科林研發 (LRCX) 2027 財年第一季法說會：創紀錄利潤率與 86 億美元營收指引

核心要點

• 科林研發 (Lam Research) 公布 9 月季度營收創歷史新高，達 82.2 億美元，季增 22%，年增 54%，連續第五個季度刷新營收紀錄。
• Non-GAAP 毛利率達 52.4%，營業利益率首次突破 40%，達 40.2%，稀釋後每股盈餘 (EPS) 亦創歷史新高，達 2.20 美元。
• 管理層預估 12 月季度營收為 86 億美元，上下浮動 4 億美元，毛利率預計為 52.5%，營業利益率為 40.5%（以各自指引的中位數計算）。
• 科林研發預計 2026 日曆年晶圓製造設備 (WFE) 支出將落在 1,500 億美元中段區間。其 2027 日曆年的初步展望為 超過 2,000 億美元，主由 DRAM 成長領軍，其次為先進晶圓代工／邏輯與 NAND。
• 記憶體佔 系統營收的 46%。NAND 營收季增約 40%，而 DRAM 與晶圓代工營收雙雙創下新高。
• 近期零組件短缺限制了 12 月季度的營收展望。管理層預計延遲的 12 月出貨將遞延至 3 月季度，並表示無塵室空間與產業供應限制可能會在 2027 年繼續限制成長。

主要財務數據

科林研發買回了約 10 億美元 的股票，並支付了 3.25 億美元 的股利。本季度公司在已授權的 股票買回計畫中尚有 30 億美元 的剩餘額度，並繼續以長期將 85% 的自由現金流回饋給股東為目標。

業務與營運績效

記憶體、晶圓代工與邏輯

記憶體占 系統營收的 46%，與上季持平，但按金額計成長 120%（與去年同期相比）。非揮發性記憶體上升至 系統營收的 26%（高於之前的 23%），受惠於轉向 256 層及以上 NAND 的轉換與對 384 層級元件的初始支出。

DRAM 占 系統營收的 20%，低於上季的 23%，儘管金額創下歷史新高。支出仍集中在支援 DDR5、LPDDR5 和高頻寬記憶體 (HBM) 的 1-alpha、1-beta 及 1-gamma 技術擴充與升級。

晶圓代工上升至 系統營收的 46%（高於前一季的 44%）。成長反映了先進製程投資、先進封裝以及中國在成熟製程上的支出。邏輯及其他業務下降至 8%（低於前一季的 10%），大部分支出用於先進製程節點。

AI 驅動的技術密度

管理層表示，AI 相關元件的複雜度正在提高 NAND、DRAM、晶圓代工/邏輯與先進封裝的沉積與蝕刻強度。科林研發預計 2026 日曆年的系統營收將成長 遠超過年增 40%，並預計將連續第三年超越晶圓廠設備 (WFE) 市場表現。

公司強調針對不同的 AI 儲存工作負載出現了 QLC、TLC 和 SLC 級別的 NAND。管理層估計，NAND 位元每向 SLC 級別元件轉移一個百分點，就需要增加約 2% 的已安裝晶圓產能 以滿足相同的位元需求。然而，該公司表示現在評估潛在的 SLC 產品組合還為時過早。

科林研發將其先進圖形化產品相關的預期累計營收調升至約 本十年末達 25 億美元，約高出 10 億美元，高於 其先前預期。管理層指出客戶採用度增加與產量上升，包括 HBM 製造。

先進封裝與客戶支援

先進封裝仍是主要的成長驅動力。管理層維持其預期，即大約 2026 日曆年成長 70%，並預期該業務在 2027 年將進一步成長。

科林研發已向超過 20 家客戶出貨面板電鍍系統，基板尺寸涵蓋 300x300 公厘至 600x600 公厘。管理層預期，其最終在面板級封裝的市佔率將超越目前在晶圓級封裝的地位，不過也表示面板級封裝的採用仍處於非常早期階段。

客戶支援事業群 (CSBG) 營收預期將成長 2026 日曆年成長超過 30%，為科林研發先前分析師日所設定長期成長目標的兩倍以上。管理層預測 2027 日曆年成長超過 15%，獲得備品、升級方案、服務、設備智慧化 (Equipment Intelligence) 與 Dextro 協作機器人的支撐。

地理區域營收比重

中國佔 總營收的 35%，高於 6 月季度的 26%，反映出本土與跨國客戶的支出。管理層預計中國的營收貢獻將在 12 月季度顯著下降，主因是本土客戶支出減少。

韓國維持在 20% 的營收比重。台灣則下滑至 18%，低於上一季的 27%。

管理層財務預測

管理層表示，若非短期供應限制，營收展望本可稍微更高。儘管面臨客戶組合帶來的逆風，毛利率預計仍將適度改善。Non-GAAP 稅率預計將維持在 9 月季度的水準，約為 16.5%，且 12 月季度及未來可能繼續維持該水準。

針對 2027 日曆年，科林研發對晶圓廠設備 (WFE) 的初步預測為 超過 2,000 億美元，相較於 2026 年預期的 1,500 億美元中段水準。預計 2027 年將有 10 座新的第一線 (Tier 1) 晶圓廠上線營運，不過其貢獻將在全年逐步顯現。

風險與觀察重點

• 供應鏈限制： 特定零組件限制了 12 月季度的營收展望。管理層預期能解決這些瓶頸，但並未量化受影響的營收金額。
• 無塵室產能：有限的無塵室空間限制了 2026 年的產業成長，且在 2027 年可能仍是限制因素。
• 設備供應鏈的執行情況：2027 年晶圓廠設備 (WFE) 前景取決於設備到貨狀況、晶圓廠建設時程，以及客戶招聘與培訓員工的能力。
• 中國營收占比：隨著中國本土客戶支出下滑，預計中國市場的貢獻將在 12 月季度顯著下降。
• NAND 技術轉型：升級需要客戶將設備停機，這可能在產能利用率高的時期帶來排程上的挑戰。

分析師 Q&A 重點

• 受限出貨的時間點：管理層表示，因時程問題或零組件供應狀況而未能於 12 月出貨的產品，預計將遞延至 3 月季度。
• DRAM 架構改變：科林研發 (Lam) 預期，無論客戶是繼續推進現有 DRAM 架構微縮，還是邁向 3D DRAM，蝕刻與沉積強度都會增加。更快向 3D 結構過渡可能會進一步提升該強度。
• NAND 升級週期：從 128 層轉換至 200 層等級元件的升級預計將持續至 2027 年，隨後將進一步升級至 300 層與 400 層元件。升級與新設備均對 NAND 成長做出了顯著貢獻。
• 毛利率軌跡：管理層預計近期毛利率將維持在當前水準附近。未來幾年內，科林研發將繼續透過差異化產品、就近客戶製造、成本改善與定價策略，以 50% 中段區間為目標。
• WFE 上行空間與限制因素：管理層指出，若不受限制，2027 年的需求將高於目前對 WFE 的預測，但拒絕量化差距。公司計劃在下一季法說會上調整細化展望。
• 2027 年表現超越大盤的驅動因素：管理層強調了沉積與蝕刻強度的提升、先進封裝、全環繞閘極 (GAA) 架構、背部供電、先進微影以及潛在的 NAND 產能擴充。

完整法說會逐字稿

---

完整財報電話會議逐字稿

管理層陳述

Operator

Good day, and welcome to the Lam Research Corporation September 2026 Earnings Conference Call. [Operator Instructions] After today's presentation, there will be an opportunity to ask questions. [Operator Instructions]. Please also note, today's event is being recorded. I would now like to turn the conference over to Ram Ganesh, Vice President, Investor Relations. Please go ahead.

Ram Ganesh

Thank you, and good afternoon, everyone. Welcome to the Lam Research Quarterly Earnings Conference Call. With me today are Tim Archer, President and Chief Executive Officer; and Doug Bettinger, Executive Vice President and Chief Financial Officer. .

During today's call, we will share our overview on the business environment, and we'll review our financial results for the September 2026 quarter and our outlook for the December 2026 quarter. The press release detailing our financial results was distributed a little after 1:00 p.m. Pacific Time. The release and the accompanying presentation slides for today's call can be also found on the Investors section of the company's website.

Today's presentation and Q&A include forward-looking statements based on our current beliefs, expectations and assumptions. These statements are subject to risks and uncertainties, and actual results could differ materially from those expressed or implied in such statements.

For a discussion of factors that could cause actual results to differ materially, please refer to the risk factors in our most recently filed periodic reports on Form 10-K, on Form 10-Q and subsequent filings with the SEC and the cautionary statement in the accompanying presentation slides.

Today's discussion of our financial results will be presented on a non-GAAP financial basis unless otherwise specified. A detailed reconciliation between GAAP and non-GAAP results can be found in the accompanying presentation slides. This call is scheduled to last until 3:00 p.m. Pacific Time. A replay of this call will be made available later this afternoon on our website.

And with that, I'll hand the call over to Tim.

Timothy Archer

Thanks, Ram, and thank you to everyone joining us today. In the September quarter, Lam again delivered record revenue and profitability, gross margin, operating margin and EPS all reached new highs with operating margin topping 40% for the first time. Our performance throughout the year reflects the broad strength of our product portfolio and a relentless focus on operational execution as AI continues to drive unprecedented semiconductor demand and increasing device complexity, including the midpoint of our December quarter guidance, we expect calendar year 2026 systems revenue to grow well over 40% year-on-year, rising deposition and etch intensity across all device types and advanced packaging solutions is continuing to drive greater opportunity for Lam.

Over the past few years, increasing device complexity and a push towards greater 3D scaling have driven a 4-point expansion of our Served Available Market, or SAM, as a percentage of WFE. When combined with share gains from products addressing this growing SAM, we expect 2026 to mark our third consecutive year of WFE outperformance.

For calendar year 2026, we expect WFE in the mid-$150 billion range. Our preliminary 2027 WFE outlook is for more than $200 billion of spending with percentage growth led by DRAM, followed by leading-edge foundry logic and NAND.

As we move into calendar year 2027, we see clean room space availability and supply chain constraints for certain components continuing to limit chip makers and equipment companies' ability to fully meet end demand. We are actively working with our suppliers to accelerate their capacity expansion plans and are confident Lam is well positioned to deliver on the strong demand we see ahead.

Underpinning our outlook for WFE growth is the continued rapid adoption of AI by enterprises and consumers. It is reported that average weekly token usage is up roughly 250 fold in the past 18 months. As AI moves increasingly towards autonomous Agentic workflows, demand for compute, memory and storage solutions continues to expand.

For example, in NAND, AI performance and economics are driving demand for multiple tiers of storage, each optimized for different workloads. QLC today provides cost-effective storage for rapidly growing AI data volumes. TLC storage is used as a high-bandwidth tier for context memory and KB cash workloads. And SLC class NAND is emerging for low latency, high-speed access in Agentic AI.

We believe adoption of an SLC storage tier could become a meaningful driver of NAND wafer capacity. Converting from TLC to SLC class devices reduces bits per device by roughly 3x. Said differently, for each percentage point mix shift of NAND bits to SLC class devices, approximately 2% more installed wafer capacity will be needed to support the same bit demand. Incremental growth in bit demand would further amplify wafer capacity requirements.

We expect all 3 storage tiers to grow, increasing the value and strategic importance of NAND. We believe Lam is best positioned to benefit from this inflection. We have the industry's largest installed base of NAND systems and a comprehensive product portfolio spanning all critical deposition and etch steps.

Any increase in NAND bit demand whether satisfied through installed base upgrades or capacity additions will generate additional need for Lam's tools and technology. Foundry logic and DRAM are seeing a similar device level evolution to satisfy AI's emerging performance requirements. From advanced patterning and backside power to next-generation packaging, we see billions of dollars of potential SAM expansion and share gain opportunities for Lam in the years ahead.

We are seeing the impact of our targeted investments in foundry logic and DRAM products in our calendar year 2026 results. These 2 market segments have accounted for approximately 80% of Lam's systems revenue growth year-to-date. A good example is our focus on advanced patterning requirements for foundry, logic and DRAM. With each technology node bringing more stringent EUV pattern fidelity and defectivity requirements, Lam's Acara etch, conductor etch and ether dry resist tools are being paired to deliver unprecedented scaling capability.

Importantly, these tools are also helping improve the productivity of DRAM scaling at a time when maximizing fab output is a critical priority for our customers. Multiple chipmakers are adopting Acara for its ability to etch very high aspect ratio features with exceptional profile control, while at the same time, delivering up to a 50% reduction in wafer-to-wafer and lot-to-lot variability and up to a 50% gain in wafer output per fab footprint.

Similarly, Lam's ether dry resist solution has demonstrated sub-10-nanometer resolution at the highest pattern density in fidelity with high NA EUV. It helps achieve this result while requiring 10% lower exposure dose, improving scanner productivity and reducing patterning cost per wafer. Ether solutions are already ramping in HBM high-volume manufacturing with low NA EUV, and we see additional insertion opportunities ahead across both DRAM and foundry logic customers.

As a result, we now expect cumulative EPA-related revenue to reach approximately $2.5 billion by the end of the decade, roughly $1 billion above our prior expectations. Backside Power is another example of where Lam's strength and deposition and etch is evident. Separating power and signal routing allows for thicker backside metallization to reduce resistance and improve power efficiency.

Implementation increases process complexity, driving the need for more deposition, plating and etch steps. As customers ramp backside power at 2-nanometer and below, our production-proven vector, Sabre and flex tools are well suited to the task. Vector provides industry-leading footprint density and wafer-to-wafer repeatability with its unique quad architecture.

Sabre is the global industry benchmark for electroplating through its high productivity and superior on-water performance. Flex delivers high throughput dielectric etch with low defect levels leveraging its differentiated small volume chamber design and confined plasma. New materials such as molybdenum are also being introduced to enable more efficient electrical connections between the backside and front side of the wafer.

For this emerging application, our Halo tool is gaining traction by enabling both plasma and thermal processes within a single chamber while delivering industry-leading productivity. We are gaining share in this expanding backside power SAM opportunity and recently secured tool of record wins for all 4 products at a leading customer.

In advanced packaging, our market-leading Syndion product is winning advanced logic and HPM TSV applications. Syndion's differentiated hardware design enables rapid process switching to deliver highly uniform TSB profiles, smooth sidewalls and high-volume productivity. Recently, we attained 4 new positions across multiple customers.

Our Stryker ALD product also achieved multiple TSV liner wins due to the film's superior step coverage and low etch rate. Beyond TSV applications, we secured additional Vector and TS 3D wins for high-energy bonding films and interdi gapfill applications by delivering lower defectivity for thick film deposition. With its high deposition in etch intensity, advanced packaging has become a major growth driver for Lam.

For next-generation AI and high-performance computing devices, customers are increasingly turning their sites towards panel-level packaging to scale manufacturing. Our panel level packaging portfolio continues to gain momentum across fan-out packaging, glass core substrates and emerging substrate and interposer architectures.

To date, we have shipped panel plating systems to more than 20 customers worldwide across panel sizes ranging from 300 by 300-millimeter to 600 by 600-millimeter. Customers continue to select our Kalisto platform for its differentiated colinearity and planting uniformity performance across bumps, pillars and redistribution layers.

Kalisto is also well positioned to benefit from increasing adoption of glass core substrates through emerging through glass via applications. Given our customer engagement, installed base growth and technology leadership, -- we expect to achieve panel-level packaging market share that surpasses the strong position we already hold in current wafer-level packaging. And finally, turning to CSBG. We are set to exit calendar year 2026 with solid momentum with annual revenue expected to grow more than 30% year-over-year. This is more than twice the rate of our long-term target set at our last Analyst Day.

Growth is broad-based across upgrades, spares and services supported by high fab utilization and customers' continued focus on installed base efficiency. The value we are delivering with Equipment Intelligence and Dextro cobots recently helped us secure performance-based service contracts with multiple memory manufacturers. We expect CSBG momentum to continue in calendar year 2027 with revenue growing over 15% year-over-year.

So to wrap up, -- we believe the industry is in the early stages of a multiyear investment cycle with customer visibility extending through 2028, increasing manufacturing complexity, new materials integration and advanced packaging are driving higher deposition and etch intensity across device segments. Our technology leadership, growing installed base and strategic investments in product innovation and manufacturing capacity position Lam to capitalize on these inflections and drive sustained outperformance. Thank you, and here is Doug.

Douglas Bettinger

Great. Thank you, Tim. Good afternoon, everyone, and thank you for joining our call today during what I know is a very busy earnings season. We were pleased with our continued strong execution in the September quarter. resulting in our fifth consecutive quarter of record revenue as well as record gross margin, operating margin and earnings per share. September quarter revenue was $8.22 billion, which was up 22% sequentially and up 54% from the same period in 2025.

The deferred revenue balance at quarter end was $2.66 billion, which was an increase of $227 million from the June quarter. This grew primarily due to increased sales volume and it was partially offset by a decrease in customer down payments. From a market segment perspective, September quarter systems revenue in memory was 46%, which was consistent with the prior quarter.

On a dollar basis, this represented a record level for memory, and it was up 120% from the same period in 2025. Within memory, nonvolatile memory accounted for 26% of our systems revenue, up from the June quarter's level of 23%. On a dollar basis, NAND revenue increased approximately 40% sequentially as the industry continued spending on conversions to 256 layer and above class devices.

We are seeing initial spending on 384 layer class devices. DRAM revenue reached a record growing sequentially on a dollar basis, but down a little bit on a percentage basis from 23% of systems revenue in the June quarter to 20% in the September quarter. DRAM spending remained focused on technology additions and upgrades across the 1-alpha, 1-beta and 1-gamma nodes that is enabling DDR5, LPDDR5 and more importantly, hot bandwidth memory.

Foundry also achieved record revenue. Its strength -- excuse me, its share of systems revenue increased to 46% from 44% in the June quarter. Strength in the quarter was driven by investment in leach nodes, including advanced packaging as well as mature node spending in China. And finally, Logic and Other were 8% of our systems revenue in the September quarter, slightly down from the prior quarter's level of 10%.

Most of the spending here was targeted at leading-edge process nodes. Let me now discuss the regional composition of our total revenue. China contributed 35% of revenue, up from the June quarter level of 26%. We saw strong investments here by both domestic and multinational customers. In the December quarter, we expect China will represent a significantly lower percentage of our total revenue compared with the September quarter. Korea's contribution to revenue in the September quarter remained unchanged from the June quarter level at 20%.

The Taiwan region contributed 18% of revenue down from the June quarter at 27%. The Customer Support Business Group generated a fourth consecutive quarter of record revenues at nearly $2.5 billion in the September quarter. This was up slightly from the June quarter and 40% higher than the same period in 2025. Each component of CSBG remained strong with the prior quarter.

Turning to the gross margin performance. The September quarter came in at an all-time record of 52.4%, improving from the June quarter level of 52% due to favorable customer and product mix. Additionally, we continue to balance our cost and pricing postures. Operating expenses for the June quarter were $1 billion up from the prior quarter amount of $916 million. The increase was primarily driven by higher incentive compensation due to our increased profitability levels.

We are also investing incremental resources to support growing programs like Akara, Halo, panel form factors and Dextro. R&D accounted for nearly -- or excuse me, approximately 2/3 of our total operating expenses. Operating margin for the September quarter was another record at 40.2% compared to the June quarter level of 38.4%. This improvement reflected stronger gross margins and continuing operating leverage as our business scales.

The non-GAAP tax rate was 16.5% for the quarter compared to 11% in the June quarter and in line with our expectations. The increase was primarily driven by a higher U.S. tax rate on foreign earnings as well as expected shifts in the geographic mix of income. We expect the tax rate to remain in this range for the December quarter and likely beyond.

I'd encourage everyone to make sure they have their earnings models updated to this higher tax rate. Other income expense for the September quarter was approximately $16 million in income compared with $19 million in expense in the June quarter. The improvement in OI&E was due to foreign exchange fluctuations, higher interest income as well as favorability in our venture portfolio.

And as we've talked about in the past, you should expect to see some variability in OI&E quarter-to-quarter dictated by market factors. For the September quarter, diluted earnings per share were a record at $2.20. The diluted share count was roughly 1.26 billion shares, which was relatively unchanged with the June quarter.

Let me pivot to capital return. We allocated approximately $1 billion to share repurchases, and we paid $325 million in dividends in the September quarter. During the quarter, we entered into an accelerated share repurchase structure that will continue to execute in the December quarter. We returned approximately 41% of free cash flow in the quarter and 59% in the calendar year-to-date.

I would point out that we also used $750 million to pay down maturing debt during the year. Our long-term plans remain to return 85% of free cash flow to shareholders. We have $3 billion remaining on our Board-authorized share repurchase program. Our cash and short-term investments totaled $7.5 billion at the end of the September quarter, which was up from $5.6 billion at the end of the June quarter.

The factors behind the increase were cash from operating activities partly offset by our capital return activities as well as capital expenditure. The September quarter was also a record for our free cash flow, coming in at $3.24 billion in the quarter. And as I mentioned last quarter, we've been increasing our cash levels a little bit to support growing business volumes as well as capital investments in this growing demand environment.

Day sales outstanding were 63 days in the September quarter, a decrease from 72 days in the June quarter. Inventory at the September quarter end totaled $4.6 billion, which was an increase of $274 million from the June quarter as we're building inventory to meet the expected future demand for our products.

Nonetheless, inventory turns continued to strengthen to 3.4x versus 3x in the prior quarter. I would just mention this was the highest level of inventory turns in nearly 7 years. Our noncash expenses for the September quarter included approximately $115 million in equity compensation, $106 million in depreciation and $16 million for amortization.

Capital expenditures in the September quarter were $258 million, which was up from the June quarter at $189 million. Capital spending grew as we continue to increase our lab infrastructure and expand our manufacturing facilities globally. We ended the September quarter with approximately 24,300 regular full-time employees, which was a growth of approximately 1,900 people from the prior quarter.

Headcount increased primarily in our factory and field teams to support expanding manufacturing activities and growing tool installations. We also added head count in R&D. Now let's turn to our non-GAAP guidance for the December 2026 quarter. We're expecting revenue of $8.6 billion, plus or minus $400 million. I would mention that this revenue -- this revenue guide would have been a bit higher, if not for some near-term supply chain constraints that we are actively managing.

We're expecting gross margin of 52.5% plus or minus 1 percentage point. This is improving a little bit in spite of headwinds related to customer mix. We're expecting gross -- operating margin of 40.5%, plus or minus 1 percentage point; and finally, earnings per share of $2.33, plus or minus $0.15, based on a share count of approximately 1.253 billion shares.

So let me wrap up. We continue to execute our strategy for outperformance delivering another quarter of record revenue and profitability. Operating margin exceeded 40% for the first time in our history, and we're guiding for this to expand slightly. We're increasing R&D investment to strengthen the already strong product portfolio that we have. We're also investing in the infrastructure needed to support the growing demand we expect in the future.

These activities continue to deliver a return on invested capital level that is comfortably above the competition. Operator, that concludes our prepared remarks. Tim and I would like to open up the call for questions.

Operator

[Operator Instructions] Today's first question comes from Timothy Arcuri at UBS. Please go ahead.

分析師問答

Timothy Arcuri

Doug, I don't think I've ever seen a quarter where you grew systems 35% Q-on-Q. So kudos on that. You did talk about the near-term constraints I guess you can see that from payables. I think payables were up like $1.3 billion. So can you just give us a sense? Is it fair to say this is costing you like $300 million to $400 million. Is that like a reasonable number to assume that, that's costing you in the guidance for December. And do you get that back in March? So do you get back on this sort of 10% plus or minus systems growth in March, does that actually -- do you actually get that back?

Douglas Bettinger

Yes, Tim, that's a good question. I'm not going to put a specific number on it. In the script, I said it would have been a little bit higher. So it would have been a little bit higher. And yes, Tim, to the extent that it's something we would have shipped in December because of timing, we will ship that in the March quarter.

Timothy Arcuri

Okay, Doug. And then -- can you just talk about -- there was a call hosted by a big memory company today, and they were talking about 4 square. And they were saying that they're going to prioritize 3D DRAM rather than 4 square and your stock traded down, which was a little bit odd to me. It seems that people don't really understand what your position is there. So Tim, I wonder if maybe you can talk about that. Can you talk about how you do for -- in like a 3D world versus like a 4F square world? And sort of what you're seeing in terms of your customers' DRAM road maps? .

Douglas Bettinger

Sure. Obviously, we don't know exactly what the reaction was on why people would have reacted in that way because we've said for a long time that as all device architectures go 3D, etch and deposition intensity rises. Now I would say that in some ways, interestingly, the longer it takes to get there, the more we're seeing momentum for some of our other scaling activities.

And so we focused on this -- even in this call, advanced patterning, if you want to extend the current architecture in which devices are getting -- patterns are getting even smaller. That's why we're starting to see this uptick in costs like ether, processes like a car for conductor etch. The processes that are needed to etch and create those smaller structures. So I would say in the short term, etch and depth intensity arising -- and if really, we're making a leap break from 6 square to 3D DRAM, then etch and depth intensity is going to rise even faster even sooner than we would have thought, and that's a great thing.

Operator

And our next question comes from Melissa Weathers at Deutsche Bank.

Melissa Weathers

I wanted to also ask on the supply constraints. I think you said you're working with your slicing to make sure some of those bottlenecks get resolved. Can you just give a little more color on what actions you're taking? Could you consider any co-investments or help on the capital side of those guys, just what needs to happen to relax those constraints. .

Douglas Bettinger

Listen Melissa. Yes, I mean we're always working with our supply chain to make sure they're ready for the demand environment that we see. But given how strong things have been, we're seeing a little bit of tightness. I'm not going to get into specifically where it is. except I would tell you that what we're managing, just like we always do, we'll continue to do that, but it just got a little bit tight in the quarter.

I have every expectation will effectively manage this as well as anybody in the industry will -- and yes, if we needed to make some level of investments, we'd be open to that. But frankly, a lot of our suppliers are well capitalized and can manage on their own from a financial standpoint. Tim, would you add anything?

Timothy Archer

No, I think that's correct. I mean for those of us who around during the COVID shortages, this is nothing like that. We learned a lot of lessons during that period. And in fact, just in the last 2 months, we've hosted a couple of big events with our global supply -- global suppliers to really lay out what we have in terms of expectations for growth into the next couple of years.

And where you can run into problems is -- within our own products, as we talked about etch and depth position intensity increasing in certain parts of our business, like we talked about advanced packaging, for instance, really being a major driver of growth for Lam. You can find certain suppliers that are seeing percentage growth of our demand to them rising even much faster than Lam itself is growing.

And so yes, I think we have been very clearly communicating that to those suppliers and really asking for transparency to the help they need to keep up with the pace that we're driving in the industry. And I think we'll be caught up very shortly.

Melissa Weathers

And then on the NAND spending cycle and that upgrade opportunity that you guys have been talking about, you pulled it in last quarter. And then it's nice to hear about the 384 layer devices that I think are starting to -- you guys are shipping to. So -- can you update us on where you -- like what inning you think we're at in that NAND upgrade cycle? What trends are you seeing? And yes, maybe any forward-looking comments on the SLC and some of those new opportunities.

Timothy Archer

Yes. I think in terms -- there's still room to go on the upgrade cycle. And in fact, when you're in a period of high demand actually taking tools offline to do the upgrade, sometimes it's also a scheduling challenge for customers and for us. But I'd say that we're working our way through that, and we said that we would be finishing that in 2027. And so -- but we also pointed out that then you just roll to sort of the next upgrade cycle. And so remember, that upgrade was from the 128 layer to kind of into the 200 layer class devices.

And as you just mentioned, people are already moving into the 300 and 400 layer devices, we've been engaged on those in R&D for quite some time. But I think the real thing we wanted to point out today is that 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.

And that actually, just like it did in DRAM with HBM and the impact on bits that could be manufactured per wafer, -- you're starting to see the performance needs of AI start to create a need for higher performance, low-latency devices. And I think we've just set a customer took in some detail about that opportunity. I think that's going to drive NAND wafer capacity demand and intensity for Lam equipment.

Operator

And our next question today comes from Krish Sankar with TD Cowen .

Sreekrishnan Sankarnarayanan

I had 2 of them, Doug or Tim, when I look at your Q4 guide, I understand you're talking about mid $150 billion WFE, but the implied WFE for Q4 is like it was $180 billion or so. I'm just wondering, are there any episodic revenues in Q4? Or is some of the price increases that the semi front-end semi equipment companies are doing impacting Q4 and kind of making the WFE number larger. And then I have a quick follow-up. .

Douglas Bettinger

Yes, Krish. No, there's nothing unique or one-off or episodic to use your word in the December it's just consistent with customer demand, frankly. And like I alluded to, it would have been a little bit higher, if not for some of the supply chain constraints, but there's not anything one-off. .

Sreekrishnan Sankarnarayanan

Got it. Doug. And then I guess a quick follow-up. I think in your prepared comments, you mentioned China would be significantly lower in December versus September, kind of wondering, can you quantify that? How will the mix look like? Because it seems like September quarter was a surprise on the upside with both multinationals and domestics coming in.

Douglas Bettinger

Yes, Krish, it wasn't a surprise. It was exactly as expected. And yes, I mean the domestic customers attempt in September were strong, but so were the global multinationals. I expect it in December will be meaningfully down, driven by the domestic customers in China. I expect in the December quarter, the multinationals in China likely continue to be reasonably strong, but the percent of revenue in China will be meaningfully down in the December quarter. .

Operator

Next question comes from Harlan Sur at JPMorgan.

Harlan Sur

On CSBG, 36% growth for 9 months of this year, full year outlook greater than 30%, very, very strong. The growth tailwinds, I assume higher spares as customer utilizations are at record levels, higher sales of higher dollar value upgrades for NAND and maybe strength in Reliance, right, as the mature sectors of the markets start to come back.

And of course, you have your intelligent services offerings. Can you guys just rank order these growth drivers? And more importantly, on a year-over-year basis, how much higher are CSBG operating margins? I'm just wondering if mix and volume dynamics within CSBG are contributing to the strong overall operating margin profile?

Timothy Archer

Yes. I'll let Doug talk about the rank order in terms of financial impact in the short term. But if I were to rank order them in terms of my excitement, clearly, as we've talked about, we see equipment intelligence, the cobot driven maintenance, the performance-based service contracts we talked about as really kind of the new part of our CSBG business that's really catching on.

And we're in an environment right now where just as there was an earlier question about if people are trying to shrink devices even faster to get more bits or more performance out, that creates an intolerance to variability in the process, whether that's in the process control of the maintenance.

And that's exactly where equipment intelligence and cobots come into play. So we're seeing strong pull for that element, and that's something new for our CSBG business -- and so I'm very excited about that becoming a bigger growth driver for us than it has been in the past. And I'll let Doug talk about upgrade spares and everything else. .

Douglas Bettinger

Yes, Harlan, we don't specifically give you hard numbers on each of the individual components -- what I would tell you relative to strength of each of them, though, given how strong utilization in the industry spares and service are just coming along as you would expect. I've always said spares is the biggest individual component, and that continues to be true. But I'm not going to be specific on the other ones. What you alluded to is right, given where we add in the NAND conversion/upgrade cycle, that portion of the business continues to be strong.

And yes, Reliant did quite nicely last quarter, given the strength of investment in mature node spending. And I expect you'll see all of these trends continue into next year. .

Harlan Sur

Yes. I appreciate that. And then as a follow-up, the supply tightness in leading-edge foundry, logic is creating tightness in mature and specialty analog or microcontroller, foundry supply here is tightening very rapidly, right? Looks like mature foundry suppliers, internal IDM factories, they kind of all increasing capacity -- you just said that this is reflected in your Reliant business. But if I exclude China, are you anticipating this segment of your customer base to grow this year? .

Douglas Bettinger

Probably not in a meaningful way this year, Harlan. I do expect as we get into next year that mature node spending outside of China will be strong because of the dynamics you were just referring to. I'm not sure that it's really sticking out quite yet in 2026. .

Operator

Our next question today comes from Srini Pajjuri with RBC Capital Markets. .

Srinivas Pajjuri

Thank you. Tim, I'm looking at your systems business growth. I think last year, you were up 40%. And then this year, you're probably going to grow 45%, both well about the WFE growth. So looking out to next year, given your $200 billion plus outlook, roughly 30%, let's say, I suspect you expect to outgrow that market next year as well. Can you talk about what are the drivers? Is it driven by any particular end market like NAND or DRAM or any product line or anything else you can comment on would be helpful.

Timothy Archer

Yes. I guess I would -- at a very general level, I would say it's driven by the increase in deposition and etch intensity across all devices. And if there are a few things I'd point out just as maybe obvious trends in the market. Lam's exposure in the advanced packaging space is extremely high.

And I think advanced packaging continues to play an even more vital role in just about every device that's feeding into AI today. And so that's, I think, a great position. We will see outperformance and we talk even about that extending in the panel level. But it really is in every device. I mean you see NAND layers growing, you see new architectures coming demand.

You see -- we talked a little bit about the emergence of SLC Class NAND you're starting to hear from our customers. foundry/logic, we've talked about success and gate all around. We invested 4 or 5 years ago in an entire set of products focused on improving our position in foundry, logic and DRAM. And I commented on the fact that that today, those products are driving a significant portion of our growth here in the years in which foundry, logic and DRAM are leading the charge on growth. And so we're extremely happy with that.

NAND is yet to emerge as to its full potential from our perspective -- and I think you're starting to hear about it and the strategic role that can play in AI. And so I don't know, it's hard to project exactly. We'd love to see all the devices grow, but I think the single greatest theme is regardless of of which device we're talking about deposition that will play a greater role in 2027.

And if you think about -- we had this model, which was the $100 billion of AI CapEx being about $10 billion of WFE, but the important thing is that $10 billion next year, Lam's opportunity and that $10 billion would be even greater than it is this year because of the device evolution. And so I think that's intensity is just the way to think about Lam's opportunity.

Srinivas Pajjuri

Right. And then maybe on the gross margins, Doug, you're guiding actually a slight increase despite -- you highlighted some customer mix headwinds. I'm suspecting that's related to China. So I guess, on a normalized basis, as we look out to the next few quarters, once China normalizes, how should we think about gross margins? I mean, in the past, you did talk about some pricing actions. And of course, you have the cost and efficiency improvements as well. So just curious as to how we should think about gross margins as we go into next year. .

Douglas Bettinger

Yes. Listen, in the near term, Srini, I think you're going to be plus or minus the level that we're at. We did suggest over the next several years, we see an opportunity to expand gross margin in the mid-50s. So kind of triangulate yourself between these 2 points. But in the near term, it will bounce around generally where it is. And yes, the things that contributed to our very strong gross margin as well as our expectation of how we're going to drive improvement into the future, our new product introductions, right? .

We've got some very nice technically differentiated new products like Akara and Ether. We're excited about where that's all going to trend from a profitability standpoint. We've talked about our close to customer strategy relative to how we're manufacturing closer to where the fabs are and that's contributed and still have a little bit of use left to squeeze there. And then we've also talked about the fact that we're working on pricing. All of those things have contributed to getting us the all-time record level that we printed in the September quarter and levers we're going to keep trying to pull as we go into the future to get to that mid-50s that we've said we are going to drive towards.

Operator

Our next question today comes from Vivek Arya with Bank of America Securities. .

Vivek Arya

For the first one, if I look at last year, you were able to kind of consistently expand your WFE expectations, I think you started the year at mid-130s. Now it's like mid-150s. So I'm curious what helped you increase WFE expectations, right, last year? And then what are the constraints next year that you think can be resolved to help do better than the 30% or so industry growth. I know you mentioned some constraints in the near term. I'm curious if they are the same constraints that limit industry growth for next year or they are different. Just what can help you do better than this 30% industry WFE growth? .

Douglas Bettinger

Yes, Vivek, I'll start, and then I'll let Tim add. As we've walked through this year, you're absolutely right. We started mid-130s, and then we went to a little bit more than 140 and then a little more than $150 million and then we just took it to mid-150s. Basically, what we saw happening across all aspects of our customer base is everybody squeezed a little more out of the clean rooms that they had and/or pulled projects and a little bit quicker than maybe they originally intended.

That's generally what happened. And frankly, we did exactly the same thing. And with the more than 200 number, we're giving you our initial starting point more than 200. We normally don't give you this much specificity this early in the year, and we'll give you a sharper pencil on that when we get to the earnings call next quarter. .

Vivek Arya

Got it. And for my follow-up, Doug, you mentioned something about an accelerated repurchase plan. I think you can all guess what that is due to, but would help to hear from you also and if you could help give some sense of what the magnitude is, how long you plan to continue with? I know you mentioned you have $3 billion left in the buyback. But just how much is the ASR component, how long it continues?

Douglas Bettinger

Yes. I mean, the way these accelerated share repurchase structures work is you enter into them, you get some shares upfront, and then they execute over time depending on where the share price is and volatility and things like that. I'm not going to disclose actually the start or end date because there's ranges around it. I don't like people to know what we do and how we do it, except I will reinforce for you. It's going to continue to execute in December. .

And then we will likely be doing other things in the market relative to the buyback, right? Our medium long-term goals remained unchanged, which is to return 85% of free cash flow to shareholders.

Operator

And our next question today comes from Atif Malik with Citi. .

Atif Malik

The first 1 for Tim. Tim, I thought your panel-level packaging comments are extremely bullish you're talking about a share that will exceed the wafer-level packaging. Can you just kind of double click in terms of the industry ramp on panel level packaging and your products are the parts the same as wafer level like Sabre that will drive this growth? .

Timothy Archer

Yes. I mean at a fundamental level, they're similar. I mean, I think part of our optimism in this space is, one, it makes sense. I mean, obviously, we talked on the last -- I think, on the last earnings call about how packages themselves are getting larger, which, in many cases, is -- they're too big for wafer-level packaging.

And so panels make sense as you try to assemble bigger and bigger system-level solutions. So we saw that quite a while ago, and we've been investing in panel for several years. And I would say we're still going in the very, very early stages. I mean it's -- wafer level packaging is the predominant method -- but as we talked about the shipments to more than 20 customers, you can think of those as really the planting the seeds for an industry that I think is going to ultimately assume a fair bit of the packaging work for the industry.

The processes that we talked about, we don't call it Sabre in this case, I mean, but there's a plating aspect of this and some other processes that I think we'll do extremely well in within panel level.

Atif Malik

Great. And then Doug, on the $200 billion plus basis type equipment outlook for next year, your rank orders DRAM, foundry/logic and NAND, which is the same as 90 days ago. But has the gap between DRAM and foundry/logic expectations narrowed given some of the announcements on the new logic fabs from a foundry and tariff fab like?

Douglas Bettinger

It, they really haven't. And as you might expect, all of these projects perhaps that people are now publicly hearing about, we knew about in advance of that. So it really hasn't changed much. .

Operator

Our next question today comes from Jim Schneider at Goldman Sachs.

James Schneider

I was wondering as you look into 2027, can you maybe frame for us some of the moving pieces that could adjust the WFE outlook higher or lower? How much of it really is your customers' individual spending plans for the different technology or product areas versus just the general availability of clean room space overall? .

Douglas Bettinger

Yes, Jim, I'll start, and then I'll let Tim add. When we look into next year, we see 10 new Tier 1 fabs coming online between the beginning of the year and the end of the year and they will layer in over time. So obviously, that's a contributor. This year has been primarily clean room constrained. In that in, you see, like we said, some emerging supply chain constraints that we're confident we're going to be able to work through effectively, but the whole industry is going to have to work through very similar things. .

And so all this at the end of the day has to get balanced. Every equipment company needs to be able to supply things. clean rooms need to come on, on schedule. People need to hire and train to make sure they're already -- the industry is good at doing this, and we're great at doing it. But all of those things have to come into play when you're thinking about what the next year is going to look like. .

Timothy Archer

Yes. And I think in my prepared remarks, I talked about the fact that we have visibility through 2028. That's really from the demand side. And so I think that where we're saying it might just be a little bit early to give such a precise number on WFE is simply because we think by the January call, which is the normal time we would get that level of detail. We'll have a better view of clean room schedules of supply chain constraints and really be able to hone in on that.

But it's -- those are the things we're looking at. And just as this year, we were able to work through those. And I think as we get closer, we get a better sense for how the industry is responding. .

Operator

Our next question comes from Stacy Rasgon with Bernstein Research. .

Stacy Rasgon

For the first one, like I understand that this more than $200 billion WFE next year is a constrained number -- do you guys have a point of view on what that number would be if it was not constrained, if clean rooms were available. Supply chain was not as tight as it is. Where do you think -- I guess what I imagine where do you think the demand for WFE in 2027 versus the supply number that you've given? .

Douglas Bettinger

Listen, Stacy, you always ask these interesting questions -- you know we're not going to add. But I love the questions nonetheless. Listen, things are going to continue to be constrained, we think, in the next year by the things that we described. I'm not going to put a number on what the unconstrained demand would be. We probably not probably. It would be higher, but I'm not going to give you a precise number. And like I said, we'll fine-tune what we think the number is and give you a little more specificity on the next quarter call also. .

Stacy Rasgon

Okay. That's helpful. For my follow-up, I focus on the NAND upgrades relative to the services. It sounded like upgrades were strong. I don't know if that meant that they grew sequence or not. I assume they did because the NAND revenues grew a bunch, but the the CSBG revenues were kind of flattish, and I thought the upgrades were in CSBG. So if that's the case, I guess what I'm asking is, did upgrades drive the NAND growth? And if they did, what else was going on in CSBG to absorb the growth such that the overall was flat. .

Douglas Bettinger

Yes. Listen, it's Stacy. Upgrades are a part of NAND, also CSBG, but NAND also, when you go through these conversions, you need to buy new equipment also. So obviously, there was a contribution in demand spending from new equipment also. .

Stacy Rasgon

Was it bigger than the upgrade piece -- it sounds like it was. .

Douglas Bettinger

Both were meaningful contributors.

Operator

And our next question today comes from Blayne Curtis at Jefferies. .

Blayne Curtis

I actually want to ask on the NAND side. You mentioned SLC. That could be a big transition. Is that something you're seeing from customers? Or is that still a potential transition? .

Timothy Archer

I think it's something that you're starting to hear from customers talk about. I mean, fundamentally, this idea of customizing memory and storage hierarchy to the different workloads, I think, is now pretty well understood. And so I think that this is emerging as 1 of the potential solutions for low latency NAND access.

And so I think if you go look just in the last recent days and weeks, there's been discussion with SLC plan, NAND emerging as part of that stack. A little bit longer. I mean for a little bit longer time, there's been discussion also things like high bandwidth flash and the role it can play.

And so we're really starting to see NAND move from sort of a singular application. We've always thought of to now 1 in which perhaps the device performances themselves will need to be customized. And that will have implications for us, but it will also -- from a process perspective, but really, what we were highlighting here was the impact it would have on wafer capacity needs and therefore, perhaps an uptick in NAND equipment demand.

Blayne Curtis

And then I want to ask on advanced packaging. I think you said last time it would grow over 70% for the calendar year. I was curious what the update is. And then when you look at it's probably the smaller end to mention it, but in terms of that $200 billion outlook, like any thoughts on advanced packaging growth next year? .

Douglas Bettinger

Yes. Blayne, 70% for this year is still the right number, and I would be stunned if advanced packaging didn't grow nicely next year. .

Operator

Next question comes from Shane Brett at Morgan Stanley. .

Shane Brett

So my first question is on foundry/logic. At your 2025 Investor Day, you talked about your foundry logic SAM per wafer increasing 2x. Just how far are we towards that 2x with the shift from 3-nanometer, 2-nanometer or rephrase that, how high did our SAM grow from 3-nanometer nanometer? And do you think your share gains would accelerate with the transition to 1.4.

Timothy Archer

Yes. That was a -- that 2x was a multigenerational shift. And so it's a little bit hard to put exactly where we are in that, maybe halfway through, if I thought about that road map that we gave there. But really, it's -- as we've moved forward, though, we've seen just new emerging applications. I talked about a couple of them today.

Clearly, advanced patterning itself is becoming ever more critical. We talked about the growth in ether. We talked about the use of Acara for conductor etch backside power really starting to get a lot more attention being a very etch and deposition intensive application we just see a number of elements of that improvement. And then I commented about how excited I was about at packaging. I mean it's advanced packaging is playing a huge role within foundry/logic and our exposure to the WFE there. And so I just think foundry/logic is turning into a good story.

The thing we've highlighted over the last couple of calls, though, just to remind you, is the role we play and gate all around our strength in ALD in our strength in surface treatment. Those are areas where I think even at the Investor Day, we might not have realized just how strong the uptake would be for things like the Argos surface treatment tool. So I feel like the story just kind of keeps getting better and better, the more complex the devices get across all 3 device types, foundry/logic DRAM.

Shane Brett

Got it. And for my follow-up, it will be on advanced packaging. So at SEMICON West, 1 of your colleagues had a great presentation on breaking the integration well and advanced packaging for the AIR with 1 of the highlights being key challenges to hybrid bonding adoption at scale. Just how does Lam enable hybrid bonding and is more hybrid bonding better or worse for our Advanced Packaging share? .

Timothy Archer

Well, I think that the -- we participate in the elements of the process that help enable hybrid bonding. And so obviously, as a deposition in X company, we don't do the bonding ourselves, but we talked about a number of films, whether it's in the dielectric space or in the metal space that ultimately play an important role in achieving successful hybrid bonding.

And so there is an opportunity for Lam through that transition. But I think that overall, look, the reality is success in advanced packaging, whatever version it is, will lead to greater adoption of advanced packaging. And ultimately, given our very high share position, that's a good thing for us. Hybrid bonding, you know it's a very etching deposition-intensive process compared to other parts of the semiconductor manufacturing process itself.

Operator

And our next question comes from Vijay Rakesh with Mizuho.

Vijay Rakesh

Sorry about that. So Tim, had some interesting comments on the NAND side. You mentioned every 1% move to SLC drives about -- requires about 2% increase in wafer capacity. So as you look at next year, with the move to SLC and potentially some HBF, any thoughts on how you think what percent of the market moves to SLC?Is it like 5%? Or how would you see that? .

Timothy Archer

I think it's too early for us to make any comment on that. And it would be probably something our customers could -- should actually be commenting on. But what you can rest assured is that the -- in that sense, the intensity of Lam equipment and manufacturing those devices and those wafers would go up. .

Vijay Rakesh

Got it. And then a quick follow-up. On the advanced packaging side, you mentioned, obviously, you have -- there's a whole Copas and EMT coming -- and I guess, Lam has huge opportunities on the metalization preclean PSV, et cetera. Just wondering if it's a way to size that opportunity as you have done before for every 100,000 wafers of EP or Copas, -- like how would you size that opportunity for Lam? .

Timothy Archer

Yes. I think maybe we'll take that as an action for work just being done, maybe sizing -- we're not prepared to do that today. .

Douglas Bettinger

Operator, we're going to take 1 more question. .

Operator

Absolutely. And our final question today comes from Brian Chin at Stifel.

Brian Chin

A few quick questions here. Maybe, Doug, firstly, days of inventory kind of running a little bit over maybe a quarter at the moment, clearly has drawn down. You said the turns have been really strong. Ideally, where would you want that to be? .

Douglas Bettinger

Yes. Listen, we're super comfortable where it is right now. As the company scales up, you do get some level of efficiency Offsetting that, though, right now, we are trying to build inventory in anticipation of the growing demand we see into the next year. But I think we collectively feel pretty good about where we're at at 3.4x. .

Brian Chin

Okay. Great. Maybe just the last question here. Is the $1 billion increase in your drivers this revenue outlook reflective more of adding logic and DRAM customers in terms of the adoption of that platform? Or is it more towards existing customers pulling forward introduction of the timing of Drive Resist .

Timothy Archer

I think it's a little bit of both. I mean there have been some additional customer adoptions, but really, it is -- as you really become PGR for something like we mentioned, HBM manufacturing and the volumes are growing there. The number of tools and therefore, the business just continues to expand. And so we feel like we're positioned in production for some of the fast-growing parts of the market, and that's why we were able to take the number up .

Douglas Bettinger

Thanks, Brian. Listen, thanks, everybody, for joining the call today. we're really excited about where the business is performing. We're really excited about what things look like into the next year. And I think we're probably the most excited about a third consecutive year of outperformance and an expectation that we're going to continue to drive those things into the next calendar year. So I know we'll see a lot of you guys in NDRs and conferences in the coming quarter, and we look forward to seeing you .

Timothy Archer

Thanks for joining the call. .

Operator

Thank you. That concludes today's conference call. We thank you all for attending today's presentation. You may now disconnect your lines, and have a wonderful day.