<p><strong>英伟达</strong>(NVDA -0.12%) 被认为是人工智能 (AI) 硬件市场的先锋，这一点是有道理的，因为这家芯片设计公司的图形处理单元 (GPU) 使云计算公司和其他公司能够训练 AI 模型并运行推理应用程序。</p>
<p>英伟达 GPU 的并行计算能力使其非常适合同时执行大量计算，这正是训练 AI 模型所需的。此外，由于这些芯片能够快速使用训练好的模型进行预测和决策，它们在 AI 推理中的应用也正在获得关注。</p>
<p>毫不奇怪，英伟达在 AI 芯片市场上已经建立了稳固的立足点。它在 AI 数据中心加速器市场的估计市场份额高达 80%，远超其竞争对手。然而，英伟达的主导地位离不开其代工合作伙伴 <strong>台积电</strong>(TSM 1.60%)，后者才是真正的 AI 芯片市场的关键角色。</p>
<p>让我们来看看为什么台积电在 AI 芯片领域的影响力大于英伟达。</p>
<p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MjYyNTEvYXJ0aWZpY2lhbC1pbnRlbGxpZ2VuY2UtY2VudHJhbC1wcm9jZXNzaW5nLXVuaXQuanBnJnc9NzAw?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" width="700" height="393" original-src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MjYyNTEvYXJ0aWZpY2lhbC1pbnRlbGxpZ2VuY2UtY2VudHJhbC1wcm9jZXNzaW5nLXVuaXQuanBnJnc9NzAw"/></p>
<p>图片来源：Getty Images。</p>
<h2>台积电的主导代工地位使其成为 AI 芯片的首选制造商</h2>
<p>台积电在全球范围内运营半导体制造厂，用于根据客户提供的设计制造芯片。值得注意的是，台积电并不设计自己的芯片。它只是为那些没有生产设施的无厂半导体公司制造芯片。</p>
<p>英伟达就是一家利用台积电设施制造其 AI 芯片的公司。股权研究和经纪公司 Bernstein 估计，英伟达今年可能占台积电营收的五分之一，较几年前的 5% 到 10% 有显著增长。这并不令人惊讶，因为英伟达一直在积极寻求确保更多的台积电芯片制造能力。</p>
<p>总部位于台湾的商业报纸《经济日报》今年早些时候指出，英伟达据报道已确保在 2025 年获得超过 70% 的台积电先进芯片封装能力，以满足其 AI GPU 的强劲需求。然而，英伟达并不是唯一一家排队利用台积电工厂的公司。</p>
<p><strong>苹果</strong>(AAPL 0.07%) 是另一家主要客户，预计到 2025 年其对台积电营收的贡献将与英伟达相当。这家消费电子巨头利用台积电制造用于 iPhone 和 iPad 等热门设备的处理器，并据报道已预订了代工巨头的 2 纳米 (nm) 产能，以大规模生产下一代 iPhone 的芯片。</p>
<p>值得注意的是，苹果据报道已预订了 2023 年台积电的所有 3nm 供应，以制造 iPhone 和其他设备的处理器。现在，苹果希望增强其设备的本地 AI 能力，预计将转向 2nm 节点，以便能够提供更多的计算能力并提高能效。</p>
<p>然而，苹果并不是唯一的公司，另一家智能手机芯片设计公司 <strong>高通</strong> 也预计将基于台积电的 2nm 工艺节点生产芯片。另一方面，英伟达在 AI 加速器市场的同行也在与台积电合作制造先进芯片。</p>
<p>例如，<strong>Marvell Technology</strong> 据报道将采用台积电的亚 3nm 工艺节点来制造其定制 AI 处理器的下一代产品，这些产品在云计算巨头中需求旺盛，以降低成本。同时，<strong>AMD</strong> 也在通过台积电制造其为服务器和个人电脑 (PC) 提供动力的中央处理单元 (CPU) 和 GPU。</p>
<p>显然，台积电在 AI 芯片市场中是一个强有力的参与者。其工厂制造的芯片广泛应用于智能手机、个人电脑和数据中心等多种应用，而所有这些市场都因 AI 的发展而有望实现持续增长。重要的是，台积电正在采取措施确保能够满足所有这些市场的巨大需求。</p>
<h2>积极的扩张计划将帮助其满足蓬勃发展的 AI 芯片需求</h2>
<p>台积电 2025 年的资本支出预测为 380 亿至 420 亿美元，较 2024 年的 300 亿美元支出大幅增加。它将把 2025 年资本支出的 70% 投资于用于制造 AI 芯片的先进工艺技术，这并不令人惊讶。</p>
<p>此外，该公司还有积极的长期扩张计划。它已规划在美国投资 1650 亿美元以建设更多工厂，同时还在台湾和欧洲建设工厂。这些扩张举措将使台积电能够利用 AI 芯片市场的令人印象深刻的长期增长。</p>
<p>根据一项估计，全球 AI 芯片市场的年增长率可能在 2033 年之前达到 31%，这意味着台积电有能力在未来几年内维持其出色的增长。毫不奇怪，分析师预计台积电的增长将会加速。</p>
<p><img src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/4a338ae742672367def978bbfe3e4ece.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" width="720" height="380" original-src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/4a338ae742672367def978bbfe3e4ece.png"/></p>
<p>TSM 当前财年的每股收益预测数据来自 YCharts</p>
<p>这就是为什么现在大量购买这只 AI 股票是个好主意，因为它似乎被低估了。台积电的收益预计今年将增长 34%，几乎是 <strong>标准普尔 500</strong> 指数平均收益预期增长的五倍。考虑到该股的市盈率为 28 倍，投资者在台积电的潜在上涨空间上获得了良好的交易。</p>

英伟达

台积电

半导体 3 倍做多 - Direxion

<p>英伟达在人工智能硬件市场中处于领先地位，拥有 80% 的人工智能数据中心加速器市场份额。然而，台积电是关键参与者，为英伟达以及苹果和高通等其他公司制造芯片。台积电的主导地位得益于其全球制造工厂和大量资本投资，包括预计在 2025 年投入 380 亿至 420 亿美元以满足日益增长的人工智能芯片需求。全球人工智能芯片市场预计到 2033 年将以 31% 的年增长率增长，这为台积电的持续增长奠定了基础</p>

<p><strong>Nvidia</strong>(NVDA -0.12%) is considered a pioneer in the artificial intelligence (AI) hardware market, and rightly so, as the chip designer's graphics processing units (GPUs) have allowed cloud computing companies and others to train AI models and run inference applications.</p>
<div class="lb-trans"><p><strong>英伟达</strong>(NVDA -0.12%) 被认为是人工智能 (AI) 硬件市场的先锋，这一点是有道理的，因为这家芯片设计公司的图形处理单元 (GPU) 使云计算公司和其他公司能够训练 AI 模型并运行推理应用程序。</p>
</div><p>The parallel computing power of Nvidia's GPUs makes them ideal for performing a large number of calculations simultaneously, which is precisely what's required for training AI models. Also, these chips are now gaining traction in AI inference as well, thanks to their ability to quickly make predictions and decisions using the trained model.</p>
<div class="lb-trans"><p>英伟达 GPU 的并行计算能力使其非常适合同时执行大量计算，这正是训练 AI 模型所需的。此外，由于这些芯片能够快速使用训练好的模型进行预测和决策，它们在 AI 推理中的应用也正在获得关注。</p>
</div><p>Not surprisingly, Nvidia has established a solid foothold in the AI chip market. It towers above its competitors with an estimated market share of 80% in AI data center accelerators. However, Nvidia's dominance wouldn't have been possible without its foundry partner <strong>Taiwan Semiconductor Manufacturing</strong>(TSM 1.60%), which is the real kingpin of the AI chip market.</p>
<div class="lb-trans"><p>毫不奇怪，英伟达在 AI 芯片市场上已经建立了稳固的立足点。它在 AI 数据中心加速器市场的估计市场份额高达 80%，远超其竞争对手。然而，英伟达的主导地位离不开其代工合作伙伴 <strong>台积电</strong>(TSM 1.60%)，后者才是真正的 AI 芯片市场的关键角色。</p>
</div><p>Let's look at the reasons why TSMC is a bigger power player than Nvidia in AI chips.</p>
<div class="lb-trans"><p>让我们来看看为什么台积电在 AI 芯片领域的影响力大于英伟达。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2cuZm9vbGNkbi5jb20vZWRpdG9yaWFsL2ltYWdlcy84MjYyNTEvYXJ0aWZpY2lhbC1pbnRlbGxpZ2VuY2UtY2VudHJhbC1wcm9jZXNzaW5nLXVuaXQuanBnJnc9NzAw" alt="" original-src="https://g.foolcdn.com/image/?url=https%3A%2F%2Fg.foolcdn.com%2Feditorial%2Fimages%2F826251%2Fartificial-intelligence-central-processing-unit.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>TSMC's dominant foundry position makes it the go-to manufacturer of AI chips</h2>
<div class="lb-trans"><h2>台积电的主导代工地位使其成为 AI 芯片的首选制造商</h2>
</div><p>TSMC operates semiconductor fabrication plants across the globe, which are used to manufacture chips based on designs provided by its customers. It is worth noting that TSMC doesn't design its own chips. It simply makes chips for fabless semiconductor companies that don't have production facilities of their own.</p>
<div class="lb-trans"><p>台积电在全球范围内运营半导体制造厂，用于根据客户提供的设计制造芯片。值得注意的是，台积电并不设计自己的芯片。它只是为那些没有生产设施的无厂半导体公司制造芯片。</p>
</div><p>Nvidia is one such company that utilizes TSMC's facilities for manufacturing its AI chips. Equity research and brokerage firm Bernstein estimates that Nvidia could account for over a fifth of TSMC's top line this year, up significantly from around 5% to 10% a couple of years ago. That's not surprising, as Nvidia has been aggressively looking to secure more of TSMC's chipmaking capacity.</p>
<div class="lb-trans"><p>英伟达就是一家利用台积电设施制造其 AI 芯片的公司。股权研究和经纪公司 Bernstein 估计，英伟达今年可能占台积电营收的五分之一，较几年前的 5% 到 10% 有显著增长。这并不令人惊讶，因为英伟达一直在积极寻求确保更多的台积电芯片制造能力。</p>
</div><p>Taiwan-based business newspaper Economic Daily News pointed out earlier this year that Nvidia has reportedly secured more than 70% of TSMC's advanced chip packaging capacity for 2025 in a bid to meet the robust demand for its AI GPUs. However, Nvidia is not the only company that's in line to utilize TSMC's fabs.</p>
<div class="lb-trans"><p>总部位于台湾的商业报纸《经济日报》今年早些时候指出，英伟达据报道已确保在 2025 年获得超过 70% 的台积电先进芯片封装能力，以满足其 AI GPU 的强劲需求。然而，英伟达并不是唯一一家排队利用台积电工厂的公司。</p>
</div><p><strong>Apple</strong>(AAPL 0.07%) is another major customer, and its contribution toward TSMC's top line is expected to be identical to that of Nvidia's in 2025. The consumer electronics giant taps TSMC to manufacture the processors that go into popular devices such as iPhones and iPads, and it has reportedly pre-booked the foundry giant's 2-nanometer (nm) capacity to mass produce chips for its next-generation iPhones.</p>
<div class="lb-trans"><p><strong>苹果</strong>(AAPL 0.07%) 是另一家主要客户，预计到 2025 年其对台积电营收的贡献将与英伟达相当。这家消费电子巨头利用台积电制造用于 iPhone 和 iPad 等热门设备的处理器，并据报道已预订了代工巨头的 2 纳米 (nm) 产能，以大规模生产下一代 iPhone 的芯片。</p>
</div><p>It is worth noting that Apple had reportedly booked all of TSMC's 3nm supply in 2023 to make processors for the iPhone and other devices. And now that Apple is looking to bolster the on-device AI capabilities of its devices, it is expected to move to the 2nm node so that it can pack more computing power and increase energy efficiency.</p>
<div class="lb-trans"><p>值得注意的是，苹果据报道已预订了 2023 年台积电的所有 3nm 供应，以制造 iPhone 和其他设备的处理器。现在，苹果希望增强其设备的本地 AI 能力，预计将转向 2nm 节点，以便能够提供更多的计算能力并提高能效。</p>
</div><p>Apple, however, has company, as another smartphone chip designer -- <strong>Qualcomm</strong> -- is expected to produce chips based on TSMC's 2nm process node as well. On the other hand, Nvidia's peers in the AI accelerator market are also partnering with TSMC to manufacture advanced chips.</p>
<div class="lb-trans"><p>然而，苹果并不是唯一的公司，另一家智能手机芯片设计公司 <strong>高通</strong> 也预计将基于台积电的 2nm 工艺节点生产芯片。另一方面，英伟达在 AI 加速器市场的同行也在与台积电合作制造先进芯片。</p>
</div><p><strong>Marvell Technology</strong>, for instance, is reportedly going to adopt TSMC's sub-3nm process nodes to manufacture the next generation of its custom AI processors, which are in tremendous demand from cloud computing giants to reduce costs. Meanwhile, <strong>AMD</strong> is getting its central processing units (CPUs) and GPUs that power both servers and personal computers (PCs) manufactured by TSMC as well.</p>
<div class="lb-trans"><p>例如，<strong>Marvell Technology</strong> 据报道将采用台积电的亚 3nm 工艺节点来制造其定制 AI 处理器的下一代产品，这些产品在云计算巨头中需求旺盛，以降低成本。同时，<strong>AMD</strong> 也在通过台积电制造其为服务器和个人电脑 (PC) 提供动力的中央处理单元 (CPU) 和 GPU。</p>
</div><p>Clearly, TSMC is the power player in the AI chip market. Its plants manufacture chips that go into a wide variety of applications, ranging from smartphones to PCs to data centers, and all of these markets are on track to record secular growth because of AI. Importantly, TSMC is taking steps to ensure that it can meet the incredible demand from all of these markets.</p>
<div class="lb-trans"><p>显然，台积电在 AI 芯片市场中是一个强有力的参与者。其工厂制造的芯片广泛应用于智能手机、个人电脑和数据中心等多种应用，而所有这些市场都因 AI 的发展而有望实现持续增长。重要的是，台积电正在采取措施确保能够满足所有这些市场的巨大需求。</p>
</div><h2>An aggressive expansion plan should help it satisfy the booming AI chip demand</h2>
<div class="lb-trans"><h2>积极的扩张计划将帮助其满足蓬勃发展的 AI 芯片需求</h2>
</div><p>TSMC's 2025 capital expenditure forecast of $38 billion to $42 billion points toward a significant increase over its 2024 outlay of $30 billion. It is going to invest 70% of its 2025 capex on advanced process technologies that are used for making AI chips, which isn't surprising.</p>
<div class="lb-trans"><p>台积电 2025 年的资本支出预测为 380 亿至 420 亿美元，较 2024 年的 300 亿美元支出大幅增加。它将把 2025 年资本支出的 70% 投资于用于制造 AI 芯片的先进工艺技术，这并不令人惊讶。</p>
</div><p>Moreover, the company has aggressive long-term expansion plans as well. It has outlined an investment of $165 billion in the U.S. to build more plants, while it is also building factories in Taiwan and Europe. These expansionary moves should enable TSMC to capitalize on the AI chip market's impressive long-term growth.</p>
<div class="lb-trans"><p>此外，该公司还有积极的长期扩张计划。它已规划在美国投资 1650 亿美元以建设更多工厂，同时还在台湾和欧洲建设工厂。这些扩张举措将使台积电能够利用 AI 芯片市场的令人印象深刻的长期增长。</p>
</div><p>According to one estimate, the global AI chipset market could clock an annual growth rate of 31% through 2033, which means that TSMC has the ability to sustain its terrific growth for years to come. Not surprisingly, analysts are expecting a pick-up in TSMC's growth going forward.</p>
<div class="lb-trans"><p>根据一项估计，全球 AI 芯片市场的年增长率可能在 2033 年之前达到 31%，这意味着台积电有能力在未来几年内维持其出色的增长。毫不奇怪，分析师预计台积电的增长将会加速。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/4a338ae742672367def978bbfe3e4ece.png" alt="" original-src="https://media.ycharts.com/charts/4a338ae742672367def978bbfe3e4ece.png"/></p>
<p>TSM EPS Estimates for Current Fiscal Year data by YCharts</p>
<div class="lb-trans"><p>TSM 当前财年的每股收益预测数据来自 YCharts</p>
</div><p>That's why it would be a good idea to buy this AI stock hand over fist right now, as it seems undervalued. TSMC's earnings are expected to jump by 34% this year, which is nearly five times the projected increase in the <strong>S&amp;P 500</strong> index's average earnings. With the stock trading at 28 times earnings, investors are getting a good deal on TSMC based on the potential upside it could deliver.</p>
<div class="lb-trans"><p>这就是为什么现在大量购买这只 AI 股票是个好主意，因为它似乎被低估了。台积电的收益预计今年将增长 34%，几乎是 <strong>标准普尔 500</strong> 指数平均收益预期增长的五倍。考虑到该股的市盈率为 28 倍，投资者在台积电的潜在上涨空间上获得了良好的交易。</p>
</div>

Everyone's Watching Nvidia -- but This AI Supplier Is the Real Power Player

- Nvidia 在人工智能硬件市场占据 80% 的市场份额。  
- TSMC 是主要的芯片制造商，预计 2025 年将为 Nvidia 和 Apple 提供相同的收入贡献。  
- TSMC 计划在 2025 年投资 380 亿至 420 亿美元以满足 AI 芯片需求。  

虽然所有人都在关注 NVIDIA，但这个 AI 供应商才是真正的实力玩家