Entegris Inc (ENTG) 2025 Q4 法說會逐字稿

Welcome to the Entegris fourth-quarter 2025 earnings conference call. (Operator Instructions)

歡迎參加 Entegris 2025 年第四季財報電話會議。（操作說明）

I would now like to turn the call over to Jeff Schnell, VP of Investor Relations.

現在我將把電話交給投資者關係副總裁傑夫·施內爾。

Good morning, everyone. Earlier today, we announced the financial results for the fourth quarter of 2025. Before we begin, I would like to remind listeners that our comments today will include some forward-looking statements. These statements involve a number of risks and uncertainties, and actual results could differ materially from those projected in the forward-looking statements. Additional information regarding these risks and uncertainties is contained in our most recent annual report and subsequent quarterly reports that we have filed with the SEC. Please refer to the information on the disclaimer slide in the presentation.

各位早安。今天早些時候，我們公佈了2025年第四季的財務業績。在開始之前，我想提醒各位聽眾，我們今天的評論將包含一些前瞻性陳述。這些聲明涉及諸多風險和不確定性，實際結果可能與前瞻性聲明中預測的結果有重大差異。有關這些風險和不確定性的更多信息，請參閱我們向美國證券交易委員會提交的最新年度報告和後續季度報告。請參閱簡報中免責聲明投影片上的資訊。

On this call, we will also refer to non-GAAP financial measures as defined by the SEC and Regulation G. You can find reconciliation tables in today's news release as well as on the IR page of our website at entegris.com. On the call today are Dave Reeder, our CEO; and Linda LaGorga, our CFO.

本次電話會議中，我們也將提及美國證券交易委員會（SEC）和G條例定義的非GAAP財務指標。您可以在今天的新聞稿以及我們網站entegris.com的投資者關係頁面上找到相關調節表。今天參加電話會議的有我們的執行長Dave Reeder和財務長Linda LaGorga。

With that, I'll hand the call over to Dave.

這樣，我就把電話交給戴夫了。

Thank you, and good morning. Before we dive into results, I want to take a moment to welcome Jeff to our team. We're excited to have him on board leading our IR efforts as our new Head of Investor Relations.

謝謝，早安。在深入探討結果之前，我想花一點時間歡迎傑夫加入我們的團隊。我們很高興他能加入我們，擔任新的投資者關係主管，並領導我們的投資者關係工作。

Our solid fourth-quarter results demonstrate disciplined execution and a consistent focus on delivering on our commitments. Fourth-quarter revenue, gross margin, adjusted EBITDA margin, and non-GAAP EPS were all at the high end or above our guidance range.

我們第四季穩健的業績顯示我們執行力強，並始終專注於履行我們的承諾。第四季營收、毛利率、調整後 EBITDA 利潤率和非 GAAP 每股盈餘均達到或超過我們預期範圍的高端水準。

For the full year, total revenue was approximately flat compared to 2024, excluding divestitures. Our unit-driven revenue grew approximately 2% in 2025, in line with wafer starts for the market and was led by CMP consumables, liquid filtration, and selective etch.

全年總收入與 2024 年基本持平（不包括資產剝離）。2025 年，我們的單位驅動收入成長了約 2%，與市場晶圓開工量一致，主要由 CMP 耗材、液體過濾和選擇性蝕刻帶動。

Our CapEx-driven revenue declined 7% in 2025, consistent with the decline in industry fab construction CapEx, where we are most highly correlated. The fab CapEx slowdown was most evident in our FOUP and fluid handling product lines within our APS division.

2025 年，我們資本支出驅動的收入下降了 7%，這與產業晶圓廠建設資本支出的下降趨勢一致，而我們與該產業的相關性最高。晶圓廠資本支出放緩在我們APS部門的FOUP和流體處理產品線最為明顯。

Looking ahead to 2026, the industry backdrop appears more constructive. I'll touch on the semi market in more detail in a bit, but there are a few areas where we expect notable improvement compared to 2025 that should benefit Entegris.

展望 2026 年，產業前景似乎更加樂觀。稍後我會更詳細地談談半導體市場，但與 2025 年相比，我們預計有幾個領域會有顯著改善，這將有利於 Entegris。

First, we expect to benefit in 2026 from the node transitions in both logic and memory. In logic, increased demand for 2-nanometer devices is expected to meaningfully drive wafer output throughout 2026. In memory, NAND transitions are progressing, migrating from low 250 layers to approximately 300 layers. Additionally, next-generation DRAM and HBM products are expected to be rolled out this year, and all these transitions create accretive content per wafer opportunities for Entegris.

首先，我們預計在 2026 年將受益於邏輯和記憶體領域的節點過渡。在邏輯電路領域，預計到 2026 年，對 2 奈米元件的需求成長將顯著推動晶圓產量成長。在記憶體領域，NAND 快閃記憶體正在轉型，從 250 層左右遷移到大約 300 層。此外，新一代 DRAM 和 HBM 產品預計將於今年推出，所有這些轉變都將為 Entegris 創造每片晶圓增加內容的機會。

Next, we expect industry MSI growth to increase in 2026 led by continued strong growth in advanced logic and DRAM, improving demand for NAND and stable demand for mainstream logic. Finally, we expect industry fab construction spending to grow in 2026, reversing a significant decline in 2025. This is meaningful for Entegris because two-thirds of our CapEx-related revenue is correlated to fab construction

接下來，我們預期在先進邏輯和DRAM持續強勁成長的帶動下，2026年微單產業成長將有所提高，NAND需求將有所改善，主流邏輯需求也將維持穩定。最後，我們預期 2026 年工業晶圓廠建設支出將有所成長，扭轉 2025 年的大幅下降趨勢。這對Entegris來說意義重大，因為我們三分之二的資本支出相關收入都與晶圓廠建設有關。

Last quarter, my first as CEO, I shared my initial priorities for Entegris. Let me provide an update on those priorities. First is deepening customer intimacy. This includes supporting our customers' technology road maps. Success in this area translates into securing key positions of record, PORs, in new nodes, which will expand our served available market and increase both revenue and content per wafer. For logic devices at the most advanced node, we've secured strong POR positions and solid share in key product lines such as CMP consumables, advanced deposition and implant materials, liquid purification and filtration, and wafer handling products.

上個季度，也就是我擔任執行長的第一個季度，我分享了我對 Entegris 的初步工作重點。讓我來報告一下這些優先事項的最新進展。首先是加深與客戶的親密關係。這包括支援我們客戶的技術路線圖。在該領域取得成功意味著在新節點中佔據關鍵的記錄位置（POR），這將擴大我們服務的可用市場，並增加每個晶圓的收入和內容。對於最先進的邏輯元件，我們在關鍵產品線（如 CMP 耗材、先進沉積和注入材料、液體淨化和過濾以及晶圓處理產品）中佔據了強大的 POR 地位和穩固的市場份額。

In addition to this, the team is focused on winning incremental share and PORs in subsequent advanced nodes. For advanced memory, we are gaining traction in DRAM and HBM, in particular, for products associated with CMP consumables and advanced packaging solutions. And for next-generation NAND devices, we have also achieved strong POR wins with solid share across key NAND-specific product lines, including deposition materials, CMP, and selective etch applications.

除此之外，團隊也專注於在後續的高階節點中贏得更多市場份額和POR。對於先進記憶體，我們在 DRAM 和 HBM 領域，特別是與 CMP 耗材和先進封裝解決方案相關的產品，獲得了越來越多的市場份額。對於下一代 NAND 設備，我們也取得了強大的 POR 勝利，在關鍵的 NAND 專用產品線（包括沉積材料、CMP 和選擇性蝕刻應用）中佔據了穩固的市場份額。

Our second priority is improving utilization by ramping our new facilities in Taiwan and Colorado while rationalizing our existing manufacturing footprint. Our Taiwan facility continues to ramp production, and our Colorado facility is expected to substantially complete key customer product qualifications in 2026. And in the fourth quarter, we exited our Chester, Pennsylvania, facility, and we expect to rationalize at least one additional facility in the first half of 2026.

我們的第二個優先事項是提高利用率，透過加快我們在台灣和科羅拉多州的新工廠的建設，同時合理化我們現有的生產佈局。我們在台灣的工廠正在持續提高產量，預計我們在科羅拉多州的工廠將於 2026 年基本完成關鍵客戶的產品認證。第四季度，我們關閉了位於賓州切斯特的工廠，我們預計在 2026 年上半年至少還會關閉一家工廠。

As I discussed last quarter, we have completed the multiyear manufacturing CapEx investment cycle that began in 2022. As a result, we expect 2026 CapEx to decline to $250 million. Longer term, we expect CapEx to return to historical levels of approximately 7% to 8% of sales. The additional manufacturing capacity we've built, combined with our current manufacturing base, enables us to deliver significantly more than $1 billion in incremental revenue with limited further investment.

正如我上個季度所討論的，我們已經完成了從 2022 年開始的多年製造業資本支出投資週期。因此，我們預計 2026 年資本支出將下降至 2.5 億美元。從長遠來看，我們預計資本支出將恢復到歷史水平，約為銷售額的 7% 至 8%。我們新增的生產能力，加上我們現有的生產基地，使我們能夠在有限的進一步投資下，實現超過 10 億美元的增量收入。

Our third priority is improving free cash flow. Thanks to the team's execution, free cash flow margin, which is free cash flow divided by sales, improved meaningfully, reaching 12.7% in 2025, in line with our target. Higher operating cash flow in combination with reduced CapEx is expected to increase free cash flow again in 2026. This will support debt reduction and enable us to reduce net leverage to below 3.5 times exiting 2026. Underscoring our commitment, free cash flow is now part of our short-term and long-term incentive plans.

我們的第三項優先事項是改善自由現金流。由於團隊的出色執行，自由現金流利潤率（自由現金流除以銷售額）得到了顯著改善，到 2025 年將達到 12.7%，符合我們的目標。預計2026年，更高的營運現金流加上更少的資本支出將再次增加自由現金流。這將有助於減少債務，並使我們能夠在 2026 年底將淨槓桿率降低到 3.5 倍以下。為了彰顯我們的承諾，自由現金流現已納入我們的短期和長期激勵計畫。

Our fourth priority is increasing local-for-local manufacturing, particularly for China. This provides us with critical strategic flexibility and enhanced ability to serve our global customers. We expect approximately 85% of our China revenue in Q1 will be supplied by our Asia facilities, with that proportion increasing through 2026.

我們的第四個優先事項是增加本地製造，特別是針對中國的本地製造。這為我們提供了重要的策略靈活性，並增強了我們服務全球客戶的能力。我們預計第一季中國收入的約 85% 將由我們在亞洲的工廠供應，並且到 2026 年這一比例將不斷增加。

Turning our thoughts to the semiconductor market. We expect mid-ish single-digit industry MSI or wafer starts growth in 2026. As a reminder, about 75% of our revenue is unit driven and is correlated to MSI.

讓我們把目光轉向半導體市場。我們預計 2026 年微創半導體或晶圓製造業將實現中等個位數的成長。再次提醒，我們約 75% 的收入是由銷售驅動的，並且與 MSI 相關。

Looking closer at semi end markets. Advanced logic is positioned for significant growth again in 2026, driven largely by AI-enabled applications. Fab utilization rates in advanced logic are already near 100%, and our customers are aggressively investing in additional capacity. Beyond the benefits of strong unit growth, as 2-nanometer significantly ramps wafer output this year, this node provides an additional tailwind as it carries both higher content per wafer and strong share for Entegris.

深入分析半導體市場。高階邏輯技術預計在 2026 年再次實現顯著成長，這主要得益於人工智慧應用的發展。先進邏輯晶片的晶圓廠利用率已接近 100%，我們的客戶正在積極投資擴大產能。除了強勁的單位成長帶來的好處之外，由於 2 奈米製程製程今年大幅提高了晶圓產量，該節點還帶來了額外的順風，因為它每片晶圓的含量更高，並為 Entegris 帶來了強大的市場份額。

In mainstream logic, feedback suggests inventory levels are now healthy. While we're seeing early signs of improvement and mainstream MSI still remains well below the 2022 peak, the overall end market recovery is slow and mixed. We also note that ongoing memory shortages may weigh on the industry's ability to supply some mainstream end markets.

依照主流邏輯，回饋顯示庫存水準目前處於健康狀態。雖然我們看到了一些改善的早期跡象，主流微星市場仍遠低於 2022 年的峰值，但整體終端市場復甦緩慢且喜憂參半。我們也注意到，持續的記憶體短缺可能會影響該產業向一些主流終端市場供貨的能力。

NAND, NAND continues to benefit from strong AI-driven demand and pricing trends. This is expected to translate into more than 20% bit growth in 2026, driven primarily by the shift to higher layer, higher capacity NAND rather than a significant increase in MSI.

NAND快閃記憶體持續受惠於人工智慧驅動的強勁需求和價格趨勢。預計到 2026 年，這將轉化為超過 20% 的位元成長，主要驅動力是向更高層、更高容量的 NAND 轉變，而不是 MSI 的大幅成長。

While NAND MSI is expected to rise modestly in 2026, we expect to additionally benefit from a double-digit increase in NAND content per wafer as customers move to higher layer count advanced nodes and introduce new materials such as moly and selective etch. If demand remains robust, flash memory makers will likely need to add significant fab capacity, setting the stage for higher NAND MSI growth in 2027. DRAM is expected to see solid MSI growth in 2026.

雖然預計 2026 年 NAND MSI 將小幅增長，但我們預計隨著客戶轉向更高層數的先進節點並引入鉬和選擇性蝕刻等新材料，每片晶圓的 NAND 含量將實現兩位數增長，從而進一步受益。如果需求保持強勁，快閃記憶體製造商可能需要大幅增加晶圓廠產能，為 2027 年 NAND MSI 的更高成長奠定基礎。預計 DRAM 在 2026 年將穩健成長。

Pricing trends and underlying demand remains strong in both HBM and DDR5. Tight supply in HBM, DDR5, and in advanced packaging are all expected to drive the need for additional fab capacity heading into 2027.

HBM 和 DDR5 的價格趨勢和潛在需求依然強勁。預計到 2027 年，HBM、DDR5 和先進封裝的供應緊張都將推動對額外晶圓廠產能的需求。

While 75% of our revenue is related to MSI, 25% is tied to industry CapEx. There are two primary drivers of our CapEx revenue, fab construction-related spending, which correlates with approximately two-thirds of our CapEx sales and the remaining one-third related to WFE. Fab construction CapEx is expected to grow modestly this year after a high single-digit decline last year with a more meaningful acceleration anticipated in 2027 as construction begins on new fabs. Additionally, we expect WFE to deliver strong growth in 2026.

雖然我們 75% 的收入與 MSI 相關，但 25% 與行業資本支出有關。我們的資本支出收入主要有兩個驅動因素：晶圓廠建設相關支出（約占我們資本支出銷售額的三分之二）和與晶圓廠設備相關的支出（約占我們資本支出銷售額的三分之一）。晶圓廠建設資本支出預計在去年出現高個位數下降後，今年將小幅增長，隨著新晶圓廠建設的開始，預計 2027 年將出現更顯著的增長。此外，我們預計 WFE 在 2026 年將實現強勁成長。

Overall, AI continues to be an important growth driver for the semi market, and we are seeing an increased benefit from this trend. Today, more than 60% of Entegris' revenue comes from advanced logic and advanced memory. AI is, of course, not the majority of these advanced nodes, but it is an important part and the most significant growth driver.

總體而言，人工智慧仍然是半導體市場的重要成長驅動力，我們正在從這一趨勢中獲得更多益處。如今，Entegris 60% 以上的收入來自先進邏輯電路和先進記憶體。當然，人工智慧並非這些先進節點的主體，但它是重要組成部分，也是最重要的成長驅動力。

In closing, we ended 2025 with momentum. We are cautiously optimistic about the industry conditions entering 2026. We continue to focus on winning key PORs and new nodes, driving higher Entegris content per wafer and revenue.

總之，我們以強勁的勢頭結束了2025年。我們對2026年業界狀況持謹慎樂觀態度。我們將繼續專注於贏得關鍵的POR和新節點，從而提高每片晶圓的Entegris含量和收入。

The growth we expect this year should improve utilization, thus increasing free cash flow and reducing leverage. And as devices become more complex, our expertise in material science and materials purity becomes increasingly critical, helping customers enhance performance and achieve optimal yields. As a result, we expect to significantly grow our content per wafer and outperform the market, and we will continue to focus on execution and delivering on our commitments.

我們預計今年的成長將提高利用率，從而增加自由現金流並降低槓桿率。隨著設備變得越來越複雜，我們在材料科學和材料純度方面的專業知識變得越來越重要，可以幫助客戶提高性能並實現最佳產量。因此，我們預計每片晶圓的含量將大幅提高，並超越市場表現，我們將繼續專注於執行和履行我們的承諾。

Before handing over to Linda, I wanted to share that given the CFO transition, we are rescheduling our Capital Markets Day from this May to the fall of this year. We'll share more details on this as soon as we can. And finally, I want to thank Linda for her many contributions and lasting impact on Entegris. We wish her all the best in the future.

在將發言權交給琳達之前，我想宣布，鑑於財務長的更迭，我們將把資本市場日從今年五月改期到今年秋季舉行。我們會盡快分享更多細節。最後，我要感謝琳達對 Entegris 的諸多貢獻和深遠影響。我們祝福她未來一切順利。

With that, let me turn the call over to Linda.

那麼，我把電話交給琳達吧。

Good morning. Q4 sales were $824 million at the high end of guidance, down 3% year over year and up 2% sequentially. Gross margin on a GAAP basis was 43.8% and 44% on a non-GAAP basis in the fourth quarter, also at the high end of guidance. The sequential increase in gross margin was primarily driven by increased production volumes across our manufacturing facilities.

早安.第四季銷售額為 8.24 億美元，達到預期上限，年減 3%，季增 2%。第四季度，以美國通用會計準則計算的毛利率為 43.8%，以非美國通用會計準則計算的毛利率為 44%，均達到預期上限。毛利率的環比成長主要得益於我們各製造工廠產量的增加。

Back to the Q4 P&L.

回到第四季損益表。

Operating expenses on a GAAP basis were $256 million in Q4. Operating expenses on a non-GAAP basis in Q4 were $188 million. Adjusted EBITDA in Q4 was 27.7% of revenue, above our guidance.

第四季以美國通用會計準則計算的營運費用為 2.56 億美元。第四季非GAAP準則下的營運費用為1.88億美元。第四季調整後 EBITDA 佔營收的 27.7%，高於我們的預期。

The GAAP tax rate in Q4 was 10%, and the non-GAAP tax rate was 15.4%. GAAP diluted EPS was $0.32 per share in the fourth quarter. Non-GAAP EPS was $0.70 per share, above our guidance.

第四季 GAAP 稅率為 10%，非 GAAP 稅率為 15.4%。第四季GAAP稀釋後每股收益為0.32美元。非GAAP每股收益為0.70美元，高於我們的預期。

Sales for Materials Solutions in Q4 were $362 million. Sales were flat year over year and up 4% sequentially. Sequential growth was driven primarily by advanced deposition materials, supported by demand for moly deposition within NAND.

第四季材料解決方案的銷售額為 3.62 億美元。銷售額與去年同期持平，季增4%。連續成長主要由先進的沉積材料驅動，而 NAND 中對鉬沉積的需求也支撐了這一增長。

Adjusted operating margin for MS was 20.9% for the quarter. The year-on-year decline in margin was driven by slightly lower production volumes and strategic investments. The strong sequential increase in margin was driven by increased production volumes and product mix.

摩根士丹利本季的調整後營業利益率為 20.9%。利潤率年減是由於產量略有下降和策略投資所致。利潤率的強勁環比成長是由產量增加和產品組合優化所驅動的。

Sales for Advanced Purity Solutions in Q4 were $465 million, down 5% year on year and up 1% sequentially. The year-over-year sales decline was driven by fluid handling and FOUPs, partially offset by strong growth in liquid filtration, which had another record quarter.

第四季度，Advanced Purity Solutions 的銷售額為 4.65 億美元，年減 5%，季增 1%。年比銷售額下降主要受流體處理和FOUP（前端、 ...

Sequential growth in liquid filtration and gas purification was partially offset by lower FOUP sales. The adjusted operating margin for APS was 24.8% for the quarter. The year-on-year decline in margin was driven by costs related to the ramp of our Taiwan and Colorado manufacturing sites and lower production volumes. The sequential decrease in margin was primarily driven by unfavorable product mix and timing of operating expenses.

液體過濾和氣體淨化業務的連續成長被FOUP銷售額的下降部分抵消。APS本季的調整後營業利益率為24.8%。利潤率較去年同期下降的原因是台灣和科羅拉多州生產基地產能提升相關的成本以及產量下降。利潤率環比下降的主要原因是產品組合不利以及營運費用發生的時間安排不當。

Moving on to cash flow. Full year free cash flow was $404 million, representing a free cash flow margin of 12.7% in 2025, nearly a 300 basis point increase year over year. This improvement was driven by our team's disciplined focus on working capital, including accounts receivable and decreased year-on-year inventory growth. CapEx for 2025 was $299 million, approximately 9% of sales.

接下來分析現金流。全年自由現金流為 4.04 億美元，2025 年自由現金流利潤率為 12.7%，較上年增長近 300 個基點。這項改善得益於我們團隊對營運資金的嚴格控制，包括應收帳款和年比庫存成長的下降。2025 年資本支出為 2.99 億美元，約佔銷售額的 9%。

A quick overview of our capital structure. During the fourth quarter, we paid down $150 million of the term loan from cash on hand. And for the full year, we paid down $300 million of the term loan. At quarter end, our gross debt was approximately $3.7 billion, and our net debt was $3.4 billion. Net leverage ended the year at 3.8 times. As Dave said, we are targeting net leverage of below 3.5 times by the end of 2026.

公司資本結構概覽。第四季度，我們用手頭現金償還了 1.5 億美元的定期貸款。全年下來，我們償還了 3 億美元的定期貸款。截至季末，我們的總債務約為 37 億美元，淨債務為 34 億美元。年末淨槓桿率為 3.8 倍。正如戴夫所說，我們的目標是到 2026 年底將淨槓桿率降至 3.5 倍以下。

Moving on to our Q1 outlook. We expect our Q1 sales to range from $785 million to $825 million, reflecting an increase of approximately 4% to the midpoint year over year. Gross margin of 44.5% to 45.5%, both on a GAAP and non-GAAP basis.

接下來是第一季展望。我們預計第一季銷售額將在 7.85 億美元至 8.25 億美元之間，較上年同期成長約 4%（取中間值）。毛利率為 44.5% 至 45.5%，不論以 GAAP 或非 GAAP 計算。

We recently completed an assessment of the useful lives of our assets. This gross margin guidance includes the positive impact from the useful life accounting change of approximately 100 basis points in Q1 on gross margin.

我們最近完成了對資產使用壽命的評估。此毛利率預期包含了第一季毛利率因使用壽命會計變更而產生的約 100 個基點的正面影響。

We expect GAAP operating expenses of approximately $229 million and non-GAAP operating expenses of approximately $181 million. EBITDA margin to range from 26.5% to 27.5%. Net interest expense of approximately $47 million. We expect our non-GAAP Q1 tax rate to be approximately 15%. We expect GAAP EPS between $0.43 to $0.51 per share. Non-GAAP EPS between $0.70 and $0.78 per share. And we expect depreciation of approximately $36 million in Q1.

我們預計 GAAP 營運費用約為 2.29 億美元，非 GAAP 營運費用約為 1.81 億美元。EBITDA利潤率預計在26.5%至27.5%之間。淨利息支出約 4700 萬美元。我們預計第一季非GAAP稅率約為15%。我們預計 GAAP 每股收益在 0.43 美元至 0.51 美元之間。非GAAP每股收益在0.70美元至0.78美元之間。我們預計第一季折舊額約為 3,600 萬美元。

Looking slightly further ahead, based on our current visibility, we expect Q2 sales to increase 1% to 3% sequentially from Q1, in line with normal industry seasonality. I'd like to provide a few modeling items for the full year of 2026.

展望未來，根據我們目前的預期，我們預計第二季銷售額將比第一季環比成長 1% 至 3%，符合正常的行業季節性規律。我想提供一些 2026 年全年的模型專案。

We expect net interest expense will be approximately $190 million, the non-GAAP tax rate to be approximately 15%, diluted share count of approximately 152 million shares for Q1 and approximately 153 million shares for the full year, CapEx of $250 million, and depreciation of approximately $150 million, reflecting the recently completed assessment of the useful lives of our assets.

我們預計淨利息支出約為 1.9 億美元，非 GAAP 稅率約為 15%，第一季稀釋後股份數約為 1.52 億股，全年稀釋後股份數約為 1.53 億股，資本支出為 2.5 億美元，折舊約為 1.5 億美元，這反映了我們最近完成的資產使用壽命評估。

Before we begin Q&A, I would like to thank the finance team, the leadership team and the Board for their partnership over the past three years. I am proud of the work we have done to strengthen the foundation of the business and position the company to capitalize on future opportunities. It's been a privilege to be CFO, and I am confident in Entegris' path forward.

在開始問答環節之前，我要感謝財務團隊、領導團隊和董事會在過去三年的合作。我為我們所做的工作感到自豪，這些工作鞏固了公司的業務基礎，並使公司能夠抓住未來的機會。能夠擔任財務長是我的榮幸，我對 Entegris 的未來發展道路充滿信心。

With that, operator, let's open the line for questions.

好了，接線員，我們現在開始接受提問。

(Operator Instructions) Mike Harrison, Seaport Research Partners.

（操作說明）Mike Harrison，Seaport Research Partners。

Best wishes to Linda, and welcome, Jeff, to the team. Dave, I appreciate you walking through your detailed thoughts there on underlying market growth in 2026. It sounds like if we roll that all together, you're looking at something in the mid-single-digit range for growth overall.

祝琳達一切順利，歡迎傑夫加入團隊。Dave，感謝你詳細闡述了你對 2026 年潛在市場成長的看法。聽起來，如果把所有這些因素綜合起來，整體成長率大概會在個位數中段。

But I'm curious, historically, Entegris would talk about growing 3 to 6 percentage points faster than the underlying market. As you look at the opportunities that you're seeing, you mentioned advanced nodes in 2-nanometer as well as growing content per wafer in NAND. I'm just curious, are you expecting an environment in 2026 where you can get back to growing in that 3% to 6% range faster than underlying markets?

但我很好奇，從歷史上看，Entegris 經常談到其成長速度比基礎市場快 3 到 6 個百分點。您在展望未來機會時，提到了 2 奈米製程的先進節點以及 NAND 晶圓單片容量的成長。我只是好奇，您是否預期到 2026 年，您的業務成長速度能夠恢復到 3% 到 6% 的水平，並且成長速度要高於基礎市場？

Hi, Mike. Good to speak to you again. When we think about 2026, we do think the industry backdrop is a little bit more constructive than it was in 2025. And specifically, if you think about kind of the areas in which we grow revenue, we do about 40% of our revenue is from advanced logic, about 30% from mainstream logic and then the remainder from memory. And so when you look at 2026, it feels like advanced logic is pretty fully utilized as we add capacity there, we get the benefit of both additional growth plus more content.

嗨，麥克。很高興再次與您交談。當我們展望 2026 年時，我們認為產業背景比 2025 年要好一些。具體來說，如果你想想我們收入成長的領域，我們大約 40% 的收入來自高階邏輯電路，大約 30% 來自主流邏輯電路，其餘部分來自記憶體。因此，展望 2026 年，隨著我們增加產能，高階邏輯將充分利用，我們既能獲得額外的成長，又能獲得更多的內容。

Mainstream looks kind of mixed. So we think we're performing on a cylinder for advanced logic. We think mainstream looks mixed but stable. And then we think memory can perform. So think of it as kind of performing on three of our four cylinders.

主流觀點看起來有點混雜。所以我們認為我們正在對一個用於高級邏輯的圓柱體進行操作。我們認為主流市場情勢喜憂參半，但整體穩定。然後我們認為記憶體可以發揮作用。所以你可以把它想像成只用了四個氣缸中的三個來運作。

The additional piece to then layer on top of it is CapEx, and CapEx was not terribly constructive in '25, but we do think that CapEx could be more constructive in '26, particularly the portion related to fab CapEx.

接下來需要疊加的部分是資本支出，而資本支出在 2025 年並沒有起到特別大的建設性作用，但我們認為資本支出在 2026 年可能會更有建設性，特別是與製造資本支出相關的部分。

So when you look at that industry backdrop and you think about outperformance, I'll just add a couple of more points to that. One, we typically get the most outperformance when we have node transitions because that drives additional content per wafer. So that's our -- typically our biggest driver of outperformance. And so while both logic and NAND node transitions look solid, we don't really control the timing and pace of that.

所以，當你審視產業背景並思考如何超越市場表現時，我只想補充幾點。第一，我們通常在節點過渡時獲得最大的效能提升，因為這會為每個晶圓增加更多內容。所以，這通常是我們取得優異績效的最大驅動因素。因此，雖然邏輯節點和 NAND 節點的過渡看起來都很穩定，但我們實際上無法控制其時間和速度。

And then, of course, as I mentioned, the CapEx piece is, particularly fab CapEx is relatively volatile. So when we look at '26, we look at our first-quarter guide, plus 4% at midpoint, slightly greater than 6% at the high end of our guidance range. And of course, we gave you a little bit of color for Q2. We feel like the setup is constructive to the extent the node transitions, both logic and memory happen, we feel like we can get back to outperformance and then the CapEx piece looks to be a little bit more second half weighted.

當然，正如我所提到的，資本支出部分，特別是製造資本支出，波動性相對較大。因此，當我們展望 2026 年時，我們會參考第一季的業績指引，並在中點處加上 4%，略高於我們指引範圍上限的 6%。當然，第二季我們也加入了一些色彩元素。我們認為，就節點轉換、邏輯和記憶體的實現而言，這種設定是建設性的，我們感覺我們可以恢復到超越效能的水平，然後資本支出部分看起來會更集中在下半年。

So I gave you a lot of details there for contents, Mike, but did you have a follow-up?

麥克，我給你很多內容細節，但你還有後續問題嗎？

Yes. That's very helpful. A follow-up is just kind of on the margin trajectory for the year. Your guidance for Q1 calls for a little bit of sequential contraction in EBITDA margin. I assume that's just seasonality, but anything you can share in terms of how we should think about margin trajectory in '26? Presumably, you're getting back to more normal production rates yourselves and seeing some benefits from ramping the Taiwan facility. So I appreciate some details there.

是的。那很有幫助。後續報道只是今年發展軌跡中的邊緣項目。您對第一季的業績指引是 EBITDA 利潤率將略有環比下降。我猜這只是季節性因素，但您能否分享一下我們應該如何看待 2026 年的利潤率走勢？想必你們的生產速度也正在逐步恢復正常，並且從提高台灣工廠的產能中也看到了一些好處。我很欣賞你提供的細節。

Yeah. So thanks, Mike. Thanks for that question. Let me bring it up to gross margin. I know you mentioned EBITDA, but I think it's important to go back to when we think about our gross margin, first, it's really stabilized in the current range.

是的。謝謝你，麥克。謝謝你的提問。讓我來談談毛利率。我知道你提到了 EBITDA，但我認為當我們考慮毛利率時，首先重要的是要回到毛利率本身，它目前已經非常穩定了。

We had mentioned we called a trough at the second half of last year. And you could see based on the Q1 guidance that stabilization. The key as we drive margins, and this will drive through to the bottom into EBITDA is the volume leverage. And so as we said, there's a constructive environment going into this year. As we see more production going through our facilities, that's going to go into our gross margin and see that improvement. That includes ramping Taiwan this year and continuing to ramp.

我們之前提到過，我們在去年下半年遇到了低潮。從第一季業績指引可以看出，市場趨於穩定。提高利潤率的關鍵在於利用銷售槓桿，而這最終將影響 EBITDA。所以正如我們所說，今年的情況是正面的。隨著我們工廠產量的增加，這將提高我們的毛利率，我們將看到毛利率的改善。這包括今年在台灣加大投入，並持續加大投入。

And then as Dave mentioned in his remarks, we did rationalize one facility, and we plan to rationalize another one in this first half of the year. So again, all those dynamics, volume leverage, combined with Taiwan ramping, combined with some rationalization is going to help us improve gross margin with that increased production and drive down to EBITDA.

正如戴夫在演講中提到的那樣，我們已經對一家工廠進行了合理化改造，並且我們計劃在今年上半年對另一家工廠進行合理化改造。所以，所有這些動態因素，包括產量槓桿作用、台灣產能提升以及一些合理化措施，都將幫助我們提高毛利率，提高產量，並降低 EBITDA。

Timothy Arcuri, UBS.

提摩西‧阿庫裡，瑞銀集團。

Dave, for the full year, you said CapEx is going to be up modestly. We know WFE is going to be up low to mid-20s or probably lower than that. What about MSI for the year? I don't think I heard you give a target for MSI for the year.

戴夫，你說過全年資本支出將小幅成長。我們知道WFE會在20%左右，甚至可能更低。今年的MSI排名如何？我好像沒聽到你給MSI今年的目標。

MSI, we think, is mid-ish single digits. Just still early days, and obviously, we've got Chinese New Year that's happening next week in Q1 versus the end of January. But when you look at MSI -- so there's some Q1 dynamics in there. But when you look at MSI overall for the year, current estimates are kind of mid-ish single digits.

我們認為 MSI 值在個位數的中等水準。現在還為時過早，而且很明顯，下週就是第一季的中國新年，而不是一月底。但如果你看看 MSI——你會發現其中有一些第一季的動態。但從全年來看，MSI 目前的預期值大概是個位數的中等水準。

Agree with your commentary on WFE. That looks like it's going to be strong this year. That's about a third of our CapEx-related revenue. And then two-thirds of our CapEx-related revenue is tied to fab construction CapEx. And when you think about that portion, it looks like there's probably not a lot of that in the first half with it picking up perhaps even significantly in the second half and then, of course, setting up well for 2027.

我同意你對WFE的評論。看來今年會很強勁。這大約占我們資本支出相關收入的三分之一。而且，我們三分之二的資本支出相關收入都與晶圓廠建設資本支出有關。仔細想想，上半年可能沒有太多這樣的情況，下半年可能會大幅增加，當然，這也為 2027 年做好了充分的準備。

Did you have a follow-up, Tim?

提姆，你還有後續問題嗎？

I do, Dave. Yes. So if I just add that all together and I run the ratios, you're probably -- I mean, your market is probably up somewhere close to 10%, probably high single digits at least. So do you think you can outgrow that by a significant margin? I mean, is that a good level to say that you should grow revenue at least high singles, probably even low doubles to get to your outperformance metrics for the year?

是的，戴夫。是的。所以，如果我把所有這些加起來，然後計算出各種比率，你的市場可能——我的意思是，你的市場可能上漲了接近 10%，至少也可能是接近 10%。所以你認為你能大幅超越那個水準嗎？我的意思是，這個水平是否足以說明，為了達到年度業績目標，收入至少應該增長到較高的單邊數，甚至可能要增長到較低的雙邊數？

Hey, Tim. I think I've given you a lot of the elements here. I think we'll probably stand pat for guiding one quarter at a time. We gave you a little visibility with respect to second quarter. Normal seasonality would imply kind of sequential growth of 1% to 3% from first quarter and based on order patterns, we feel pretty good about that range right now. And so we'll continue to give you more visibility as we see it.

嘿，提姆。我想我已經把這裡的大部分要素都告訴你了。我認為我們可能會繼續維持現狀，採取一次只專注於一個季度的策略。我們已向大家簡要介紹了第二季的情況。正常的季節性因素意味著從第一季開始，季度將持續成長 1% 到 3%，根據訂單模式來看，我們目前對這個成長範圍相當滿意。因此，我們將繼續為您提供更多相關資訊。

I think the wildcard that we kind of see right now is how does that fab CapEx kind of layer in throughout the course of the year. And then how do we kind of participate in that portion of the revenue. That's the piece that we're really watching right now, and it's moved pretty significantly month to month. So that's the hesitancy or perhaps the conservatism that you're hearing in my voice. I want to see how that plays out a little bit.

我認為我們現在看到的未知數是，這些巨額資本支出將如何貫穿全年。那麼，我們如何參與這部分收入的分配呢？這是我們目前真正關注的重點，而且它每個月的變化都相當顯著。所以，你從我的語氣中聽到的猶豫或說保守，就是這個意思。我想看看事情會如何發展。

Christopher Parkinson, Wolfe Research.

克里斯托弗·帕金森，沃爾夫研究公司。

You mentioned last quarter more of a concerted selling effort directed to mainstream customers. And I was wondering if you could give us a quick update on what's underway there?

您在上個季度提到，將加強向主流客戶群銷售。我想請您簡單介紹一下那邊正在進行的工作？

Thanks, Chris. We look at our customers in quite a lot of detail, particularly kind of our top 50-ish customers. And so when we look at that customer list and we look at mainstream, we then kind of break them down into their corresponding portions of mainstream. And so you've got kind of mainstream logic. You've got some mainstream in there that's associated with some specialty manufacturing, for example, silicon carbide as well as some other nodes.

謝謝你，克里斯。我們會非常詳細地了解我們的客戶，特別是我們前 50 名左右的客戶。因此，當我們查看客戶名單並查看主流客戶時，我們會將他們細分為主流客戶的各個組成部分。所以，這就是一種主流邏輯。其中包含一些主流材料，這些材料與一些特殊製造過程相關，例如碳化矽以及其他一些節點。

And so when we look at that universe, I'll start with kind of the latter. Silicon carbide was a headwind for us in 2025. I'm talking on a year-over-year basis from '24 to '25. We think that is now stable and perhaps even improving slightly, albeit slowly in '26. So we think the silicon carbide headwind where we have a very nice solution for the CMP process. We feel like that will not be a headwind for us, at least expectation-wise in 2026. So we think that will be constructive and helpful. And we continue to gain even more share in that process.

所以，當我們審視那個宇宙時，我會從後者開始。2025年，碳化矽對我們來說是個不利因素。我指的是 2024 年到 2025 年的年比數據。我們認為目前情況穩定，甚至可能略有改善，儘管在 2026 年進展緩慢。所以我們認為碳化矽是逆風，但我們為 CMP 製程提供了一個非常好的解決方案。我們認為這至少在 2026 年不會對我們造成不利影響。所以我們認為這將是建設性的、有益的。而且我們在這個過程中不斷獲得更大的份額。

When we look at the other mainstream, and I'm referring to mainstream logic, mainstream logic has a number of needs across our entire product portfolio. And so our efforts in mainstream logic become more about providing all of those solutions, not just individual product lines to each of those mainstream customers. So when we look across those customers, we're trying to more deeply penetrate their wallet across our complete product portfolio, whereas in some of those mainstream logic customers, we're only selling individual product lines.

當我們檢視其他主流觀點時，我指的是主流邏輯，主流邏輯在我們整個產品組合中都有許多需求。因此，我們在主流邏輯方面的努力更多是提供所有這些解決方案，而不僅僅是為每個主流客戶提供單獨的產品線。因此，當我們審視這些客戶時，我們正努力透過我們完整的產品組合更深入地滲透到他們的錢包中，而對於一些主流邏輯客戶，我們只銷售個別產品線。

Did you have a follow-up, Chris?

克里斯，你還有後續問題嗎？

Yeah. And sorry, I should have said, this is Harris Fein on for Chris. For the second question, I mean, for a while now, there's been a lot of headlines on China competition. I guess, it would be helpful to hear if you're seeing anything in terms of changing behaviors or any sort of step-up in competitive intensity? And if so, where are you seeing it?

是的。抱歉，我應該說，這裡是哈里斯費恩代替克里斯發言。關於第二個問題，我的意思是，一段時間以來，關於中國競爭的新聞報導很多。我想，如果您觀察到任何行為上的變化或競爭強度有所提升，請告訴我，這將很有幫助？如果屬實，你在哪裡看到的？

Good question. When we look at the China market, we think the fundamentals of the China market are very similar to the rest of the world. In other words, they care about yield and performance. And so when you think through products in our space that improve yield and performance, you think of the Entegris products that do both. It's, one, continuing to deliver purity both at point of use and at source and then, of course, having high-purity materials that enter the process pure.

問得好。當我們審視中國市場時，我們認為中國市場的基本面與世界其他地區非常相似。換句話說，他們關心的是產量和性能。因此，當您考慮我們領域中能夠提高產量和性能的產品時，您就會想到 Entegris 的產品，它們能夠同時做到這兩點。第一，持續確保使用點和源頭的純度；第二，當然，也要確保進入生產過程的高純度材料是純淨的。

And so those two products, which is really product portfolios that Entegris has built upon, that improves yield and performance. And that's competitive irrespective of kind of where you are around the world.

因此，Entegris 真正建立的這兩款產品，也就是提高產量和效能的產品組合。無論你身處世界何處，這種競爭都存在。

Now then when you kind of hone in specifically in China, because they compete fiercely in China, our biggest obstacle in China is being able to guarantee to those customers that we can assure supply. So can we guarantee supply to those customers? And when we can guarantee supply to those customers, we find that they revert back to yield and performance being important.

現在，當我們把目光聚焦到中國市場時，由於中國的競爭非常激烈，我們在中國面臨的最大障礙是能夠向客戶保證我們可以保證供貨。那我們能保證向這些客戶供貨嗎？當我們能夠保證向這些客戶供貨時，我們發現他們又會重新重視產量和性能。

And so what you saw us do in 2025 was you saw us put a really concentrated effort into qualifying more manufacturing overseas, specifically for the China market as well as the rest of Asia. We got up to about 85% of products -- at least in first quarter, about 85% of our revenue for China, we're expecting to supply from region. And so we're able to guarantee that supply. We're going to continue to work on that throughout 2026, probably getting to a number around or even greater than 90%.

因此，你們看到我們在 2025 年所做的，就是集中精力使更多海外製造業獲得資質，特別是針對中國市場以及亞洲其他地區。我們大約有 85% 的產品——至少在第一季度，我們預計中國市場約 85% 的收入——將從該地區供應。因此，我們能夠保證供應。我們將繼續在 2026 年實現這一目標，可能會達到 90% 左右甚至更高。

And so I think as we continue to be able to qualify more products for Asia manufacturing, we then get to guarantee supply to those customers, then we get to compete in that market like we do around the world. And when we can compete fairly in those markets, we tend to do pretty well.

因此，我認為，隨著我們不斷有更多產品獲得亞洲製造的認證，我們就能保證向這些客戶供貨，然後我們就能像在世界各地一樣在這個市場上競爭。當我們能夠在這些市場中公平競爭時，我們往往能取得不錯的成績。

Charles Shi, Needham.

石正麗，李約瑟。

Hi, Dave. Good results. I want to ask you about NAND. I think we've spoken about this for a while, NAND sentiment-wise, pricing-wise, business-wise for your customers have inflected, but it doesn't appear that it's inflecting for you yet. Wondering what's your best prediction as of today? When do you think that business is going to pick up?

嗨，戴夫。結果良好。我想問你一些關於NAND的問題。我認為我們已經討論過這個問題一段時間了，NAND閃存在情感上、定價上、商業上，對你的客戶來說都已經產生了影響，但對你來說似乎還沒有產生影響。想知道你今天做出的最佳預測是什麼？你認為生意什麼時候會好轉？

And by the way, I did notice in your prepared remarks, double-digit content gain for this year, even on the back of pretty flattish or maybe single-digit MSI should do well for your NAND, but I just don't really feel like I see that in your March quarter guide or June quarter guide. Is it like more of a second half driver? And why it's so delayed versus your customers?

順便說一句，我注意到您在準備好的發言稿中提到，今年的內容增長將達到兩位數，即使MSI的增長相當平緩甚至可能只有個位數，這對您的NAND來說應該會很不錯，但我感覺在您的三月季度指南或六月季度指南中並沒有看到這一點。它更像是下半場型車手嗎？為什麼我們的產品交付速度比客戶的慢這麼多？

Thank you for the question. With respect to NAND, we think the underlying demand like you remains very strong. In fact, you started to see pricing kind of firm for NAND in the early second to mid-second half of 2025. You saw the pricing continue to perform well throughout the latter half then of '25 and then continue to grow through '26. We actually think that increased wafer starts on NAND has actually been very, let's call it, measured.

謝謝你的提問。關於NAND閃存，我們認為像您這樣的潛在需求仍然非常強勁。事實上，從 2025 年下半年初到中期，NAND 的價格開始趨於穩定。然後，您可以看到價格在 2025 年下半年繼續表現良好，並在 2026 年繼續增長。我們認為，NAND 晶圓開工量的增加其實是非常，我們姑且稱之為，穩定成長的。

And so we think incremental wafer starts for NAND will remain measured because what we're actually starting to see is we're starting to see some node transitions on NAND, where you get a premium where the NAND producers get a premium pricing for bit density. And so we're finally starting to see some of those node migrations that we expected on NAND going from kind of, call it, 250-ish layer count to roughly 300-ish layer count. So as you kind of grow that layer count kind of 20% bit density growth on a year-over-year basis, it's a premium product for them. We get benefit from those incremental layers, but it effectively consumes capacity.

因此，我們認為 NAND 晶圓的增量開工量將保持緩慢，因為我們實際上開始看到 NAND 的一些節點過渡，在這些節點過渡中，NAND 生產商會因位元密度而獲得溢價。因此，我們終於開始看到一些我們預期的節點遷移，NAND 的層數從大約 250 層增加到大約 300 層。因此，隨著層數逐年增加，位密度逐年增加 20%，對他們來說，這是一款高階產品。我們從這些增量層中受益，但這實際上會消耗容量。

And so I think what you're hearing from us is we like the incremental layers. Incremental layers brings higher content per wafer for Entegris, but the actual increased wafer starts -- we're waiting for the NAND producers to effectively drive those wafer starts.

所以我覺得你們從我們這裡聽到的意思是，我們喜歡這種循序漸進的方式。對於 Entegris 而言，增量層技術可以提高每片晶圓的容量，但實際晶圓開工量的增加——我們還在等待 NAND 生產商有效地推動這些晶圓開工量。

So this is the trade-off and environment right now that when we look at Q1, we think we've got a solid guide for Q1. We've got an indication kind of for second quarter that we feel quite comfortable with. And then we'll leave it for the NAND producers to determine the rate and pace, both of the layer count as well as incremental wafer growth.

所以，這就是我們目前面臨的權衡和環境，當我們展望第一季時，我們認為我們已經對第一季有了可靠的指導。我們對第二季的業績已經有了一個比較樂觀的預期。然後，我們將把層數和晶圓增量的成長速度和步伐，留給 NAND 生產商來決定。

Did you have a follow-up, Charles?

查爾斯，你還有後續問題嗎？

Thanks, Dave. Yes, I do. The second question, thanks for the China color, the amount of supply supporting the China market. But I wonder if you have a view on how your China business is going to grow in '26? And if you can, what was the China growth number for 2025?

謝謝你，戴夫。是的，我願意。第二個問題，感謝您提供的中國色彩，支撐中國市場的供應量是多少？但我很想知道您對貴公司在中國的業務在 2026 年的發展前景有何看法？如果可以的話，請問中國2025年的成長預期是多少？

Sure. So in terms of the China business growth in '26 -- I'll let Linda in a moment talk about the '25 growth. But the areas that we expect to grow in '26 for China, one, we think some of the CapEx-related areas will grow in 2026, specifically FM and perhaps FOUPs. We expect LMC or liquid filtration to perform in 2026 in the China market as well as some of the CMP products. There's probably a couple of others in there, but we expect the China market to have growth in '26, and we think it's kind of underpinned by the areas that I mentioned.

當然。所以，關於 2026 年中國業務的成長——稍後我會讓琳達談談 2025 年的成長情況。但我們預計 2026 年中國市場將成長的領域包括：我們認為一些與資本支出相關的領域將在 2026 年成長，特別是設施管理 (FM) 和 FOUPs。我們預計 LMC 或液體過濾技術以及一些 CMP 產品將在 2026 年在中國市場取得良好表現。可能還有其他一些因素，但我們預計中國市場在 2026 年將實現成長，我們認為這在一定程度上是由我提到的那些領域所支撐的。

Linda, do you have the China growth number for '25?

琳達，你有2025年中國的經濟成長數據嗎？

Yeah. So China has remained '24 and '25, approximately 21% of (technical difficulty) actual dollars are down slightly. But as we've talked about before, and Dave highlighted some of the reasons why our China customers like our products, we've been able to maintain very solid performance in China.

是的。所以中國仍然是“24”和“25”，大約 21%（技術難度）實際美元略有下降。但正如我們之前討論過的，戴夫也強調了中國客戶喜歡我們產品的一些原因，我們在中國一直保持著非常穩定的表現。

Melissa Weathers, Deutsche Bank.

梅麗莎‧韋瑟斯，德意志銀行。

I wanted to touch on something you flagged in the prepared remarks, the potential impact of memory shortages and pricing on the electronics market and any decreased production we could see from that. So I know you're calling for mid-ish single-digit growth, which mid-ish, that's a new word for me that I've learned today. But could you help us -- what are you embedding in that outlook with respect to like any demand destruction from the memory shortages?

我想談談您在準備好的發言稿中提到的一點，即記憶體短缺和價格對電子市場的潛在影響，以及由此可能導致的產量下降。我知道你們呼籲實現個位數的中等成長，而「中等」這個詞，我今天才學到。但是您能否幫我們解答一下——您在展望未來時，考慮到了內存短缺可能造成的任何需求下降？

Yeah. I think when we think about our first-quarter guide, obviously, we didn't factor anything into the first quarter. We didn't really factor anything into kind of our at least indication for second quarter at this stage. I think really what we were doing was we were just flagging it as potential. We're expecting mainstream to be stable this year, perhaps even slightly improving versus 2025.

是的。我認為，當我們考慮第一季業績指引時，顯然，我們並沒有將任何因素納入第一季的預測。現階段，我們並沒有將任何因素納入第二季業績的預測。我認為我們當時所做的，實際上只是把它標記為一種潛在可能性。我們預計今年主流市場將保持穩定，甚至可能比 2025 年略有改善。

But a lot of the mainstream logic, a lot of that production is reliant on some form of memory. And so we're really just, at this stage, calling it out as a flag to watch for the second half of 2026. I think that's where the impact would be if there was any.

但很多主流邏輯，很多生產過程都依賴某種形式的記憶。因此，現階段我們只是將其視為 2026 年下半年需要關注的一個訊號。我認為如果有影響的話，那應該就在這裡。

Did you have a follow-up, Melissa?

梅麗莎，你還有後續問題嗎？

Yes, I did. On the capacity shutdowns that you've done in the fourth quarter and that you might do in the first half of this year? I'm sorry if I missed it, but have you given any timing on when we could expect those closures to impact gross margins?

是的，我做了。關於您在第四季實施的產能關閉措施，以及您今年上半年可能實施的產能關閉措施？如果我錯過了相關信息，請見諒，但您有沒有給出這些門市關閉預計何時會對毛利率產生影響的時間表？

Yeah. Let me broaden the question out to gross margin, and then I'll answer your question specifically. In the third quarter, which was my first as CEO, we guided kind of a trough for gross margin between 43% and 44%. Third quarter gross margin was 43.6%. We were able to increase that to 44% in fourth quarter.

是的。我先把問題擴大到毛利率，然後再具體回答你的問題。第三季度，也就是我擔任執行長的第一個季度，我們預計毛利率將維持在 43% 到 44% 的低潮水準。第三季毛利率為43.6%。第四季度，我們成功將這一比例提高到了 44%。

In first quarter on slightly lower volume, we're still guiding you on a normalized basis to kind of 44%. And so we feel like at this point, incremental volume growth for us will drive incremental gross margin from these levels. So from that perspective, we feel quite good about it.

第一季成交量略有下降，但我們仍以正常水準預測，成交量約為 44%。因此，我們認為，在目前這個階段，銷售量的逐步成長將推動毛利率的逐步提高。所以從這個角度來看，我們感覺相當不錯。

We were able to rationalize one facility in the fourth quarter of this year -- excuse me, of 2025. As we go forward into 2026, we'll get some modest benefit from that in terms of utilization. When we think about what I mentioned in the script, which was we're expecting to rationalize another facility in the first half of 2026, then you would expect to get some minor benefit on a go-forward basis through the remainder of '26.

我們得以在今年第四季（抱歉，應該是 2025 年第四季）對一家工廠進行合理化改造。展望 2026 年，我們在利用率方面將從中獲得一些適度的好處。當我們考慮到我在劇本中提到的內容，即我們預計在 2026 年上半年對另一個設施進行合理化改造，那麼你就會期望在 2026 年剩餘的時間裡獲得一些小的收益。

And we'll continue to go. We'll continue to both ramp our facilities in Taiwan. We'll qualify our facility in Rockrimmon, and we'll continue to look at our manufacturing footprint and the rate and pace at which the utilization is improving and make the decisions that you would expect us to make. So for all those reasons, we feel quite good about the trajectory that we're on, and we feel like the execution is in front of us to perform.

我們將繼續前進。我們將繼續擴大在台灣的設施規模。我們將對位於羅克里蒙的工廠進行資格認證，並繼續關注我們的生產佈局以及利用率提高的速度和節奏，並做出你們期望我們做出的決定。綜上所述，我們對目前的發展軌跡感到非常滿意，並且覺得機會就在眼前，只需全力以赴即可。

Elizabeth Sun, Citi.

伊麗莎白·孫，花旗銀行。

The first one, I guess, it's -- Dave, you briefly talked about AI is 5% of the wafer starts market, but I believe your content is much higher for AI-related products. So I'm just wondering if you have looked from the perspective of like how much is AI as a percentage of your total revenue?

首先，我想說的是——戴夫，你之前簡要地提到人工智慧只佔晶圓新創公司市場的 5%，但我相信你關於人工智慧相關產品的內容要多得多。所以我想知道您是否從人工智慧佔總收入百分比的角度來看待？

Well, we tried to give you some -- at least some indication and some color in the prepared commentary. We mentioned that about 60% of our revenue in 2025 was driven by advanced nodes, so advanced manufacturing. That's both logic and memory. And then as you think about going forward, AI is a big part of that growth in advanced manufacturing.

嗯，我們試圖在準備好的評論中給你一些提示和一些細節。我們提到，到 2025 年，我們約 60% 的收入將由先進節點（即先進製造）驅動。這既涉及邏輯，也涉及記憶。展望未來，人工智慧是先進製造業發展的重要組成部分。

And for example, advanced logic, we designate that as 7 nanometers and below. And then the last kind of 2 generations of memory, the newest one plus current manufacturing. That's how we define it advanced manufacturing nodes.

例如，對於先進邏輯電路，我們將其定義為 7 奈米及以下。最後還有兩代記憶體，即最新款加上現有型號。這就是我們對先進製造節點的定義。

So from that perspective, it's about 60% of our total revenue as a company. We expect that to grow going forward because we expect both incremental capacity to come online, mostly to satisfy AI. And then, of course, we expect memory to continue to grow as AI drives more growth through memory.

從這個角度來看，它約占我們公司總收入的 60%。我們預期未來這數字還會成長，因為我們預期會有更多新增產能投入使用，主要是為了滿足人工智慧的需求。當然，隨著人工智慧透過記憶體推動更多成長，我們預期記憶體需求也會持續成長。

Did you have a follow-up, Elizabeth?

伊莉莎白，你還有後續問題嗎？

Yes, I do. Thanks for that. And on advanced packaging, I understood it has been a smaller part of your total revenue, but I think I heard in your prepared remarks that you are expecting some POR wins or some -- and you are doing some efforts on HBM side. So I was wondering what's your expectation for advanced packaging revenue this year?

是的，我願意。謝謝。至於先進封裝，我了解到它在你們的總收入中所佔比例較小，但我認為我從你們的準備發言稿中聽到，你們預計會贏得一些 POR 訂單，或者——而且你們正在 HBM 方面做出一些努力。所以我想知道您對今年先進封裝產業的收入有何預期？

Yeah. Advanced packaging is an area that grew nicely for us in 2025, representing roughly $100 million plus/minus. We expect that to continue to grow nicely in 2026. And this is an area where we've made a little bit more concerted effort to grow across a number of product lines. So those product lines are sampling now, and we have some others that will sample later in the year for the advanced packaging market. So we're not expecting as much benefit in '26 as perhaps we could get in '27 and beyond.

是的。先進包裝是我們2025年發展勢頭良好的領域，預計市場規模將達到1億美元左右。我們預計到 2026 年，這一數字將繼續保持良好的成長勢頭。在這個領域，我們已經投入了更多精力，力求在多個產品線上成長。所以這些產品線目前正在進行樣品測試，我們還有一些其他產品線將在今年稍後針對先進包裝市場進行樣品測試。因此，我們預計 2026 年不會像 2027 年及以後那樣獲得那麼多收益。

But it is a growing part of the market. It is starting to look more and more like some of the more advanced nodes in terms of its complexity and the challenges that our customers face, and it's an area that ultimately will play well with some of our product portfolio. So you're going to see a little bit more of a focused effort from us in the space, and we're cautiously optimistic.

但這部分市場佔有率正在不斷成長。就其複雜性和客戶面臨的挑戰而言，它越來越像一些更先進的節點，而且它最終將與我們的一些產品組合很好地配合。所以你們將會看到我們在這個領域投入更多精力，我們對此持謹慎樂觀態度。

Edward Yang, Oppenheimer.

愛德華楊，奧本海默。

Just wanted to touch again your leverage to memory market trends. Obviously, a lot of excitement there. Can you just first remind us your ballpark total revenue exposure to memory overall and where it could go in an up cycle? And maybe also clarify the trade-off between what you were talking about layer count benefit versus wafer counts? Like with CMP, I would think that you'd be relatively indifferent, but perhaps in other parts of the business, you get more or less revenue.

想再談談您對記憶體市場趨勢的影響。顯然，那裡非常熱鬧。您能否先簡單介紹一下您在記憶體業務上的整體總收入佔比，以及在經濟週期上漲階段記憶體業務可能達到的水平？或許還可以澄清一下您之前提到的層數優勢與晶圓數量之間的權衡取捨？就像 CMP 一樣，我認為你會相對漠不關心，但或許在業務的其他方面，你會獲得更多或更少的收入。

Sure. So Memory is about 30% of our total revenue. It's roughly split equally, and I'm standing back and squinting on an approximate basis. So say about half of it is NAND. About half of it is DRAM. As you know, DRAM has performed very well in 2025, very high utilization rates across DRAM, more of DRAM moving from kind of individual sales of DDR5 into HBM. And so as that migration happens, there's some incremental content associated with that, but it's not the same as what you would get, for example, from an incremental wafer.

當然。因此，記憶體業務約占我們總收入的 30%。大致平分，我站在遠處瞇著眼睛，大致判斷了一下。假設其中大約一半是 NAND 快閃記憶體。其中大約一半是DRAM。如你所知，DRAM 在 2025 年表現非常出色，DRAM 的利用率非常高，DRAM 的銷售方式也從 DDR5 的獨立銷售轉向 HBM。因此，隨著這種遷移的發生，會產生一些增量內容，但這與例如從增量晶圓中獲得的內容並不相同。

And so there is incremental content when you go from stand-alone DDR5 to HBM, but there would be more total benefit if you would start and generate more total wafer starts. But the technology capacity for DRAM is pretty fully utilized exiting '25, and we see it remaining that way through 2026.

因此，從獨立DDR5到HBM，內容會增加，但如果能啟動更多晶圓並產生更多晶圓啟動，則整體收益會更大。但到 2025 年底，DRAM 的技術產能已基本充分利用，我們預計這種情況將持續到 2026 年。

The other half roughly of this 30% of revenue is NAND. NAND is probably around 85% utilization. That's down a bit from where it was in peak in the 2022 type time frame. But what you're seeing in NAND is you're seeing that 15% available capacity, you're starting to see it get absorbed by incremental wafer count -- or excuse me, by incremental layer count.

這30%的收入中，另一半大約來自NAND快閃記憶體。NAND快閃記憶體的利用率可能在85%左右。這比 2022 年左右的峰值有所下降。但你在 NAND 中看到的是，那 15% 的可用容量，開始被晶圓數量的增加所吸收——或者更準確地說，是被層數的增加所吸收。

And so as those incremental layers happen, we're relatively indifferent on whether you're absorbing that capacity on an incremental layer basis or on a wafer basis. I think on a general statement, we would say we would be relatively indifferent. I think the reality is we would probably get slightly more incremental benefit from a wafer, from a full wafer start, but we do get benefit from both.

因此，隨著這些增量層的出現，我們對於你是按增量層還是按晶圓吸收該容量都相對無所謂。總的來說，我認為我們會說我們對此比較漠不關心。我認為現實情況是，從晶圓生產或整片晶圓生產開始，我們可能會獲得略微更多的增量收益，但兩者都能帶來收益。

Did you have a follow-up, Edward?

愛德華，你還有後續問題嗎？

Yes, I do. So Dave, you mentioned focusing on winning new PORs. And I was just wondering has your go-to-market approach changed there? Are you cross-selling more intradivision between -- and also interdivision between Material Solutions and Advanced Purity Solutions?

是的，我願意。戴夫，你剛才提到要專注於贏得新的業主關係。我想知道你們在那裡的市場推廣策略是否有所改變？你們是否在部門內部（例如材料解決方案部和高級純度解決方案部）以及部門之間（例如材料解決方案部和高級純度解決方案部）進行更多交叉銷售？

Yeah. I think you've seen us really continue a lot of the good sales focus that was in place before I joined. We've kind of continued that momentum since I've joined the company. I think what perhaps we've been able to focus a bit more on now is we have been able to focus a bit more on selling the complete portfolio of products. We've always engaged very well on a technology road map basis. So that's something we don't want to change. We want to continue those best efforts where we're focused on our customers' node transitions and technology road map, many of which are kind of several years out.

是的。我認為你們已經看到，我們繼續保持了我加入之前就已建立的許多良好的銷售策略。自從我加入公司以來，我們一直保持著這種發展勢頭。我認為我們現在或許能夠更專注於銷售全系列產品。我們一直以來都以技術路線圖為基礎進行非常良好的合作。所以這是我們不想改變的。我們希望繼續努力，專注於客戶的節點過渡和技術路線圖，其中許多路線圖的實現時間都還有好幾年。

But then as we do that, we also want to layer that in that road map and that engagement on road map. We want to layer in our other product lines where we bring best-in-class filtration, best-in-class purification, best-in-class wafer handling and fluid management and bring that together with some of these longer technology road maps such as the CMP process, the deposition process, the etch process. So you've seen us try to make a concerted effort with not only continuing the good engagement on the technology road map, many times of which it's looking out several years, but then also bring more of the other product lines along in that engagement and discussion for today.

但是，在這樣做的時候，我們也希望將這些內容融入路線圖和路線圖參與度中。我們希望將我們其他的產品線融入其中，這些產品線提供一流的過濾、一流的淨化、一流的晶圓處理和流體管理，並將這些與一些更長遠的技術路線圖結合起來，例如 CMP 製程、沉積製程、蝕刻製程。所以你們已經看到，我們努力做出協調一致的努力，不僅繼續就技術路線圖進行良好的溝通（很多時候，技術路線圖的規劃都著眼於未來幾年），而且還讓其他產品線更多地參與到今天的溝通和討論中來。

Bhavesh Lodaya, BMO.

Bhavesh Lodaya，BMO。

Congrats on a nice quarter. And certainly, all the best to Linda, and welcome, Jeff as well. Dave, you shared a lot of color for the short term. So maybe a longer-term question, if I look at where MSI -- overall MSI stands today, we are still under prior peaks, around 13% lower, it seems. And we don't see a week go by without news of higher and higher CapEx spends in AI data centers. If end market growth kind of hangs in there, I'm curious on your view as to how that plays for MSI over the next three years -- next three to four years as these capacities are brought online?

恭喜你本季業績出色。當然，也祝福琳達一切順利，同時歡迎傑夫的加入。戴夫，你分享了很多短期內能帶來好運的色彩。所以，從長遠來看，如果我看一下 MSI 目前的狀況——總體而言，MSI 目前仍低於之前的峰值，似乎低了大約 13%。幾乎每週都有新聞報導人工智慧資料中心的資本支出不斷增加。如果終端市場成長保持穩定，我很想知道您認為這對 MSI 未來三年（或未來三到四年，隨著這些產能陸續上線）的發展會有什麼影響？

Thanks, Bhavesh. Well, you're right, it is somewhat of a bifurcated market where we're talking about maybe we need to start new fabs, but yet you're looking at MSI and you're looking at some underutilization in some areas of the market where you're not fully utilized as of yet. So let me -- maybe the best way to answer this question is to kind of break it down into its constituent parts.

謝謝你，Bhavesh。沒錯，這是一個有點兩極分化的市場，一方面我們討論可能需要新建晶圓廠，另一方面，看看微星，你會發現市場某些領域的產能利用率不足，尚未得到充分利用。所以，或許回答這個問題最好的方法就是把它分解成各個組成部分。

If you think about advanced logic, which we define as 7 nanometers and below, advanced logic is pretty fully utilized, particularly advanced logic below 5 nanometers. And so you're seeing a lot of incremental capacity and focus on incremental capacity for the most advanced nodes, and that's growing total capacity for this 7-nanometer and below advanced logic category. But it's slow, and it takes time to grow 2-nanometer base, and it takes time to then also transition to new nodes, for example, like 1.4 nanometer.

如果你考慮先進邏輯，我們將其定義為 7 奈米及以下，那麼先進邏輯已經得到了相當充分的利用，特別是 5 奈米以下的先進邏輯。因此，我們看到最先進的節點的產能大幅提升，並且更加重視產能的提升，這使得 7 奈米及以下先進邏輯類別的總產能不斷增長。但是速度很慢，發展 2 奈米的基礎需要時間，而且過渡到新的節點（例如 1.4 奈米）也需要時間。

But that space is pretty fully utilized or that category, 7 nanometers and below is pretty fully utilized. And so what you need is you need more capacity to grow MSI. DRAM is also very highly utilized. So if you want to grow a lot of MSI and DRAM, you need incremental capacity. And I think you're starting to see some of the producers of DRAM think through where and how do you add that incremental capacity and over what time frame.

但是，7 奈米及以下這個尺寸範圍已經相當充分利用了。所以，你需要的是更大的發展空間來支持MSI的發展。DRAM 的利用率也非常高。因此，如果您想大量開發 MSI 和 DRAM，則需要逐步增加容量。我認為，一些DRAM生產商已經開始思考在哪裡以及如何增加增量產能，以及在多長時間內增加產能。

And so I think you're seeing the market kind of recognize that DRAM is very tight. There's not a lot of incremental available capacity and how do you best -- how do you best drive incremental capacity? Is it through incremental tools and perhaps efficiencies? Is it through groundbreaking? I think you'll see the market kind of make some moves on this front through 2026.

所以我覺得你會看到市場開始意識到DRAM供應非常緊張。可用的新增產能並不多，那麼如何才能最好地推動新增產能呢？是透過漸進式工具和效率提升來實現的嗎？這是突破性的嗎？我認為到 2026 年，市場將在這方面有所行動。

NAND, there is some available utilization. My best guess is that the utilization that's available for NAND will be consumed by layer count -- some combination of layer count and MSI growth, but I think it's more layer count than MSI. I think ultimately, you get some of both. But I would say, at least at this point in time, I think it's more layers with lesser amounts of incremental wafer growth or MSI growth. And then I think ultimately, that market perhaps will have to look at adding capacity, but I think that's probably second half, perhaps even latter part of second half type of decision before we see what's happening there.

NAND快閃記憶體還有一些可用空間。我最好的猜測是，NAND 的可用容量將被層數消耗掉——層數和 MSI 增長的某種組合，但我認為層數比 MSI 增長的影響更大。我認為最終你會兩者兼得。但我認為，至少就目前而言，我認為層數更多，晶圓增量增長或 MSI 增長量更少。然後我認為，最終，該市場或許不得不考慮增加產能，但我認為這可能是下半年，甚至可能是下半年後期才會做出的決定，之後我們才能看到那裡的情況。

And then we're now kind of on to the crux of your question, which is mainstream. Mainstream is the bulk of the logic market in terms of MSI. And mainstream has been slow to recover and mixed. The good news is that feedback on mainstream inventory, it looks like inventory levels are relatively healthy. But the rate and pace of growth on the mainstream part or MSI growth on the mainstream part of the market, that's a bit unclear right now. And that is the bulk of MSI for logic. I'm talking total logic, the bulk of the capacity sits in mainstream. And I think that's the part that people are looking at and wondering what's the rate and pace of growth in mainstream because it's such a big part of logic MSI.

而現在，我們終於觸及了你問題的核心，也就是主流問題。就 MSI 而言，主流是邏輯市場的大部分。主流市場復甦緩慢且情勢不一。好消息是，根據主流庫存的回饋，庫存水準似乎相對健康。但是，主流市場或MSI在主流市場的成長速度和步伐，目前還不太清楚。這就是 MSI 邏輯部分的主體。我說的是完全合乎邏輯的，大部分容量都集中在主流領域。我認為這正是人們關注的重點，他們想知道主流市場的成長速度和節奏，因為這是邏輯 MSI 的重要組成部分。

Did you have a follow-up, Bhavesh?

Bhavesh，你還有後續問題嗎？

Yes. In your equation here, so one side of growth comes from utilization, the other side comes from, I would say, content gains or a mix benefit as more layers come in. Where do you put outperformance metrics in this? Do you count the content gains and outperformance? Or would outperformance be over and above these two things?

是的。在你的等式中，成長的一邊來自利用率，另一邊，我認為來自內容成長，或是隨著更多層次的加入而帶來的混合效益。這裡該把超額收益指標放在哪裡？您是否將內容收益和超額收益計算在內？或者說，超額收益是指除了這兩點之外的任何其他因素？

Yeah. We think of outperformance really from a revenue perspective. And so to the extent that you grow layer count, it actually increases content per wafer. And so from that perspective, that incremental content per wafer would show up in revenue, and we would count that as outperformance to the market because it would be incremental kind of to what the normal market would see. So that's how we would think about that.

是的。我們認為超額報酬主要從營收角度來看。因此，隨著層數的增加，實際上每片晶圓的內容量也會增加。因此從這個角度來看，每片晶圓的增量內容將體現在收入上，我們將其視為超越市場的表現，因為它相對於正常市場而言是增量的。這就是我們看待這個問題的方式。

John Roberts, Mizuho.

約翰·羅伯茨，瑞穗銀行。

Linda, best wishes, and welcome again, Jeff is also here. Could you talk about the weaker parts of the business? So it looks like, again, FOUPs were probably down. Gas filtration was probably down. Is that all related just to the new fab construction activity being down? And did they continue down in the first quarter?

琳達，祝你一切順利，再次歡迎你，傑夫也在這裡。能談談公司比較弱的環節嗎？所以看來，FOUP 可能又下降了。氣體過濾系統可能故障。這是否都與新廠建設活動減少有關？他們在第一季是否繼續下滑？

Yeah. So when you look at 2025, fab construction CapEx was down high single digits, call it, 7%-ish. Our CapEx-related business was down about the same amount. And it was really driven by two product lines, both within APS, one being fluid management and the other one being FOUPs. Those were the two that were down the most and very much in line with that fab construction CapEx.

是的。因此，展望 2025 年，工廠建設資本支出下降了接近兩位數，大概下降了 7% 左右。我們的資本支出相關業務也下降了大致相同的幅度。而這其實是由 APS 旗下的兩條產品線所推動的，一條是流體管理，另一條是 FOUP。這兩個項目的降幅最大，而且與該工廠的建設資本支出非常吻合。

So when we think about 2026 and you think about perhaps fab construction CapEx being at least flat and perhaps at this point, up slightly, though to be determined how much through the course of '26. You're seeing a corresponding recovery in both those two businesses. So our expectations for the year is that fluid management and FOUPs will have a better year versus 2025 on the basis of fab construction CapEx at a minimum being flat and most likely being up, but to be determined how that kind of layers in throughout the year.

所以，當我們展望 2026 年，並考慮到工廠建設資本支出可能至少會持平，甚至可能略有上升，儘管在 2026 年期間上升多少還有待確定。你可以看到這兩個行業的業務都在相應復甦。因此，我們對今年的預期是，流體管理和 FOUPs 將比 2025 年表現更好，因為晶圓廠建設資本支出至少會持平，而且很可能會上升，但這種影響如何貫穿全年還有待確定。

Did you have a follow-up, John?

約翰，你還有後續問題嗎？

Well, I'll follow up with -- how about the March quarter?

那我再補充一下──三月的季度狀況如何？

Sorry, you cut out on me what -- the question was what again?

抱歉，你剛才突然中斷了我的對話──問題是，什麼來著？

Will they be down in the March quarter or more flattish in the March quarter?

3月季度業績會下滑還是會維持穩定？

I think at this point, and I'll speak a little bit on a sequential basis. We'll see ultimately how much kind of revenue for that -- those specific product lines fall in first quarter versus second quarter. But I think it's fair to say that we're already starting to see some recovery in those product lines from an order pattern perspective for 2026.

我認為現在我可以按順序稍微講一下了。最終我們將看到這些特定產品線在第一季和第二季分別能帶來多少收入。但我認為可以公平地說，從 2026 年的訂單模式來看，我們已經開始看到這些產品線出現一些復甦跡象。

And then ultimately, the timing of whether it's first quarter or second quarter, that will be determined by delivery and our customers, but I do think that we've seen order patterns improve for those product lines as we're very early here in 2026.

最終，具體是第一季度還是第二季度，將取決於交付情況和客戶需求，但我認為，在 2026 年伊始，我們已經看到這些產品線的訂單模式有所改善。

This does conclude the Q&A portion of today's call. I would now like to hand it back to Jeff Schnell for any additional or closing remarks.

今天的電話會議問答環節到此結束。現在我把發言權交還給傑夫·施內爾，讓他補充或作總結發言。

Great. Thank you for joining the call today and your continued interest in Entegris. Please reach out if you have any follow-ups.

偉大的。感謝您今天參加電話會議，並感謝您一直以來對 Entegris 的關注。如有任何後續事宜，請與我們聯絡。

Thank you. This concludes today's Entegris fourth-quarter 2025 earnings conference call. Please disconnect your line at this time, and have a wonderful day.

謝謝。本次 Entegris 2025 年第四季財報電話會議到此結束。請您暫時斷開電話，祝您有美好的一天。