Illumina, Inc. ILMN

Key data

The business

Illumina sells DNA sequencers and, more importantly, the single-use flow cells and reagents those sequencers consume with every run. Razor and blades: the instrument sells once, the chemistry sells forever. It owns roughly 80% of high-throughput sequencing, and once a lab validates its assays on Illumina chemistry, ripping it out means re-validating everything, so the installed base is sticky. The flagship NovaSeq X, launched in 2023, drives the cost of a genome toward $200 and is the engine meant to pull labs onto higher-volume kits. Buyers are research universities, pharma, and clinical testing labs.

Two overhangs shape the story. GRAIL, the cancer-test acquisition Illumina was forced to unwind under antitrust orders, was spun off in 2024 and was the source of the billions in impairments that turned 2022 through 2024 into a bloodbath on the income statement; that is now behind it. And in 2025 China placed Illumina on its "unreliable entity list," closing a market that had been a high-single-digit slice of sales. In January 2026 it bought SomaLogic to push into proteomics.

The numbers

Revenue reaccelerated to its best rate in the set even as reported profit slipped.

Q2 2026 revenue rose 9.4% over the year before, the fastest here (Q4 2025 revenue and EPS backed out from the full-year figures). Net income fell to $207M from $235M only because a year ago the quarter carried $92M of other income against $15M this time; operating income actually rose to $245M from $214M, and non-GAAP EPS of $1.31 beat the $1.23 consensus. The seasonal shape matters: Q4 is the strong quarter, so the flat-looking $1.2B in Q2 is the real signal.

FY2025 revenue sits below the FY2022 peak: four years without growth, while earnings swung from billions in GRAIL losses back to $850M and $5.45. Consensus reaches $5.26B of revenue and $7.02 of EPS by FY2028, roughly 7% and 9% a year from here. At 40 times trailing and 31 times those 2028 numbers, you pay a growth multiple for mid-single-digit growth. The whole case rests on the consumable engine, so watch the split.

Product revenue up 7.7% with gross margin holding, despite tariffs and pricier memory chips the filing calls out, is the tell that the NovaSeq X installed base is starting to burn kits. The one thing that settles the bet is whether product growth stays high single digits for two more prints, which turns a comp into a flywheel.

Management

Insiders sold $505M across 90 sales in the last twelve months and bought nothing. The heaviest seller is Keith Meister of Corvex, the activist who sits on the board: three sales totaling $86M in August alone. Plan status is not disclosed, so read it as a board member trimming hard into strength rather than a scheduled drip. Capital allocation turned aggressive, $742M of stock repurchased in 2025 against $116M the year before, plus a fresh 2026 program authorized in April, buying back shares at about 40 times earnings, a rich price to pay with your own cash. Non-GAAP EPS has beaten consensus each of the last four quarters, by 6% to 15%.

How it fails or surprises you

The margin squeeze bites. The 10-Q names tariffs, costlier memory chips, and Middle East instability as active drags on 2026 gross margin. If those compound while China stays shut, the 66% gross margin that funds a fifth of revenue in R&D erodes, and the earnings recovery stalls even with revenue growing. Watch gross margin below 65% on the next two prints.

The reacceleration was a comp, not a trend. This is the fact the read explains least: revenue rose 9.4% in Q2 while GAAP profit fell 12%. If the growth is instrument placements pulled forward rather than durable kit consumption, product revenue reverts to low single digits by 2027 and the 31x multiple has nothing behind it. Product growth slipping under 5% would prove it.

The consumable flywheel compounds (right tail). If the NovaSeq X installed base keeps lifting kit volumes and SomaLogic opens proteomics as a second razor, revenue can beat the 7% consensus and margin can climb past 67%. The market is paying for a fixed hardware cycler, not a compounder; two prints of double-digit product growth would force the re-rate.

Closing thoughts

The next two quarters settle it. Product-revenue growth and gross margin will show whether the NovaSeq X installed base is really burning kits at scale or whether Q2 was just a comp. A print with product growth holding high single digits and margin above 66% converts the story from broken-and-fixed to compounder; a print with product growth back under 5% and margin sliding says the Q2 pop was a comp and the stock is expensive at 40x. The left tail is a China-plus-tariff margin grind that plays out slowly: survivability matters, and net debt is about one turn of EBITDA with free cash flow covering the buyback, so it endures a bad year. The downside linchpin risks a return to the flat, no-growth box the last four years lived in; the upside is worth a durable double-digit grower bought near its cheapest cash-profit multiple in a decade.

The bet is still that labs keep reading more DNA and keep buying Illumina's machines and the kits they burn, and that the NovaSeq X pulls enough of them onto those kits to grow again. What breaks it is product revenue reverting and gross margin cracking under tariffs and the China ban. The pair that tells you first: product-revenue growth and gross margin on the next two quarters.