[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"content-doc-dcioffd7a921":3},{"user":4,"document":8,"mainDocument":27,"columnUrl":29,"subscription":30,"footer":42,"text":77},{"isAuthenticated":5,"isAdmin":5,"displayName":6,"avatarUrl":6,"nid":6,"groupLevel":7},false,"",-10,{"id":9,"fullTitle":10,"subTitle":6,"url":11,"columnId":12,"columnName":13,"columnUrl":14,"summary":6,"contentHtml":15,"mainContentHtml":6,"posterUrl":16,"createDate":17,"displayDate":18,"displayDateSlash":19,"pageviews":20,"tags":21,"hidden":5,"isSubContent":5,"replyDocOrTargetId":6,"contentType":23,"videoId":6,"liveVideoUrl":6,"duration":24,"price":24,"priceText":25,"priceBadgeText":25,"priceBadgeClass":26,"freeForMinGroupLevel":24,"redirectUrl":6,"readyToStream":5},"dcioffd7a921","从推理到训练：Meta(META)宣布自研芯片战略升级，CFO称定制芯片是“核心支柱”","\u002Fdoc\u002Fdcioffd7a921","col18178739ee","美股资讯","\u002Fcol\u002Fcol18178739ee","\u003Cp> \u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">美股投资网获悉，尽管Meta Platforms Inc.(META)近期与顶级芯片制造商达成了重大交易，但该公司首席财务官Susan Li在本周三仍明确表示，公司正致力于扩展定制芯片的应用边界。她指出，由于Meta的部分工作负载具有极高的定制化属性，自研芯片能够更好地适配内部特定算法需求。目前，Meta已经在其核心的排名与推荐系统领域实现了定制芯片的大规模部署，而未来的战略重点将是把这一能力逐步扩展至人工智能模型的训练领域。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">尽管并非传统云计算提供商，Meta却是全球用于训练和运行人工智能模型的最大数据中心运营商之一。过去几周内，该公司已与行业领导者英伟达(NVDA)及竞争对手AMD(AMD)达成多项巨额协议，采购用于支撑人工智能工作负载的芯片与设备。与此同时，这家社交媒体母公司仍在持续推进其内部人工智能处理器的研发进程。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">Susan Li在发言中强调，Meta正通过采购不同类型芯片来适配多样化任务需求。\"基于当前认知与实际需求，我们正在系统性评估每个应用场景最适合的芯片方案，\"她指出，\"而定制芯片始终是这一战略布局中的核心支柱。\"\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">这一表态标志着Meta自研芯片项目(MTIA)进入了关键的进阶阶段。自2023年首次公开MTIA计划以来，Meta最初的研发重心主要集中在推理环节，旨在提升Facebook和Instagram推荐系统的运算效率并降低对英伟达通用GPU的依赖。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">随着生成式AI浪潮的爆发，Meta对算力的需求呈现指数级增长，仅仅满足于推理环节已不足以支撑其大模型战略。苏珊·利的最新声明向市场释放了明确信号尽管当前业界对顶级AI训练芯片的研发门槛存在疑虑，但Meta仍坚定地将“自研训练芯片”视为其基础设施转型的终极目标。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">然而，通往算力自主化的道路并非坦途。近期市场有消息称，Meta在研发最前沿训练芯片的过程中遭遇了一定的技术瓶颈，甚至有传闻指出其部分高性能项目可能面临节奏调整。为了平衡即时的高性能算力缺口与长期自研目标，Meta目前采取了灵活的多元化供应策略。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">一方面，Meta被爆出已与谷歌达成协议，通过租用谷歌的TPU资源来加速现阶段大模型的开发进程;另一方面，公司依然保持与英伟达的深度采购关系。苏珊·利所强调的“随着时间推移逐步扩展”，暗示了Meta将采取稳扎稳打的过渡模式，即先在特定定制化任务中取得突破，再最终攻克通用大模型训练的算力高地。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">从行业视角来看，Meta的造芯始末反映了超大规模云厂商在AI时代的共同逻辑——全栈自研。通过将芯片架构与Llama等自有模型深度耦合，Meta不仅有望在长期运营中大幅削减硬件采购与能耗成本，更能避免在供应链波动中受制于人。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">虽然从推荐系统的推理跨越到复杂模型的训练面临着巨大的架构挑战，但Meta凭借其庞大的应用场景和充沛的现金流，正试图重新定义互联网巨头与硬件供应商之间的权力天平。\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">最专业的美股资讯推荐美股大数据 \u003Ca href=\"https:\u002F\u002FStockwe.com\u002F\">https:\u002F\u002FStockwe.com\u002F\u003C\u002Fa>\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp>\u003Cspan style=\"font-size: large;\">如何识别美股市场异常波动？美国机构主力资金买卖情况，出货和吸筹，使用美股投资网VIP会员，2008年成立于美国硅谷，由前纽约证券交易所分析师Ken创立，联合多位摩根斯坦利分析师，谷歌 Meta工程师利用AI和大数据，配合十多年美股实战经验和业内量化模型，建立了一个股市数据库 \u003Ca href=\"https:\u002F\u002FStockWe.com\u002F\">https:\u002F\u002FStockWe.com\u002F\u003C\u002Fa> 每天处理千万级股票数据：捕捉期权大单，实时主力资金流向、机构持仓变化、川普突发新闻，精准交易信号第一时间发到您手机APP！\u003C\u002Fspan>\u003C\u002Fp>\r\n\u003Cp> \u003C\u002Fp>","https:\u002F\u002Fwww.tradesmax.com\u002Fimages\u002Fa_Stock\u002FM\u002FMETA\u002FMETA.jpg","2026-03-05T19:52:06","2026.03.05","2026\u002F03\u002F05",33997,[22],"META","Article",0,"免费","success",{"id":9,"fullTitle":10,"subTitle":6,"url":11,"columnId":12,"columnName":13,"columnUrl":14,"summary":6,"contentHtml":15,"mainContentHtml":6,"posterUrl":16,"createDate":17,"displayDate":18,"displayDateSlash":19,"pageviews":20,"tags":28,"hidden":5,"isSubContent":5,"replyDocOrTargetId":6,"contentType":23,"videoId":6,"liveVideoUrl":6,"duration":24,"price":24,"priceText":25,"priceBadgeText":25,"priceBadgeClass":26,"freeForMinGroupLevel":24,"redirectUrl":6,"readyToStream":5},[22],"\u002Fcol\u002Fstocknews",{"visible":5,"marketingHtml":31,"services":32,"recentDocuments":41},"\u003Cfigure class=\"image\">\u003Ca href=\"https:\u002F\u002Fstockwe.com\u002Fdoc\u002Fdcio537efad5\" target=\"_blank\" rel=\"noopener noreferrer\">\u003Cimg style=\"display:block;margin-left:auto;margin-right:auto;\" src=\"https:\u002F\u002Fstockwewebfiles.blob.core.windows.net\u002Fweb-202408-stk\u002F1586109431mceclip0.jpg\">\u003C\u002Fa>\u003C\u002Ffigure>\u003Cdiv class=\"text-center\">\u003Ch2 class=\"card-title mx-auto\">\u003Cbr>\u003Ca target=\"_blank\" rel=\"noopener noreferrer\" href=\"https:\u002F\u002Fstockwe.com\u002Fdoc\u002Fdcio537efad5\">案例介绍：英伟达深度研究报告\u003C\u002Fa>\u003C\u002Fh2>\u003C\u002Fdiv>",[33,37],{"productId":34,"serviceName":35,"priceText":36},"prod_PPxdDdK87QaiLv","月付","$12.95美元",{"productId":38,"serviceName":39,"priceText":40},"prod_PPxeMs3bix1da5","年付","$149.00美元",[],{"links":43,"images":68,"summaryHtml":73,"aboutTitle":74,"aboutHtml":75,"copyrightHtml":76},[44,47,50,53,56,59,62,65],{"label":45,"url":46},"深度报告","\u002Fcol\u002FdepthReport",{"label":48,"url":49},"VIP会员","\u002Fplan",{"label":51,"url":52},"期权推荐","\u002FOption",{"label":54,"url":55},"低价暴涨股","\u002FPenny",{"label":57,"url":58},"常见问题","https:\u002F\u002Fstockwe.com\u002FFAQ",{"label":60,"url":61},"美股课程","\u002Fcol\u002Fvideos",{"label":63,"url":64},"免责声明","\u002Fdisclaimer",{"label":66,"url":67},"联系我们","\u002FContactUs",[69,70,71,72],"https:\u002F\u002Fstockwebsiteblob.blob.core.windows.net\u002Fweb-202509-stk\u002FUploaderzic2tuwsol2_2025_09_11_18_21_07.gif","https:\u002F\u002Fstockwebsiteblob.blob.core.windows.net\u002Fweb-202509-stk\u002FUploadercakzdvydksw_2025_09_03_09_00_56.png","https:\u002F\u002Fstockwebsiteblob.blob.core.windows.net\u002Fweb-202509-stk\u002FUploadergtjyagwvoyk_2025_09_14_08_32_05.png","https:\u002F\u002Fstockwebsiteblob.blob.core.windows.net\u002Fweb-202509-stk\u002FUploader3u0tt4jhlqh_2025_09_23_22_30_48.png","邮箱: buy@TradesMax.com 美国电话 626-378-3637","公司介绍","\u003Cp class=\"MsoNormal\">美股大数据 StockWe.com 是一个美国领先的金融和美股信息大数据提供商，紧盯华尔街金融市场和行情，2008年成立于美国硅谷，创始人是前纽约证券交易所资深分析师Ken，联合多位摩根斯坦利分析师，谷歌 Meta工程师利用AI和大数据，配合十多年美股实战经验和业内量化交易模型，每天处理千万级股票数据：挖掘潜力大牛股，捕捉期权异动大单，实时主力资金流向、机构持仓变化、川普突发新闻，精准买卖信号第一时间发到您手机APP。\u003C\u002Fp>","专业美股投资者都在这里",{"loading":78,"search":79,"searchPlaceholder":79,"hotContent":80,"draft":81,"noData":82,"searchNoData":83,"edit":84,"editVideo":85,"courseContent":86,"more":87,"buyNow":88,"subscribeNow":89,"encoding":90,"paidContent":91},"Loading...","搜索","热门内容","草稿","目前没有任何内容公布","当前检索内容没有数据","编辑","编辑视频","课程内容","更多","立即购买后观看","- 立即订阅 -","视频编码中...","付费内容"]