Lam Printed Its Best Quarter Ever. The Stock Is Down a Third.
Record revenue, a first-ever 40% operating margin and a $200 billion 2027 WFE call — delivered into a tape that has spent three months selling the entire complex
LRCX · Earnings Call · 2026-10-21
The split screen
Lam's September quarter was, by the company's own reckoning, the best in its history: revenue of $8.22 billion, up 22% sequentially and 54% year-on-year — a fifth straight record — with gross margin of 52.4% and operating margin of 40.2%, the first time that line has cleared 40%. “Gross margin, operating margin, and EPS all reached new highs, with operating margin topping 40% for the first time.” — Tim Archer · 2026-10-21 Guidance for December is $8.6 billion ±$400 million at 52.5% gross and 40.5% operating margin.
And the stock is down 32.9% over the last 90 trading days, 39.8% below its June 30 peak of $433.33, leaving a $326 billion market cap. That single fact frames everything else. Management spent the quarter arguing that demand exceeds its ability to ship; the market spent the quarter de-rating the entire semiconductor capital equipment complex. The global tape is unambiguous about it: the AI data center theme that returned +53% as a cluster over 360 days has given back 5.6% over 90 days with **48 decliners against 6 advancers**. Lam's own record print is swimming against that current.
What is genuinely new: the storage hierarchy splits
The freshest idea on this call was not about layer counts. It was that AI is fragmenting NAND into distinct tiers — QLC for bulk capacity, TLC as the high-bandwidth storage tier for context memory and KV cache, and SLC emerging for low-latency agentic inference. The arithmetic matters: each percentage point of NAND bits shifting to SLC-class devices requires roughly 2% more installed wafer capacity for the same bit demand. That is a capacity multiplier sitting on top of bit growth.
While we used to just talk about layer counts as the driver for Lam equipment demand and wafer capacity needs, now you're really seeing this separation of storage tiers and each of the devices within NAND starting to be more specialized to different workloads within AI.
Note what this does to the old script. For five quarters the NAND story was a ~$40 billion installed-base upgrade to 200-plus class devices, and Lam has repeatedly said it was arriving faster than planned. On this call, Tim Archer effectively rolled the frame forward: the upgrade finishes in 2027, "and then you just roll to sort of the next upgrade cycle." The company needed a post-upgrade narrative and supplied one.
It also had to swat down a same-day scare. A large memory maker reportedly said it would prioritize 3D DRAM over 4F², and Lam shares fell. Archer's rebuttal was structurally sound: “if really we're making a leap right from 6F² into 3D DRAM, then etch and dep intensity is going to rise even faster, even sooner than we would have thought, and that's a great thing.” — Tim Archer · 2026-10-21 Whether the market believes that a 3D transition is unambiguously bullish for etch intensity is a separate question from whether it is true.
The SAM expansion case: panel, backside, Aether
Three concrete programs got more specific. Panel-level packaging: shipments to more than 20 customers across 300mm to 600mm panel sizes, with management now expecting panel-level share to exceed its wafer-level position. “we do not call it Sabre in this case, but there is a plating aspect to this and some other processes that I think will do extremely well within panel level.” — Tim Archer · 2026-10-21 That is a share-gain claim, not a market-growth claim — the more meaningful kind.
Backside power moved from roadmap to tool-of-record wins for all four products (Vector, Sabre, Flex, Halo) at one leading customer. Lam's keyword history shows Backside power cycling through its top-30 since 2024; this quarter it finally has wins attached. And Hybrid bonding surfaced as a fresh top-10 theme, with management positioning deposition and etch films as the enabler rather than the bonder. Aether — the dry resist platform — had its cumulative revenue outlook lifted to roughly $2.5 billion by decade end, about $1 billion above the prior target, ramping in HBM high-volume manufacturing alongside Akara in conductor etch.
The constraint flip — and the early 2027 call
Lam raised its calendar 2026 WFE estimate to the mid-$150 billion range and, unusually, pre-announced 2027 at more than $200 billion, with growth led by DRAM, then leading-edge foundry logic, then NAND. Doug Bettinger was explicit that this is early: “We normally do not give you this much specificity this early in the year, and we will give you a sharper pencil on that when we get to the earnings call next quarter.” — Doug Bettinger · 2026-10-21 Archer added visibility "through 2028." Compare January, when the same CFO refused to size anything: “It's hard for us to put a number on it, and I'm gonna decline to do that as we sit here right now.” — Doug Bettinger, Executive Vice President and Chief Financial Officer · 2026-01-28 Nine months later the number exists and is explicitly labelled conservative.
The tell is that the guidance is supply-limited, not demand-limited. “In the script, I said it would have been a little bit higher.” — Doug Bettinger · 2026-10-21 The bottleneck has also migrated: a year ago it was Clean rooms alone; now it is both clean rooms (ten new tier-one fabs layering in through 2027) and supply chain constraints on certain components. When Stacy Rasgon asked what unconstrained demand would look like, Bettinger declined — twice. The number is a floor, but it is a floor with a marketing function.
What quietly faded
Three soft spots are worth naming. First, CSBG: a fourth consecutive record at nearly $2.5 billion, up 40% year-on-year and tracking above 30% for the calendar year — but only "up slightly" sequentially, with NAND growth now split between new tools and upgrades rather than upgrades alone. The pure installed-base annuity framing is thinning.
Second, gross margin. In April, investors were told to “keep it roughly in the levels that we just guided you to in June. This is going to kind of level out at where it's at, I think, for the rest of the year.” — Douglas Bettinger, Executive Vice President and Chief Financial Officer · 2026-04-22 It did not level out. Gross margin has added roughly 10 points over 14 years and just printed 52.4%, with a mid-50s target now the explicit multi-year goal. That is a rare case of a company undershooting its own guidance in the right direction.
Third, China: 35% of revenue in September, up from 26%, and guided meaningfully down in December on domestic demand. This is now the third consecutive year in which "China down next year" has been promised and then not delivered — a point Tim Arcuri made bluntly on the year-ago call.
The bottom line
Underneath the noise, the operating machine is intact: operating margin has widened roughly 16 points over seven years, on a revenue base that has grown about 270% in seven years, funded by R&D that has risen about 172% over six years.
The market is pricing a cycle. The company is describing a build. Both have evidence: a 40% drawdown in 90 days says investors have seen this movie and know how it ends, while a $200 billion-plus 2027 WFE call, tool-of-record wins in backside power and panel-level packaging, and a NAND capacity multiplier from SLC tiering say the movie is longer this time. The honest resolution is January, when the >$200 billion gets its sharper pencil — and when we find out whether the constraint is the customer's clean room or Lam's own confidence.