<p>尽管 <strong>微软</strong> 最近在数据中心建设方面放弃了一些租约，但 <strong>亚马逊</strong>(AMZN 2.01%) 和 <strong>谷歌</strong>(GOOGL 2.79%)(GOOG 2.56%) 似乎准备全速前进。</p>
<p>微软仍计划在本财年内在人工智能（AI）基础设施资本支出（capex）上花费约 800 亿美元，但其财年将在 6 月结束，距离现在只有几个月。然而，它正在暂停一些早期项目，显然是因为它的需求与其人工智能合作伙伴 OpenAI 的需求正在朝不同方向发展。就 OpenAI 而言，它希望扩展自己的能力；这也是 “星际计划” 的一部分，该计划计划在未来几年内在人工智能数据中心上花费 5000 亿美元。</p>
<p>然而，亚马逊和谷歌都计划在 2025 年大举投资。谷歌最近重申，今年将在数据中心资本支出上花费 750 亿美元，而谷歌计划花费约 1000 亿美元。关税的潜在影响并没有改变他们的计划。</p>
<p>在本月给股东的信中，亚马逊首席执行官安迪·贾西称人工智能是 “我们所知的一切的千载难逢的重塑”，并表示它 “发展的速度比几乎任何技术都要快”。</p>
<p>与此同时，在最近于拉斯维加斯举行的谷歌云下一代'25 大会上，谷歌首席执行官桑达尔·皮查伊表示 “人工智能的机会是前所未有的巨大”。</p>
<h2>数据中心支出</h2>
<p>历史表明，亚马逊和谷歌的支出将会得到回报。亚马逊在资本支出上有着悠久的历史，以建立其业务。它从零开始建立了整个仓储和物流网络，以加快所售商品的交付。这虽然代价不菲，但帮助公司成为今天的电子商务巨头。</p>
<p>随后，它在云计算方面也采取了同样的做法，基本上创造了基础设施即服务行业，推出了亚马逊网络服务（AWS），这现在是其最盈利的业务。许多分析师最初质疑公司在建设 AWS 方面的支出计划，并怀疑它是否会成为一个盈利的业务。</p>
<p>谷歌也在其谷歌云业务的建设上进行了大量前期投资，并经历了初期的亏损。然而，这项工作的成果在上个季度开始显现，当时谷歌云部门达到了盈利的拐点，运营收入飙升 142%，达到 21 亿美元。</p>
<p>早在 2017 年，<strong>高盛</strong> 的分析师就认识到亚马逊 “加速投资周期与收入再加速之间的历史关系”。他们还指出，亚马逊的股票在这些密集投资周期后表现优于大盘。</p>
<p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MTQ1MzYvZ2V0dHlpbWFnZXMtMTgwOTY2MDE0OC5qcGcmdz03MDA?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MTQ1MzYvZ2V0dHlpbWFnZXMtMTgwOTY2MDE0OC5qcGcmdz03MDA"/></p>
<p>图片来源：Getty Images。</p>
<p>在给股东的信中，亚马逊指出，数据中心投资具有吸引人的自由现金流（FCF）和投资资本回报率（ROIC）特征，这些资产的使用寿命为 15 到 20 年或更长。它还预测，随着更多芯片选项在 <strong>英伟达</strong> 之外变得可用，人工智能基础设施的定价将会下降。亚马逊还预计，推理将成为未来人工智能成本的最大驱动因素，而不是今天的模型训练。</p>
<p>随着推理预计将变得越来越重要，亚马逊和谷歌都开发了专门为推理设计的自定义人工智能芯片。亚马逊表示，其新款 Trainium2 芯片的性价比比当前的图形处理单元（GPU）提高了 30% 到 40%。其最大的目标之一是降低客户的推理成本，亚马逊相信这最终将导致整体人工智能支出的增加。</p>
<p>与此同时，谷歌刚刚推出了其第七代人工智能芯片 Ironwood。它表示，这款新芯片是为 “推理时代” 设计的，具有更强的计算能力和内存容量。这是谷歌首款专门为推理设计的芯片，旨在处理 “提供主动生成洞察和解释” 的模型。这也是其迄今为止最节能的芯片。</p>
<h2>购买股票的时机</h2>
<p>亚马逊和谷歌正在大力投资人工智能，从长远来看，这些投资应该会得到回报，尤其是在微软可能放缓支出的情况下。对云计算和人工智能服务的需求正在推动强劲增长，因为这些公司帮助客户创建自己的人工智能模型和应用程序，并在其平台上运行人工智能工作负载。</p>
<p>这两家公司也一直处于开发自定义人工智能芯片的前沿，以帮助降低人工智能基础设施成本。随着人工智能越来越向推理转移，这两家公司通过开发功耗更低、专门处理这些任务的芯片来降低整体成本。</p>
<p>人工智能也正在渗透到他们的其他业务中。亚马逊一直在利用人工智能提高其物流和仓储操作的效率，并为客户提供更好的产品推荐。谷歌在其最新的 Gemini 2.5 模型上取得了重大进展，迅速在人工智能竞赛中迎头赶上；这将有助于其搜索和广告业务，以及一些突破性的人工智能工具，例如其 Veo 2 文本转视频生成器。</p>
<p>随着最近市场的抛售，这两只股票的交易估值都很有吸引力。如果历史有任何指示，这两者都将是长期的人工智能赢家，使它们成为可靠的长期投资。</p>

谷歌-A

谷歌-C

亚马逊

<p>亚马逊和谷歌正在大力投资人工智能，亚马逊计划在数据中心资本支出上花费 1000 亿美元，而谷歌则计划花费 750 亿美元。这两家公司在投资方面有着成功的历史，带来了盈利，正如亚马逊网络服务和谷歌云所示。它们还在开发定制的人工智能芯片，以降低成本并提高效率。分析师认为，这些投资将会带来回报，尤其是在云计算和人工智能服务需求增长的背景下。随着微软减缓支出，现在可能是购买亚马逊和谷歌股票的好时机</p>

<p>While <strong>Microsoft</strong> has backed off some leases recently with its data center buildout, both <strong>Amazon</strong>(AMZN 2.01%) and <strong>Alphabet</strong>(GOOGL 2.79%)(GOOG 2.56%) look prepared to go full steam ahead.</p>
<div class="lb-trans"><p>尽管 <strong>微软</strong> 最近在数据中心建设方面放弃了一些租约，但 <strong>亚马逊</strong>(AMZN 2.01%) 和 <strong>谷歌</strong>(GOOGL 2.79%)(GOOG 2.56%) 似乎准备全速前进。</p>
</div><p>Microsoft still plans to spend around $80 billion on infrastructure capital expenditures (capex) for artificial intelligence (AI) this fiscal year, but its fiscal year ends in June, only a couple of months from now. However, it's pausing some early-stage projects, apparently because its needs and those of its AI partner OpenAI are moving in different directions. For its part, OpenAI is looking to build out its own capacity; it's part of Project Stargate, which plans to spend $500 billion on AI data centers over the next few years.</p>
<div class="lb-trans"><p>微软仍计划在本财年内在人工智能（AI）基础设施资本支出（capex）上花费约 800 亿美元，但其财年将在 6 月结束，距离现在只有几个月。然而，它正在暂停一些早期项目，显然是因为它的需求与其人工智能合作伙伴 OpenAI 的需求正在朝不同方向发展。就 OpenAI 而言，它希望扩展自己的能力；这也是 “星际计划” 的一部分，该计划计划在未来几年内在人工智能数据中心上花费 5000 亿美元。</p>
</div><p>However, Amazon and Alphabet both plan to spend big in 2025. Alphabet recently reiterated that it would spend $75 billion in data center capex this year, while Alphabet plans to spend around $100 billion. The potential impact of tariffs is not changing their plans.</p>
<div class="lb-trans"><p>然而，亚马逊和谷歌都计划在 2025 年大举投资。谷歌最近重申，今年将在数据中心资本支出上花费 750 亿美元，而谷歌计划花费约 1000 亿美元。关税的潜在影响并没有改变他们的计划。</p>
</div><p>In a letter to shareholders this month, Amazon CEO Andy Jassy called AI &#34;a once-in-a-lifetime reinvention of everything we know,&#34; and said that that it's &#34;moving faster than almost anything technology has ever seen.&#34;</p>
<div class="lb-trans"><p>在本月给股东的信中，亚马逊首席执行官安迪·贾西称人工智能是 “我们所知的一切的千载难逢的重塑”，并表示它 “发展的速度比几乎任何技术都要快”。</p>
</div><p>Meanwhile, at the recent Google Cloud Next '25 conference in Las Vegas, Alphabet CEO Sundar Pichai said &#34;the opportunity with AI is as big as it gets.&#34;</p>
<div class="lb-trans"><p>与此同时，在最近于拉斯维加斯举行的谷歌云下一代'25 大会上，谷歌首席执行官桑达尔·皮查伊表示 “人工智能的机会是前所未有的巨大”。</p>
</div><h2>Data center spending</h2>
<div class="lb-trans"><h2>数据中心支出</h2>
</div><p>History suggests that Amazon and Alphabet's expenditures will pay off. Amazon has a long history of spending big on capex to build its business. It built an entire warehousing and logistics network from scratch in order to speed up delivery of the goods it sold. This was pricey, but helped turn the company into the e-commerce behemoth it is today.</p>
<div class="lb-trans"><p>历史表明，亚马逊和谷歌的支出将会得到回报。亚马逊在资本支出上有着悠久的历史，以建立其业务。它从零开始建立了整个仓储和物流网络，以加快所售商品的交付。这虽然代价不菲，但帮助公司成为今天的电子商务巨头。</p>
</div><p>It then turned around and did the same thing with cloud computing, basically inventing the infrastructure-as-a-service industry with Amazon Web Services (AWS), which is now its most profitable business. Many analysts initially questioned the company's spending plans for building out AWS and doubted it would become a profitable business.</p>
<div class="lb-trans"><p>随后，它在云计算方面也采取了同样的做法，基本上创造了基础设施即服务行业，推出了亚马逊网络服务（AWS），这现在是其最盈利的业务。许多分析师最初质疑公司在建设 AWS 方面的支出计划，并怀疑它是否会成为一个盈利的业务。</p>
</div><p>Alphabet also built out its Google Cloud business spending with a lot in up-front costs, and endured initial losses. However, the fruits of this labor began to shine through last quarter when the Google Cloud segment hit a profitability inflection point, with operating income soaring 142% to $2.1 billion.</p>
<div class="lb-trans"><p>谷歌也在其谷歌云业务的建设上进行了大量前期投资，并经历了初期的亏损。然而，这项工作的成果在上个季度开始显现，当时谷歌云部门达到了盈利的拐点，运营收入飙升 142%，达到 21 亿美元。</p>
</div><p>Back in 2017, analysts at <strong>Goldman Sachs</strong> recognized a &#34;historical relationship between accelerated investment periods and revenue reacceleration&#34; at Amazon. They also noted that Amazon's stock outperformed following these cycles of intensive investment.</p>
<div class="lb-trans"><p>早在 2017 年，<strong>高盛</strong> 的分析师就认识到亚马逊 “加速投资周期与收入再加速之间的历史关系”。他们还指出，亚马逊的股票在这些密集投资周期后表现优于大盘。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MTQ1MzYvZ2V0dHlpbWFnZXMtMTgwOTY2MDE0OC5qcGcmdz03MDA" alt="" original-src="https://g.foolcdn.com/image/?url=https%3A%2F%2Fg.foolcdn.com%2Feditorial%2Fimages%2F814536%2Fgettyimages-1809660148.jpg&amp;op=resize&amp;w=700"/></p>
<p>Image source: Getty Images.</p>
<div class="lb-trans"><p>图片来源：Getty Images。</p>
</div><p>In its letter to shareholders, Amazon noted that data center investments have attractive free cash flow (FCF) and return on invested capital (ROIC) profiles, and that these assets have useful lives of 15 to 20 years or more. It also predicted that AI infrastructure pricing will come down, especially as more chip options become available outside of <strong>Nvidia</strong>. Amazon also expects inference to become the biggest driver of AI costs in the future, compared to model training today.</p>
<div class="lb-trans"><p>在给股东的信中，亚马逊指出，数据中心投资具有吸引人的自由现金流（FCF）和投资资本回报率（ROIC）特征，这些资产的使用寿命为 15 到 20 年或更长。它还预测，随着更多芯片选项在 <strong>英伟达</strong> 之外变得可用，人工智能基础设施的定价将会下降。亚马逊还预计，推理将成为未来人工智能成本的最大驱动因素，而不是今天的模型训练。</p>
</div><p>With inference projected to become increasingly important, both Amazon and Alphabet have developed their own custom AI chips designed specifically for inference. Amazon said its new Trainium2 chip has a 30% to 40% better price-to-performance ratio than current graphic processing units (GPUs). One of its biggest goals is making inference less expensive for customers, which it believes will ultimately lead to more overall AI spending.</p>
<div class="lb-trans"><p>随着推理预计将变得越来越重要，亚马逊和谷歌都开发了专门为推理设计的自定义人工智能芯片。亚马逊表示，其新款 Trainium2 芯片的性价比比当前的图形处理单元（GPU）提高了 30% 到 40%。其最大的目标之一是降低客户的推理成本，亚马逊相信这最终将导致整体人工智能支出的增加。</p>
</div><p>Meanwhile, Alphabet just introduced its seventh-generation AI chip, Ironwood. It said the new chip has been designed for the &#34;age of inference,&#34; with increased computation power and memory capacity. This is Alphabet's first chip designed specifically for inference, and was created to handle models that &#34;provide the proactive generation of insights and interpretation.&#34; It's also its most energy-efficient chip to date.</p>
<div class="lb-trans"><p>与此同时，谷歌刚刚推出了其第七代人工智能芯片 Ironwood。它表示，这款新芯片是为 “推理时代” 设计的，具有更强的计算能力和内存容量。这是谷歌首款专门为推理设计的芯片，旨在处理 “提供主动生成洞察和解释” 的模型。这也是其迄今为止最节能的芯片。</p>
</div><h2>Time to buy the stocks</h2>
<div class="lb-trans"><h2>购买股票的时机</h2>
</div><p>Amazon and Alphabet are investing heavily in AI, and over the long run these investments should pay off, especially with Microsoft potentially slowing its spending. Demand for cloud computing and AI services is driving strong growth, as these companies help customers create their own AI models and apps and run AI workloads on their platforms.</p>
<div class="lb-trans"><p>亚马逊和谷歌正在大力投资人工智能，从长远来看，这些投资应该会得到回报，尤其是在微软可能放缓支出的情况下。对云计算和人工智能服务的需求正在推动强劲增长，因为这些公司帮助客户创建自己的人工智能模型和应用程序，并在其平台上运行人工智能工作负载。</p>
</div><p>Both companies have also been at the forefront of developing custom AI chips to help lower AI infrastructure costs. As AI moves more toward inference, both companies are lowering the overall cost by developing chips that consume less power and are designed specifically to handle these tasks.</p>
<div class="lb-trans"><p>这两家公司也一直处于开发自定义人工智能芯片的前沿，以帮助降低人工智能基础设施成本。随着人工智能越来越向推理转移，这两家公司通过开发功耗更低、专门处理这些任务的芯片来降低整体成本。</p>
</div><p>AI is also permeating the rest of their businesses. Amazon has been using AI to become more efficient in its logistics and warehouse operations, and to make better product recommendations to its customers. Alphabet has made great strides with its newest Gemini 2.5 model, quickly catching up in the AI race; this should help its search and ad businesses, as should some groundbreaking AI tools, such as its Veo 2 text-to-video generator.</p>
<div class="lb-trans"><p>人工智能也正在渗透到他们的其他业务中。亚马逊一直在利用人工智能提高其物流和仓储操作的效率，并为客户提供更好的产品推荐。谷歌在其最新的 Gemini 2.5 模型上取得了重大进展，迅速在人工智能竞赛中迎头赶上；这将有助于其搜索和广告业务，以及一些突破性的人工智能工具，例如其 Veo 2 文本转视频生成器。</p>
</div><p>With the recent market sell-off, both stocks are trading at attractive valuations. If history is any indication, both will be long-term AI winners, making them both solid long-term investments.</p>
<div class="lb-trans"><p>随着最近市场的抛售，这两只股票的交易估值都很有吸引力。如果历史有任何指示，这两者都将是长期的人工智能赢家，使它们成为可靠的长期投资。</p>
</div>

Amazon and Alphabet Bet Big on AI. Why History Says It's Time to Buy Both Stocks

- 亚马逊和谷歌计划在 2025 年大幅增加数据中心投资。  
- 亚马逊预计支出 75 亿美元，谷歌计划支出约 1000 亿美元。  
- 这两家公司正在开发定制 AI 芯片，以降低基础设施成本。  

亚马逊和谷歌-A 在人工智能上大举押注。历史表明现在是购买这两只股票的好时机