<p>目前市場上沒有比人工智能和半導體股票更熱門的交易了，這些股票正在迅速上漲。在過去一個月的走勢之後，我認為值得退一步，誠實地問一個問題：這波上漲是否來得太快、太猛，還是説背後的故事如此強大，以至於唯一的錯誤就是不買這些熱門股票？</p>
<p>這筆交易雙方的數據確實引人注目。讓我們來逐步分析。</p>
<h3>讓所有人都在談論的走勢</h3>
<p>首先來看 <strong>半導體 ETF (SMH)</strong>。SMH 在過去 28 天中上漲了 24 天。在這 28 天裏，SMH 上漲了 <strong>50%</strong>！</p>
<p>這是一種單向的動量，往往會隨着時間的推移吸引越來越多的資本——後來的投資者紛紛湧入，空頭被擠壓，期權市場的做市商忙於對沖。當交易者不斷追逐這些股票時，這種情況就會變成自我實現的預言。</p>
<h3>那麼，為什麼多頭如此自信？</h3>
<p>這裏變得真正有趣了……</p>
<p><strong>高盛</strong> 估計，五大 AI 巨頭在 2026 年的資本支出將達到 <strong>7550 億美元</strong>，同比增長 83%。而摩根士丹利則將這一估計進一步提高，預計該集團今年的支出將超過 8000 億美元，並在 2027 年達到 <strong>1.1 萬億美元</strong>。</p>
<p>為了讓 1.1 萬億美元更具可比性，這一數字將超過美國預計的國防支出！</p>
<p>當然，關鍵問題一直是：所有這些支出是否產生了真正的收入？</p>
<p>[text_ad]</p>
<h3>收入故事開始追趕上來</h3>
<p>根據《金融時報》，<strong>Anthropic</strong> 的年化收入預計將超過 <strong>450 億美元</strong>——比去年年底的 90 億美元的年化水平增長了五倍。在幾個月內實現五倍增長。</p>
<p>現在，Anthropic 是一傢俬營公司，僅僅是一個數據點。但這個數據點反映了更廣泛的趨勢：AI 應用正在從 “酷炫的演示” 階段轉向大規模的企業部署。公司們正在付費。</p>
<h3>美國銀行的警告</h3>
<p>但 SMH 的動量並不是唯一值得密切關注的事情。</p>
<p>美國銀行指出了一個更具結構性的問題：他們所稱的 “AI 大十”——七大巨頭加上 <strong>博通 (AVGO)</strong>、<strong>美光 (MU)</strong> 和 <strong>超威半導體 (AMD)</strong> 的集中度，正接近歷史上通常會出現重大市場頂部的峯值水平。</p>
<p><img src="https://imageproxy.pbkrs.com/https://cabot.brightspotgocdn.com/dims4/default/1d3f2d7/2147483647/strip/true/crop/1060x717+0+0/resize/464x314?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://cabot.brightspotgocdn.com/dims4/default/1d3f2d7/2147483647/strip/true/crop/1060x717+0+0/resize/464x314"/></p>
<p>這是一種不同於 “圖表看起來過度擴展” 的警告。集中風險是一個結構性問題。當少數幾家公司佔據了標準普爾 500 指數總市值的過大份額時，市場失去了緩衝。一旦從這些公司中僅僅幾隻股票發生輪動，不僅會傷害這些股票，還會對整個市場產生漣漪效應。</p>
<h3>市場的其他部分</h3>
<p>另一個擔憂是，儘管 AI 和半導體股票正在迅速上漲，但 “其他” 市場的表現卻只是一般。</p>
<p>正如 @bespokeinvest 在週二所指出的：“昨天，標準普爾 500 指數在另一個歷史高點收盤，幾乎比其 50 日移動平均線高出 8%，但該指數的新 52 周低點數量超過了新 52 周高點數量（42 比 39），而且超過其 50 日移動平均線的股票比例降到了 50% 以下（48.6%）。”</p>
<h3>同時存在的兩個真相</h3>
<p>那麼，這將我們置於何處？我認為我們正面臨一種情況，其中兩件事情可以同時成立。</p>
<p>AI 交易已經擴展。SMH 的走勢是歷史性的。集中度水平發出警告。任何嚴重的回調觸發因素——來自巨頭的令人失望的財報、聯邦儲備的語氣變化、地緣政治衝擊——都可能迅速使這一羣體下跌 15% 至 20%。</p>
<p>儘管存在風險，但基礎設施投資是真實的，收入加速正在發生，而進行支出的公司並沒有表現出恐慌。</p>
<p>退一步來看，我的看法是，這是一個你需要接觸的主題，但在經歷如此巨大的 AI 股票上漲後全力以赴可能會在音樂停止時造成極大的痛苦。</p>
<p>[author_ad]</p>

嘉實上證科創板芯片ETF

國聯安半導體ETF

國泰中證全指集成電路ETF

<p>人工智能和半導體股票正經歷顯著的上漲勢頭，半導體 ETF（SMH）在 28 天內上漲了 50%。高盛預測人工智能領域的資本支出將大幅增加，預計到 2026 年將達到 7550 億美元。然而，美國銀行警告稱，頂尖人工智能股票的集中風險可能導致市場不穩定。儘管存在風險，人工智能應用的收入增長正在加速，表明有強勁的基本趨勢。建議投資者在考慮投資該領域時要謹慎，以防潛在的回調</p>

<p>There is no hotter trade in the market right now than AI and Semiconductor stocks, which have been racing higher. And after the move we’ve seen over the past month, I think it’s worth stepping back and asking an honest question: Has this run come too far, too fast – or is the underlying story so powerful that the only mistake would be not buying these hot stocks?</p>
<div class="lb-trans"><p>目前市場上沒有比人工智能和半導體股票更熱門的交易了，這些股票正在迅速上漲。在過去一個月的走勢之後，我認為值得退一步，誠實地問一個問題：這波上漲是否來得太快、太猛，還是説背後的故事如此強大，以至於唯一的錯誤就是不買這些熱門股票？</p>
</div><p>The data on both sides of this trade is genuinely compelling. Let’s walk through it.</p>
<div class="lb-trans"><p>這筆交易雙方的數據確實引人注目。讓我們來逐步分析。</p>
</div><h3>The Move That Has Everyone Talking</h3>
<div class="lb-trans"><h3>讓所有人都在談論的走勢</h3>
</div><p>Start with the <strong>Semiconductor ETF (SMH)</strong>. The SMH has been up 24 out of the last 28 days. And during those 28 days, the SMH has rallied <strong>50%</strong>!</p>
<div class="lb-trans"><p>首先來看 <strong>半導體 ETF (SMH)</strong>。SMH 在過去 28 天中上漲了 24 天。在這 28 天裏，SMH 上漲了 <strong>50%</strong>！</p>
</div><p>This is the kind of one-directional momentum that tends to attract more and more capital as it unfolds – latecomers piling in, shorts getting squeezed, and options market makers scrambling to hedge. It becomes a self-fulfilling prophecy as traders chase these stocks higher and higher.</p>
<div class="lb-trans"><p>這是一種單向的動量，往往會隨着時間的推移吸引越來越多的資本——後來的投資者紛紛湧入，空頭被擠壓，期權市場的做市商忙於對沖。當交易者不斷追逐這些股票時，這種情況就會變成自我實現的預言。</p>
</div><h3>So Why Are the Bulls So Confident?</h3>
<div class="lb-trans"><h3>那麼，為什麼多頭如此自信？</h3>
</div><p>Here’s where it gets genuinely interesting …</p>
<div class="lb-trans"><p>這裏變得真正有趣了……</p>
</div><p><strong>Goldman Sachs</strong> estimates that the five largest AI hyperscalers will spend <strong>$755 billion</strong> on capital expenditures in 2026 alone – an increase of 83% year over year. And Morgan Stanley pushes that estimate even higher, seeing the group top $800 billion this year on their way to <strong>$1.1 trillion</strong> in 2027.</p>
<div class="lb-trans"><p><strong>高盛</strong> 估計，五大 AI 巨頭在 2026 年的資本支出將達到 <strong>7550 億美元</strong>，同比增長 83%。而摩根士丹利則將這一估計進一步提高，預計該集團今年的支出將超過 8000 億美元，並在 2027 年達到 <strong>1.1 萬億美元</strong>。</p>
</div><p>To put $1.1 trillion in context, that figure would exceed the projected American spending on national defense!</p>
<div class="lb-trans"><p>為了讓 1.1 萬億美元更具可比性，這一數字將超過美國預計的國防支出！</p>
</div><p>The critical question, of course, has always been: Is all this spending generating real revenue?</p>
<div class="lb-trans"><p>當然，關鍵問題一直是：所有這些支出是否產生了真正的收入？</p>
</div><p>[text_ad]</p>
<div class="lb-trans"><p>[text_ad]</p>
</div><h3>The Revenue Story Is Starting to Catch Up</h3>
<div class="lb-trans"><h3>收入故事開始追趕上來</h3>
</div><p>According to the <em>Financial Times</em>, <strong>Anthropic’s</strong> annualized revenue is on track to cross <strong>$45 billion</strong> – a fivefold increase from the $9 billion run rate it was carrying at the end of last year. A fivefold increase. In a matter of months.</p>
<div class="lb-trans"><p>根據《金融時報》，<strong>Anthropic</strong> 的年化收入預計將超過 <strong>450 億美元</strong>——比去年年底的 90 億美元的年化水平增長了五倍。在幾個月內實現五倍增長。</p>
</div><p>Now, Anthropic is a private company and a single data point. But it’s a data point that speaks to something broader: AI applications are moving past the “cool demo” stage and into enterprise deployment at scale. Companies are paying.</p>
<div class="lb-trans"><p>現在，Anthropic 是一傢俬營公司，僅僅是一個數據點。但這個數據點反映了更廣泛的趨勢：AI 應用正在從 “酷炫的演示” 階段轉向大規模的企業部署。公司們正在付費。</p>
</div><h3>A Warning from Bank of America</h3>
<div class="lb-trans"><h3>美國銀行的警告</h3>
</div><p>But the SMH momentum isn’t the only thing worth watching carefully.</p>
<div class="lb-trans"><p>但 SMH 的動量並不是唯一值得密切關注的事情。</p>
</div><p>Bank of America has flagged something arguably more structural: concentration in what they’re calling the “AI Big 10” – the Magnificent 7 plus <strong>Broadcom (AVGO)</strong>, <strong>Micron (MU)</strong>, and <strong>Advanced Micro Devices (AMD)</strong> – is now approaching the kind of peak levels that have historically preceded meaningful market tops.</p>
<div class="lb-trans"><p>美國銀行指出了一個更具結構性的問題：他們所稱的 “AI 大十”——七大巨頭加上 <strong>博通 (AVGO)</strong>、<strong>美光 (MU)</strong> 和 <strong>超威半導體 (AMD)</strong> 的集中度，正接近歷史上通常會出現重大市場頂部的峯值水平。</p>
</div><p><img src="https://imageproxy.pbkrs.com/https://cabot.brightspotgocdn.com/dims4/default/1d3f2d7/2147483647/strip/true/crop/1060x717+0+0/resize/464x314!/quality/90//query-dXJsPWh0dHBzOi8vazItcHJvZC1jYWJvdC5zMy51cy1lYXN0LTEuYW1hem9uYXdzLmNvbS9icmlnaHRzcG90LzE4LzVkLzFjNzBkZTI1NGRkODgyNGFkYzU2ZmFjODBmMmMvcGVhay1jb25jZW50cmF0aW9uLWJ1YmJsZXMucG5n" alt="" original-src="https://cabot.brightspotgocdn.com/dims4/default/1d3f2d7/2147483647/strip/true/crop/1060x717+0+0/resize/464x314!/quality/90/?url=https%3A%2F%2Fk2-prod-cabot.s3.us-east-1.amazonaws.com%2Fbrightspot%2F18%2F5d%2F1c70de254dd8824adc56fac80f2c%2Fpeak-concentration-bubbles.png"/></p>
<p>This is a different kind of warning than “the chart looks extended.” Concentration risk is a structural problem. When a small handful of names account for an outsized share of total S&amp;P 500 market cap, the market loses its cushion. A rotation out of just a few of these names doesn’t just hurt those stocks – it sends ripples through everything.</p>
<div class="lb-trans"><p>這是一種不同於 “圖表看起來過度擴展” 的警告。集中風險是一個結構性問題。當少數幾家公司佔據了標準普爾 500 指數總市值的過大份額時，市場失去了緩衝。一旦從這些公司中僅僅幾隻股票發生輪動，不僅會傷害這些股票，還會對整個市場產生漣漪效應。</p>
</div><h3>The Rest of the Market</h3>
<div class="lb-trans"><h3>市場的其他部分</h3>
</div><p>Another concern is that while the AI and Semiconductor stocks are racing higher, “the rest” of the market is doing just ok.</p>
<div class="lb-trans"><p>另一個擔憂是，儘管 AI 和半導體股票正在迅速上漲，但 “其他” 市場的表現卻只是一般。</p>
</div><p>As noted by @bespokeinvest on Tuesday: “Yesterday, on a day that the S&amp;P 500 closed at another all-time high, nearly 8% above its 50-DMA, there were more new 52-week lows in the index than new 52-week highs (42 to 39), and the % of stocks above their 50-DMAs dropped below 50% (48.6%).”</p>
<div class="lb-trans"><p>正如 @bespokeinvest 在週二所指出的：“昨天，標準普爾 500 指數在另一個歷史高點收盤，幾乎比其 50 日移動平均線高出 8%，但該指數的新 52 周低點數量超過了新 52 周高點數量（42 比 39），而且超過其 50 日移動平均線的股票比例降到了 50% 以下（48.6%）。”</p>
</div><h3>Two Truths at the Same Time</h3>
<div class="lb-trans"><h3>同時存在的兩個真相</h3>
</div><p>So where does that leave us? I think we’re looking at a situation where two things can be true at once.</p>
<div class="lb-trans"><p>那麼，這將我們置於何處？我認為我們正面臨一種情況，其中兩件事情可以同時成立。</p>
</div><p>The AI trade is extended. The SMH move is historic. Concentration levels are flashing caution. Any serious pullback trigger – a disappointing earnings report from a hyperscaler, a shift in Fed tone, a geopolitical shock – could send this group down 15–20% in a hurry.</p>
<div class="lb-trans"><p>AI 交易已經擴展。SMH 的走勢是歷史性的。集中度水平發出警告。任何嚴重的回調觸發因素——來自巨頭的令人失望的財報、聯邦儲備的語氣變化、地緣政治衝擊——都可能迅速使這一羣體下跌 15% 至 20%。</p>
</div><p>And while there are risks, the infrastructure investment is real, the revenue acceleration is happening and the companies doing the spending are not acting like they’re spooked.</p>
<div class="lb-trans"><p>儘管存在風險，但基礎設施投資是真實的，收入加速正在發生，而進行支出的公司並沒有表現出恐慌。</p>
</div><p>Stepping back, my read is this is a theme you need to have exposure to, but to go all in after such a monster run in AI stocks could be extremely painful if/when the music stops.</p>
<div class="lb-trans"><p>退一步來看，我的看法是，這是一個你需要接觸的主題，但在經歷如此巨大的 AI 股票上漲後全力以赴可能會在音樂停止時造成極大的痛苦。</p>
</div><p>[author_ad]</p>
<div class="lb-trans"><p>[author_ad]</p>
</div>

Has the AI Trade Gotten Too Hot?

CabotWealthNetwork

嘉實中證全指集成電路ETF

易方達中證半導體材料設備主題ETF

半導體 ETF - VanEck Vectors

半導體 3 倍做多 - Direxion

費城交易所 半導體 ETF - iShares

標普半導體 ETF - SPDR

國泰CES半導體芯片行業ETF

招商中證半導體產業ETF

國泰中證半導體材料設備主題ETF

國聯安科創芯片設計ETF

華夏國證半導體芯片ETF

高盛

摩根士丹利

美國銀行

博通

美光科技

標普 500 指數

GS Hldg R

MORGAN STANLEY DEP SHS EA REP 1/1000 SHS NON CUM PFD O

MORGAN STANLEY DEP SHS REPSTG 1/1000TH PFD SER Q

MORGAN STANLEY DEP SHS REPS 1/1000 7.125% PFD E

MORGAN STANLEY NON-CUM DEP SH REP 1/1000 PFD SER I

MORGAN STANLEY 4.875 DEP SHS REP 1/1000 NCUM PFD SR L

MORGAN STANLEY 6.5%DEP SHS REPSTG 1/1000 NON CUM SER P

MORGAN STANLEY DEP SHS REPS 1/1000 PFD SER 'A'

MORGAN STANLEY DEP SHS REPR 1/1000TH INT PFD SER 'F' FX

MORGAN STANLEY DEP SHS REP 1/1000 NON-CUM PFD STK SER K

BANK OF AMERICA CORPORATION DEP SHS REP 1/1000TH INT NON CUM PFD QQ

BANK OF AMERICA CORPORATION DEP SHS REP 1/1200TH PRF SERIES '5'

BANK OF AMERICA CORPORATION 7.25% NON-CUM PERP CONV PFD SER L

BANK OF AMERICA CORPORATION 5.875%NON CUM DEP SHS REP1/1000TH SER HH

BANK OF AMERICA CORPORATION DEP SHS REP 1/1200TH PRF SERIES '2'

BANK OF AMERICA CORPORATION 5% DEP SHS REPR 1/1000TH NON CUM PFD LL

BANK OF AMERICA CORPORATION DEP SHARES REP 1/1000TH PERP PFD SER 'E'

BANK OF AMERICA CORPORATION INCOME CAPITAL OBLIG NTS DUE 15/12/2066

BANK OF AMERICA CORPORATION 4.375% DP SH RP 1/1000TH IN NON CU PF NN

BANK OF AMERICA CORPORATION DEP SH REP 1/1200TH PRF SER 1 USD25

BANK OF AMERICA CORPORATION 5.375%DEP SHS REP 1/1000TH NON CUM SR KK

BANK OF AMERICA CORPORATION 6%NON CUM DEP SHS EACH REP 1/1000 SER GG

BANK OF AMERICA CORPORATION DEP SHS REP 1/1200 PFD SER 4 USD25

BANK OF AMERICA CORPORATION DEP SHS REP 1/1000TH NON CUM PFD SER SS

BANK OF AMERICA CORPORATION 4.125%DEP SH REP 1/1000 PFD SER PP USD25

Bank of America Corp

人工智能交易是否已經過熱？