电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链_老虎社区_美港股上老虎

电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链

东吴证券09-10

（以下内容从东吴证券《电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链》研报附件原文摘录）投资要点Rubin CPX切入百万Token痛点，重塑推理架构基础。过去一年，随着生成式AI进入规模化落地阶段，行业对“长上下文”的需求快速上升。无论是企业级知识库问答、代码生成，还是多模态长视频生成，均需要模型在极大输入序列下保持推理准确性与计算效率。然而，现有GPU在应对...

网页链接

免责声明：本文观点仅代表作者个人观点，不构成本平台的投资建议，本平台不对文章信息准确性、完整性和及时性做出任何保证，亦不对因使用或信赖文章信息引发的任何损失承担责任。

暂无评论

热议股票

7x24

全部

{"i18n":{"language":"zh_CN"},"data":{"share":"https://www.laohu8.com/m/news/2566627142?lang=zh_CN&edition=full","thumbnail":"","is_english":false,"pubTime":"2025-09-10 14:05","share_image_url":"https://static.laohu8.com/9a95c1376e76363c1401fee7d3717173","id":"2566627142","market":"us","top_or_hot":-1,"title":"电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链","media":"东吴证券","content":"<div>\n<p>（以下内容从东吴证券《电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链》研报附件原文摘录）投资要点Rubin CPX切入百万Token痛点，重塑推理架构基础。过去一年，随着生成式AI进入规模化落地阶段，行业对“长上下文”的需求快速上升。无论是企业级知识库问答、代码生成，还是多模态长视频生成，均需要模型在极大输入序列下保持推理准确性与计算效率。然而，现有GPU在应对...</p>\n\n<a href=\"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025091014054795204a86&s=b\">网页链接</a>\n\n</div>\n","source":"tencent","html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\n电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链\n</h2>\n\n<h4 class=\"meta\">\n\n\n2025-09-10 14:05 北京时间&nbsp;&nbsp;&nbsp;<a href=http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025091014054795204a86&s=b><strong>东吴证券</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>（以下内容从东吴证券《电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链》研报附件原文摘录）投资要点Rubin CPX切入百万Token痛点，重塑推理架构基础。过去一年，随着生成式AI进入规模化落地阶段，行业对“长上下文”的需求快速上升。无论是企业级知识库问答、代码生成，还是多模态长视频生成，均需要模型在极大输入序列下保持推理准确性与计算效率。然而，现有GPU在应对...</p>\n\n<a href=\"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025091014054795204a86&s=b\">网页链接</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","isBrief":false,"type":0,"news_type":1,"symbol":"LU0238689110.USD","symbol_name":"贝莱德环球动力股票基金","start_time":0,"source_url":"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025091014054795204a86&s=b","article_id":"2566627142","we_media_id":null,"thumbnails":[],"rights":{"source":"tencent","url":"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN2025091014054795204a86&s=b","rn_cache_url":null,"customStyle":"body{padding-top:10px;}#news_title{font-weight:bold;#titleStyle#;}#news_description span{font-size:12px;#descriptionStyle#;}.footer-note{#statement#}","selectors":".mod-LoadTzbdNews, body","filters":".relate-stock, .hot-list, .recom-box, .wx-sou","directOrigin":true},"url":"https://stock-news.laohu8.com/highlight/detail?id=2566627142","pubTimestamp":1757484334,"columns":[],"sourceInfo":{"source_id":"tencent","name":"腾讯"},"weMediaInfo":null,"summary":"投资要点Rubin CPX切入百万Token痛点，重塑推理架构基础。英伟达于2025年9月发布的Rubin CPX，正是为解决这一痛点而生，标志着NVIDIA将推理场景的架构优化推向新高度。根据公司规划，Rubin CPX预计将在2026年底上市，与Rubin GPU与Dynamo、TensorRT-LLM等软件工具链一体化部署。从产业角度来看，Rubin CPX的推出不仅是NVIDIA产品线的升级，更意味着海外算力基础设施进入“上下文与生成分工协作”的新阶段。","collect":0,"end_time":0,"defaultTopTitle":"qq.com","property":[],"viewcount":null,"language":"zh","relate_stocks":{"LU0238689110.USD":"贝莱德环球动力股票基金","LU0109391861.USD":"富兰克林美国机遇基金A Acc","NVDD":"1倍做空NVDA ETF-Direxion","LU1429558221.USD":"Natixis Loomis Sayles US Growth Equity RA USD","LU2237957902.USD":"NIKKO AM GLOBAL EQUITY \"F\" (USD) ACC","LU2083900584.USD":"HSBC GIF US INCOME FOCUSED \"AM2\" (USD) INC","LU0417517546.SGD":"Allianz US Equity Cl AT Acc SGD","LU1066051811.HKD":"HSBC GIF GLOBAL EQUITY VOLATILITY FOCUSED \"AM2\" (HKD) INC","LU2361044949.HKD":"WELLINGTON US QUALITY GROWTH \"A\" (HKD) ACC","IE00B19Z8W00.USD":"FTGF CLEARBRIDGE US LARGE CAP GROWTH \"A\" INC","LU0069063385.USD":"SUSTAINABLE GLOBAL THEMATIC PORTFOLIO \"A\" (USD) ACC","LU2471134879.HKD":"INVESCO GLOBAL EQUITY INCOME ADVANTAGE \"A\" (HKD) INC","LU2125154778.USD":"ALLSPRING GLOBAL EQUITY ENHANCED INCOME \"A\" (USD) INC","IE00B1XK9C88.USD":"PINEBRIDGE US LARGE CAP RESEARCH ENHANCED \"A\" (USD) ACC","NVD2.UK":"2X NVIDIA ETP","LU1992135399.USD":"Allianz Global Intelligent Cities AT Acc USD","LU1280957306.USD":"THREADNEEDLE (LUX) US CONTRARIAN CORE EQUITIES \"AUP\" (USD) INC","SNVD.UK":"LS -1X NVIDIA","SGXZ51526630.SGD":"大华环球创新基金A Acc SGD","NVD3.UK":"LS 3X NVIDIA","LU2361045086.USD":"WELLINGTON US QUALITY GROWTH \"A\" (USD) ACC","NVIW.SI":"NVDA 3xLongSG261006","SG9999004303.SGD":"Nikko AM Shenton Global Opportunities SGD","NVDS":"1.5倍做空NVDA ETF-Tradr","NVDA":"英伟达","IE00BN29S564.USD":"JANUS HENDERSON BALANCED \"A3\" (USD) INC","NVDS.UK":"LS -1X NVIDIA","LU0820562030.AUD":"ALLIANZ INCOME AND GROWTH \"AMH2\" (AUDHDG) H2 INC","NVDU":"2倍做多NVDA ETF-Direxion","HK0000306685.HKD":"TAIKANG KAITAI CHINA NEW OPPORTUNITIES FUND \"A\" (HKD) INC","LU2023251221.USD":"ALLIANZ GLOBAL SUSTAINABILITY \"AM\" (USD) INC","IE00BZ1G4Q59.USD":"LEGG MASON CLEARBRIDGE US EQUITY SUSTAINABILITY LEADER \"A\"(USD) INC (A)","3NVD.UK":"LS 3X NVIDIA","LU2089985449.USD":"MANULIFE GF GLOBAL SEMICONDUCTOR OPPORTUNITIES \"AA\" (USD) ACC","LU1242518931.SGD":"Fullerton Lux Funds - Asia Absolute Alpha A Acc SGD","LU1935043536.SGD":"MANULIFE GF GLOBAL MULTI-ASSET DIVERSIFIED INCOME \"AA\" (SGDHDG) INC A","LU1989764748.USD":"东方汇理环球颠覆性机遇A2 Acc","NVDX":"2倍做多NVDA ETF-T-Rex","IE0009356076.USD":"JANUS HENDERSON GLOBAL TECHNOLOGY AND INNOVATION \"A2\" (USD) ACC","LU2290526834.HKD":"BGF NEXT GENERATION TECHNOLOGY \"A2\" (HKDHDG) ACC","NVD":"2倍做空NVDA ETF-GraniteShares","IE00BZ199S13.USD":"BNY MELLON MOBILITY INNOVATION \"B\" (USD) ACC","2NVD.UK":"2X NVIDIA ETP","NVDY":"NVDA期权收益策略ETF-YieldMax","LU0320765059.SGD":"FTIF - Franklin US Opportunities A Acc SGD","LU0203201768.USD":"AB SICAV I - ALL MARKET INCOME PORTFOLIO \"AX\" (USD) INC","LU1815336760.USD":"THREADNEEDLE (LUX) GLOBAL TECHNOLOGY \"AUP\" (USD) INC","LU1917777945.USD":"安联专题基金Cl AT Acc","LU0820561909.HKD":"ALLIANZ INCOME AND GROWTH \"AM\" (HKD) INC"},"translate_title":"Electronics Industry Review Report: The era of millions of Tokens is coming, Rubin CPX reshapes the inference architecture and industrial chain","themeId":null,"isJumpTheme":false,"ttsUrl":null,"symbols_score_info":{"NVDS.UK":0.6,"NVDY":0.6,"NVIW.SI":0.6,"3NVD.UK":0.6,"NVD":0.6,"SNVD.UK":0.6,"NVD3.UK":0.6,"NVDA":0.9,"NVDD":0.6,"NVDX":0.6,"NVD2.UK":0.6,"NVDS":0.6,"2NVD.UK":0.6,"NVDU":0.6},"content_text":"（以下内容从东吴证券《电子行业点评报告：百万Token时代来临，Rubin CPX重塑推理架构与产业链》研报附件原文摘录）投资要点Rubin CPX切入百万Token痛点，重塑推理架构基础。过去一年，随着生成式AI进入规模化落地阶段，行业对“长上下文”的需求快速上升。无论是企业级知识库问答、代码生成，还是多模态长视频生成，均需要模型在极大输入序列下保持推理准确性与计算效率。然而，现有GPU在应对超长上下文时普遍存在内存带宽瓶颈与计算冗余，导致算力利用率不足。英伟达于2025年9月发布的Rubin CPX，正是为解决这一痛点而生，标志着NVIDIA将推理场景的架构优化推向新高度。Rubin CPX是一款专为“百万级上下文”推理场景设计的专用加速处理器（Contextual Processing eXtension）。其设计目标是通过硬件与架构优化，提升对超长上下文场景的吞吐与能效，并在机架级系统，如VeraRubin NVL144CPX中与Rubin GPU及Vera CPU协同构成面向大规模推理的整体平台。上下文与生成任务分工，实现算力利用率与效率提升。在大模型推理过程中，可大体分为两类任务：一是“上下文分析”，即对超长输入序列进行编码、筛选与压缩，以便后续生成环节调用；二是“生成任务”，即基于上下文信息进行逐Token的预测输出。二者在计算负载和性能需求上差异显著：上下文分析更依赖并行化处理和带宽利用，而生成任务则要求对计算延迟与单步性能进行极致优化。英伟达Rubin CPX定位为“上下文处理加速器”，负责对海量输入做高通量注意力与前置计算；而Rubin系列通用GPU则负责生成/输出阶段的持续带宽密集型任务。官方展示的Vera Rubin NVL144CPX机架中，144个Rubin CPX（context）配合144个Rubin GPU（generation）与36个Vera CPU（调度/通用），共同提供完整服务能力，进而实现资源的高效利用，推理成本降低以及推理响应加速。Rubin CPX30PFLOPS，机架8EFLOPS算力，2026年落地路径明确。从技术参数看，Rubin CPX单卡提供约30PFLOPS（NVFP4精度）的算力，并搭载128GB GDDR7显存，同时内置视频编解码能力，可满足多模态场景需求。官方展示的Vera Rubin NVL144CPX系统，由144张CPX、144张Rubin GPU和36个Vera CPU共同构成，其整体性能指标达到8ExaFLOPS算力、100TB高速内存与1.7PB/s内存带宽，相比上一代GB300NVL72系统在上下文处理效率上实现数倍提升。根据公司规划，Rubin CPX预计将在2026年底上市，与Rubin GPU与Dynamo、TensorRT-LLM等软件工具链一体化部署。海外算力链受益加速，长上下文推理带来新增长动能。从产业角度来看，Rubin CPX的推出不仅是NVIDIA产品线的升级，更意味着海外算力基础设施进入“上下文与生成分工协作”的新阶段。随着百万Token推理与长视频生成成为AI应用的标配需求，硬件和软件的耦合度显著提高，算力产业链的价值量同步上升。无论是GPU、存储、网络，还是配套的高速PCB、光模块与封装工艺，相关厂商都有望深度受益。我们认为Rubin CPX的量产与落地，将成为全球算力需求加速释放的重要信号，产业链公司中长期成长空间将更加清晰。产业链相关公司：PCB/CCL：沪电股份、胜宏科技、生益电子、深南电路、景旺电子、广合科技、生益科技、南亚新材；铜缆：沃尔核材、博创科技、华丰科技；光芯片/光器件：博创科技、仕佳光子、太辰光、长光华芯、源杰科技；服务器代工：工业富联、华勤技术风险提示：供应链波动风险，下游需求不及预期，行业竞争加剧。","kind":"news","is_publish_news":true,"is_publish_highlight":false,"is_publish_live":false,"is_publish_wemedia":null,"editions":null,"column":"","sentiment":"0","news_tag":"","news_rank":0,"symbols":[],"gpt_button":0,"need_auth":false,"code":"91000000","status":"200"},"commentList":[],"hasMoreComments":false,"newsSizeData":{"likeSize":0,"commentSize":0,"repostSize":0,"favoriteSize":1,"likeStatus":false,"favoriteStatus":false},"isCrawlerRequest":true}