新思科技 (SNPS) 2026 Q3 法說會逐字稿

Ladies and gentlemen, welcome to the Synopsys earnings conference call for the third-quarter fiscal year 2026. (Operator Instructions) As a reminder, today's call is being recorded.

各位女士、先生，歡迎參加新思科技（Synopsys）2026 會計年度第三季財報電話會議。（接線員指示）提醒各位，今天的電話會議將被錄音。

At this time, I would like to turn the conference over to Tushar Jain, Vice President. Please go ahead.

此刻，我想將會議交給副總裁 Tushar Jain。請開始。

Good afternoon, everyone. Welcome to Synopsys' third-quarter fiscal year 2026 earnings call. With us today are Sassine Ghazi, President and CEO of Synopsys; and Shelagh Glaser, CFO. Before we begin, I'd like to remind everyone that during the course of this conference call, Synopsys will discuss forecasts, targets and other forward-looking statements regarding the company and its financial results.

各位下午好。歡迎參加新思科技 2026 會計年度第三季財報電話會議。今天與會者包括新思科技總裁暨執行長 Sassine Ghazi，以及財務長 Shelagh Glaser。在開始之前，我想提醒各位，在本次電話會議期間，新思科技將討論與公司及其財務結果相關的預測、目標及其他前瞻性陳述。

While these statements represent our best current judgment about future results and performance as of today, our actual results are subject to many risks and uncertainties that could cause actual results to differ materially from what we expect. In addition to any risks that we highlight during this call, important factors that may affect our future results are described in our most recent SEC reports and today's earnings press release.

雖然截至今日，這些陳述代表我們對未來結果與表現的最佳當前判斷，但我們的實際結果仍受多項風險與不確定性影響，可能導致實際結果與我們的預期出現重大差異。除本次電話會議中我們可能提及的任何風險外，可能影響我們未來結果的重要因素，已載於我們最新的 SEC 報告以及今日的財報新聞稿中。

In addition, we will refer to certain non-GAAP financial measures during the discussion. Reconciliations to their most directly comparable GAAP financial measures and supplemental financial information can be found in the earnings press release, financial supplement and 8-K that we released earlier today. All of these items, plus the most recent investor presentation, prepared remarks and Investor Day information can be found on our website at www.synopsys.com.

此外，在討論中我們將引用若干非 GAAP 財務衡量指標。與其最直接可比的 GAAP 財務衡量指標之調節表，以及補充財務資訊，可見於我們今日稍早發布的財報新聞稿、財務補充資料與 8-K。上述所有資料，加上最新的投資人簡報、事先準備的講稿與投資人日資訊，皆可於我們網站 www.synopsys.com 查閱。

With that, I'll turn the call over to Sassine Ghazi.

接下來，我把電話交給 Sassine Ghazi。

Good afternoon. Synopsys delivered an outstanding third-quarter with revenue, non-GAAP operating margin and EPS all exceeding the high end of guidance. These results reflect broad-based strength, including outperformance in EDA and Ansys and continued growth in IP. We are raising our full year revenue, non-GAAP operating margin and EPS guidance.

各位下午好。新思科技第三季表現出色，營收、非 GAAP 營業利益率與每股盈餘（EPS）皆超越財測區間上緣。這些成果反映出廣泛的強勁動能，包括 EDA 與 Ansys 的優於預期表現，以及 IP 的持續成長。我們上調全年營收、非 GAAP 營業利益率與 EPS 的財測指引。

In addition, we expect EDA growth to accelerate in Q4 and to deliver double-digit growth for the full year. The key takeaway from Q3 is that the fundamentals across our portfolio are strengthening. EDA is accelerating. Design IP has returned to growth and Ansys is performing strongly while beginning to create new growth opportunities across the combined portfolio.

此外，我們預期 EDA 成長將在第四季加速，並在全年實現雙位數成長。第三季的關鍵重點是：我們整體產品組合的基本面正在轉強。EDA 正在加速。設計 IP 已恢復成長，而 Ansys 表現強勁，同時也開始在合併後的產品組合中創造新的成長機會。

This quarter marked one year since the Ansys acquisition close. And in Q3, we launched our first joint Synopsys and Ansys solutions, Multiphysics Fusion. I want to recognize our global team for executing with focus and agility to integrate our world-class capabilities. The combination has strengthened our competitive position, expanded our opportunity and enabled us to deliver differentiated solutions addressing the physics challenges of modern chip design.

本季也標誌著 Ansys 併購案完成交割滿一週年。在第三季，我們推出首批新思科技與 Ansys 的聯合解決方案：Multiphysics Fusion（多物理融合）。我要肯定我們全球團隊以專注與敏捷的執行力，整合我們世界級的能力。此一結合強化了我們的競爭地位、擴大了機會版圖，並使我們能提供具差異化的解決方案，以因應現代晶片設計的物理挑戰。

Industry trends are aligned to our strategy and our strengths as the leading provider of engineering solutions from silicon to systems. Unprecedented design complexity driven by the demands of AI is fueling the need for the IP and design solutions necessary to deliver next-generation AI compute, infrastructure and physical AI systems. These trends are evident in our Q3 results.

產業趨勢與我們的策略一致，也與我們作為從矽到系統工程解決方案領導供應商的優勢相契合。AI 需求所驅動的前所未有設計複雜度，正推升對 IP 與設計解決方案的需求，以交付下一代 AI 運算、基礎設施與實體 AI 系統。這些趨勢在我們第三季的結果中清楚可見。

Starting with Design Automation, which achieved healthy growth in Q3, underpinned by strength in EDA, including record hardware revenue. We expect this momentum to continue with EDA growth accelerating to double digits in Q4 and for the full year. Design activity is highest among AI and high-performance compute customers, developing increasingly specialized chips with multi-die architectures, more complex packaging and system requirements.

先從設計自動化（Design Automation）談起，第三季實現穩健成長，主要受 EDA 的強勁表現支撐，其中包括創紀錄的硬體營收。我們預期這股動能將延續，EDA 成長在第四季加速至雙位數，並在全年維持雙位數成長。設計活動最為活躍的客戶群來自 AI 與高效能運算領域，他們正在開發日益專用化的晶片，採用多晶粒（multi-die）架構、更複雜的封裝與系統需求。

These are all areas where Synopsys leads. AMD's recently launched Instinct MI455X GPU is a good example. To deliver this highly sophisticated new product series, AMD leveraged Synopsys' 3DIC Compiler, the industry's only exploration-to-sign-off platform for multi-die and advanced package co-design and optimization.

這些都是新思科技領先的領域。AMD 近期推出的 Instinct MI455X GPU 就是一個很好的例子。為了交付這一高度精密的新產品系列，AMD 採用新思科技的 3DIC Compiler——業界唯一從探索到簽核（exploration-to-sign-off）的平台，可用於多晶粒與先進封裝的協同設計與最佳化。

The complex software and system requirements of AI compute also drive demand for our hardware-assisted verification solutions. We secured 12 new and 66 repeat HAV customer wins in the quarter. As I mentioned, the launch of Multiphysics Fusion was a major EDA milestone in Q3 and creates a new growth opportunity for Synopsys.

AI 運算的複雜軟體與系統需求，也帶動對我們硬體輔助驗證（HAV）解決方案的需求。本季我們取得 12 個新的 HAV 客戶贏單，以及 66 個既有客戶的重複贏單。如我所提到，Multiphysics Fusion 的推出是第三季 EDA 的一項重大里程碑，並為新思科技創造新的成長機會。

Multiphysics Fusion combines Synopsys and Ansys technology in the industry's only solution with thermal analysis fully integrated into the chip design flow. Customers, including NVIDIA, Cisco, MediaTek and Samsung Foundry have validated up to 10 times faster design closure and three times faster runtime. This drives greater value for our customers and also for our products.

Multiphysics Fusion 結合新思科技與 Ansys 技術，是業界唯一將熱分析完整整合至晶片設計流程中的解決方案。包括 NVIDIA、Cisco、MediaTek 與 Samsung Foundry 在內的客戶已驗證，其設計收斂（design closure）速度最高可提升 10 倍，且執行時間（runtime）可加快 3 倍。這為客戶帶來更高價值，也同樣提升我們產品的價值。

We expect these add-on capabilities to begin contributing to EDA growth in 2027. Agentic AI is another growth vector for Synopsys, and we demonstrated strong progress in Q3. At the DAC conference, with NVIDIA, we showcased a fully autonomous long-running design verification agent that can orchestrate the entire chip verification cycle and deliver up to 50 times faster time to validated RTL while achieving 20% additional coverage improvement.

我們預期這些附加能力將在 2027 年開始對 EDA 成長做出貢獻。代理式 AI（Agentic AI）是新思科技的另一個成長向量，我們在第三季展現了強勁進展。在 DAC 大會上，我們與 NVIDIA 展示了一個可長時間自主運行的設計驗證代理（agent），能協調整個晶片驗證週期，將達到已驗證 RTL 的時間縮短最多 50 倍，同時額外提升 20% 的覆蓋率。

With Microsoft and AMD, we introduced the first autonomous EDA workflows on Microsoft Discovery that can automate debug, implementation and design closure. Early engagements show up to 40% reduction in debug cycle time, saving weeks of engineering effort while improving design quality. We're seeing strong customer interest in our agentic AI platform with more than 30 active customer engagements underway. Early feedback has made clear that as these agents take on more engineering work, they orchestrate our underlying EDA tools at a significantly higher rate.

我們也與 Microsoft 與 AMD 一同在 Microsoft Discovery 上推出首個自主式 EDA 工作流程，可自動化除錯、實作與設計收斂。早期合作顯示，除錯週期時間最多可降低 40%，在提升設計品質的同時，節省數週的工程投入。我們看到客戶對代理式 AI 平台的興趣強勁，目前已有超過 30 個活躍的客戶合作案在進行中。早期回饋清楚顯示，隨著這些代理承擔更多工程工作，它們以顯著更高的頻率協調我們底層的 EDA 工具。

That allows customers to run more design and verification workloads, creating an incremental growth opportunity for Synopsys as we capture our fair share of the value that these agentic workflows and foundational tools provide for our customers. Turning to Ansys. One year into our integration, Ansys continues to see strong demand.

這使客戶能執行更多設計與驗證工作負載，並在我們取得代理式工作流程與基礎工具為客戶所創造價值的合理份額時，為新思科技帶來額外的成長機會。接著談 Ansys。在整合滿一年之際，Ansys 仍持續看到強勁需求。

From semis to aerospace to industrial and more, companies are embracing digital engineering. Across industries, Ansys simulation is accelerating innovation while reducing development risk and cost. For example, a leading automaker is using Ansys SimAI to achieve roughly 98% prediction accuracy and move crash analysis to near real time.

從半導體到航太到工業等各領域，企業正積極採用數位工程。跨產業而言，Ansys 模擬正在加速創新，同時降低開發風險與成本。例如，一家領先的汽車製造商正使用 Ansys SimAI，達到約 98% 的預測準確率，並將碰撞分析推進到近乎即時。

And a leading heavy equipment manufacturer achieved more than 10x faster motor design. We're applying AI to extend our S&A leadership and further automate the simulation of complex systems. This includes expanding our portfolio of GPU-accelerated Ansys applications. And in Q3, our largest Ansys deal was for GPU-accelerated Ansys CFD to support a company-wide digital twin at a multinational electronics component maker.

另一家領先的重型設備製造商則將馬達設計速度提升至 10 倍以上。我們正運用 AI 來延伸我們在模擬與分析（S&A）領域的領導地位，並進一步自動化複雜系統的模擬。這包括擴充我們以 GPU 加速的 Ansys 應用產品組合。而在第三季，我們最大的 Ansys 訂單是提供 GPU 加速的 Ansys CFD，以支援一家跨國電子元件製造商的全公司數位分身（digital twin）。

Turning to Design IP, which grew sequentially and year-over-year on broad AI infrastructure demand. As AI drives demand for higher bandwidth, faster connectivity and more complex system architectures, our interface, memory and die-to-die IP portfolio sits at the center of the stack. And our Q3 results showed it. We won more than 95% of PCIe 7 opportunities, including a subsystem win at a marquee enterprise storage customer.

轉向設計 IP，其在廣泛的 AI 基礎設施需求帶動下，無論按季或按年皆成長。隨著 AI 推動對更高頻寬、更快連線與更複雜系統架構的需求，我們的介面、記憶體與 die-to-die IP 產品組合位於整個技術堆疊的核心。而我們第三季的結果也證明了這一點。我們贏得了超過 95% 的 PCIe 7 機會，包括在一家指標性企業級儲存客戶取得子系統的勝利。

In LPDDR6, silicon proven across multiple nodes and foundries, we've secured 25 design wins year-to-date. Our die-to-die business is on pace to double year-over-year, and we now have more than 100 cumulative design wins. The industry continues to rely on Synopsys for silicon-proven quality and unrivaled scale. Our standards-based build once, sell many IP model remains foundational to our growth strategy.

在 LPDDR6 方面，我們的矽驗證已在多個製程節點與晶圓代工廠完成驗證，今年迄今已取得 25 項設計導入（design win）。我們的 die-to-die 業務正朝按年倍增的步伐前進，目前累計設計導入已超過 100 項。產業持續仰賴 Synopsys 提供經矽驗證的品質與無可匹敵的規模。我們以標準為基礎的「一次建置、多次銷售」IP 商業模式，仍是我們成長策略的基石。

We'll continue to invest and grow this business, what I call Factory 1, which benefits from strong chip start activity and solid traction across industries. For example, in automotive, we've sustained a 90% plus design win rate for three consecutive quarters as ADAS platforms refresh on to five- and three-nanometer. In mobile, consumer and edge AI, our USB IP has now crossed $2 billion in lifetime bookings with Tier 1 design wins already moving to the leading-edge node.

我們將持續投資並擴大這項業務，也就是我所稱的 Factory 1；它受惠於強勁的晶片啟動（chip start）活動，以及在各產業的穩健進展。例如在車用領域，隨著 ADAS 平台更新至 5 奈米與 3 奈米，我們已連續三個季度維持 90% 以上的設計導入率。在行動裝置、消費性與邊緣 AI 方面，我們的 USB IP 歷史累計訂單金額已突破 20 億美元，且第一級（Tier 1）設計導入已開始移轉至先進製程節點。

As AI expands beyond digital infrastructure into physical products, demand for silicon will continue to expand, providing a tailwind for our standards-based IP business. The higher growth opportunity in IP lies with a growing set of AI customers who are asking for deeper collaboration and IP solutions optimized to their specific workloads and architectures.

隨著 AI 從數位基礎設施擴展到實體產品，對矽的需求將持續擴大，為我們以標準為基礎的 IP 業務帶來順風。IP 更高成長的機會，來自一群不斷擴大的 AI 客戶，他們要求更深度的合作，以及針對其特定工作負載與架構最佳化的 IP 解決方案。

To meet that demand, we are expanding into differentiated IP subsystems and enabling custom silicon solutions. Customers ranging from hyperscalers, ASIC vendors, foundries and classic semiconductors want to partner with Synopsys to accelerate their chip development efforts and leverage our IP and engineering expertise to build increasingly differentiated custom silicon.

為了滿足該需求，我們正擴展至具差異化的 IP 子系統，並促成客製化矽解決方案。從超大規模業者（hyperscalers）、ASIC 供應商、晶圓代工廠到傳統半導體公司等各類客戶，都希望與 Synopsys 合作，以加速其晶片開發工作，並運用我們的 IP 與工程專業，打造差異化程度日益提高的客製化矽。

This is our Factory 2 model for customized IP. It moves us up the value chain from licensing alone to licensing plus royalties and positions us to capture the fast-growing custom silicon opportunity. This is a large focus, and we are making strong progress. We are in active discussions with multiple Factory 2 customers, and I look forward to sharing more at Investor Day.

這就是我們用於客製化 IP 的 Factory 2 模式。它使我們在價值鏈上從單純授權，提升到「授權加權利金（royalties）」，並讓我們得以掌握快速成長的客製化矽機會。這是我們的重點投入方向之一，我們也正取得強勁進展。我們正與多家 Factory 2 客戶進行積極洽談，並期待在投資人日分享更多內容。

To summarize, I want to thank the entire Synopsys team for their continued focus, innovation and execution. Q3 reinforced the strength of our strategy and our confidence in a strong finish to the year. AI is driving demand for advanced silicon, system-level engineering and AI-powered design. Our leadership portfolio positions us to capture a greater share of R&D investment across industries. We remain focused on translating our technology leadership into sustainable growth and margin expansion.

總結來說，我要感謝整個 Synopsys 團隊持續專注、創新與落實執行。第三季再次印證我們策略的強度，並強化我們對全年強勁收官的信心。AI 正推動對先進矽、系統層級工程，以及 AI 驅動設計的需求。我們的領先產品組合使我們能在各產業中取得更高比例的研發投資。我們仍專注於將技術領先轉化為可持續的成長與利潤率擴張。

Now over to Shelagh.

接下來交給 Shelagh。

Thank you, Sassine. We delivered an outstanding Q3, achieving revenue of $2.477 billion, non-GAAP operating margin of 41.6% and non-GAAP EPS of $3.91, all beating the high end of our guidance range. With broad-based strength across the business, the revenue outperformance was driven by EDA as well as strength in the Ansys business.

謝謝你，Sassine。我們第三季表現出色，營收達 24.77 億美元，非 GAAP 營業利益率為 41.6%，非 GAAP 每股盈餘（EPS）為 3.91 美元，皆高於我們指引區間的上緣。在業務全面走強的帶動下，營收超預期主要由 EDA 驅動，同時 Ansys 業務也展現強勁表現。

Backlog remains very strong at $10.9 billion, modestly down quarter-over-quarter due to the divestiture of the processor IP solutions business that closed in Q3. With the strength in Q3, strong cash flow generation and continued momentum into Q4, we are raising our full year revenue, non-GAAP operating margin, EPS and cash flow guidance.

積壓訂單（backlog）仍非常強勁，達 109 億美元；由於我們在第三季完成處分處理器 IP 解決方案業務，積壓訂單按季小幅下降。在第三季的強勁表現、強勁的現金流創造，以及延續至第四季的動能之下，我們上調全年營收、非 GAAP 營業利益率、EPS 與現金流指引。

I'll now review our third-quarter results. All comparisons are year-over-year unless otherwise stated. We generated total revenue of $2.477 billion, up approximately 42%, including Ansys revenue of approximately $711 million. As Sassine noted, one year into the combination, Ansys continues to perform strongly. We are also ahead of the schedule on the cost synergy commitments we made at close and have repaid the term loans earlier than planned.

我現在回顧第三季業績。除非另有說明，所有比較皆為按年（year-over-year）。我們的總營收為 24.77 億美元，約成長 42%，其中包含約 7.11 億美元的 Ansys 營收。如 Sassine 所提，合併滿一年後，Ansys 仍持續展現強勁表現。我們也在結案時承諾的成本綜效（cost synergy）方面進度超前，並且比原計畫更早償還定期貸款（term loans）。

Total GAAP costs and expenses were $2.119 billion with GAAP earnings per share of $2.84. Q3 GAAP EPS includes a gain associated with the sale of the processor IP solutions business that closed in the quarter. Total non-GAAP costs and expenses were $1.446 billion, on the lower end of our guided range, as we continue to improve operational efficiency and deliver Ansys cost synergies ahead of schedule, resulting in non-GAAP operating margin of 41.6%.

GAAP 總成本與費用為 21.19 億美元，GAAP 每股盈餘為 2.84 美元。第三季 GAAP EPS 包含與本季完成出售處理器 IP 解決方案業務相關的處分收益。非 GAAP 總成本與費用為 14.46 億美元，位於我們指引區間的低端；我們持續提升營運效率，並提前實現 Ansys 成本綜效，使非 GAAP 營業利益率達 41.6%。

Non-GAAP earnings per share were $3.91, ahead of our guidance, underscoring our strong operational execution in the quarter. Now on to our segments. Design Automation segment revenue was approximately $2 billion. As a reminder, this excludes the Optical Solutions Group, which was divested in Q4 '25. Within the Design Automation segment, Q3 EDA revenue increased 8.5% year-over-year, reflecting robust EDA software performance and another record quarter for hardware-assisted verification solutions.

非 GAAP 每股盈餘為 3.91 美元，高於我們的指引，凸顯我們本季強勁的營運執行力。接下來看各事業部門。設計自動化（Design Automation）部門營收約 20 億美元。提醒一下，這不包含光學解決方案事業群（Optical Solutions Group），該事業群已於 2025 年第四季完成處分。在設計自動化部門內，第三季 EDA 營收按年成長 8.5%，反映 EDA 軟體表現強勁，以及硬體輔助驗證解決方案再創單季新高。

Design Automation adjusted operating margin was 45.2%. The Design IP segment returned to growth with revenue of $474 million, up approximately 11% year-on-year. Consistent with our outlook, this represents continued sequential growth in the IP segment as we've repositioned the portfolio to focus on the highest value opportunities. Design IP adjusted operating margin was 26.5%.

設計自動化調整後營業利益率為 45.2%。設計 IP（Design IP）部門營收回到成長軌道，達 4.74 億美元，按年約成長 11%。符合我們的展望，這代表 IP 部門在我們重新定位產品組合、聚焦最高價值機會後，仍持續呈現按季成長。設計 IP 調整後營業利益率為 26.5%。

Turning to cash. Free cash flow was $746 million in Q3, and we ended the quarter with cash and short-term investments of $3.6 billion. Total debt at the end of Q3 was approximately $10 billion. Now to guidance for the full year. We are raising our total revenue guidance by $50 million at the midpoint, driven by strength in Design Automation segment led by EDA.

接著看現金。第三季自由現金流為 7.46 億美元，季末現金與短期投資為 36 億美元。第三季末總負債約為 100 億美元。接下來是全年指引。我們將總營收指引在中位數上調 5,000 萬美元，主要由設計自動化部門的強勁表現所帶動，其中以 EDA 為主。

As Sassine stated, momentum in EDA remains strong, and we expect double-digit organic EDA revenue growth in Q4 and for the full year 2026. We continue to expect the IP business to grow sequentially in Q4. This results in a revenue range of $9.69 billion to $9.74 billion. Within that, Ansys revenue contribution is expected to be approximately $2.98 billion, up $20 million versus our prior guidance.

如 Sassine 所述，EDA 動能仍然強勁，我們預期 2026 年第四季及全年 EDA 有機營收將達到雙位數成長。我們仍預期 IP 業務在第四季將按季成長。因此，全年營收區間為 96.9 億至 97.4 億美元。其中，預期 Ansys 的營收貢獻約為 29.8 億美元，較先前指引上調 2,000 萬美元。

Next, expenses. Total GAAP costs and expenses are expected to be between $8.667 billion and $8.742 billion. This includes an increase in expected charges for fiscal year 2026 in relation to our previously announced restructuring program as we continue to accelerate our committed synergies. Total non-GAAP costs and expenses are expected to be between $5.67 billion and $5.70 billion and non-GAAP operating margin of 41.5% at the midpoint, a 50 basis point raise to our previous guidance.

接下來談費用。預計 GAAP 總成本與費用將介於 86.67 億美元至 87.42 億美元之間。其中包含與我們先前宣布的重組計畫相關、對 2026 會計年度預期費用的增加，因為我們持續加速落實已承諾的綜效。預計非 GAAP 總成本與費用將介於 56.7 億美元至 57.0 億美元之間，且非 GAAP 營業利益率在中點為 41.5%，較我們先前指引上調 50 個基點。

GAAP earnings are expected to be between $3.84 to $4.08 per share. We expect non-GAAP earnings of $15.04 to $15.10 per share, a $0.31 increase at the midpoint from our prior guidance due to higher revenue and increased operational efficiency. We are raising our cash flow from operations guidance by $500 million to approximately $2.8 billion on strong cash collections and reducing our CapEx guidance to approximately $225 million, resulting in free cash flow of approximately $2.6 billion, an increase of $600 million versus our previous guidance.

預計 GAAP 每股盈餘將介於 3.84 美元至 4.08 美元之間。我們預期非 GAAP 每股盈餘為 15.04 美元至 15.10 美元，中點較先前指引增加 0.31 美元，主要因營收較高與營運效率提升。在強勁現金回收的帶動下，我們將營運活動現金流指引上調 5 億美元至約 28 億美元，並將資本支出（CapEx）指引下調至約 2.25 億美元，因此自由現金流約為 26 億美元，較先前指引增加 6 億美元。

Now to targets for the fourth-quarter: total revenue between $2.53 billion and $2.58 billion, total GAAP costs and expenses between $2.225 billion and $2.3 billion, total non-GAAP cost and expenses between $1.45 billion and $1.48 billion, GAAP earnings of $0.60 to $0.85 per share, and non-GAAP earnings of $4.10 to $4.16 per share. Our press release and financial supplement include additional targets and GAAP to non-GAAP reconciliations.

接著是第四季目標：總營收介於 25.3 億美元至 25.8 億美元之間，GAAP 總成本與費用介於 22.25 億美元至 23.0 億美元之間，非 GAAP 總成本與費用介於 14.5 億美元至 14.8 億美元之間，GAAP 每股盈餘 0.60 美元至 0.85 美元，以及非 GAAP 每股盈餘 4.10 美元至 4.16 美元。我們的新聞稿與財務補充資料包含其他目標，以及 GAAP 與非 GAAP 的調節表。

Thanks to our global Synopsys team for another strong quarter. These results reflect strong execution across the business, continued demand for our technology and disciplined operating performance as we build the foundation for the next phase of growth. We look forward to seeing many of you at our September Investor Day to discuss the compelling long-term opportunity we have as a mission-critical partner for our customers.

感謝我們全球 Synopsys 團隊再度交出強勁的一季。這些結果反映我們在各項業務上的強力執行、對我們技術的持續需求，以及在建立下一階段成長基礎時所展現的嚴謹營運表現。我們期待在 9 月的投資人日與各位中的許多人見面，討論我們作為客戶關鍵任務合作夥伴所擁有的具吸引力長期機會。

With that, I'll turn it over to the operator for questions.

接下來我把時間交給接線員進行提問。

(Operator Instructions) Jason Celino, KeyBanc Capital Markets.

(接線員指示) Jason Celino，KeyBanc Capital Markets。

Really good results here. I think what really stuck out to me was the 8% EDA growth, which was stable with last quarter despite the harder comp. You mentioned it's supposed to accelerate to double digits in Q4 and the full year. I mean, how would you describe what's driving that incremental acceleration? Is it design start activity? Is it agentic? Is it just better monetization strategies? Just help us understand.

這裡的結果真的很不錯。我覺得最讓我印象深刻的是 EDA 成長 8%，儘管比較基期更具挑戰，仍與上季持平。你提到第四季以及全年應該會加速到雙位數。我的意思是，你會如何描述推動這個額外加速的因素？是設計案啟動（design start）活動嗎？是 agentic（代理式）嗎？還是只是更好的變現策略？請幫我們理解一下。

Thanks, Jason, for the question. Yes, we are very excited about the overall performance in EDA and the acceleration to wrap up the year with double-digit growth, exactly what we have committed to in terms of our segment growth. What's driving the increase in confidence in our business in EDA is driven by multiple factors, the complexity of chip design, the move to advanced package 3DIC, the example I mentioned in my prepared remarks like an AMD expansion and use of 3DIC Compiler, and there are a number of other customers that are designing these advanced package are using our technology.

謝謝你的提問，Jason。是的，我們對 EDA 的整體表現以及在年底以雙位數成長收官的加速態勢感到非常振奮，這正是我們對該事業部門成長所承諾的目標。我們對 EDA 業務信心提升的原因來自多項因素：晶片設計複雜度提高、轉向先進封裝 3DIC；以及我在事先準備的發言中提到的例子，例如 AMD 擴大採用並使用 3DIC Compiler；此外還有許多其他正在設計這些先進封裝的客戶也在使用我們的技術。

AI is definitely a tailwind. As customers are rethinking of how to reengineer their chip design engineering, it's requiring different methods for that engineering. So that's driving another tailwind for us. And hardware, we had a record revenue year on hardware. So all in all, that 8-plus percent is organic growth for EDA that we're fairly excited about.

AI 絕對是一股順風。當客戶重新思考如何重新設計其晶片設計工程流程時，這需要不同的工程方法。因此這也為我們帶來另一股順風。另外在硬體方面，我們在硬體上創下年度營收新高。總而言之，這個 8% 以上的成長是 EDA 的有機成長，我們對此相當振奮。

Yes. And I'd just point out, Jason, that the 8.5% Q3 EDA growth that you saw was against a really tough compare. Q3 '25 was 16%. And so as Sassine said, it really just shows the strength of the business to be able to perform against that tough compare and then have the full year continue to have double digit.

是的。Jason，我也想補充指出，你看到的第三季 EDA 成長 8.5% 是在非常艱難的比較基期之下達成的。2025 年第三季是 16%。因此如 Sassine 所說，這確實顯示了業務的韌性，能在這樣的高基期下仍有表現，並且全年仍可維持雙位數成長。

Yes. And then, Sassine, you mentioned something. You talked about your customers having to reengineer their processes. We're seeing a lot of innovation happening. Everyone always talks about these new AI models. I think yesterday, there was an example. OpenAI talked about the development of their new chip. They talked about using their own AI models to accelerate the design process.

是的。另外，Sassine，你提到一件事。你談到客戶必須重新設計他們的流程。我們看到很多創新正在發生。大家總是在談這些新的 AI 模型。我想昨天有個例子。OpenAI 談到他們新晶片的開發。他們提到使用自家的 AI 模型來加速設計流程。

Maybe can you speak to when you hear examples like this, because a lot of your customers have used their own models in the past, what is involved in the reengineering of a design process? How much is supplemental or incremental or replacement of something that might be existing? I don't know if that makes sense but --

也許你可以談談，當你聽到這樣的例子時——因為你們很多客戶過去也用過自家的模型——在重新設計設計流程時，具體會涉及哪些內容？其中有多少是補充性的、增量性的，或是取代既有的某些做法？我不確定這樣問是否清楚，但--

Yes, of course, of course. Yes. Thank you for the question. For at least 1.5 years now to 2 years, we've been talking about how AI is reshaping how engineering is done. The investment that Synopsys has been making and leading in delivering agent engineers to our customers in rethinking the workflow, including AI models that will absolutely participate and contribute to that reengineering of engineering, in every case, the underlying requirement is more and more and more of our software because if you're using an AI model or agents that the customer is developing, it doesn't matter.

是的，當然，當然。是的。謝謝你的問題。至少在過去 1.5 年到 2 年間，我們一直在談 AI 如何重塑工程的進行方式。Synopsys 一直在投入並領先，為客戶提供代理式工程師（agent engineers），協助重新思考工作流程，其中也包含 AI 模型，這些模型絕對會參與並貢獻於工程再造；而在每一種情況下，底層的需求都是我們的軟體越來越多，因為即使客戶正在開發 AI 模型或代理（agents），其實都無關緊要。

You still need the ground truth physics in order to -- for that model to be able to operate with confidence and delivering to the best outcome. So the example that you mentioned actually is a very good example and a great opportunity for Synopsys on the EDA front. Not to mention, it's a huge opportunity on the IP as well because that's rethinking the whole architecture as you customize the silicon.

你仍然需要「真實物理」（ground truth physics），才能讓該模型有信心地運作，並交付最佳結果。所以你提到的例子其實是一個非常好的例子，也是在 EDA 方面對 Synopsys 而言很棒的機會。更不用說，在 IP 方面也是巨大的機會，因為當你客製化矽晶時，這等於是在重新思考整個架構。

Joe Vruwink, Baird.

Joe Vruwink，Baird。

I wanted to go back a few years, but at your 2024 Investor Day, you shared an outlook back then around how 30% of EDA software demand might end up coming from multi-die efforts by next year. I'm curious how that figure might be tracking and maybe what you see after 2027. I think we all appreciate there's been quite a lot of recent attention even this week on HBM.

我想回顧幾年前的內容：在你們 2024 年投資人日，你們當時分享了一個展望，提到到明年為止，EDA 軟體需求中可能有約 30% 會來自多晶粒（multi-die）相關的工作。我很好奇這個數字目前的進展如何，以及你們在 2027 年之後可能看到什麼趨勢。我想大家都理解，最近甚至就在本週，市場對 HBM 的關注度相當高。

How might EDA content change for you when thinking about DRAM processes moving to logic, that sort of thing? And is it actually strengthening maybe relative to what you thought a few years ago?

當你思考 DRAM 製程往邏輯製程靠攏、諸如此類的趨勢時，EDA 內容對你而言可能會如何改變？而且相較於你幾年前的看法，這是否其實正在變得更強勁？

Yes, Joe, actually, that forecast has accelerated. As I mentioned in the prepared remarks as well, the die-to-die wins that we've had in the last 12 months have doubled. And the reason for that is this whole advanced package and 3DIC architecture. That drives a significant opportunity in IT. Die to die is one example; then, of course, all the other interfaces that are required to stitch the system together as a final product at our customer.

是的，Joe，實際上，那個預測已經加速了。正如我在事先準備的發言中也提到的，我們在過去 12 個月拿下的 die-to-die 勝利已經翻倍。原因在於整個先進封裝與 3DIC 架構。這在 IT 方面帶來了顯著的機會。Die-to-die 是其中一個例子；當然，還有所有其他為了把系統整合成客戶最終產品所需的介面。

And on the EDA side is 3DIC Compiler, and this is where the emphasis of the joint solution with Ansys is essential. You cannot build these systems without taking into account physics, thermal, structure, fluid into the chip design phase. So the 30% back then was the figure we thought was a stretch, but it absolutely accelerated given all the investments our customers are making to build these efficient custom silicon.

在 EDA 端則是 3DIC Compiler，而這正是與 Ansys 的聯合解決方案之所以至關重要的地方。若不把物理、熱、結構、流體等因素納入晶片設計階段，就無法打造這些系統。因此，當時我們認為 30% 是一個偏勉強的數字，但在客戶為打造高效率客製化矽晶所投入的各項投資推動下，確實大幅加速。

Okay. That's great. I wanted to ask, the double-digit organic growth in EDA, if you think about splitting that up between software and hardware, is the performance in your software business where you would like it to be here at year-end? It seems like hardware has remained a very large and strong driver. I'm more curious about the software performance.

好的。很棒。我想問一下，EDA 的有機成長達到兩位數，如果把它拆分成軟體與硬體，你覺得在年末時你們軟體業務的表現是否達到你希望的水準？看起來硬體一直是非常大且強勁的驅動因素。我比較好奇的是軟體的表現。

We're very happy with the software component actually. Very, very pleased. The part that actually I'm most excited about is the delivery of our joint solution with Ansys. And that gives us the platform to deliver to where the future of engineering challenges is heading. As we just talked about, the advanced package, the need for physics into electronics gives us even the confidence to look at the trajectory and continue on delivering these double-digit growth.

其實我們對軟體部分非常滿意。非常、非常滿意。我最興奮的部分其實是我們與 Ansys 的聯合解決方案的交付。這讓我們擁有一個平台，能夠交付並對應未來工程挑戰的走向。如同我們剛剛談到的，先進封裝、以及把物理導入電子設計的需求，讓我們更有信心去看待這條成長軌跡，並持續交付這種兩位數成長。

Charles Shi, Needham & Company.

Charles Shi，Needham & Company。

First thing, Shelagh, I have -- maybe this is for Sassine. I have a high level -- maybe a long-term question. Sassine, we know that going back probably more than 10 years, you played a very pivotal role in terms of infusing AI. I know back then, it was not a large language model, probably more like reinforcement learning type of AI into the Synopsys tool flows, DSO.ai, all those great products.

第一個問題，Shelagh，我有——也許這題是問 Sassine。我有一個高層次——可能是長期的問題。Sassine，我們知道回溯到大概 10 多年前，你在導入 AI 方面扮演了非常關鍵的角色。我知道當時還不是大型語言模型，可能更像是強化學習類型的 AI，被導入到 Synopsys 的工具流程、DSO.ai，以及那些很棒的產品中。

But AI has advanced so much over the last 10 years, especially last three. So the question I constantly hear from investors is about is there any risk for AI to actually disrupt the commercial EDA business. And I think one -- at least one school of thought we're thinking, could there be an end-to-end, what they call, AI native chip design that bypasses all the commercial EDA tools, especially maybe with some AI models trained by the commercial EDA generated synthetic data.

但過去 10 年 AI 進步非常大，尤其是最近三年。所以投資人一直問我的問題是：AI 是否有任何風險會真正顛覆商業 EDA 業務。我想至少有一派想法在思考：會不會出現端到端、所謂 AI 原生的晶片設計，繞過所有商業 EDA 工具——尤其是如果有些 AI 模型是用商業 EDA 產生的合成資料來訓練的話。

Is that a real threat, in your opinion, to the overall EDA industry at all? Or where do you see where AI can be more substitutive or complementary to the commercial EDA business?

你認為這對整體 EDA 產業而言算是真正的威脅嗎？或者你怎麼看 AI 在商業 EDA 業務中，哪些地方可能更具替代性、哪些地方更具互補性？

Yes. Thank you, Charles. I don't need to go back or go as far as 10 years ago. You're right. Around 2017, we introduced and invested and saw great results with DSO.ai centered around reinforcement learning. If you look at the last three years, the focus was around copilot, generative AI, move to agents. Right now, we're talking about autonomous designs.

是的。謝謝你，Charles。我不需要回溯到 10 年前那麼久。你說得對。大約在 2017 年，我們推出並投入資源在以強化學習為核心的 DSO.ai，並看到了很好的成果。如果看過去三年，重點在於 copilot、生成式 AI、以及走向 agents。而現在，我們談的是自主設計（autonomous designs）。

As you start looking at autonomous workflow, the most important thing is accuracy and determinism. Customers will not invest hundreds of millions of dollars in a product without having the confidence that it's going to work. The portfolio we have with the sign-off leadership is essential to building these autonomous workflows. We are participating with our customers on how to achieve an autonomous workflow.

當你開始看自主化工作流程時，最重要的是準確性與確定性（determinism）。客戶不會在沒有信心其能夠運作的情況下，投入數億美元在一個產品上。我們在 sign-off 領域的領先產品組合，對於建立這些自主化工作流程至關重要。我們正與客戶一起參與，探討如何達成自主化工作流程。

It's not like it's happening without our participation. We're proactively engaged with them to reengineer how they're looking at the future of engineering with AI being the center of that evolution. I am not worried at all that, at some point that, that model can do the end to end without our participation because you have to remember, these models are not static. They're constantly changing. They constantly need to learn.

這並不是在沒有我們參與的情況下就會發生的事。我們正主動與他們合作，重新設計他們如何看待工程的未來，而 AI 會是這場演進的核心。我完全不擔心某個時間點會出現那種模型能在沒有我們參與下完成端到端，因為你必須記得，這些模型不是靜態的。它們一直在變。它們也一直需要學習。

So the opportunity is the opposite. It's not a threat. It's a significant demand for our software to train, to inference, to constantly enable that faster design to deal with the complexity, and we're at the center of it.

所以機會其實是相反的。這不是威脅。這會帶來對我們軟體的顯著需求，用於訓練、推論，並持續促成更快的設計以應對複雜度，而我們正處於核心位置。

Maybe, Sassine, on the agentic AI agent opportunities, how should we think about any uplift to the overall revenue growth, especially EDA growth? And can you give us an update on the changing the business model more to the subscription plus consumption, at least for the agents? Any discussion with your customers so far?

Sassine，關於 agentic AI／AI 代理的機會，我們應該如何看待它對整體營收成長、尤其是 EDA 成長的任何上行空間（uplift）？另外，你能否更新一下把商業模式更多轉向「訂閱加用量」（subscription plus consumption），至少在代理（agents）方面的進展？到目前為止，和客戶有任何討論嗎？

Yes, number of engagements. Actually, to the first question you had, as our customers are exploring whether to use an agent from Synopsys or an agent plus -- from Synopsys plus their own agents to keep their special sauce inside their workflow, the customer workflow or the customer agent, the need is for more licenses. We are defining with our customers multiple ways on how to engage from a subscription of our agents, subscription of our workflow as well as a consumption measure as these agents and the new workflow is consuming more software.

是的，有多項合作洽談。其實回到你第一個問題，當客戶在探索要使用 Synopsys 的代理，或是使用 Synopsys 的代理再加上他們自己的代理，以便把他們的獨門配方保留在其工作流程中——也就是客戶工作流程或客戶代理——其需求會是更多的授權（licenses）。我們正與客戶一起定義多種合作方式：包含訂閱我們的代理、訂閱我們的工作流程，以及在這些代理與新工作流程消耗更多軟體時，採用用量計費（consumption measure）。

We'll highlight more of how we're thinking how to model the long-term growth with this context that I just described in a few weeks, September 30. But absolutely, we're in advanced discussions with number of customers with different flavors of using their agents, our agents, a hybrid of both and multiple model optionality that they're thinking about.

我們會在幾週後、9 月 30 日，更進一步說明在我剛描述的背景下，我們如何思考用模型化方式來呈現長期成長。但毫無疑問地，我們正與多家客戶進行深入討論，涵蓋不同使用樣態：用他們的代理、用我們的代理、兩者混合，以及他們正在考量的多種模型選項。

Lee Simpson, Morgan Stanley.

Lee Simpson，Morgan Stanley。

Great. Maybe just a couple of quick ones on IP actually. Maybe just preempting the Analyst Day. I wondered, Sassine, if you can maybe just give us a little bit of outline on the speed of shift to the Factory 2 opportunity you outlined with licensing and royalties. And then secondly, it's been about a year, I think, since we've seen the Intrinsic ID acquisition.

很好。我想就 IP 再快速問兩個問題。也許算是先為分析師日做個鋪陳。Sassine，我想請你大致說明一下，你所提到的 Factory 2 機會在授權與權利金（royalties）方面的轉移速度。第二個問題是，我想大概已經一年了，自從我們看到 Intrinsic ID 的收購案之後。

And I think at the time, you talked about security IP as being a new vector of growth in IP. Just wanted to hear if you could maybe outline the size of that opportunity, how you've seen engagements go and where in particular you think deployments will happen.

我想當時你提到安全性 IP 是 IP 領域一個新的成長向量。想請你概述一下這個機會的規模、你們看到的客戶互動進展如何，以及你特別認為部署會發生在哪些地方。

Thank you, Lee. On Factory 2, actually, all you need to look at is the momentum in every hyperscaler investing and are at various stages of delivering their own custom silicon. These chips will not happen without our interface IP. These customers need our interface IP in order to build their own chips and to connect to the ecosystem, be it the memory provider or if they're using -- if they are investing in their XPU, but they need a networking chip from their supplier, connecting them together come through our interface IP. So that's the opportunity.

謝謝你，Lee。關於 Factory 2，其實你只要看看每一家超大規模雲端業者（hyperscaler）投資的動能，以及他們各自處於交付自家客製化矽晶片的不同階段即可。沒有我們的介面 IP，這些晶片就不可能實現。這些客戶需要我們的介面 IP 才能打造自家的晶片並連接到生態系，不論是連到記憶體供應商，或是他們在投資自家的 XPU，但需要由供應商提供的網路晶片，彼此之間的連接都要透過我們的介面 IP。這就是機會所在。

What we could see as well and the reason we started talking about Factory 2 is the need to customize these standards and to accelerate the delivery of these customizations. Synopsys is in a unique position given our scale and the knowledge, the skills, the market position to deliver on that customization and acceleration. We are in advanced discussions with number of these customers to change the business model from the traditional IP license plus some NRE to a license plus a royalty.

我們也看到、也是我們開始談 Factory 2 的原因，是需要對這些標準進行客製化，並加速交付這些客製化版本。Synopsys 憑藉我們的規模、知識、技能與市場地位，具備獨特優勢來實現這些客製化與加速。我們正與其中多家客戶進行深入討論，將商業模式從傳統的 IP 授權加上一些 NRE，轉變為授權加上權利金（royalty）。

I look forward in a few weeks to, again, help you model what does that look like. But I cannot be more excited about the investment that we've made and the agility in pivoting in that direction while absolutely continue on leading and investing in Factory 1 because that's another significant opportunity that will continue.

我期待在幾週後，再次協助你們建模這會是什麼樣子。但我對我們所做的投資，以及在保持敏捷、轉向這個方向的同時，仍然持續領先並投資於 Factory 1 感到無比興奮，因為那也是另一個將持續存在的重大機會。

As for security, the reason we started looking at the portfolio and we divested the processor IP is to focus on the areas of growth. Security is one of them. As you said, we don't split it out as a separate item, but we have a great market position in security and will only continue on becoming more and more important given you need to secure the chip not only at the software level, as much as you can do at the hardware level, and that's where our security portfolio comes in.

至於安全性，我們之所以開始檢視產品組合並剝離處理器 IP，是為了聚焦在成長領域。安全性就是其中之一。如你所說，我們沒有把它拆分成單獨項目，但我們在安全性方面擁有很強的市場地位，而且它只會變得越來越重要，因為你需要保護晶片，不僅在軟體層面，也要盡可能在硬體層面做到，而這正是我們安全性產品組合發揮作用之處。

Josh Tilton, Wolfe Research.

Josh Tilton，Wolfe Research。

This is Arsenije on for Josh. Just, Sassine, first, you talked about seeing strong early customer interest in Multiphysics Fusion following the launch. And just historically, chip design and simulation often were kind of handled by different engineering teams and different workflows and tools. I guess when you're bringing thermal analysis directly into the design flow, how are customers structurally, I guess, approaching that convergence?

我是 Arsenije，代 Josh 提問。Sassine，首先你提到 Multiphysics Fusion 發布後，早期客戶興趣很強。而從歷史上看，晶片設計與模擬通常由不同的工程團隊、不同的工作流程與工具來處理。我想問，當你把熱分析直接帶入設計流程時，客戶在組織層面上是如何看待並推動這種整合的？

Are you seeing those teams start to work more closely together and consolidate around more common workflows? Or does adoption still kind of require navigating distinct engineering teams and maybe separate budgets?

你們是否看到這些團隊開始更緊密合作，並整合到更共通的工作流程上？還是說採用仍需要在不同工程團隊之間協調，甚至可能涉及不同的預算？

Yes, you're absolutely right. These teams or the skills and expertise of engineering were separate domains. They had a different handshake as the design steps are moving from the synthesis to the physical design to sign-off, et cetera. The requirements to have codesign is essential to reduce margin and deliver to these competitive products.

是的，你說得完全正確。這些團隊，或工程的技能與專長，原本是分屬不同領域。當設計步驟從綜合（synthesis）走到實體設計（physical design）再到簽核（sign-off）等階段時，他們之間的交接方式也不同。為了降低裕度並交付具競爭力的產品，協同設計（codesign）的需求至關重要。

So absolutely, leading customers are requiring, to take thermal into account during the design implementation to take structure, stress as they're building these 3DIC into the architecture of the chip, not only the synthesis or the implementation of the multiple chiplets. So yes, absolutely. And this is where we have invested, that the implementation engineer that's sitting in the upfront part of the design flow are able without too much effort to be able to bring in sign-off accuracy early in the design flow.

因此，領先的客戶確實要求在設計實作期間就把熱因素納入考量，並在把這些 3DIC 納入晶片架構時，同時考量結構與應力，而不只是多個 chiplet 的綜合或實作。所以是的，完全如此。這也是我們投資的方向：讓坐在設計流程前段的實作工程師，不需要太多額外工作，就能在設計早期引入具簽核等級精度的分析。

And that's exactly where we see the opportunity of the combined portfolio and delivering to the fusion of physics with electronics. From a budget point of view, by the way, the second part of your -- I'm sorry, from the second part of your question, from a budget point of view, as we have committed one plus one will be greater than two. What it means even if it's coming from an EDA budget, the joint solution will capture an upside in revenue to the existing separate point tools.

而這正是我們看到的機會：結合產品組合，實現物理與電子的融合。順帶一提，關於你問題的第二部分——抱歉——從預算角度來看，我們承諾 1+1 會大於 2。這表示即使預算來自 EDA，這個聯合解決方案也將在既有分散的單點工具之上，帶來額外的營收上行空間。

Got it. That's helpful. And then just, Shelagh, just clarifying just one topic specifically on Ansys, the raise $20 million to the $2.98 billion. Last quarter, you helped us with $12.5 million contribution from that accounting dynamic and it was $60 million for the full year. Is it still $60 million for the full year and that raise on that $20 million increase in Ansys is just core upside from good execution there?

了解。這很有幫助。另外，Shelagh，我想釐清一個特別關於 Ansys 的議題：你們把預估上調了 2,000 萬美元到 29.8 億美元。上季你提到那個會計動態帶來 1,250 萬美元的貢獻，全年是 6,000 萬美元。全年仍然是 6,000 萬美元嗎？而這次 Ansys 上調的 2,000 萬美元只是因為執行良好帶來的核心上行嗎？

Yes. So last quarter, we talked about the accounting change, and so we made that accounting change, and we'll be making that accounting change and in the guidance that we're giving incorporates that. So we're seeing strength in the Ansys business, including the channel business.

是的。所以上季我們談到會計變更，我們已經做了那項會計變更，而我們在提供的指引中也已納入該變更的影響。因此我們看到 Ansys 業務表現強勁，包括通路業務。

Got it. And what was it, I guess, in Q3, for Ansys' accounting change and it's still $60 million for the full year? Or is it a different number this time?

了解。那我想問，Q3 來自 Ansys 會計變更的影響是多少？全年仍是 6,000 萬美元嗎？還是這次是不同的數字？

It's still in that same range for the full year.

全年仍在同一個區間。

Got it. And the quarter, it was $12.5 million or --

了解。那單季是 1,250 萬美元還是--

We didn't disclose the in quarter.

我們沒有揭露單季的數字。

Joe Quatrochi, Wells Fargo.

Joe Quatrochi，Wells Fargo。

You talked about a tale of two markets with AI versus non-AI and EDA. I'm just wondering if the acceleration that you're seeing, is AI becoming a larger piece or in offsetting kind of the non-AI? Or have you started to see also some acceleration from the non-AI part of your business as well?

你提到 AI 與非 AI、以及 EDA 呈現兩個市場的分化。我想問你們看到的加速，是否代表 AI 正在占比變得更大、或是在抵消非 AI 的疲弱？或者你們也開始看到非 AI 的業務部分出現一些加速？

Yes, Joe, the -- so the reason for our assertion to begin with, what we do on a quarterly basis, that these are internal measures that we have is we track chip starts. And the reason we have a good coverage on design starts is our IP portfolio. There is no customer that is planning a new chip start that we don't engage very early on through our IP portfolio. And of course, EDA will follow.

是的，Joe——我們之所以一開始會做出那樣的判斷，是因為我們每季都會做一些內部衡量：我們追蹤晶片開案（chip starts）。而我們之所以能很好地覆蓋設計開案（design starts），是因為我們的 IP 產品組合。沒有任何客戶在規劃新的晶片開案時，我們不會透過 IP 產品組合在非常早期就介入。當然，EDA 也會隨之跟進。

The observation is on the non-AI, in the last two quarters, it has stabilized. What it means was we were observing a slowdown in design start in the non-AI segment. In the last couple of quarters, stabilization, so it's not declining anymore. Now on the flip side for AI, where we have been seeing and continue on seeing an acceleration in design start, which is a great balance for the opportunity that we have.

觀察到的是，在非 AI 方面，過去兩個季度已經趨於穩定。也就是說，我們之前看到非 AI 區隔的設計開案在放緩。在最近幾個季度出現穩定，因此不再下滑。另一方面，在 AI 方面，我們一直看到、且仍持續看到設計開案加速，這為我們的機會帶來很好的平衡。

That's helpful. And then maybe as a follow-up, just wondering if you could give any puts and takes on the RPO. It was down a touch this quarter. Was there any impact from the divestiture? And did it come in as expected with your plan?

這很有幫助。接著可能追問一下，想請你談談 RPO 的一些利弊取捨。本季略有下滑。是否有受到資產剝離的影響？以及是否如你們的計畫預期？

Yes, it came in as expected. And as I said in my prepared remarks, the modest change is really due to the divestiture of the processor IP business that happened inside the quarter. As you recall, we were close to closing the deal when we did last earnings. It closed a few days after we did the earnings. So that's why you're seeing it this quarter.

是的，結果符合預期。而且如我在事先準備的發言中所說，這個小幅變動主要是因為本季內完成了處理器 IP 業務的剝離。你應該記得，我們在上次財報時已接近完成交易。在我們公布財報後幾天交易就完成了。所以你才會在本季看到這個影響。

Siti Panigrahi, Mizuho.

Siti Panigrahi，瑞穗。

Sassine, it's really a good quarter. Congratulations. Going back to the Ansys and Synopsys, the integrated products that you launched. I think you talked about that Multiphysics Fusion is not expected to contribute to EDA growth until 2027. So can you talk about the adoption of pipeline trajectory that we should expect from now and then? And is that '27 contribution more likely to show up as an incremental EDA growth?

Sassine，這一季表現真的很好。恭喜。回到你們推出的 Ansys 與 Synopsys 整合式產品。我記得你提到 Multiphysics Fusion 預期要到 2027 年才會對 EDA 成長有貢獻。那你能談談從現在到那時，我們應該期待的採用與管線（pipeline）走勢軌跡嗎？另外，2027 年的貢獻更可能以 EDA 的增量成長呈現嗎？

Or as share gains from your Ansys simulation base? Any color like what you are seeing in terms of pricing benefit, value to the customer on that? That will be great.

還是來自你們 Ansys 模擬客群基礎的市占提升？能否補充一些你們在定價效益、以及對客戶價值方面所看到的情況？那就太好了。

Yes. So from a value to the customer, the 10 times faster design closure or a 3x faster SPICE accurate multiphysics timing, that is a significant value to customers. What does it mean for the customer? Less iteration, better design, faster. So our customers in these early engagements have validated and they're in early stages of deployment.

可以。就客戶價值而言，設計收斂速度快 10 倍，或是 SPICE 精度的多物理時序快 3 倍，對客戶都是非常顯著的價值。這對客戶意味著什麼？更少的迭代、更好的設計、更快的速度。因此我們在這些早期合作中的客戶已經驗證了效果，目前也正處於早期部署階段。

The moment we move to production, which those customers will move to production, we'll start seeing the revenue upside. As we've said from the beginning, FY26, we will not have much contribution in the joint solution. It was a year of execution, delivering to these products. As we look at FY27, absolutely, it will contribute to our growth in EDA.

一旦我們進入量產階段——而這些客戶會進入量產——我們就會開始看到營收上行。如同我們一開始就說過的，FY26 在這個聯合解決方案上的貢獻不會太多。那是一個執行之年，把這些產品交付出來。展望 FY27，毫無疑問，它將對我們 EDA 的成長做出貢獻。

We are absolutely committed as well to the $400 million synergy in year four. So as we meet in September, we'll be able to start talking about '27 and the longer-term contribution of this differentiated solution.

我們也同樣堅定承諾在第四年達成 4 億美元的綜效。因此等到我們 9 月再見面時，我們就能開始談 2027 年以及這個差異化解決方案的長期貢獻。

That's helpful. And then a quick follow-up. I know Mike Ellow has been there now a few quarters. How is he driving the sales organization? Any kind of specific changes he is contemplating or has been working towards as you roll out this integrated product?

這很有幫助。再快速追問一下。我知道 Mike Ellow 已經加入幾個季度了。他是如何推動銷售組織的？在你們推出這個整合式產品的同時，他是否正在思考或推動任何具體變革？

Yes. Mike has been doing a great job in leveraging what the organization does incredibly well and the areas that we need to increase our investment from portfolio, go-to-market point of view, et cetera, et cetera. The priority as we look at FY27 and beyond is how to engage the customers in the areas of differentiation and ensure we have the right investment with the customer.

是的。Mike 做得非常好，善用組織本來就做得非常出色的部分，同時也針對我們需要加大投資的領域——從產品組合、上市策略（go-to-market）等角度——來推進。展望 FY27 及之後的優先事項，是如何在差異化領域與客戶互動，並確保我們對客戶投入正確的資源。

We're enabling the customers successfully so we capture the monetization opportunity. So Mike has been spending a lot of energy internally with the team to prepare the organization for IP Factory 2, for the AI monetization, for the joint solutions. All this is a significant time where Mike is spending as we look at '27 and beyond.

我們正在協助客戶成功，從而把握變現機會。因此 Mike 在內部投入大量精力，與團隊一起為 IP Factory 2、AI 變現、以及聯合解決方案做好組織準備。隨著我們看向 2027 年及以後，這些都是 Mike 正在投入大量時間的重大工作。

He's a great hire.

他是一個很棒的招募。

He is. We're happy having him.

確實。我們很高興有他加入。

Jay Vleeschhouwer, Griffin Securities.

Jay Vleeschhouwer，Griffin Securities。

Sassine, for you first. This may be an imperfect analogy. But how would you compare the prospective migration or adoption benefit of the new multiphysics cohort and cohorts to follow, might compare with the ICC to Fusion Compiler transition? That obviously didn't include Ansys component at the time, but that was your last big architectural or generational change in your stack. So how would you think this one now that it's underway might compare to that? And then a follow-up.

Sassine，先問你。這可能是一個不太完美的類比。但你會如何比較：新的多物理（multiphysics）這一批以及後續批次的潛在遷移或採用效益，與 ICC 轉向 Fusion Compiler 的轉換相比如何？當時顯然不包含 Ansys 的組件，但那是你們技術堆疊上一次重大的架構或世代變革。那麼這一次既然已經啟動，你會覺得它可能與當時相比如何？然後我還有一個追問。

Yes. Jay, thank you. Actually, a very good question. What we've done with Fusion was bringing our sign-off capability and strength into the design physical implementation phase because the customers at the time were having to iterate late stage to correlate or to sign off the chip. It's exactly the same here, but you're including physics.

好的。Jay，謝謝。其實這是個非常好的問題。我們在 Fusion 所做的，是把我們的 sign-off 能力與優勢帶入設計的實體實作（physical implementation）階段，因為當時客戶必須在後期反覆迭代，才能做關聯或完成晶片 sign-off。這裡完全一樣，只是你把物理效應也納入其中。

Now the RedHawk, the HFSS, the rest of the portfolio of Ansys, how do you bring not only proximation into the design phase, how do you bring the actual engines into the design phase so you reduce iteration, you have a convergent flow, et cetera? The investment we made in the Fusion platform, because remember, the Fusion platform is not about bringing tools together with a common interface or a user interface.

現在有 RedHawk、HFSS，以及 Ansys 產品組合的其他部分，你要如何不只是把近似（approximation）帶進設計階段，而是把真正的引擎帶進設計階段，從而減少迭代、形成收斂式流程等等？我們在 Fusion 平台上的投資——因為請記得，Fusion 平台不是把工具用共同介面或使用者介面拼在一起而已。

We made a significant investment at the data model level. So we -- so our R&D team can write code, optimize on the same exact infrastructure and data model. And that's where Shankar and team have been investing quite heavily from day one of the integration start to deliver to exactly the same value that we were able to build on with Fusion.

我們在資料模型層面做了重大投資。因此我們的研發團隊可以在完全相同的基礎架構與資料模型上撰寫程式碼並進行最佳化。而這正是 Shankar 與團隊自整合開始第一天起就大力投入的地方，目標是交付與我們在 Fusion 上所建立的同等價值。

Okay. Secondly, the reengineering of engineering concept that you've talked about now since Converge last year remains a very interesting concept and prospect for you and your customers. The question is what are the ingredients that customers need to make that work. Is it a set of variables or changes they need to make? Is there a single magic bullet that they can implement to make that happen?

了解。第二，你們自去年 Converge 以來一直在談的「工程再造（reengineering of engineering）」概念，對你們與客戶而言仍是非常有趣的概念與前景。問題是，客戶要讓它運作需要哪些要素？他們需要調整一組變數或做出哪些改變嗎？是否存在一個可以落地的單一「靈丹妙藥」，讓這件事發生？

The demos that you did jointly, for example, at DAC with AMD and Microsoft around Discovery seemed quite interesting as perhaps a catalyst for some of those changes. So maybe talk about what actually has to happen for the whole reengineering thing to really happen.

例如你們在 DAC 與 AMD、Microsoft 共同做的、圍繞 Discovery 的展示，看起來相當有意思，或許能成為推動其中一些改變的催化劑。所以也請談談，要讓整個工程再造真正發生，實際上需要做到哪些事情。

Yes. We talked about Fusion as an essential part to have the entire complete set from a spec all the way to sign-off. So that's an essential component that Synopsys has that complete set of assets. Now from bringing together a workflow that can leverages the AI speed from a copilot, generative AI all the way to autonomy is where we're putting significant investment and racing.

好的。我們談到 Fusion 是不可或缺的一部分，因為它能提供從規格（spec）一路到 sign-off 的完整集合。因此 Synopsys 擁有那套完整資產是關鍵要件。接著，在把工作流程整合起來、並能利用 AI 速度——從 copilot、生成式 AI 一直到自主化（autonomy）——這一塊，就是我們正在投入大量資源並加速競逐的方向。

The two demos we've done -- and thank you for mentioning the Microsoft, AMD, which was a demonstration of a full autonomous EDA workflow on the Microsoft Discovery. Think how powerful that is, where you can start from a spec and you have a cognitive layer that can orchestrate multiple tasks agents in order to deliver to the outcome of the spec.

我們做過的兩個展示——也謝謝你提到與 Microsoft、AMD 的合作，那是一個在 Microsoft Discovery 上展示完整自主 EDA 工作流程的示範。想想這有多強大：你可以從規格出發，並且有一個認知層能夠協調多個任務代理，以交付符合該規格的成果。

At the same time, what we've done at DAC actually is a demonstration with NVIDIA on an autonomous long-running agent capability. So that's part of -- essential part of reengineering the workflow. What I'm most excited about, as you double-click into what I just described, is not only the assets that we have, the consumption of these assets is exponential in order to allow for an autonomous flow to deliver to the promise that the customer is looking for, which is an efficient design, better power, better performance, better cost, and that's exactly what we're doing.

同時，我們在 DAC 實際上也與 NVIDIA 做了一個關於自主、可長時間運行的代理能力的展示。所以那是——工作流程再造的關鍵部分。當你深入檢視我剛才描述的內容時，我最興奮的不僅是我們擁有的資產；為了讓自主流程兌現客戶所期待的承諾——也就是更高效率的設計、更好的功耗、更好的效能、更好的成本——這些資產的使用量正以指數級成長，而這正是我們正在做的。

Gary Mobley, StoneX.

Gary Mobley，StoneX。

Sassine, you mentioned in your prepared remarks, the largest deal in the quarter for Ansys was a GPU-based digital twin. Was that a large deal because it is accelerated compute, runs an accelerated compute versus CPU-based compute? I'm asking because I'm just trying to get a sense of whether the efficiency from an accelerated compute digital twin system accrues to NVIDIA or accrues to you as well.

Sassine，你在事先準備的發言中提到，本季 Ansys 最大的一筆交易是基於 GPU 的數位孿生。那筆交易之所以大，是因為它是加速運算、以加速運算來執行，而不是以 CPU 為基礎的運算嗎？我這樣問是因為我想了解：加速運算的數位孿生系統所帶來的效率提升，價值是歸於 NVIDIA，還是你們也能同樣受益。

Thank you, Gary, for the question. So simulation is a perfect application for GPU acceleration because these are jobs you can, for the most part, not only paralyze. You can achieve a significant speed up. So the speed up does not stop or has limitation at 10 or 15 times. In CFD, we're seeing 40, 50, 60 times speed up. The bottleneck for our customers is the time to results, the time to accurate results. And the investment actually started with the GPU acceleration even before the acquisition of Ansys by Synopsys.

謝謝你，Gary，謝謝你的問題。所以模擬是 GPU 加速的完美應用，因為這些工作在很大程度上不僅可以並行化。你也能達到顯著的加速。因此，加速並不會在 10 倍或 15 倍就停止或受到限制。在 CFD 方面，我們看到 40、50、60 倍的加速。客戶的瓶頸在於取得結果的時間、取得精準結果的時間。而這項投資其實在 Synopsys 收購 Ansys 之前就已從 GPU 加速開始。

After the acquisition, we just accelerated even further the commitment, the investment and of course, the NVIDIA investment and alignment towards this opportunity only helped. In terms of who gets the value, we sell and we capture the entire value and uplift of the GPU. NVIDIA, of course, in the back end, they -- you need GPU to run from the traditional CPU to a GPU. They benefit that way. So the large deal is large because of the benefit to the customer and the speed up we were able to deliver to the customer.

收購之後，我們只是進一步加速了投入與承諾；當然，NVIDIA 在這個機會上的投資與一致性也只會帶來助益。至於價值由誰取得，我們銷售並掌握 GPU 所帶來的全部價值與增益。NVIDIA 當然在後端——你需要 GPU 才能從傳統 CPU 轉到 GPU 來運行。他們以這種方式受益。所以這筆大單之所以大，是因為它為客戶帶來的效益，以及我們能交付給客戶的加速幅度。

And my follow-up, I wanted to ask about Korea. Revenue generated from Korea appears to be trending up, close to 20% this year. That's a standout for sure. Is that a reflection of the strength of the memory market? Is it a reflection of maybe, I guess, you retaining more market share at Samsung and what many have speculated? Maybe you can just give some color there.

我還有一個追問，想請教關於韓國的情況。來自韓國的營收看起來呈上升趨勢，今年接近 20%。這確實很突出。這是反映記憶體市場的強勁嗎？還是反映例如——我想——你們在三星保住了更多市占率，正如許多人所推測的那樣？也許你可以提供一些背景說明。

I'm not sure about the speculation. One thing I can tell you is our relationship with Samsung, with SK hynix, with the broad market in Korea has been a great collaboration in an area and a region that are leading with a very essential part of the AI infrastructure, and we are leading with those customers and not limited to those two in the broader region itself in -- across the portfolio.

我不確定那些推測。但我可以告訴你一件事：我們與三星、與 SK hynix、以及與韓國更廣泛市場的關係，一直是非常好的合作；那個領域與地區在 AI 基礎設施中一個非常關鍵的部分上處於領先，而我們也與那些客戶一起領先——而且不僅限於那兩家——在更廣泛的地區內，涵蓋整個產品組合。

So you need to think of it from IP with our custom HBM and HBM engagements with the lead customers to EDA and to Ansys. So very pleased with the performance that we have.

所以你需要從 IP 的角度來看，包括我們與領先客戶在客製化 HBM 與 HBM 專案上的合作，到 EDA，再到 Ansys。因此我們對目前的表現非常滿意。

Kelsey Chia, Citi.

Kelsey Chia，花旗。

Great to see the IT business getting back on track. I believe the team outlined a long-term growth of about mid-teens for IP business several years ago. I mean since then, we have seen significant acceleration in the chip design activity, particularly among hyperscalers. And it seems that the non-AI portion of the business is also stabilizing. Given that backdrop, is it unreasonable to think that IP business could achieve a much stronger growth than the mid-teens laid out over the next couple of quarters? And if so, are there any factors that could prevent that growth algorithm from being set higher?

很高興看到 IT 業務回到正軌。我記得團隊幾年前曾提出 IP 業務的長期成長約為十幾個百分點（mid-teens）。自那之後，我們看到晶片設計活動顯著加速，特別是在超大規模業者之間。而且看起來非 AI 的業務部分也在趨於穩定。在這樣的背景下，認為 IP 業務在未來幾個季度能達到遠強於先前所述 mid-teens 的成長，是否不合理？如果可以的話，有哪些因素可能會阻止把這個成長演算法（growth algorithm）設定得更高？

Kelsey, thank you for the question. Your assumptions are good ones, which is more chip start, a huge opportunity with custom chips, which I refer to as Factory 2, as Factory 1 continue on delivering and expanding. And for now, the mid-teens is our long-term guide. We look forward for September, where we can share more and provide any updates as necessary.

Kelsey，謝謝你的問題。你的假設是合理的，也就是更多的晶片啟動（chip start）、客製晶片的巨大機會——我把它稱為 Factory 2——同時 Factory 1 也持續交付並擴張。目前，mid-teens 仍是我們的長期指引。我們期待在 9 月時能分享更多內容，並在必要時提供任何更新。

We have reached the end of the Q&A session. I will now turn the call back to Sassine Ghazi for closing remarks.

問答環節到此結束。我現在把電話會議交回給 Sassine Ghazi 作結語。

Thank you for all the questions. One year after the transformational acquisition of Ansys, we're executing with focus, gaining momentum and extending our leadership from silicon to systems. I look forward to seeing many of you at Investor Day in September. Thank you very much.

謝謝大家的提問。在完成對 Ansys 的變革性收購一年後，我們正專注執行、動能持續增強，並將我們的領導地位從矽到系統進一步延伸。我期待在 9 月的投資人日見到各位中的許多人。非常感謝。

This concludes today's call. Thank you for attending. You may now disconnect.

今天的電話會議到此結束。感謝各位參與。您現在可以掛線。