博通 (AVGO) 2023 Q1 法說會逐字稿

Welcome to Broadcom Inc.'s First Quarter Fiscal Year 2023 Financial Results Conference Call. At this time, for opening remarks and introductions, I would like to turn the call over to Ji Yoo, Head of Investor Relations of Broadcom Inc.

歡迎來到博通公司 2023 財年第一季度財務業績電話會議。現在，關於開場白和介紹，我想把電話轉給 Broadcom Inc 投資者關係主管 Ji Yoo。

Thank you, operator, and good afternoon, everyone. Joining me on today's call are Hock Tan, President and CEO; Kirsten Spears, Chief Financial Officer; and Charlie Kawwas, President, Semiconductor Solutions Group.

謝謝接線員，大家下午好。總裁兼首席執行官 Hock Tan 和我一起參加今天的電話會議；克爾斯滕斯皮爾斯，首席財務官；和半導體解決方案集團總裁 Charlie Kawwas。

Broadcom distributed a press release and financial tables after the market closed describing our financial performance for the first quarter fiscal year 2023. If you did not receive a copy, you may obtain the information from the Investors section of Broadcom's website at broadcom.com. This conference call is being webcast live and an audio replay of the call can be accessed for 1 year through the Investors section of Broadcom's website.

Broadcom 在收市後分發了一份新聞稿和財務表格，描述了我們 2023 財年第一季度的財務業績。如果您沒有收到副本，您可以從 Broadcom 網站 broadcom.com 的投資者部分獲取信息。此電話會議正在網絡直播，電話的音頻重播可通過 Broadcom 網站的投資者部分訪問 1 年。

During the prepared comments, Hock and Kirsten will be providing details of our first quarter fiscal year 2023 results, guidance for our second quarter as well as commentary regarding the business environment. We'll take questions after the end of our prepared comments.

在準備好的評論中，Hock 和 Kirsten 將提供我們 2023 財年第一季度業績的詳細信息、第二季度的指導意見以及有關商業環境的評論。我們將在準備好的評論結束後回答問題。

Please refer to our press release today and our recent filings with the SEC for information on the specific risk factors that could cause our actual results to differ materially from the forward-looking statements made on this call.

請參閱我們今天的新聞稿和我們最近向美國證券交易委員會提交的文件，了解可能導致我們的實際結果與本次電話會議上所做的前瞻性陳述存在重大差異的具體風險因素的信息。

In addition to U.S. GAAP reporting, Broadcom reports certain financial measures on a non-GAAP basis. A reconciliation between GAAP and non-GAAP measures is included in the tables attached to today's press release. Comments made during today's call will primarily refer to our non-GAAP financial results.

除了美國 GAAP 報告外，Broadcom 還在非 GAAP 基礎上報告某些財務指標。 GAAP 和非 GAAP 措施之間的對賬包含在今天新聞稿所附的表格中。在今天的電話會議上發表的評論將主要參考我們的非 GAAP 財務業績。

I'll now turn the call over to Hock.

我現在將電話轉給 Hock。

Thank you, Ji, and thank you, everyone, for joining us today. In our fiscal Q1 '23 consolidated net revenue, that was -- revenue was $8.9 billion, up 16% year-on-year. Semiconductor solutions revenue increased 21% year-on-year to $7.1 billion. While, as we expected, infrastructure software declined 1% year-on-year to $1.8 billion even as our core software sustained growth of 5% year-on-year.

謝謝 Ji，也謝謝大家今天加入我們。在我們 23 財年第一季度的合併淨收入中，收入為 89 億美元，同比增長 16%。半導體解決方案收入同比增長 21% 至 71 億美元。正如我們預期的那樣，儘管我們的核心軟件同比增長 5%，但基礎設施軟件同比下降 1% 至 18 億美元。

Stepping back, let me sum up what happened in Q1. From our view, infrastructure spending continues to be up particularly in service providers, even as hyperscale and enterprise sustain. Spending in technology for infrastructure has been strong, showing double-digit growth for 9 consecutive quarters.

退一步，讓我總結一下第一季度發生的事情。在我們看來，基礎設施支出繼續增加，特別是在服務提供商中，即使超大規模和企業維持。基礎設施技術支出一直強勁，連續 9 個季度呈現兩位數增長。

We continue to be booked for fiscal '23 and our lead times and visibility on semiconductors remain largely at 50 weeks. While there have been a small number of requests to push out certain orders, we know that these are the exceptions, and they have not had a material impact on our business. Because we ship linearly throughout the quarter to our customers, inventory on our books has been consistent around 80 days, and overall inventory of Broadcom products across the ecosystem remains very well managed.

我們繼續為 23 財年預訂，我們在半導體方面的交貨時間和可見度基本保持在 50 週。雖然有少量要求推出某些訂單，但我們知道這些都是例外情況，它們並未對我們的業務產生實質性影響。由於我們在整個季度內以線性方式向客戶發貨，因此我們賬面上的庫存在 80 天左右保持一致，並且整個生態系統中 Broadcom 產品的整體庫存得到了很好的管理。

We continue, needless to say, to be very disciplined in shipping our backlog only as and when needed by our end customers. With that, let me now provide more color on each of our end markets.

不用說，我們繼續嚴格遵守紀律，僅在最終客戶需要時才運送我們的積壓訂單。有了這個，讓我現在提供更多關於我們每個終端市場的顏色。

Starting with networking. Networking revenue was $2.3 billion and was up 20% year-on-year, in line with guidance, representing 32% of our semiconductor revenue. We see continued deployment of our advanced Tomahawk switches by hyperscalers in their leaf and spine architectures.

從網絡開始。網絡收入為 23 億美元，同比增長 20%，符合指引，占我們半導體收入的 32%。我們看到超大規模企業在其葉和主乾架構中繼續部署我們的高級 Tomahawk 交換機。

Even as we deliver on increased bandwidth for the hyperscalers, having said that, power remains a major challenge. So just this week, we announced the industry's first integrated silicon photonics networking solution, codename Bailly, which integrates the active optical interconnects with our next-generation Tomahawk 5 switch at 51.2 terabit per second. Bailly doubles switching performance, but it will reduce total system power.

話雖如此，儘管我們為超大規模用戶提供了更高的帶寬，但功耗仍然是一個主要挑戰。因此，就在本週，我們宣布了業界首個集成矽光子網絡解決方案，代號為 Bailly，它以每秒 51.2 太比特的速度將有源光互連與我們的下一代 Tomahawk 5 交換機集成在一起。 Bailly 使開關性能翻倍，但會降低系統總功耗。

Keep in mind that at hyperscalers, a growing portion of our switches have been deployed within the AI networks, which are separate from the traditional x86 CPU scale out running existing workloads.

請記住，在超大規模企業中，我們越來越多的交換機已部署在 AI 網絡中，這些網絡與傳統的 x86 CPU 橫向擴展運行現有工作負載是分開的。

Now this is today. Tomorrow, we generated AI using large scale -- large language, I should say, models with billions of parameters. We have to run thousands of AI engines in parallel, enabling large and synchronized bursts of data at speeds of 400 and 800 gig. The networks to support this massive processor density is critical and is important SDAI engines. Such networks have to be lossless, low latency and be able to scale.

現在是今天。明天，我們將使用大規模——大型語言，我應該說，具有數十億參數的模型來生成人工智能。我們必須並行運行數千個 AI 引擎，以 400 和 800 gig 的速度實現大量同步數據爆發。支持這種巨大處理器密度的網絡是至關重要的，也是重要的 SDAI 引擎。這樣的網絡必須是無損的、低延遲的並且能夠擴展。

So as you know, such AI networks are already been deployed at certain hyperscalers through our Jericho2 switches and Ramon fabric. In fact, in 2022, we estimated our Ethernet switch shipments deployed in AI was over $200 million.

如您所知，此類 AI 網絡已經通過我們的 Jericho2 交換機和 Ramon 結構部署在某些超大規模企業中。事實上，到 2022 年，我們估計部署在 AI 中的以太網交換機出貨量超過 2 億美元。

With the expected exponential demand from our hyperscale customers, we forecast that these could grow to well over $800 million in 2023. We anticipate this trend will continue to accelerate and mindful that we need even more higher performance networks in the future. We have been investing in a new generation of this lossless low-latency ethernet fabric designed specifically to handle such data and compute-intensive AI workloads.

鑑於我們的超大規模客戶的預期需求呈指數級增長，我們預測到 2023 年這些需求可能會增長到 8 億美元以上。我們預計這一趨勢將繼續加速，並註意到我們在未來需要更高性能的網絡。我們一直在投資新一代無損低延遲以太網結構，專為處理此類數據和計算密集型 AI 工作負載而設計。

Of course, additionally, the exciting growth prospects for generative AI are driving our compute offload accelerated business at hyperscalers. As we have indicated to you last quarter, this business achieved over $2 billion in revenue in 2022. We are on track to exceed $3 billion in revenue in our fiscal '23. In Q2, looking forward, short term, we expect these tailwinds to drive our networking revenue to grow about another 20% year-over-year.

當然，此外，生成式 AI 令人興奮的增長前景正在推動我們在超大規模上的計算卸載加速業務。正如我們上個季度向您指出的那樣，該業務在 2022 年實現了超過 20 億美元的收入。我們有望在 23 財年實現超過 30 億美元的收入。在第二季度，展望短期，我們預計這些順風將推動我們的網絡收入同比增長約 20%。

Moving on next to our server storage connectivity revenue. There was a record $1.3 billion or 18% of semiconductor revenue and up 57% year-on-year. Once again, as we discussed in preceding quarters, the rapid transition to next-generation MegaRAID solutions drove this substantial year-on-year content increase. After 4 consecutive quarters of such increases, this transition, however, is significantly complete, and we expect that in Q2 on a year-on-year basis server storage connectivity revenue will moderate towards 20% year-on-year growth.

接下來是我們的服務器存儲連接收入。半導體收入達到創紀錄的 13 億美元，佔半導體收入的 18%，同比增長 57%。正如我們在前幾個季度所討論的那樣，向下一代 MegaRAID 解決方案的快速過渡再次推動了內容的大幅同比增長。然而，在連續 4 個季度出現這種增長之後，這種轉變已經基本完成，我們預計第二季度服務器存儲連接收入同比增長將放緩至 20%。

Moving on to broadband. Revenue grew 34% year-on-year to a record $1.2 billion and represented 17% of semiconductor revenue. During this quarter, our broadband business particularly benefited from robust deployments of -- by telcos of 10G PON and cable operators of DOCSIS 3.1. These gateways have high attach rates of WiFi 6 and 6E. And in Q2, we expect the secular drivers behind broadband to sustain momentum on a sequential basis. And year-on-year, broadband will grow a solid 10%.

轉向寬帶。收入同比增長 34%，達到創紀錄的 12 億美元，佔半導體收入的 17%。在本季度，我們的寬帶業務尤其受益於 10G PON 的電信公司和 DOCSIS 3.1 的有線電視運營商的穩健部署。這些網關具有 WiFi 6 和 6E 的高連接率。在第二季度，我們預計寬帶背後的長期驅動因素將連續保持增長勢頭。與去年同期相比，寬帶將增長 10%。

Moving on to wireless. Q1 revenue of $2.1 billion represented 29% of semiconductor revenue. Demand from our North American customer drove wireless revenue up 4% year-on-year, reflecting content increases, which we had previously indicated last quarter. Sequentially, wireless was flattish compared to Q4. And seasonally, we expect wireless to be down sequentially in Q2 and down high single-digit percentage year-over-year.

繼續無線。第一季度收入為 21 億美元，佔半導體收入的 29%。我們北美客戶的需求推動無線收入同比增長 4%，反映出我們上個季度曾表示的內容增長。因此，與第四季度相比，無線業務表現平平。在季節性方面，我們預計無線業務將在第二季度連續下降，並且同比下降高個位數百分比。

Finally, Q1 industrial resale of $229 million decreased 4% year-over-year as softness in China offset strength in renewable energy and medical. And in Q2, we forecast industrial resales to be down low single-digit percentage year-on-year on continuing softness in China.

最後，由於中國的疲軟抵消了可再生能源和醫療的強勢，第一季度工業轉售額為 2.29 億美元，同比下降 4%。在第二季度，由於中國持續疲軟，我們預測工業轉售將同比下降低個位數百分比。

So in summary, Q1 semiconductor solutions revenue was up 21% year-on-year. And in Q2, we expect semiconductor revenue growth of high single-digit percentage year-on-year.

因此綜上所述，Q1 半導體解決方案收入同比增長 21%。而在第二季度，我們預計半導體收入同比將實現高個位數百分比增長。

Turning to software. In Q1, infrastructure software revenue of $1.8 billion declined 1% year-on-year and represented 20% of total revenue. While core software revenue grew 5% year-on-year, the Brocade business declined because of lumpiness in enterprise consumption in this very narrow vertical of SAN storage. For core software, consolidated renewal rates averaged 119% of expiring contracts. And within our strategic accounts, we averaged 129%, and within the strategic accounts, annualized bookings of $536 million included $197 million, which represent 37% of cross-selling of our portfolio of products to these same core strategic customers. Over 90% of the renewal value represented recurring subscription and maintenance.

轉向軟件。第一季度，基礎設施軟件收入為 18 億美元，同比下降 1%，佔總收入的 20%。雖然核心軟件收入同比增長 5%，但博科業務卻出現下滑，原因是企業在這個非常狹窄的 SAN 存儲垂直領域的消費不穩定。對於核心軟件，綜合續訂率平均為到期合同的 119%。在我們的戰略客戶中，我們平均為 129%，在戰略客戶中，5.36 億美元的年化預訂量包括 1.97 億美元，占我們向這些相同核心戰略客戶交叉銷售產品組合的 37%。超過 90% 的續訂價值代表定期訂閱和維護。

Now by way of comparison, over the last 12 months, consolidated renewal rates averaged 119% over expiring contracts, and in our strategic accounts, we averaged 134%. Because of these, our ARR, the indicator of forward revenue at the end of Q1 was $5.3 billion, which is up 3% from a year ago. In Q2, we expect our infrastructure software segment revenue to be up low to mid-single-digit percentage year-on-year as the stable core software growth continues to be partially offset now by weakness in Brocade.

現在通過比較，在過去 12 個月中，綜合續約率平均比到期合同高 119%，而在我們的戰略客戶中，我們的平均續約率為 134%。因此，我們的 ARR，第一季度末的遠期收入指標為 53 億美元，比去年同期增長 3%。在第二季度，我們預計我們的基礎設施軟件部門收入將同比增長低至中個位數百分比，因為穩定的核心軟件增長現在繼續被 Brocade 的疲軟部分抵消。

So in summary, we are guiding consolidated Q2 revenue for the company to be $8.7 billion, up 8% year-on-year. Before Kirsten, tells you more about our financial performance for the quarter, let me provide a brief update on our pending acquisitions of VMware.

因此，總而言之，我們指導公司第二季度的綜合收入為 87 億美元，同比增長 8%。在 Kirsten 向您詳細介紹我們本季度的財務業績之前，讓我簡要介紹一下我們對 VMware 的未決收購。

We continue to make progress with our various regulatory filings around the world, having now received legal merger clearance in Brazil, South Africa and Canada and foreign investment control clearance in Germany, France, Austria, Denmark, Italy and New Zealand. As we stated on our last earnings call, we continue to anticipate that the time line for the review process will be extended in other key regions, especially given the size of this transaction.

我們在世界各地的各種監管備案方面繼續取得進展，現已在巴西、南非和加拿大獲得合法合併許可，並在德國、法國、奧地利、丹麥、意大利和新西蘭獲得外國投資控制許可。正如我們在上次財報電話會議上所說，我們繼續預計審查過程的時間線將在其他關鍵地區延長，特別是考慮到此次交易的規模。

Having said that, we continue to expect the transaction to close within our fiscal 2023. We believe the combination of Broadcom and VMware is about enabling enterprises to accelerate innovation and expand choice by addressing their most complex technology challenges in this multi-cloud era. And we are confident regulators will see this when they conclude their review.

話雖如此，我們仍然預計該交易將在我們的 2023 財年內完成。我們相信 Broadcom 和 VMware 的結合將使企業能夠通過解決他們在這個多雲時代最複雜的技術挑戰來加速創新和擴大選擇。我們相信監管機構在結束審查時會看到這一點。

Finally, Broadcom recently published its third annual ESG report available on our corporate citizenship website, which discusses the company's ESG initiatives. As a global technology leader, we recognize Broadcom's responsibility to have a positive impact on our customers, employees and communities. Through our product and technology innovation and operational excellence, we remain committed to this mission.

最後，Broadcom 最近在我們的企業公民網站上發布了第三份年度 ESG 報告，其中討論了公司的 ESG 舉措。作為全球技術領導者，我們認識到 Broadcom 有責任對我們的客戶、員工和社區產生積極影響。通過我們的產品和技術創新以及卓越的運營，我們始終致力於實現這一使命。

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

有了這個，讓我把電話轉給克爾斯滕。

Thank you, Hock. Let me now provide additional detail on our financial performance. Broadcom had another great quarter with robust financials. Consolidated revenue was $8.9 billion for the quarter, up 16% from a year ago. Gross margins were 74% of revenue in the quarter, about 10 basis points higher than we expected. Operating expenses were $1.1 billion, down 1% year-on-year. R&D of $929 million was also down 1% year-on-year, primarily from streamlined project and other variable spending, offset in part by higher people costs resulting from increased headcount as we are hiring.

謝謝你，霍克。現在讓我提供有關我們財務業績的更多詳細信息。 Broadcom 又迎來了一個業績強勁的季度。本季度綜合收入為 89 億美元，同比增長 16%。本季度毛利率佔收入的 74%，比我們預期高約 10 個基點。營運開支為 11 億美元，按年下跌 1%。 9.29 億美元的研發費用也同比下降 1%，這主要是由於精簡項目和其他可變支出，部分被我們招聘人員增加導致的人員成本增加所抵消。

Operating income for the quarter was $5.4 billion and was up 17% from a year ago. Operating margin was 61% of revenue, up approximately 50 basis points year-on-year. Adjusted EBITDA was $5.7 billion or 64% of revenue. This figure excludes $127 million of depreciation.

本季度營業收入為 54 億美元，同比增長 17%。營業利潤率為收入的 61%，同比增長約 50 個基點。調整後的 EBITDA 為 57 億美元，佔收入的 64%。該數字不包括 1.27 億美元的折舊。

Now a review of the P&L for our 2 reportable segments. Revenue for our semiconductor solutions segment was $7.1 billion and represented 80% of total revenue in the quarter. This was up 21% year-on-year. As Hock discussed, this came from strength across all of our semiconductor end markets.

現在回顧一下我們 2 個可報告部門的損益表。我們半導體解決方案部門的收入為 71 億美元，佔本季度總收入的 80%。同比增長 21%。正如 Hock 所討論的那樣，這來自於我們所有半導體終端市場的實力。

Gross margins for our semiconductor solutions segment were approximately 69%, down approximately 160 basis points year-on-year driven primarily by product mix within our semiconductor end markets. Operating expenses were $802 million in Q1, down 2% year-on-year. R&D was $716 million in the quarter, down 1% year-on-year. Q1 semiconductor operating margins were 58%. So while semiconductor revenue was up 21%, operating profit grew 23% year-on-year.

我們的半導體解決方案部門的毛利率約為 69%，同比下降約 160 個基點，這主要是由於我們的半導體終端市場的產品組合所致。第一季度運營費用為 8.02 億美元，同比下降 2%。本季度研發為 7.16 億美元，同比下降 1%。第一季度半導體營業利潤率為 58%。因此，雖然半導體收入增長了 21%，但營業利潤同比增長了 23%。

Moving to the P&L for our infrastructure software reportable segment. Revenue for infrastructure software was $1.8 billion, down 1% year-on-year and represented 20% of revenue. Gross margins for infrastructure software were 91% in the quarter, and operating expenses were $346 million in the quarter, down 1% year-over-year. Infrastructure software operating margin was 72% in Q1, and operating profit was stable year-on-year.

轉向我們的基礎設施軟件可報告部分的損益表。基礎設施軟件收入為 18 億美元，同比下降 1%，佔收入的 20%。本季度基礎設施軟件的毛利率為 91%，本季度運營費用為 3.46 億美元，同比下降 1%。第一季度基礎設施軟件營業利潤率為72%，營業利潤同比穩定。

Moving to cash flow. Free cash flow in the quarter was $3.9 billion, representing a 16% increase year-over-year. Free cash flow represented 44% of revenues in Q1 '23 consistent with what we achieved the same quarter last year. We spent $103 million on capital expenditures.

轉向現金流。本季度自由現金流為 39 億美元，同比增長 16%。自由現金流佔 23 年第一季度收入的 44%，與我們去年同期取得的成績一致。我們在資本支出上花費了 1.03 億美元。

Days sales outstanding were 33 days in the first quarter compared to 30 days in the fourth quarter. We ended the first quarter with inventory of $1.9 billion, down 1% from the end of the prior quarter or 78 days on hand. Overall, inventory of Broadcom's products across the ecosystem, as Hock indicated, remains well managed.

第一季度銷售未完成天數為 33 天，而第四季度為 30 天。我們在第一季度結束時的庫存為 19 億美元，比上一季度末或庫存 78 天下降 1%。總體而言，正如 Hock 所指出的那樣，Broadcom 在整個生態系統中的產品庫存管理得很好。

We ended the first quarter with $12.6 billion of cash and $39.3 billion of gross debt, of which $1.1 billion is short term. During the quarter, we repaid $260 million in senior notes that were due on maturity. The weighted average coupon rate and years to maturity of our fixed rate debt is 3.61% and 10.2 years, respectively.

我們在第一季度結束時擁有 126 億美元的現金和 393 億美元的總債務，其中 11 億美元是短期債務。本季度，我們償還了 2.6 億美元到期的優先票據。我們固定利率債務的加權平均票面利率和到期年限分別為 3.61% 和 10.2 年。

Turning to capital allocation. In the quarter, we paid stockholders $1.9 billion of cash dividends. Consistent with our commitment to return excess cash to shareholders, we repurchased $1.2 billion of our common stock and eliminated $333 million of common stock for taxes due on vesting of employee equity, resulting in the repurchase and elimination of approximately 2.7 million AVGO shares. The non-GAAP diluted share count in Q1 was 434 million. As of the end of Q1, $11.8 billion was remaining under the share repurchase authorization. Excluding the potential impact of any share repurchases, in Q2, we expect the non-GAAP diluted share count to be 438 million.

轉向資本配置。本季度，我們向股東支付了 19 億美元的現金股息。與我們向股東返還多餘現金的承諾一致，我們回購了 12 億美元的普通股，並消除了 3.33 億美元的普通股，以支付員工股權歸屬的應繳稅款，從而回購和消除了約 270 萬股 AVGO 股票。第一季度的非 GAAP 稀釋後股份數為 4.34 億股。截至一季度末，股份回購授權剩餘118億美元。排除任何股票回購的潛在影響，在第二季度，我們預計非 GAAP 攤薄股數為 4.38 億股。

Based on current business trends and conditions, our guidance for the second quarter of fiscal 2023 is for consolidated revenues of $8.7 billion and adjusted EBITDA of approximately 64.5% of projected revenue. In forecasting such profitability, we expect gross margins to be up approximately 150 basis points sequentially on product mix and R&D spending to be up sequentially on continuing hiring of engineers and seasonal payroll tax step-ups.

根據當前的業務趨勢和狀況，我們對 2023 財年第二季度的指引是合併收入為 87 億美元，調整後的 EBITDA 約為預計收入的 64.5%。在預測這種盈利能力時，我們預計產品組合的毛利率將連續上升約 150 個基點，而研發支出將因繼續僱用工程師和季節性工資稅增加而連續增加。

That concludes my prepared remarks. Operator, please open up the call for questions.

我準備好的發言到此結束。接線員，請打開電話詢問。

(Operator Instructions) Our first question will come from the line of Harsh Kumar with Piper Sandler.

（操作員說明）我們的第一個問題將來自 Harsh Kumar 和 Piper Sandler 的台詞。

Congratulations on yet another solid quarter and guide. Hock, you mentioned generative models in your commentary. I wanted to understand the difference between what you're doing in AI so far versus maybe what our understanding of generative is. You talked about $200 million in Ethernet related to AI. Is that largely generative because we've heard other companies say that, for large part, the generative models are using [a ton of band]. And then you talked about $2 billion in compute offload going to sort of $3 billion. My understanding was that was mostly for video processing. Maybe help us think about how we think of Avago's place or Broadcom's place in the generative process.

祝賀又一個穩定的季度和指南。 Hock，你在評論中提到了生成模型。我想了解您目前在 AI 中所做的事情與我們對生成的理解之間的區別。你談到了與 AI 相關的 2 億美元以太網。這在很大程度上是生成性的，因為我們聽說其他公司說，在很大程度上，生成模型正在使用 [a ton of band]。然後你談到了 20 億美元的計算卸載將達到 30 億美元。我的理解是，這主要是為了視頻處理。也許可以幫助我們思考我們如何看待 Avago 的位置或 Broadcom 在生成過程中的位置。

Well, thank you for that question and opportunity to clarify why we highlighted -- or why I highlighted it very purposefully. In 2022, generative is just barely starting to kick off, but there exist AI 1networks within the hyperscalers, particularly in fairly significant volume. And what we're trying to say is, very similar to CPUs, traditional CPUs in traditional workloads in those same data centers, we've constrained on performance of those silicon CPUs and Moore's Law. We're starting to see scale out by positioning rows and rows of servers, CPUs and networking them together to work closely in parallel.

好吧，謝謝你提出這個問題和機會來澄清我們為什麼強調 - 或者為什麼我非常有目的地強調它。到 2022 年，生成才剛剛開始，但超大規模企業中存在 AI 1 網絡，尤其是相當大的數量。我們想說的是，與 CPU 非常相似，在相同數據中心的傳統工作負載中使用傳統 CPU，我們限制了這些矽 CPU 的性能和摩爾定律。我們開始看到通過定位一排又一排的服務器、CPU 並將它們聯網在一起以緊密並行工作來實現橫向擴展。

As we step up to large language models in AI, generative AI, in particular coming into play, GPUs are starting to be strung together in hundreds, soon to be thousands of racks and working in parallel. And you know how that goes. And basically, those GPUs work in parallel in fairly synchronous manner to basically run and do what we call bulk parametric exchange, basically, run GPUs together or AI engines together whether they're GPUs, AI, or TPUs or other AI engines. You run them together.

隨著我們在 AI 中採用大型語言模型，尤其是生成 AI 開始發揮作用，GPU 開始以數百個形式串在一起，很快就會成為數千個機架並並行工作。你知道那是怎麼回事。基本上，這些 GPU 以相當同步的方式並行工作，基本上運行並執行我們所說的批量參數交換，基本上，一起運行 GPU 或 AI 引擎，無論它們是 GPU、AI、TPU 還是其他 AI 引擎。你一起運行它們。

Becomes network. The network becomes now potentially a critical part of this whole AI phenomenon in hardware. To make it work, you've got to put together many racks of AI engines in parallel, very similar to what we have been doing -- hyperscalers have been doing on CPUs to make them run faster, high performance as Moore's Law comes to an end and doesn't make any difference here in the form of AI engine.

成為網絡。網絡現在可能成為硬件中整個人工智能現象的關鍵部分。為了讓它發揮作用，你必須將許多機架的 AI 引擎並行放在一起，這與我們一直在做的非常相似——超大規模計算器一直在 CPU 上做，以使它們運行得更快、性能更高，因為摩爾定律即將到來結束並不會在這里以 AI 引擎的形式產生任何區別。

They come from silicon. They have -- they face similar constraints. So network becomes a problem -- becomes a constraint. Network becomes a very key part of fulfilling generative AI dream here. And what we are saying -- what I'm saying in my comments is, last year, 2022, these are more what you call the AI workloads that are running in hyperscale, and the advent of generative AI is still relatively fresh and new. We're doing $200 million as far as we could estimate of silicon, Ethernet switches and fabric that goes into those AI networks as far as we could identify in hyperscalers.

它們來自矽。他們有——他們面臨著類似的限制。所以網絡成為一個問題——成為一種約束。在這裡，網絡成為實現生成式 AI 夢想的一個非常關鍵的部分。我們所說的——我在評論中所說的是，去年，2022 年，這些更多的是你所說的超大規模運行的 AI 工作負載，而生成 AI 的出現仍然相對較新。據我們估計，我們正在做 2 億美元的矽、以太網交換機和結構，這些矽、以太網交換機和結構將進入我們在超大規模器中識別的人工智能網絡。

With generative AI and the urgency and excitement of it coming in that we are seeing today, we are seeing that increase very, very dramatically. And we're seeing urgency in our hyperscale customers coming to us to secure products, to secure ability to put in place those very, very low -- lossless, I would call, very low latency networks that can scale. And Ethernet is what makes those networks scale.

隨著我們今天看到的生成人工智能及其緊迫性和興奮性的出現，我們看到這種增長非常非常顯著。我們看到我們的超大規模客戶迫切需要我們來保護產品，以確保能夠部署那些非常非常低的 - 我會稱之為無損的，可以擴展的非常低延遲的網絡。以太網使這些網絡具有可擴展性。

And that will come from the line of Harlan Sur with JPMorgan.

這將來自 Harlan Sur 與摩根大通的合作。

Hock, as your cloud customers are now aggressively focused on generative AI development and deployment across their data center footprints, right, this is driving strong AI-focused Ethernet switch port demand and demand for your compute offload ASICs, like TPU for this year as you mentioned. But from a new product ramp and design win funnel perspective, is this also causing your cloud customers to want to pull forward some of your future programs like Tomahawk 5 or Jericho3 next-gen switching and routing products and/or pulling the design and tape out of their next-generation compute offload AI ASIC programs?

霍克，由於您的雲客戶現在正積極專注於跨數據中心足蹟的生成人工智能開發和部署，對，這正在推動以人工智能為中心的以太網交換機端口需求以及對計算卸載 ASIC 的需求，例如今年的 TPU，因為您提及。但是，從新產品提升和設計獲勝漏斗的角度來看，這是否也會導致您的雲客戶想要推進您未來的一些計劃，例如 Tomahawk 5 或 Jericho3 下一代交換和路由產品和/或退出設計和流片他們的下一代計算卸載 AI ASIC 程序？

Yes, we're seeing all of the foregoing by the way, and that happened over the last 90 days. We have seen a lot of that urgency, a lot of that, you might call it, excitement. But you hit it right on, yes, which is accounting for the color in my commentary about both generative AI-based networks and pushing us to develop a new generation altogether of Ethernet switching that can support this kind of very compute and data-intensive workloads. So that's one side of it.

是的，順便說一句，我們看到了所有上述情況，而且這發生在過去 90 天裡。我們已經看到了很多這種緊迫感，很多，你可以稱之為興奮。但是你說對了，是的，這就是我對基於人工智能的生成網絡的評論中的顏色，並推動我們開發新一代的以太網交換，以支持這種計算和數據密集型工作負載.這就是它的一方面。

And the other side of it, you are right. We typically not want to talk much about compute offload, which is another way of saying, yes, these are very related to some of the engines that certain -- that are fairly customized dedicated to certain hyperscalers.

另一方面，你是對的。我們通常不想過多談論計算卸載，這是另一種說法，是的，這些與某些引擎非常相關——這些引擎是專門為某些超大規模定制的。

And that will come from the line of Vivek Arya with Bank of America.

這將來自美國銀行的 Vivek Arya 系列。

Hock, I'm just curious to understand just the views about the second half. If I look at the last few years, Broadcom has managed to grow semiconductor sales, right, anywhere between 5 to kind of double-digit second half half-over-half. Just the broader business environment, just kind of more of a broader business environment question, not guidance per se, what could change that trend for Broadcom in a positive or negative way this year?

霍克，我只是想了解下半場的觀點。如果我回顧過去幾年，Broadcom 已經成功地實現了半導體銷售額的增長，正確的，下半年增長了 5 到兩位數。只是更廣泛的商業環境，更廣泛的商業環境問題，而不是指導本身，今年有什麼能以積極或消極的方式改變 Broadcom 的趨勢？

In a sort of broadly conceptual, not a guidance, as you say, but trend this way, we're kind of getting rather hopeful that it would be a soft landing. There will be moderation as we are indicating this future -- in this Q2 quarter, moderating growth. But we see, nonetheless, as probably leading to a soft landing of still a year-on-year improvement in the second half.

在某種廣泛的概念上，不是像你所說的指導，而是這種趨勢，我們有點希望這將是一次軟著陸。正如我們所指出的那樣，未來會有所放緩——在這個第二季度，增長放緩。但我們認為，儘管如此，下半年仍可能實現同比改善的軟著陸。

And that will come from the line of Stacy Rasgon with Bernstein.

這將來自 Stacy Rasgon 與 Bernstein 的系列。

I just wanted to verify, did you say that you started hearing urgency from your hyperscale customers around the AI in the last 90 days? And just given that, how do I think about that in the context of lead times that are still 50 weeks? You've got like -- sounds like $1.6 billion in incremental networking growth in year-over-year in '23 from AI across both Ethernet and the ASICs. I guess given the lead times, is that more of a second half kind of thing when that contributes to the model? Or does it contribute more linearly for the year? Or I guess just how do I think about the timing of all of this in the wake of the strong demand right now just given the broader lead times?

我只是想驗證一下，您是否說過在過去 90 天內您開始從您的超大規模客戶那裡聽到有關 AI 的緊迫感？考慮到這一點，在交貨時間仍為 50 週的情況下，我如何考慮這一點？在 23 年，以太網和 ASIC 上的 AI 網絡同比增長聽起來像是 16 億美元。我想考慮到交貨時間，當這對模型有貢獻時，是否更多的是下半年的事情？還是它對這一年的貢獻更線性？或者我想在考慮到更廣泛的交貨時間後，在目前強勁的需求之後，我該如何考慮所有這些的時機？

Stacy, thank you for your question, very perceptive. And as I say, we're not -- we're trying not to -- we are not guiding you guys what happens beyond the second quarter, not the second half of this year. And...

Stacy，謝謝你的問題，非常有洞察力。正如我所說，我們不會 - 我們正在努力不 - 我們不會指導你們第二季度之後發生的事情，而不是今年下半年。和...

But you can give us some guidance for the year on this, right? So...

但是你可以在這一年給我們一些指導，對吧？所以...

No guidance, sorry. I give you a conceptual trend. How is that? But having said that, no, we're still working through timing of when our customers need those urgent -- those products in a fairly urgent manner and our ability to obviously want to be very, very helpful to help customers launch aggressively into generative AI. So we're in the midst of that.

沒有指導，抱歉。我給你一個概念趨勢。那個怎麼樣？但話說回來，不，我們仍在努力確定客戶何時需要這些緊急的產品——這些產品以相當緊急的方式以及我們顯然希望非常非常有幫助的能力來幫助客戶積極地進入生成人工智能.所以我們正處於其中。

Okay. Because like the networking implied guide for Q2 has got to be up like, call it, mid-teens sequentially. Is that -- some of that contributing? Or do I get even more, I guess, as we go beyond because we're already -- once we get through this quarter, we're already through the first half, right? So I guess it has to hit the second half, right?

好的。因為就像第二季度的網絡隱含指南一樣，必須按順序調用它。那是——其中的一些貢獻嗎？或者我會得到更多，我猜，因為我們已經 - 一旦我們度過了這個季度，我們已經度過了上半年，對吧？所以我想它必須打到下半場，對吧？

Stacy, I wish you guys will not do too much analysis, but I know that won't happen. I'm only guiding Q2. I'll let you figure out what happens in second half. I think you're probably better off at it than I am.

Stacy，我希望你們不要做太多分析，但我知道那不會發生。我只是指導 Q2。我會讓你弄清楚下半場會發生什麼。我認為你可能比我更擅長這方面。

And that will come from the line of C.J. Muse with Evercore ISI.

這將來自 C.J. Muse 與 Evercore ISI 的系列。

And I know that it might be difficult to share too much on the ongoing review from the European Commission, but I was hoping maybe you could speak a little bit about where they're concerned, i.e., mixed fiber channel, host bus adapters and other storage adapters. Do you view these as core businesses within Broadcom? Are they easy to extract out of your portfolio? And is there IP that is critical for these businesses that are clearly used by your other larger core businesses? Anything to kind of help us understand would be grateful.

我知道可能很難就歐盟委員會正在進行的審查分享太多信息，但我希望你能談談他們關心的地方，即混合光纖通道、主機總線適配器和其他存儲適配器。您是否將這些視為 Broadcom 的核心業務？從你的投資組合中提取它們容易嗎？是否存在對您的其他較大核心業務明確使用的這些業務至關重要的 IP？任何能幫助我們理解的東西都將不勝感激。

C.J., I appreciate the fact that you have been definitely reading a lot of those Reuters and Bloomberg and Lexicon reports I should say. And you equally know that I cannot and will not comment on any of this as we are working very, very positively and progressively with regulators on all the issues related to our clearance. So sorry, I can't comment. But just to let you know, we're making good progress.

C.J.，我很感激你確實閱讀了很多我應該說的路透社、彭博社和 Lexicon 的報導。你同樣知道，我不能也不會對此發表評論，因為我們正在與監管機構就與我們的許可相關的所有問題進行非常、非常積極和漸進的合作。很抱歉，我無法發表評論。但只是想讓你知道，我們正在取得很好的進展。

And that will come from the line of Vijay Rakesh with Mizuho.

這將來自 Vijay Rakesh 與 Mizuho 的血系。

Hock, just a quick question on -- you talked about generative AI. Just wondering, as you look at the workload, what percent of workload would be on generative AI, like exiting calendar '23 or '24? And also I want to hit on the silicon photonics side. I think you briefly mentioned the silicon photonics cable with integrated switch, the 51.2 terabyte switch. When do you see this ramping? And what's the power advantage on that?

霍克，只是一個簡短的問題——你談到了生成人工智能。只是想知道，當您查看工作量時，生成 AI 的工作量百分比是多少，例如退出日曆 '23 或 '24？而且我還想談談矽光子學方面的問題。我想您簡要提到了帶有集成開關的矽光子電纜，即 51.2 TB 開關。你什麼時候看到這個斜坡？那有什麼功率優勢呢？

Okay. Well, it's -- I'm sure I don't need to elaborate on what we all hear about on generative AI. And I think it's still early innings on generative AI, but we obviously are also indicating, as we are seeing a very strong and a strong sense of urgency among our customers especially in the hyperscale environment to be -- to not miss out -- not to be late in this trend. And with generative AI, as I said, with many more -- much more billions of parameters that come into the models that they're doing, you're talking about scale out of data centers driving AI engines network together in a manner that we probably have not seen before.

好的。好吧，我相信我不需要詳細說明我們都聽到的有關生成 AI 的內容。而且我認為生成人工智能仍處於早期階段，但我們顯然也在表明，因為我們看到我們的客戶非常強烈的緊迫感，尤其是在超大規模環境中——不要錯過——不要在這個趨勢中遲到了。正如我所說，對於生成式人工智能，還有更多——更多的數十億參數進入他們正在做的模型中，你在談論數據中心的橫向擴展，以我們的方式驅動人工智能引擎網絡在一起可能以前沒見過。

It's not a problem that's not solvable. It is very, very clearly solvable. As evidenced by the fact that we have and deployed technology to support AI networks even today to certain hyperscalers, where we're talking about at least hundreds, if not, thousands of AI engines, AI servers network together and working in a synchronous manner. So this is about ability to scale out in a fairly substantial manner.

不是解決不了的問題。這是非常非常清楚地可以解決的。事實證明，即使在今天，我們已經擁有並部署了支持 AI 網絡的技術，以支持某些超大規模應用程序，我們在這裡談論的是至少數百個（如果沒有的話）數千個 AI 引擎、AI 服務器網絡在一起並以同步方式工作。所以這是關於以相當大的方式擴展的能力。

And that was the color I was providing. And it's really about trying to make sure that happens and not be the bottleneck to our ability to get the best system performance. And I emphasize the word system performance of an AI data center. And where it's coming from right now is frankly how to network them and how to do those massive parametric exchange, so to speak, when you run large numbers of engines or machines in parallel, as you grind through this huge database and that we need to do.

那就是我提供的顏色。這實際上是為了確保這種情況發生，而不是成為我們獲得最佳系統性能能力的瓶頸。我強調人工智能數據中心的系統性能這個詞。坦率地說，它現在的來源是如何將它們聯網以及如何進行那些大規模的參數交換，可以這麼說，當你並行運行大量引擎或機器時，當你在這個龐大的數據庫中進行研究時，我們需要做。

So that's -- we are in early innings and which is why we think we have time to come -- to start to work on even a new generation of switches in Ethernet that are specifically designed, dedicated to these kind of workloads, which are very different from the normal workloads that we see today traditionally in data centers. And we have to address that. They have to be, as I say, literally lossless -- virtually lossless, very low latency and be able to scale into thousands of engines.

所以這就是 - 我們處於早期階段，這就是為什麼我們認為我們有時間 - 開始研究甚至是專門設計的新一代以太網交換機，專用於這些工作負載，這是非常不同於我們今天傳統上在數據中心看到的正常工作負載。我們必須解決這個問題。正如我所說，它們必須是字面意義上的無損——幾乎無損、非常低的延遲並且能夠擴展到數千個引擎。

And that's the main 3 criteria we are aware of, and we're driving solutions -- silicon solutions that enable that. We have it but we think we need to improve the performance of what we have to -- in anticipation of a trend that we foresee over the next several years. And so we're putting a lot of investment in that direction.

這是我們知道的主要 3 個標準，我們正在推動解決方案——能夠實現這一點的矽解決方案。我們擁有它，但我們認為我們需要提高我們必須做的事情的表現——因為我們預見到未來幾年的趨勢。因此，我們在這個方向上投入了大量資金。

On the silicon photonics cable, just wondering when the time of ramp and [the port advantage] there.

在矽光子電纜上，只是想知道斜坡時間和 [端口優勢] 何時出現。

Well, we intend to launch Tomahawk 5 early '24 as we indicated previously, and that's the conventional silicon-based with pluggable optics switch -- top of the rack switch Tomahawk 5, 51.2 terabit per second. Bailly, which is the fully integrated silicon photonic version, you don't fully integrate the active elements of those pluggable optics into the switch. We anticipate launching that shortly thereafter.

好吧，正如我們之前所說，我們打算在 24 年初推出 Tomahawk 5，這是傳統的基於矽的可插拔光學開關——架頂式開關 Tomahawk 5，每秒 51.2 太比特。 Bailly，這是完全集成的矽光子版本，您沒有將那些可插拔光學器件的有源元件完全集成到開關中。我們預計此後不久將推出。

Power wise, you can see silicon photonics. That's a lot. The Tomahawk 5 compared to what we have today is 2x the performance of Tomahawk 4, and -- but we believe we can do Tomahawk 5 at the same power -- close to the same power, if not, lower than the Tomahawk 4.

在功率方面，您可以看到矽光子學。好多啊。與我們今天擁有的相比，Tomahawk 5 的性能是 Tomahawk 4 的 2 倍，而且——但我們相信我們可以在相同的功率下實現 Tomahawk 5——接近相同的功率，如果不是，則低於 Tomahawk 4。

And that will come from the line of Ross Seymore with Deutsche Bank.

這將來自德意志銀行的 Ross Seymore。

I wanted to go into the compute offload number that you talked about, Hock, the $2 billion last fiscal year going to $3 billion this year. I know it's a touchy subject and so no customer specifics, of course. But generally speaking, can you just talk about the breadth and types of compute offload and how that's changing in the mix from the $2 billion last year to $3 billion this year?

Hock，我想談談你談到的計算卸載數字，上一財年的 20 億美元將在今年增加到 30 億美元。我知道這是一個敏感的話題，所以當然沒有客戶的具體信息。但總的來說，您能否談談計算卸載的廣度和類型，以及從去年的 20 億美元到今年的 30 億美元的組合如何變化？

Well, I'd rather not answer that question, Ross. Highly sensitive to some of my very limited customer base. But as I said, it includes some of the engines, the compute engines, and some are related components that support this engine.

好吧，羅斯，我不想回答這個問題。對我的一些非常有限的客戶群高度敏感。但正如我所說，它包括一些引擎、計算引擎，以及一些支持該引擎的相關組件。

Is the concentration changing? So are you broadening customers in that growth?

濃度有變化嗎？那麼，您是否在這種增長中擴大了客戶？

No, no, very concentrated.

不不不，非常專注。

And that will come from the line of Edward Snyder with Charter Equity.

這將來自 Charter Equity 的愛德華斯奈德。

Good quarter, Hock. So apparently over the last quarter, you were getting out of wireless, you're getting into wireless or handset guys are going to start doing wireless. So I wanted to get a couple of updates, so maybe you could set the record straight. First of all, even if you see a sea change in, let's say, silicon, mixed silicon baseband providers in the next year or 2, does that fundamentally change your opinion of your wireless group? And either way, actually. Does it get better? Does it get worse? Because obviously, if architectures change, it has a big impact on supply chain. And I know, historically, you've worked very closely with key players in helping develop all the other pieces of the puzzle like transceivers that are required if you're going to do your own. So maybe you could just kind of reset the bar on what you expect for, without guidance, but in general, the wireless division in the next year or 2, does it -- is that superior or greater?

好季度，霍克。所以很明顯，在上個季度，你正在擺脫無線，你正在進入無線，或者手機廠商將開始做無線。所以我想獲得一些更新，所以也許你可以澄清一下。首先，即使您看到明年或兩年內矽、混合矽基帶供應商發生翻天覆地的變化，這是否會從根本上改變您對無線團隊的看法？事實上，無論哪種方式。它會好轉嗎？它會變得更糟嗎？因為很明顯，如果架構發生變化，就會對供應鏈產生重大影響。而且我知道，從歷史上看，您與主要參與者密切合作，幫助開發所有其他難題，例如如果您要自己做就需要收發器。所以也許你可以在沒有指導的情況下重新設定你期望的標準，但總的來說，未來一兩年的無線部門，是更好還是更好？

Thanks. Good question, Ed. As you know, our wireless group, as you call it, not division, is really not one single product line or one single division. It's not one homogenous group either. There's a few key products that comprises this wireless division, all selling, you're right, you're correct, to the same application and very high-end flagship status handsets and largely focused on one key customer, our North American -- our much beloved North American OEM customer. So in that sense, it's one single focus area.

謝謝。好問題，埃德。如您所知，我們的無線組，如您所說，不是部門，實際上不是一個單一的產品線或一個部門。它也不是一個同質的群體。這個無線部門有一些關鍵產品，所有銷售，你是對的，你是對的，針對相同的應用程序和非常高端的旗艦手機，並且主要集中在一個關鍵客戶，我們的北美 - 我們深受北美OEM客戶喜愛。所以從這個意義上說，它是一個單一的重點領域。

And to answer your question, while we're on these multiple products and they tend to keep -- progress as each new generation happens may not be every year, but it happens fairly regular frequency on a cadence that is pretty predictable of the one -- each on its own cadence. It's a very, very good business for us.

並回答你的問題，雖然我們在使用這些多種產品並且它們往往會保持 - 隨著每一代新產品的出現可能不會每年都取得進展，但它以相當可預測的節奏發生相當規律的頻率 - - 每個都有自己的節奏。這對我們來說是一項非常非常好的業務。

And to answer your question directly, no, nothing meaningful has changed. Our relationship, our strategic engagement continues very much the same as it has for the last multiple years. And we see that to continue in a fairly predictable, stable manner.

直接回答你的問題，不，沒有任何有意義的改變。我們的關係，我們的戰略參與與過去多年的情況非常相似。我們看到這種情況將以一種相當可預測的、穩定的方式繼續下去。

And then just to remind if we could, 3-year road map, I mean you see stuff pretty far out, right?

然後只是提醒我們是否可以，3 年路線圖，我的意思是你看到的東西很遠，對吧？

Yes.

是的。

That will come from the line of Pierre Ferragu with New Street Research.

這將來自 New Street Research 的 Pierre Ferragu 系列。

Can you hear me well?

你能聽清楚嗎？

Yes.

是的。

Yes. Great. So I'm trying to put together a perspective of what's happening at hyperscale clients this year. So if I look at your networking division, if you grow like at least $600 million this year in AI and if you have compute offload division growth by $1 billion, that might well represent all your growth in that -- in networking. So that would mean the only thing that is really growing and that is growing a lot this year in that space is AI. And when I look at outside of Broadcom, what we've seen is memory and like the x86 CPU servers are having a very difficult time at the moment. We expect a recovery in the second half, while like the GPU segment of the market is actually in very, very good shape and growing very well and accelerating again.

是的。偉大的。因此，我試圖將今年超大規模客戶正在發生的事情放在一起。因此，如果我看一下您的網絡部門，如果您今年在人工智能方面至少增長了 6 億美元，並且如果您的計算卸載部門增長了 10 億美元，那很可能代表您在網絡方面的所有增長。因此，這意味著今年在該領域唯一真正增長並且增長很多的是人工智能。當我看向 Broadcom 外部時，我們看到的是內存，就像 x86 CPU 服務器目前正處於非常困難的時期。我們預計下半年會出現復甦，而 GPU 市場實際上處於非常非常好的狀態，增長非常好並再次加速。

So my question at the end of the day is, is it fair to say that in these large data centers this year, only AI is growing? And is that a sign of what the future will be? Or do you think like the general purpose part of infrastructure like centered around x86 or similar or general purpose CPUs still is a very good growth market?

所以我最後的問題是，今年在這些大型數據中心中，只有 AI 在增長是否公平？這是未來的標誌嗎？或者您是否認為以 x86 或類似或通用 CPU 為中心的基礎設施的通用部分仍然是一個非常好的增長市場？

You put -- you pose very, very interesting and good questions, Pierre. The problem is I do not -- my customers, hyperscale customers do not necessarily honor me by sharing all those insights that you are -- and on those questions you are asking. I do not know. I do not know. All I know -- what I do know -- because I don't sell them CPUs. I don't even sell them GPUs by the way.

你提出了 - 你提出了非常非常有趣和好的問題，皮埃爾。問題是我沒有——我的客戶，超大規模客戶不一定通過分享你的所有這些見解來尊重我——以及你提出的那些問題。我不知道。我不知道。我所知道的——我所知道的——因為我不賣給他們 CPU。順便說一下，我什至不賣給他們 GPU。

But I know what you know out there, which is in certain areas of their business, we're seeing some of these hyperscalers bringing on a sense of urgency and focus and of course, spending to be up to speed, if not, to not be left behind as we see the excitement hyped perhaps in pushing applications and put workloads in generative AI. That's what we see driving a lot of this excitement. And we -- all we are saying is we've seen some of that effect on our networking business with those hyperscalers. That's what it is.

但我知道你所知道的，在他們業務的某些領域，我們看到其中一些超大規模製造商帶來了緊迫感和專注感，當然，他們會花錢跟上速度，如果不是，當我們看到在推動應用程序並將工作負載置於生成 AI 中的興奮炒作時，我們可能會被拋在後面。這就是我們所看到的推動這種興奮的原因。我們——我們要說的是，我們已經看到了一些對我們的網絡業務的影響。就是這樣。

Beyond that, we -- unfortunately, other than the backlog we get in normal networking switches, routers and key components, we see that. And as I indicated in this last quarter's results, we continue to see sustained strength, now last quarter and continuing as we indicate this particular quarter, Q2.

除此之外，我們 - 不幸的是，除了我們在正常網絡交換機、路由器和關鍵組件中遇到的積壓之外，我們還看到了這一點。正如我在上個季度的結果中指出的那樣，我們繼續看到持續的強勁勢頭，現在是上個季度，並且正如我們指出的這個特定季度，第二季度一樣。

Beyond that, we don't get to see -- we do not want to guide what we're going to see beyond that. But right now, last quarter, this quarter, yes, traditional data centers' scale out in networking and deployment in networking continues to be strong and sustained in hyperscalers as well, I might indicate, in the enterprise.

除此之外，我們看不到——我們不想指導我們將要看到的東西。但是現在，上個季度，這個季度，是的，傳統數據中心在網絡中的擴展和網絡部署在超大規模應用程序中繼續保持強勁和持續，我可能會指出，在企業中。

Okay. Great. And just to clarify specifically on what you are doing, is that fair to assume that the majority -- a very large majority of your growth this year in networking is going to come from AI, which you have $600 million coming from AI (inaudible) and $1 billion coming from offload chips? Or is that not the right way to think about it? Just for your business. I'm not looking at anything else.

好的。偉大的。只是為了具體說明你在做什麼，假設大多數 - 今年網絡增長的絕大部分將來自人工智能，你有 6 億美元來自人工智能（聽不清）還有 10 億美元來自卸載芯片？或者這不是正確的思考方式嗎？只為您的事業。我沒有看其他任何東西。

I will not think about it at this point. It might be a bit too mature. Don't forget, generative AI is still early stage.

我現在不會考慮它。可能有點太成熟了。別忘了，生成式 AI 仍處於早期階段。

And we do have time for one final question, and that will come from the line of Karl Ackerman with BNP Paribas.

我們確實有時間回答最後一個問題，這個問題將來自法國巴黎銀行的卡爾·阿克曼 (Karl Ackerman)。

There were many great questions, quite frankly, on the networking business, which I think is quite significant for you. Maybe if I could, a clarification on that and then a broader question that I want to address on broadband. On the networking piece, I was curious if you could discuss the growth opportunity in your Tomahawk portfolio now that a peer has elected to stop investing in their switch division.

坦率地說，關於網絡業務有很多很好的問題，我認為這對你來說非常重要。也許我可以對此進行澄清，然後是我想在寬帶上解決的更廣泛的問題。在網絡部分，我很好奇你是否可以討論你的戰斧投資組合中的增長機會，因為同行已經選擇停止投資他們的交換機部門。

And then as it relates to broadband, several companies across the broadband ecosystem have guided a softer outlook due to a buildup of inventory, but quite frankly, that's been on the customer premise side. You obviously have more weighting towards fiber and sell into the infrastructure portion. And so I was hoping you could discuss how you're thinking about the growth of your fiber business within broadband both from an infrastructure side and a consumer equipment standpoint as governments begin to deploy funds for broadband infrastructure.

然後，由於涉及寬帶，寬帶生態系統中的幾家公司由於庫存增加而導致前景疲軟，但坦率地說，這是在客戶前提方面。您顯然對光纖有更多的重視，並出售給基礎設施部分。因此，我希望您能討論隨著政府開始為寬帶基礎設施部署資金，您如何從基礎設施方面和消費設備的角度考慮寬帶內光纖業務的增長。

Thank you for that question. Yes, broadband is, for us, a very, very good business and very sustaining, used to be boring. Boring is good at this point. And last quarter, Q1, as I reported, we actually grew 34% year-on-year. In my view, that's rather exceptional even though, in broadband, we have been seeing year-on-year growth now at least for the past 4, 5 quarters. But still 34% was rather exceptional. And sure enough, Q2, it normalizes to a more sedate level, but still growing. And the growth in that is simply because we are very well positioned with respect to next-generation PON, 10-gig PON, which has been deployed in big volumes now by telcos supported by their governments, countries all over Europe and even in North America, not to mention other nations beyond that. Basically, it is about reaching this key utility broadband service to every household, and we see a lot of deployment.

謝謝你提出這個問題。是的，寬帶對我們來說是一項非常非常好的業務，而且非常可持續，過去很無聊。在這一點上，無聊是件好事。正如我所報告的那樣，上個季度，第一季度，我們實際上同比增長了 34%。在我看來，即使在寬帶方面，我們現在至少在過去的 4、5 個季度中都看到了同比增長，但這是相當特別的。但仍然有 34% 相當例外。果然，第二季度，它正常化到一個更穩定的水平，但仍在增長。而這方面的增長僅僅是因為我們在下一代 PON 10-gig PON 方面處於非常有利的地位，現在已經由電信公司大量部署，這些電信公司得到了政府、歐洲國家甚至北美國家的支持，更不用說其他國家了。基本上，它是為了讓每個家庭都能享受到這一關鍵的公用事業寬帶服務，我們看到了很多部署。

And then more vertical market, we also see simultaneous with PON or fiber, as you call it, a large strong continued deployment of cable, DOCSIS 3 coaxial to the home because the cable operators, a few of them who are on the scale of the telcos and who need to maintain competitiveness as the telcos launch 10-gigabit PON, that cable has to update DOCSIS to be able to compete and not lose subscribers in the same market they compete against each other. So we see strength both in cable, DOCSIS 3.1, as I call it, and potentially next generation, not yet happening but hopefully within the next couple of years, DOCSIS 4.0.

然後是更多的垂直市場，我們也看到與 PON 或光纖同時出現，如您所說，大量強大的電纜繼續部署，DOCSIS 3 同軸到戶，因為電纜運營商，其中一些規模電信公司和需要保持競爭力的人，因為電信公司推出 10 吉比特 PON，該電纜必須更新 DOCSIS 才能在同一個市場競爭並且不會失去訂戶，他們相互競爭。因此，我們看到了電纜（我稱之為 DOCSIS 3.1）和潛在的下一代 DOCSIS 4.0（尚未出現但希望在未來幾年內出現）的優勢。

Meanwhile, PON is happening, which accounts for the strength we saw last quarter and continuing strength over the last several quarters. And content increases come to not just unit deployment of those gateways and infrastructure but also the fact that a lot of these deployments come with very high attach rates of WiFi 6 and 6E. And that provides additional boost, content increases more, what I would call it, to our revenue growth in broadband. So that's quietly still chugging along very nicely for us. All right.

與此同時，PON 正在發生，這說明了我們在上個季度看到的實力以及過去幾個季度的持續實力。內容的增加不僅來自於這些網關和基礎設施的單元部署，還來自於許多這些部署都具有非常高的 WiFi 6 和 6E 連接率這一事實。這為我們的寬帶收入增長提供了額外的推動力，內容增加了更多，我稱之為。所以這對我們來說仍然很安靜。好的。

As I'm showing no further questions in the queue at this time, I would now like to turn the call back over to Ji Yoo for any closing remarks.

由於此時隊列中沒有其他問題，我現在想將電話轉回給 Ji Yoo 以聽取任何結束語。

Thank you, Sheri. In closing, we would like to highlight that Broadcom will be attending the Morgan Stanley Technology, Media and Telecom Conference on Tuesday, March 7. Broadcom currently plans to report its earnings for the second quarter of fiscal '23 after close of market on Thursday, June 1, 2023. A public webcast of Broadcom's earnings conference call will follow at 2:00 p.m. Pacific Time.

謝謝你，雪莉。最後，我們想強調 Broadcom 將參加 3 月 7 日星期二舉行的摩根士丹利技術、媒體和電信會議。Broadcom 目前計劃在周四收盤後報告其 23 財年第二季度的收益， 2023 年 6 月 1 日。Broadcom 財報電話會議的公開網絡直播將於美國東部時間下午 2:00 進行。太平洋時間。

That will conclude our earnings call today. Thank you all for joining. Sheri, you may end the call.

這將結束我們今天的財報電話會議。謝謝大家的加入。雪莉，你可以結束通話了。

Thank you all for participating. This concludes today's program. You may now disconnect.

謝謝大家的參與。今天的節目到此結束。您現在可以斷開連接。