<p>多年来，<strong>高通</strong>（QCOM +3.65%）一直被视为一家高端智能手机芯片制造商，从移动革命中赚取了数十亿美元。然而，投资者可能正在见证一个新篇章的开始。</p>
<p>在昨天的投资者日上，该公司公布了雄心勃勃的计划，计划在短短三年内将其非手机销售额翻倍，达到 400 亿美元。虽然总收入增长似乎并不显著，但更重要的是收入构成的变化，这一变化由人工智能计算驱动，预示着多个大型市场的巨大机会。管理层预计到 2030 年，综合可寻址市场将达到 1.7 万亿美元。</p>
<p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2Nkbi5jb250ZW50LmZvb2xjZG4uY29tL2ltYWdlcy8xdW1uOXFlaC9wcm9kdWN0aW9uL2FjODAxMDVlM2VmM2M5Y2MwYzkwM2ZiNzZlNzQwYzg5N2Q3YmYwODItMTIwMHg4MDAuanBnP3c9MTIwMCZoPTgwMCZxPTc1JmF1dG89Zm9ybWF0Jnc9Mzg0MA?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" width="1200" height="800" original-src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2Nkbi5jb250ZW50LmZvb2xjZG4uY29tL2ltYWdlcy8xdW1uOXFlaC9wcm9kdWN0aW9uL2FjODAxMDVlM2VmM2M5Y2MwYzkwM2ZiNzZlNzQwYzg5N2Q3YmYwODItMTIwMHg4MDAuanBnP3c9MTIwMCZoPTgwMCZxPTc1JmF1dG89Zm9ybWF0Jnc9Mzg0MA"/></p>
<p>图片来源：Getty Images。</p>
<p>高通股票在盘前交易中上涨了 11%，尽管最初的兴奋似乎已经消退，但截至下午 1:15，股票仍上涨了近 8.5%。然而，更大的故事是，对于愿意在未来三到五年内持有该股票的散户投资者来说，隐藏的机会。</p>
<h2>没有人积极追求的机会</h2>
<p>高通在智能手机芯片制造方面的专业知识使其成为追求潜在爆炸性市场的理想候选者：边缘人工智能计算。</p>
<p>如今，大多数人工智能（AI）模型在云端运行，这解释了全球数据中心的爆炸性增长。然而，尚未大规模起飞的是在数百万设备上运行 AI 模型的能力，无论是个人还是企业拥有的设备。</p>
<p>本质上，高通的管理层预测，AI 模型在推理阶段的计算和能效将变得更加高效，推理阶段是模型根据已学习的内容生成输出的时刻。此外，将会有更精简、针对特定任务的 AI 模型为特定工作进行训练，这些模型可以驻留在设备本身上。</p>
<h2>为什么边缘 AI 很重要</h2>
<p>随着大型语言模型（LLMs）变得越来越复杂，能够执行更多通用任务，它们在小众、专业和隐私关注应用中的效用可能最终会受到限制。随着 AI 计算的普及，减少设备与通常位于远程服务器上的大型语言模型之间的延迟的需求不容小觑。需要即时照片处理的智能手机相机，或无法等待云端响应的自动驾驶汽车，都是延迟可能成为重大障碍的例子。</p>
<p>其他优势包括设备内处理的隐私性提高，以及希望将某些计算活动从昂贵的数据中心转移出去的企业的成本降低。</p>
<h2>向多个领域扩展</h2>
<p>高通预计到 2029 财年将产生 400 亿美元的非手机收入。截至 2026 财年第二季度（截至 3 月 29 日），在过去 12 个月的总销售额 440 亿美元中，非手机收入略超过 160 亿美元，其余 270 亿美元来自智能手机芯片业务。</p>
<p>为了实现 400 亿美元的年化目标，管理层需要在未来三年半内将非智能手机业务大幅提升 150%，其中边缘计算将贡献超过三分之一的收入。</p>
<p>高通进军数据中心可能每年带来 150 亿美元的收入。<strong>Meta Platforms</strong>（META 2.66%），Facebook 和 Instagram 的母公司，已同意使用高通的数据中心 CPU——Qualcomm Dragonfly™ C1000，为其下一代数据中心提供动力。该 CPU 预计将在 2028 年下半年投入生产。</p>
<h2>市场已经在定价不同的业务</h2>
<p>今天的反弹反映了市场对高通在智能手机之外的变革性 AI 雄心的信心。</p>
<p>智能手机业务是周期性的。人们通常不会每年升级手机，而是在现有手机变得过时时，通常在三到四年内进行升级。因此，传统智能手机芯片的需求正在减弱。不出所料，在过去五年中，股票仅上涨了 57%，显著落后于 <strong>纳斯达克 100</strong> 和 <strong>标准普尔 500</strong>。</p>
<p>以过去 12 个月的收益为基础，市盈率略高于 23 倍，销售额为 5 倍，对于一家不可避免地进入火热的 AI 基础设施业务的公司来说，股票看起来相当便宜。</p>
<p><img src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/7168ded6cad563835fcecf1d1ce0e368.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" width="720" height="456" original-src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/7168ded6cad563835fcecf1d1ce0e368.png"/></p>
<p>.</p>
<p>QCOM 市盈率数据来源于 YCharts。</p>
<h2>执行是关键</h2>
<p>虽然雄心壮志容易理解，但真正具有挑战性的是获得市场份额。超大规模企业正在争夺最佳可用芯片，价格最低。事实上，许多企业，如 <strong>Alphabet</strong> 和 <strong>Amazon</strong>，正在设计自己的芯片。</p>
<p>投资者必须问的真正问题是，高通能否证明自己能够实现其雄心。</p>
<p>从主要单一收入来源多元化，虽然最初令人害怕，但这是必要的。对于高通来说，未来十年的增长可能与过去大相径庭。</p>

高通

2 倍做多 QCOM ETF - GraniteShares

<p>高通公布计划到 2029 年将非手机销售额翻倍至 400 亿美元，推动因素包括边缘人工智能计算和数据中心扩展。该公司目标到 2030 年实现 1.7 万亿美元的可寻址市场，Meta 将采用其 Dragonfly CPU。投资者反应积极，股价上涨近 8.5%，尽管执行风险存在，他们仍将从周期性智能手机芯片转向人工智能基础设施视为一种变革性的增长机会</p>

<p>For years, <strong>Qualcomm</strong>(QCOM +3.65%) has been seen as a premium smartphone chipmaker that raked in billions from the mobile revolution. However, investors may be witnessing the beginning of a new chapter.</p>
<div class="lb-trans"><p>多年来，<strong>高通</strong>（QCOM +3.65%）一直被视为一家高端智能手机芯片制造商，从移动革命中赚取了数十亿美元。然而，投资者可能正在见证一个新篇章的开始。</p>
</div><p>At yesterday’s investor day, the company unveiled ambitious plans to more than double its non-handset sales in just three years, to $40 billion. While total revenue growth doesn’t seem significant, what’s huge is the changing revenue composition, driven by AI compute, which signals massive opportunities ahead across multiple large markets. Management estimates a combined total addressable market of $1.7 trillion by 2030.</p>
<div class="lb-trans"><p>在昨天的投资者日上，该公司公布了雄心勃勃的计划，计划在短短三年内将其非手机销售额翻倍，达到 400 亿美元。虽然总收入增长似乎并不显著，但更重要的是收入构成的变化，这一变化由人工智能计算驱动，预示着多个大型市场的巨大机会。管理层预计到 2030 年，综合可寻址市场将达到 1.7 万亿美元。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://g.foolcdn.com/image//query-b3A9cmVzaXplJnVybD1odHRwczovL2Nkbi5jb250ZW50LmZvb2xjZG4uY29tL2ltYWdlcy8xdW1uOXFlaC9wcm9kdWN0aW9uL2FjODAxMDVlM2VmM2M5Y2MwYzkwM2ZiNzZlNzQwYzg5N2Q3YmYwODItMTIwMHg4MDAuanBnP3c9MTIwMCZoPTgwMCZxPTc1JmF1dG89Zm9ybWF0Jnc9Mzg0MA" alt="" original-src="https://g.foolcdn.com/image/?url=https%3A%2F%2Fcdn.content.foolcdn.com%2Fimages%2F1umn9qeh%2Fproduction%2Fac80105e3ef3c9cc0c903fb76e740c897d7bf082-1200x800.jpg%3Fw%3D1200%26h%3D800%26q%3D75%26auto%3Dformat&amp;w=3840&amp;op=resize"/></p>
<p>Image source: Getty Images.</p>
<div class="lb-trans"><p>图片来源：Getty Images。</p>
</div><p>Qualcomm stock was up 11% in pre-market trading, and while the initial excitement seems to have worn off, the stock is still up nearly 8.5% by 1:15 p.m. The bigger story, though, is the hidden opportunities for retail investors willing to hold this stock over the next three to five years.</p>
<div class="lb-trans"><p>高通股票在盘前交易中上涨了 11%，尽管最初的兴奋似乎已经消退，但截至下午 1:15，股票仍上涨了近 8.5%。然而，更大的故事是，对于愿意在未来三到五年内持有该股票的散户投资者来说，隐藏的机会。</p>
</div><h2>The opportunity no one else is actively pursuing</h2>
<div class="lb-trans"><h2>没有人积极追求的机会</h2>
</div><p>Qualcomm’s expertise in smartphone chip manufacturing makes it the ideal candidate to pursue a potentially explosive segment lurking in the sidelines: edge AI compute.</p>
<div class="lb-trans"><p>高通在智能手机芯片制造方面的专业知识使其成为追求潜在爆炸性市场的理想候选者：边缘人工智能计算。</p>
</div><p>Today, most artificial intelligence (AI) models run in the cloud, which explains the explosive growth in data centers worldwide. What’s yet to takeoff at a massive level is the ability to run AI models on the millions of devices themselves, whether owned by individuals or corporations.</p>
<div class="lb-trans"><p>如今，大多数人工智能（AI）模型在云端运行，这解释了全球数据中心的爆炸性增长。然而，尚未大规模起飞的是在数百万设备上运行 AI 模型的能力，无论是个人还是企业拥有的设备。</p>
</div><p>Essentially, management at Qualcomm is predicting that AI models will become more efficient in both compute and energy use during inference, which is when the model produces an output based on what it has already learned. Additionally, there will be leaner, task-specific AI models trained for a particular job, which could then reside on the device itself.</p>
<div class="lb-trans"><p>本质上，高通的管理层预测，AI 模型在推理阶段的计算和能效将变得更加高效，推理阶段是模型根据已学习的内容生成输出的时刻。此外，将会有更精简、针对特定任务的 AI 模型为特定工作进行训练，这些模型可以驻留在设备本身上。</p>
</div><h2>Why edge AI matters</h2>
<div class="lb-trans"><h2>为什么边缘 AI 很重要</h2>
</div><p>As large language models (LLMs) become more sophisticated and capable of performing more generalized tasks, their utility for niche, specialized, and privacy-focused applications could eventually become limited. As AI compute proliferates, the need to reduce latency between the device and the large language model, usually located on a distant server, can’t be overstated. A smartphone camera that needs instant photo processing, or self-driving cars that can’t wait for a response from the cloud, are examples where any latency could be a major hindrance.</p>
<div class="lb-trans"><p>随着大型语言模型（LLMs）变得越来越复杂，能够执行更多通用任务，它们在小众、专业和隐私关注应用中的效用可能最终会受到限制。随着 AI 计算的普及，减少设备与通常位于远程服务器上的大型语言模型之间的延迟的需求不容小觑。需要即时照片处理的智能手机相机，或无法等待云端响应的自动驾驶汽车，都是延迟可能成为重大障碍的例子。</p>
</div><p>The other advantages are increased privacy for in-device processing and lower costs for enterprises looking to wean certain compute activities off expensive data centers.</p>
<div class="lb-trans"><p>其他优势包括设备内处理的隐私性提高，以及希望将某些计算活动从昂贵的数据中心转移出去的企业的成本降低。</p>
</div><h2>Expansion into multiple fields</h2>
<div class="lb-trans"><h2>向多个领域扩展</h2>
</div><p>Qualcomm expects to generate $40 billion in non-handset revenue by fiscal 2029. As of the second quarter of fiscal 2026, which ended on March 29, of the total $44 billion in trailing twelve-month sales, non-handset revenue stood at little over $16 billion, and the remaining $27 billion came from its smartphone chip business.</p>
<div class="lb-trans"><p>高通预计到 2029 财年将产生 400 亿美元的非手机收入。截至 2026 财年第二季度（截至 3 月 29 日），在过去 12 个月的总销售额 440 亿美元中，非手机收入略超过 160 亿美元，其余 270 亿美元来自智能手机芯片业务。</p>
</div><p>To reach a $40 billion annualized target, management needs to ramp up the non-smartphone business by a massive 150% over the next three and a half years, with edge computing contributing more than a third of the revenue.</p>
<div class="lb-trans"><p>为了实现 400 亿美元的年化目标，管理层需要在未来三年半内将非智能手机业务大幅提升 150%，其中边缘计算将贡献超过三分之一的收入。</p>
</div><p>Qualcomm’s foray into data centers could yield $15 billion annually. <strong>Meta Platforms</strong>(META 2.66%), the parent company of Facebook and Instagram, has agreed to use Qualcomm’s data center CPU, the Qualcomm Dragonfly™ C1000, to power its next-generation data centers. The CPU is expected to be in production in the second half of 2028.</p>
<div class="lb-trans"><p>高通进军数据中心可能每年带来 150 亿美元的收入。<strong>Meta Platforms</strong>（META 2.66%），Facebook 和 Instagram 的母公司，已同意使用高通的数据中心 CPU——Qualcomm Dragonfly™ C1000，为其下一代数据中心提供动力。该 CPU 预计将在 2028 年下半年投入生产。</p>
</div><h2>The market is already pricing a different business</h2>
<div class="lb-trans"><h2>市场已经在定价不同的业务</h2>
</div><p>Today's rally reflects confidence in Qualcomm's transformative AI ambitions beyond smartphones.</p>
<div class="lb-trans"><p>今天的反弹反映了市场对高通在智能手机之外的变革性 AI 雄心的信心。</p>
</div><p>The smartphone business is cyclical. People don’t usually upgrade their phones every year, but rather over a three- to four-year period, when existing handsets become obsolete. The result is faltering demand for legacy smartphone chips. Not surprisingly, over the last five years, the stock has gained only 57%, significantly underperforming both the <strong>Nasdaq-100</strong> and the <strong>S&amp;P 500</strong>.</p>
<div class="lb-trans"><p>智能手机业务是周期性的。人们通常不会每年升级手机，而是在现有手机变得过时时，通常在三到四年内进行升级。因此，传统智能手机芯片的需求正在减弱。不出所料，在过去五年中，股票仅上涨了 57%，显著落后于 <strong>纳斯达克 100</strong> 和 <strong>标准普尔 500</strong>。</p>
</div><p>At a little over 23 times trailing 12-month earnings and five times sales, the stock looks quite cheap for a company that is inevitably moving into the red-hot AI-infrastructure business.</p>
<div class="lb-trans"><p>以过去 12 个月的收益为基础，市盈率略高于 23 倍，销售额为 5 倍，对于一家不可避免地进入火热的 AI 基础设施业务的公司来说，股票看起来相当便宜。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://media.ycharts.com/charts/7168ded6cad563835fcecf1d1ce0e368.png" alt="" original-src="https://media.ycharts.com/charts/7168ded6cad563835fcecf1d1ce0e368.png"/></p>
<p>.</p>
<div class="lb-trans"><p>.</p>
</div><p>QCOM PE Ratio data by YCharts.</p>
<div class="lb-trans"><p>QCOM 市盈率数据来源于 YCharts。</p>
</div><h2>Execution holds the key</h2>
<div class="lb-trans"><h2>执行是关键</h2>
</div><p>While ambitions are easy to digest, what is actually challenging is gaining market share. Hyperscalers are competing for the best available chip at the lowest available price. In fact, many of them, such as <strong>Alphabet</strong> and <strong>Amazon</strong>, are designing their own chips.</p>
<div class="lb-trans"><p>虽然雄心壮志容易理解，但真正具有挑战性的是获得市场份额。超大规模企业正在争夺最佳可用芯片，价格最低。事实上，许多企业，如 <strong>Alphabet</strong> 和 <strong>Amazon</strong>，正在设计自己的芯片。</p>
</div><p>The real question investors must ask is whether Qualcomm can prove it can live up to its ambitions.</p>
<div class="lb-trans"><p>投资者必须问的真正问题是，高通能否证明自己能够实现其雄心。</p>
</div><p>Diversifying away from a predominantly single source of income, while initially scary, is a requirement. For Qualcomm, the next decade of growth could look very different from its last.</p>
<div class="lb-trans"><p>从主要单一收入来源多元化，虽然最初令人害怕，但这是必要的。对于高通来说，未来十年的增长可能与过去大相径庭。</p>
</div>

Qualcomm Just Made Its Biggest AI Bet Yet -- and Investors Are Loving It

- 高通计划在三年内将非手机收入提升至 400 亿美元，主要由 AI 计算推动。  
- 公司估计到 2030 年，总可寻址市场将达 1.7 万亿美元。  
- 高通将向数据中心领域扩展，预计将每年带来 150 亿美元收入。

Motley Fool

Direxion Daily QCOM Bear 1X Shares

Direxion Daily QCOM Bull 2X Shares

2 倍做多 Meta ETF - Direxion

Direxion Daily META Bear 1X Shares

Roundhill META WeeklyPay ETF

2 倍做多 META ETF - GraniteShares

Yieldmax Meta Option Income Strategy ETF

Graniteshares YieldBOOST Meta ETF

2 倍做多 MAG7 指数 ETF - Roundhill

纳斯达克 100 指数

标普 500 指数

谷歌-A

谷歌-C

亚马逊

高通刚刚进行了其有史以来最大的一次 AI 投资，投资者对此表示非常看好