue the moat is narrowing:

1. China. A 51.6% five-year revenue decline in one geography is not what an unassailable standard looks like.

2. Americas revenue has declined for two consecutive years ($2,521M → $2,441M → $2,406M).

3. R&D has been cut 29.8% from peak and from 30.1% to 21.3% of revenue. In a technology business facing a credible alternative platform, cutting research intensity by nine points of revenue is how you harvest a moat, not how you widen one. It is the correct financial decision if the franchise is mature; it is a dangerous one if it is contested.

Moat: real, wide in the installed base, narrowing at the margin. It supports a premium multiple over a service business. It does not support 33x forward earnings on 6% growth.

11. Management and capital allocation

CEO Jacob Thaysen. 8,600 employees (down from the pre-restructuring headcount implied by the FY23 cost base).

The capital-allocation record over the period covered is, bluntly, one of repair after destruction: the FY22 loss of $4,404M and the FY23-24 losses of $1,161M and $1,223M reflect an acquisition strategy that was reversed at enormous cost. Total shareholder capital destroyed across FY22-FY24: $6,788M of cumulative net losses. Retained earnings remain negative at −$392M.

Since then, allocation has been disciplined and unremarkable in the best sense: no dividend, no acquisitions of consequence ($10M net in FY25), capex held at $148M (3.4% of revenue), and $742M returned via buyback. Free cash flow of $931M comfortably funds it. Net leverage of ~1.1x leaves capacity.

The forward question is whether the buyback accelerates. With $931M of free cash flow, $1,633M of cash and short-term investments, and only ~1.1x net leverage, a $1.5B annual repurchase is affordable and would add ~2.5 points to the EPS growth rate. That is the single most available lever management has, and it is worth watching more closely than any product announcement.

12. Verdict, kill-criteria and flip conditions

Watch. Illumina has done exactly what a broken company should do — cut $920M of annual operating expense, write off the failed acquisition, restore a 66.8% gross margin, generate $931M of free cash flow, and start buying back stock. Operating income swung $1,697M in a single year. That work deserves credit and it is why this is Watch rather than Avoid.

But it is finished work, and the market has paid for it twice over. Revenue has declined for four consecutive years and trailing revenue still sits below the FY22 level. Consensus models 6% growth, not recovery. China — the world's most important sequencing growth market — has halved. At $198.16 the stock trades 9.4% above the street's own $181.18 consensus target, at 33.0x FY27E earnings for a 5.8% revenue grower, with 24 of 50 raters at Hold or Sell, and with the nearest technical support 11% below and the 200-day average 29% below. Our base case is $175, twelve percent below spot.

We do not own it and will not initiate here. If held, trim into strength.

Staged entry — the conditions under which we would buy:

1. No purchase above ~$165.

2. Tranche 1 (1-2%) below ~$150 — roughly 27x FY27E EPS, a level at which the multiple is defensible on the current growth profile alone and the optionality on volume inflection is free.

3. Tranche 2 (to ~3%) below ~$135 — approaching the 200-day average and the $135 street low target, at ~22x FY27E. That is the bear case, and buying the bear case is how you get paid for a franchise this durable.

Pre-registered KILL / do-not-buy criteria:

• A fifth consecutive year of revenue decline — FY26 revenue below FY25's $4,338M would confirm that the top line has no engine and would invalidate the entire consensus EPS path.
• China below ~$200M in FY26 — continuation of the competitive erosion at the same rate, and the strongest available evidence for the platform-commoditisation reading.
• Gross margin below ~64% — the 60.9% → 66.8% expansion is a substantial share of the earnings recovery; reversal removes it.
• R&D falling below ~20% of revenue — further cuts from an already-lean base would signal harvesting rather than defending, and would raise the terminal-decline probability materially.
• Buyback pace falling back toward the FY24 level ($116M) — removes ~2.5 points of the ~13% modelled EPS growth.

Pre-registered FLIP TO BUY:

• A price below ~$150 with the FY25 margin structure intact — the single most likely path, requiring nothing to change but the quote.
• Two consecutive quarters of double-digit year-over-year revenue growth — the first genuine evidence in five years that the installed base is being run harder. Q2'26 (+9.7%) is one data point; a second above ~$1,200M in Q3 would be the signal. This would justify paying today's multiple.
• China revenue stabilising or rising year-over-year — this would resolve the §3 ambiguity in favour of the benign geopolitical reading and materially raise the terminal value.
• An annualised buyback above ~$1.2B — a management decision, entirely within their control, that would add meaningfully to the EPS path without requiring any revenue help.

Where ILMN fits in the Synthos Framework Portfolio. The genomics-tools sleeve, as the infrastructure / instrument layer — target 2-3% weight, currently 0%. It is genuinely complementary to a clinical-test position rather than redundant: Illumina sells the platform and consumables to laboratories; Natera (covered separately in this batch) sells reimbursed clinical tests to physicians. Different customers, different revenue models, opposite growth rates (~6% versus ~20%) and, notably, opposite valuation errors — Illumina is a slow grower priced like a fast one, Natera is a fast grower priced for perfection. Owning both at the wrong prices is not diversification. Logged as a tracked Synthos call (Watch, do not initiate; trim if held) as of 2026-08-04 at $198.16, with a pre-registered Buy trigger at $150.

Single biggest risk: four consecutive years of revenue decline being financed at a growth multiple. If FY26 or FY27 fails to grow, there is no cost programme left to substitute for it, and the stock re-rates from 33x toward the low twenties — which is the $130 bear case, 34% below spot.

Most fragile assumption in the price: that operating margin expands a further ~300bp to carry EPS from $5.31 to $6.93 by FY28. R&D has already been cut from 30.1% to 21.3% of revenue and SG&A by a third; the easily-available leverage is spent. The remaining margin expansion must therefore come from revenue growing faster than the cost base — which is to say, the EPS forecast quietly depends on the volume growth the last four years have not produced.

Provenance and disclosures

• Traceability: 3 tagged knowledge-base claims name Illumina, across 3 thinkers, all dated 2021-2022 (breadth 3, net conviction mixed and stale). One is a five-year-old public-health observation at conviction 50; one concerns Grail, an asset the company has divested (the divestiture is evidenced in this file by the FY24 goodwill and total-asset collapse); one, at conviction 42, explicitly favours a competing sequencing technology. There is no live Synthos conviction lane on this ticker. Searched across 51,928 entries.
• Data as-of: fundamentals 2026-06-28 (Q2'26) · estimates, prices, technicals and knowledge-base claims 2026-08-04. All market data drawn exclusively from scripts/deepdive/vti_data/ILMN_data.json.
• Vendor data rejected (see §1): (a) the entire forecast EBITDA and EBIT series in est — every year carries large negative EBITDA alongside large positive net income (FY30E: EBITDA −$662.5M versus net income +$1,379.8M), arithmetically impossible; only revenueAvg, netIncomeAvg and epsAvg were used. (b) inc_q[0] interest expense of $0 for the quarter ended 2026-06-28, against $24M-$36M in every other quarter and $2,553M of outstanding debt.
• Vendor data flagged: earn_cal reports identical revenueActual of $1,159,000,000 for both 2026-02-05 and 2026-07-30, against inc_q figures of $1,160M and $1,159M — a probable stale write; inc_q was used throughout. Net debt is reported three different ways across the file ($1,135M in bal_a, ~$1,487M implied by km_ttm's enterprise value, and $920M on our own cash-plus-short-term-investments computation); we use $920M and label it as our computation, and note that the $567M spread is 1.9% of market capitalisation and changes no conclusion.
• Data gaps stated, not papered over: no consumables-versus-instruments revenue split exists in seg_prod for any year after FY18 — the single most important disclosure for a razor-and-blades business is absent, and we flag explicitly that we cannot distinguish a paused instrument cycle from a flat installed-base run-rate. FY29E rests on 8 revenue and 5 EPS contributors; FY30E rests on 7 and 2 — the FY30 EPS of $8.845 is a two-analyst figure and is not used in valuation.
• GAAP versus non-GAAP: trailing GAAP EPS is $5.41, trailing non-GAAP (from earn_cal) is $5.15. All forward multiples use the non-GAAP est series, consistent with how the market prices the name, and are labelled as such.
• Fair-value method: scenario multiples applied to consensus FY27E EPS ($6.008, 12 analysts) and FY28E EPS ($6.932, 7 analysts). Bear 22x FY27E, base 29x FY27E, bull 36x FY28E. Stated arithmetic only — no discounted-cash-flow model was built, and none should be inferred.
• Not investment advice. Independent research, educational and informational only, never personalised.
• Version: 2026-08-04-full.