<p><strong>英伟达</strong>(NVDA 1.28%) 长期以来一直是人工智能 (AI) 基础设施市场的主导者。其图形处理单元 (GPU) 是用于支持 AI 工作负载的主要芯片，而其 CUDA 软件平台则帮助其建立了广泛的护城河。然而，如果有一家意外的公司能够挑战英伟达，那就是 <strong>谷歌</strong>(GOOGL 0.54%)(GOOG 0.51%)。</p>
<p>如今，大多数运行 AI 工作负载的公司都使用英伟达的芯片。然而，谷歌转向了自家的芯片，称为张量处理单元 (TPU)，以帮助降低成本并提高性能。</p>
<h2>谷歌的 TPU 机会</h2>
<p>谷歌在内部使用 TPU 处理 AI 工作负载，但它也将这些芯片租给外部开发者。最初有报道称，OpenAI 正在租用谷歌的 TPU 来帮助运行 ChatGPT，尽管该公司后来表示，它只是测试这些芯片，并不打算在此时大规模部署。然而，这仍然是一个重要的发展。OpenAI 是英伟达最大的客户之一，也是谷歌的竞争对手，OpenAI 正在考虑谷歌的 TPU，显示出其芯片所能提供的能力和成本优势。</p>
<p>TPU 之所以如此吸引人，是因为它们专门为谷歌的基础设施设计，具有延迟、可扩展性和成本方面的优势。随着市场开始向 AI 推理转变，谷歌的 TPU 可能会成为比 GPU 更好的选择。同时，谷歌最近在四月推出了一款专门为推理设计的 TPU，名为 Ironwood。</p>
<p>谷歌的垂直整合也为其提供了潜在的优势。它是少数能够提供定制 AI 芯片、顶级云计算平台和完整软件平台的公司之一。英伟达可能在 GPU 领域占据主导地位，但它并不提供云计算平台。这是谷歌可以利用的优势，尤其是在推理中，定价和延迟是更重要的考虑因素。</p>
<p>市场背景对谷歌也很有利。预计 AI 基础设施的支出将保持强劲，英伟达预测到 2028 年全球数据中心资本支出 (capex) 将超过 1 万亿美元。同时，许多专家预测，推理将成为比模型训练更大的市场。AI 推理在技术上远没有 AI 训练那么具挑战性，这有助于削弱英伟达的 CUDA 优势。这可能导致更多公司寻找能够提供比 GPU 更低推理成本的芯片。</p>
<p>也就是说，谷歌不太可能从英伟达那里夺取大量市场份额。然而，即使在推理工作负载中获得小幅市场份额，对谷歌云来说也可能是一个巨大的推动。这是因为 TPU 的采用将直接促进谷歌云的使用。谷歌云的收入增长迅速，上个季度收入同比增长 28%，达到 123 亿美元。更重要的是，部门盈利能力已达到拐点，部门营业收入飙升 142%，达到 22 亿美元。</p>
<p>TPU 不仅通过吸引新客户来提升谷歌云的利润，还通过降低成本来实现。而且这不仅仅是芯片的问题。谷歌还开发了全光开关，显著降低了 TPU 的功耗和成本。这使得公司拥有强大的运营杠杆，而市场目前并未充分认识到这一点。</p>
<p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MjM1NDMvZ2V0dHlpbWFnZXMtMjA3ODkxNzEwOS5qcGcmdz03MDA?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" width="700" height="411" original-src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MjM1NDMvZ2V0dHlpbWFnZXMtMjA3ODkxNzEwOS5qcGcmdz03MDA"/></p>
<p>图片来源：Getty Images。</p>
<h2>谷歌股票值得购买吗？</h2>
<p>目前，谷歌的交易价格低于其他 AI 领导者，基于当前年度分析师估计的前瞻市盈率 (P/E) 为 18 倍。投资者仍然主要将其视为一家搜索公司，其核心业务可能会受到 AI 的干扰，同时忽视了其在搜索中凭借分发和全球广告网络的覆盖所拥有的巨大优势。</p>
<p>他们也低估了公司强大的 AI 产品。Gemini 在独立测试中始终被评为顶级 AI 模型之一，其应用程序的下载量激增。它还拥有市场上最好的文本转视频生成器 Veo 3，并正在将 AI 功能融入其搜索产品中。更不用说该公司还拥有全球观看次数最多的流媒体平台 YouTube，其 Waymo 机器人出租车业务也在持续增长。</p>
<p>谷歌的 TPU 是另一个巨大的潜在增长驱动因素。英伟达的领先地位非常显著，但如果谷歌能够获得像 OpenAI 这样的主要客户并扩大 TPU 的采用，它有机会在未来成为英伟达在推理领域最大的竞争对手之一。如果那样的话，谷歌的股票应该会显著上涨。</p>

<p>英伟达一直是人工智能基础设施的领导者，但谷歌正在凭借其张量处理单元（TPUs）崭露头角，成为潜在竞争对手。谷歌将 TPUs 用于自身的人工智能工作负载，并将其租赁给开发者，包括 OpenAI，这可能预示着市场的变化。TPUs 在延迟、可扩展性和成本方面具有优势，尤其是在人工智能推理方面。谷歌的垂直整合和强大的云计算平台可能进一步增强其市场地位。虽然谷歌可能不会从英伟达那里获得显著的市场份额，但即使是小幅增长也可能提升谷歌云的盈利能力和股票价值</p>

<p><strong>Nvidia</strong>(NVDA 1.28%) has long been the dominant player in the artificial intelligence (AI) infrastructure market. Its graphics processing units (GPUs) are the main chips used to power AI workloads, while its CUDA software platform has helped create a wide moat. However, if there is a surprise company that can challenge Nvidia, it's <strong>Alphabet</strong>(GOOGL 0.54%)(GOOG 0.51%).</p>
<div class="lb-trans"><p><strong>英伟达</strong>(NVDA 1.28%) 长期以来一直是人工智能 (AI) 基础设施市场的主导者。其图形处理单元 (GPU) 是用于支持 AI 工作负载的主要芯片，而其 CUDA 软件平台则帮助其建立了广泛的护城河。然而，如果有一家意外的公司能够挑战英伟达，那就是 <strong>谷歌</strong>(GOOGL 0.54%)(GOOG 0.51%)。</p>
</div><p>Today, most companies running AI workloads use Nvidia's chips. However, Alphabet has turned to its own in-house chips called Tensor Processing Units (TPUs) to help reduce costs and improve performance.</p>
<div class="lb-trans"><p>如今，大多数运行 AI 工作负载的公司都使用英伟达的芯片。然而，谷歌转向了自家的芯片，称为张量处理单元 (TPU)，以帮助降低成本并提高性能。</p>
</div><h2>Alphabet's TPU opportunity</h2>
<div class="lb-trans"><h2>谷歌的 TPU 机会</h2>
</div><p>Alphabet uses TPUs internally for AI workloads, but it also rents the chips to outside developers. It was initially reported that OpenAI was renting Alphabet's TPUs to help run ChatGPT, although the company has since said it is just testing the chips and not looking to deploy them at scale at this time. Nonetheless, this is a big development. OpenAI is one of Nvidia's largest customers and a Google rival, and the fact that it is looking at Alphabet's TPUs demonstrate the capabilities and cost advantages its chips can provide.</p>
<div class="lb-trans"><p>谷歌在内部使用 TPU 处理 AI 工作负载，但它也将这些芯片租给外部开发者。最初有报道称，OpenAI 正在租用谷歌的 TPU 来帮助运行 ChatGPT，尽管该公司后来表示，它只是测试这些芯片，并不打算在此时大规模部署。然而，这仍然是一个重要的发展。OpenAI 是英伟达最大的客户之一，也是谷歌的竞争对手，OpenAI 正在考虑谷歌的 TPU，显示出其芯片所能提供的能力和成本优势。</p>
</div><p>What makes TPUs so compelling is that they are specifically designed for Google's infrastructure, giving them advantages in latency, scalability, and cost. As the market starts to shift toward AI inference, Alphabet's TPUs could become a better choice than GPUs. Meanwhile, Alphabet recently introduced a TPU specifically designed for inference called Ironwood in April.</p>
<div class="lb-trans"><p>TPU 之所以如此吸引人，是因为它们专门为谷歌的基础设施设计，具有延迟、可扩展性和成本方面的优势。随着市场开始向 AI 推理转变，谷歌的 TPU 可能会成为比 GPU 更好的选择。同时，谷歌最近在四月推出了一款专门为推理设计的 TPU，名为 Ironwood。</p>
</div><p>Alphabet's vertical integration also gives it a potential edge. It's one of the few companies that can offer custom AI chips, a top-tier cloud-computing platform, and a full software platform. Nvidia may dominate GPUs, but it doesn't offer a cloud-computing platform. This is something that Alphabet can take advantage of, especially with inference where pricing and latency are more important considerations.</p>
<div class="lb-trans"><p>谷歌的垂直整合也为其提供了潜在的优势。它是少数能够提供定制 AI 芯片、顶级云计算平台和完整软件平台的公司之一。英伟达可能在 GPU 领域占据主导地位，但它并不提供云计算平台。这是谷歌可以利用的优势，尤其是在推理中，定价和延迟是更重要的考虑因素。</p>
</div><p>The market backdrop is also favorable for Alphabet. Spending on AI infrastructure is expected to remain strong, with Nvidia projecting global data center capital expenditures (capex) to surpass $1 trillion by 2028. At the same time, many experts are predicting that inference will become a much larger market than model training. AI inference is much less technically challenging than AI training, which helps negate some of Nvidia's CUDA advantage. This could lead to more companies looking for chips that can provide cheaper costs per inference compared to GPUs.</p>
<div class="lb-trans"><p>市场背景对谷歌也很有利。预计 AI 基础设施的支出将保持强劲，英伟达预测到 2028 年全球数据中心资本支出 (capex) 将超过 1 万亿美元。同时，许多专家预测，推理将成为比模型训练更大的市场。AI 推理在技术上远没有 AI 训练那么具挑战性，这有助于削弱英伟达的 CUDA 优势。这可能导致更多公司寻找能够提供比 GPU 更低推理成本的芯片。</p>
</div><p>That said, it isn't likely that Alphabet is going to ever to take a huge amount of share away from Nvidia. However, even small share gains in inference workloads could be a big boost for Google Cloud. That's because TPU adoption would directly feed into Google Cloud usage as well. Google Cloud has been growing revenue quickly, with revenue climbing 28% year over year last quarter to $12.3 billion. More importantly, segment profitability has hit an inflection point, with segment operating income soaring 142% to $2.2 billion.</p>
<div class="lb-trans"><p>也就是说，谷歌不太可能从英伟达那里夺取大量市场份额。然而，即使在推理工作负载中获得小幅市场份额，对谷歌云来说也可能是一个巨大的推动。这是因为 TPU 的采用将直接促进谷歌云的使用。谷歌云的收入增长迅速，上个季度收入同比增长 28%，达到 123 亿美元。更重要的是，部门盈利能力已达到拐点，部门营业收入飙升 142%，达到 22 亿美元。</p>
</div><p>TPUs help boost Google Cloud's profits, not just by bringing in new customers but by lowering costs as well. And it doesn't stop just with chips. Alphabet has also developed all-optical switches that significantly reduce the power consumption and cost of TPUs. This gives the company strong operating leverage that the market currently isn't appreciating.</p>
<div class="lb-trans"><p>TPU 不仅通过吸引新客户来提升谷歌云的利润，还通过降低成本来实现。而且这不仅仅是芯片的问题。谷歌还开发了全光开关，显著降低了 TPU 的功耗和成本。这使得公司拥有强大的运营杠杆，而市场目前并未充分认识到这一点。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MjM1NDMvZ2V0dHlpbWFnZXMtMjA3ODkxNzEwOS5qcGcmdz03MDA" alt="" original-src="https://g.foolcdn.com/image/?url=https%3A%2F%2Fg.foolcdn.com%2Feditorial%2Fimages%2F823543%2Fgettyimages-2078917109.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><h2>Is Alphabet stock a buy?</h2>
<div class="lb-trans"><h2>谷歌股票值得购买吗？</h2>
</div><p>Right now, Alphabet trades at a discount to other AI leaders, with a forward price-to-earnings (P/E) ratio of 18 times based on current year analyst estimates. Investors still largely view it as a search company whose core business could be disrupted by AI, while ignoring the big advantages in search it has with its distribution and global ad network reach.</p>
<div class="lb-trans"><p>目前，谷歌的交易价格低于其他 AI 领导者，基于当前年度分析师估计的前瞻市盈率 (P/E) 为 18 倍。投资者仍然主要将其视为一家搜索公司，其核心业务可能会受到 AI 的干扰，同时忽视了其在搜索中凭借分发和全球广告网络的覆盖所拥有的巨大优势。</p>
</div><p>They also underestimate the company's strong AI offerings. Gemini is consistently ranked as one of the top AI models in independent tests, and its app has seen downloads surge. It also has the best text-to-video generator on the market with Veo 3 and is incorporating AI features into its search offering. And let's not forget the company also owns the world's most-viewed streaming platform in YouTube, and its Waymo robotaxi business has been seeing strong momentum.</p>
<div class="lb-trans"><p>他们也低估了公司强大的 AI 产品。Gemini 在独立测试中始终被评为顶级 AI 模型之一，其应用程序的下载量激增。它还拥有市场上最好的文本转视频生成器 Veo 3，并正在将 AI 功能融入其搜索产品中。更不用说该公司还拥有全球观看次数最多的流媒体平台 YouTube，其 Waymo 机器人出租车业务也在持续增长。</p>
</div><p>Alphabet's TPUs are another big potential growth driver. Nvidia's lead is massive, but if Alphabet can land a major customer like OpenAI and expand TPU adoption, it has a chance to become one of Nvidia's biggest competitors in inference in the future. And if that happens, Alphabet's stock should move meaningfully higher.</p>
<div class="lb-trans"><p>谷歌的 TPU 是另一个巨大的潜在增长驱动因素。英伟达的领先地位非常显著，但如果谷歌能够获得像 OpenAI 这样的主要客户并扩大 TPU 的采用，它有机会在未来成为英伟达在推理领域最大的竞争对手之一。如果那样的话，谷歌的股票应该会显著上涨。</p>
</div>

Could This Surprise Company Become Nvidia's Biggest Competitor?

- Nvidia 在人工智能基础设施市场占据主导地位，主要依赖其 GPU。  
- Alphabet 通过自家 TPU 芯片降低成本并提升性能，吸引 OpenAI 测试。  
- 随着 AI 推理市场的增长，Alphabet 有望在云计算和芯片领域获得优势。

Motley Fool

谷歌-A

谷歌-C

英伟达

这家出乎意料的公司是否有可能成为英伟达最大的竞争对手？