AI泡泡要破了?美股閃崩前兆浮現,大咖示警風暴將至!

AI泡泡要破了?美股閃崩前兆浮現,大咖示警風暴將至!

Source: YouTube URL: https://www.youtube.com/watch?v=TSllcwvKXP4 Date: 2025-10-11 Duration: 52:10


📄 English Version

TL;DR

Taiwan and US stock markets continue their powerful rally, with Taiwan approaching 30,000 points and AI stocks driving unprecedented growth. However, major warning signs emerge: JPMorgan CEO warns of severe corrections within 6-24 months, while market euphoria indicators suggest short-term peaks. Taiwan’s National Stabilization Fund continues market support despite economic recovery, benefiting from Trump’s tariff policies. The “AI perpetual motion machine” expands with chip manufacturers, compute providers, and application giants in a self-reinforcing cycle. Smart investors should gradually take profits while maintaining core positions in strong performers.

Key Points

Market Overview & Warnings

Taiwan’s Economic Advantage

AI Perpetual Motion Machine Expansion

Chip Manufacturing Dominance

Compute Power Gold Rush

Application Giants’ Spending Spree

Valuation & Risk Analysis

Technical Warning Signals

Asset Class Performance

Historical Winners Analysis

Detailed Analysis

Market Psychology & Positioning

The presenter emphasizes a critical paradox: while fundamentals remain strong and AI growth undeniable, market psychology indicators flash warning signals. The appearance of predictive charts showing Taiwan’s index climbing to 32,000 points historically precedes short-term peaks. However, attempting to short or go to cash would be premature - the bubble may inflate much further before bursting.

The recommended approach mirrors the dot-com bubble wisdom: participate fully during the euphoric phase, but maintain vigilance for technical warning signals. The presenter learned from 1998-2000 that bubbles can persist far longer than rational analysis suggests, and fighting them means missing massive gains.

National Stabilization Fund’s Unusual Persistence

Taiwan’s National Stabilization Fund (NSF) typically exits after markets stabilize, historically making money on all 7-8 interventions since 2001. The current situation is anomalous - the fund continues buying despite Taiwan’s index near 30,000 points, citing ongoing tariff negotiations and semiconductor policy uncertainties.

The presenter notes that with Trump’s unpredictable policy style, these uncertainties could persist for his entire 3-year remaining term, potentially keeping the NSF active indefinitely. For investors, this represents strong institutional support - the fund has never failed to mark market bottoms, and its continued presence suggests the rally has further room to run.

The AI Perpetual Motion Machine’s Expansion

The circular capital flows create a self-reinforcing cycle:

  1. NVIDIA → invests in OpenAI
  2. OpenAI → commits compute purchases from Oracle and AMD
  3. Oracle/AMD → buy more NVIDIA chips
  4. AMD → invests back in OpenAI (10% stake)

This “perpetual motion machine” means capital continuously cycles among these companies, with each announcement driving stock prices higher. The addition of AMD effectively makes it indirectly owned by NVIDIA through the capital flow chain.

The presenter humorously acknowledges this isn’t truly perpetual - economic recession or demand slowdown could break the cycle. But the key insight is: don’t preset the top. Participate while momentum exists, exit on technical signals, re-enter after corrections.

Compute Power: The New Gold Rush

Former Bitcoin miners possess perfect infrastructure for AI compute: - Existing data centers and cooling systems - Power supply already secured - Only need GPU swap: Bitcoin mining ASICs → AI training/inference GPUs

Companies like Iris Energy (IREN) exemplify the transformation: - December 2024 quarterly revenue: $43 million - Latest quarter (Sep 2025): $187 million - Growth: 3-4x in just one year

CoreWeave (CRWF) showed similar explosive growth, with stock price appreciation from August lows to current levels representing 200%+ gains.

Risk Assessment: While current performance is spectacular, the presenter warns these companies lack the technical moats of NVIDIA/TSMC/AMD. They’re essentially commodity compute providers. If AI compute demand slows, these stocks face steeper corrections than chip manufacturers with durable competitive advantages.

Application Giants: No Signs of Slowing

Microsoft, Google, Amazon, and Meta continue aggressive AI infrastructure spending with no indication of budget constraints. Google’s Gemini and related AI applications show improving utility, suggesting strong Q3 subscriber growth ahead.

The supply chain dependency is clear: - Tier 1: Chip manufacturers (NVIDIA, AMD, TSMC) create the hardware - Tier 2: Compute providers (CoreWeave, Iris Energy) rent capacity - Tier 3: Application giants (Google, Microsoft) develop consumer products

Each tier depends on the one above it, creating a cascading demand structure that supports continued AI investment.

The Magnificent 7 Divergence

Earnings growth disparity continues widening: - Q3 2025 Magnificent 7: +14% YoY earnings growth - S&P 500 other 493 companies: +4% YoY earnings growth

Analysts have consistently underestimated this divergence for 3+ years: - Beginning of 2024: Predicted convergence → didn’t happen - Beginning of 2025: Again predicted convergence → still widening - Current 2026 estimates: Magnificent 7 revised from 0% to +3%; other 493 revised from 0% to -3%

Investment implication: Until this pattern breaks, capital concentration in tech giants and AI infrastructure makes sense. Watch for two signals of potential rotation: 1. Traditional sectors showing accelerating revenue/profit growth 2. During market corrections, lagging stocks refuse to drop and start rallying

Tesla’s Unique Position

Tesla stands out with ~200x P/E despite only 35% earnings growth expectations - a clear “dream premium” dependent on: - Full Self-Driving (FSD) subscription revenue: Recurring software income stream - Humanoid robot (Optimus): Factory automation and consumer robotics potential - Strategic pivot away from EV price competition: Can’t compete with BYD/Xiaomi on cost, focusing on software and robotics instead

Recent disappointment from the “affordable EV” announcement - instead of the promised $25,000 new model, Tesla revealed a 10% cheaper, stripped-down Model 3 without FSD. This suggests: - Short-term: Limited EV sales growth catalyst - Long-term: Strategic focus on higher-margin software and robotics

Technical Signal: The Oracle Case Study

Oracle (ORCL) provides an instructive example of how to interpret corrections in strong uptrends:

Situation: Q3 earnings showed $100M loss on Blackwell compute rental (new business with low initial margins)

Market reaction: Initially dropped 7%, creating panic

Technical analysis: - Stock had gapped up to new highs on previous strong earnings - The selloff created a gap down but didn’t fill the previous gap completely - Support held at the 20-day moving average - Market quickly recovered, erasing the black candlestick

Interpretation: As long as the gap-down doesn’t completely retrace the prior gap-up, and support holds, the uptrend remains intact. The temporary panic represented buying opportunity rather than trend change.

Warning signal to watch: Gap-up followed by immediate high-volume reversal to close near lows would indicate distribution and potential trend reversal.

Smart Position Management

The presenter emphasizes a nuanced approach to this late-stage bull market:

For long-term holders with low cost basis: - Maintain core positions without stress - Any corrections are navigable with time horizon and low entry prices

For recent entrants at current levels: - Psychological pressure will be higher during volatility - Essential to maintain cash reserves for flexibility - Consider gradual profit-taking: “sell a little every day” - When correction arrives, you’ll be glad to have cash available

Key principle: Never fully exit strong positions based on valuation fears. The market can remain irrational (expensive) far longer than you can remain solvent (in cash). Gradual rebalancing > all-or-nothing timing attempts.

What Could Break the AI Cycle?

The presenter identifies potential inflection points:

Demand-side risks: - Economic recession reducing enterprise/consumer AI spending - Diminishing returns on AI model improvements (current models “good enough”) - Regulatory constraints limiting AI development/deployment

Technical signals: - High-volume reversal candlesticks after new highs - Moving average breakdowns across multiple timeframes - Breadth deterioration: fewer stocks participating in rallies

Fundamental signals: - Earnings growth deceleration despite revenue growth (margin compression) - Guidance reductions from bellwether companies - CapEx reduction announcements from big tech

Critical insight: The inflection will likely happen without obvious news catalysts. By the time mainstream media identifies “why the market is falling,” the decline will be mostly complete. Vigilance on technical signals matters more than news flow.

The Broader Market Context

Rotation indicators to monitor:

  1. During corrections, lagging sectors don’t fall: Signals capital beginning to rotate
  2. Accelerating revenue/profit growth in non-AI sectors: Fundamental improvement driving interest
  3. Valuation normalization: As AI stocks become expensive, value investors rotate elsewhere

Current status: Premature for rotation. AI/tech leadership remains dominant. Traditional sectors (consumer goods, textiles, apparel) stuck in downtrends. Until these sectors show technical/fundamental improvement, staying with winners makes sense.

Investment Course Promotion Context

The presenter promotes a Taiwan-US stock allocation course (ending early bird pricing in 2 hours, code USA88 for additional discount), emphasizing:

Course value proposition: - Combined Taiwan and US market coverage for diversification - Both currency and industry risk mitigation - Systematic approach to identifying growth stocks - Technical and fundamental analysis integration - Private student community with instructor participation - Tools including broker comparisons and asset allocation calculators

Educational philosophy: No fixed formulas exist, but multiple analytical tools increase winning probability. Course provides systematic framework rather than prescriptive rules.

Key Quotes

“長期來看,資產真的還是要放在股市當中。如果說可能都完完全全沒有投資,真的會跟資產成長的速度是脫節的。” - “Long-term, assets really need to be in the stock market. If you completely avoid investing, you’ll truly fall behind the pace of asset growth.”

“每當這種形勢一出來之後,就代表說其實股市短線的漲幅真的是已經有一點太過誇張了。” - “Whenever these predictive charts appear, it represents that the short-term market gains have truly become somewhat excessive.”

“國安基金真的是很深準的,只要喊國安基金進場的時候,幾乎都是相對低點。” - “The National Stabilization Fund is incredibly accurate - whenever it announces entry, it’s almost always near relative lows.”

“台灣的進口金額的成長率,是遠遠超過其他的國家…基本上台灣他的成長可以說是一枝獨秀。” - “Taiwan’s import amount growth rate far exceeds other countries… Taiwan’s growth can essentially be described as uniquely outstanding.”

“既然泡沫來,即便質疑泡沫但是打不贏就加入,就是真的等到泡沫破裂我們再退場。” - “Since the bubble is here, even if you question it, if you can’t beat it, join it - truly wait until the bubble bursts, then we exit.”

“這個永動機就越滾越大,因為其實等於說這個錢,其實都一直以來,都是在這幾家公司換來換去。” - “This perpetual motion machine keeps growing bigger, because essentially the money has continuously been cycling among these companies.”

“我們永遠不知道市場最瘋狂的一個點位到底會到什麼時候。” - “We never know when the market’s craziest point will actually arrive.”

“你在4月對等關稅的時候你是受恐慌的時候做進場,其實你當下進場的時候,你是找不到什麼利多。” - “When you entered during April’s reciprocal tariff panic, you couldn’t find any positive news at that moment of entry.”

“未來可能真的股市開始要陷入一個漲多修正的時候,你一定也不會找到什麼利空。” - “When the market truly begins entering an overbought correction in the future, you definitely won’t find any negative catalysts either.”

“比起自己虧錢,我看見大家虧錢我是更難過。” - “More than my own losses, I feel sadder seeing everyone else lose money.”


📄 繁體中文版

TL;DR 極簡摘要

台股與美股持續強勁上漲,台股接近30,000點,AI股帶動空前成長。然而重大警訊浮現:摩根大通執行長警告6-24個月內將出現嚴重修正,市場狂熱指標顯示短線高點。台灣國安基金儘管經濟復甦仍持續護盤,受惠川普關稅政策。「AI永動機」擴大至晶片製造商、算力供應商、應用巨頭的自我強化循環。聰明投資人應逐步獲利了結,同時保留強勢標的核心部位。

關鍵要點

市場概況與警訊

台灣經濟優勢

AI永動機擴張

晶片製造主導地位

算力淘金熱

應用巨頭瘋狂砸錢

估值與風險分析

技術警訊

資產類別表現

歷史贏家分析

詳細分析

市場心理與定位

主講人強調一個關鍵矛盾:儘管基本面仍然強勁且AI成長無可否認,但市場心理指標閃爍警訊。預測圖表顯示台股指數攀升至32,000點,歷史上這類圖表出現時往往預示短線高點。然而,試圖放空或持有現金將為時過早 - 泡沫可能進一步膨脹後才破裂。

推薦方法呼應網路泡沫智慧:在狂熱階段充分參與,但對技術警訊保持警覺。主講人從1998-2000年學到,泡沫持續時間可能遠超理性分析所預期,對抗泡沫意味錯失巨大收益。

國安基金的異常持續

台灣國安基金通常在市場穩定後退場,自2001年以來7-8次介入均獲利出場。目前情況異常 - 儘管台股指數接近30,000點,基金仍持續買進,理由為關稅談判與半導體政策不確定性持續。

主講人指出,以川普不可預測的政策風格,這些不確定性可能持續其餘下3年任期,潛在使國安基金無限期活躍。對投資人而言,這代表強大機構支持 - 該基金從未失手標記市場底部,其持續存在顯示漲勢仍有空間。

AI永動機的擴張

循環資金流創造自我強化週期:

  1. NVIDIA → 投資 OpenAI
  2. OpenAI → 承諾向 OracleAMD 採購算力
  3. Oracle/AMD → 購買更多NVIDIA晶片
  4. AMD → 回頭投資OpenAI(10%股權)

此「永動機」意味資金持續在這些公司間循環,每個公告都推動股價上漲。AMD的加入實質上使其透過資金流鏈間接為NVIDIA所有。

主講人幽默承認這並非真正永動 - 經濟衰退或需求放緩可能打破循環。但關鍵洞察是:別預設頂部。在動能存在時參與,根據技術訊號退場,修正後再進。

算力:新淘金熱

前比特幣礦工擁有AI算力的完美基礎設施: - 既有資料中心與冷卻系統 - 電力供應已確保 - 僅需GPU互換:比特幣挖礦ASIC → AI訓練/推理GPU

Iris Energy(IREN)等公司體現轉型: - 2024年12月季營收:4,300萬美元 - 最新季度(2025年9月):1.87億美元 - 成長:一年內3-4倍

CoreWeave(CRWF)顯示類似爆炸性成長,從8月低點至當前水準的股價升幅代表200%+收益。

風險評估:儘管當前表現壯觀,主講人警告這些公司缺乏NVIDIA/TSMC/AMD的技術護城河。本質上是商品化算力供應商。若AI算力需求放緩,這些股票面臨比具持久競爭優勢的晶片製造商更劇烈的修正。

應用巨頭:無放緩跡象

微軟、Google、Amazon與Meta持續積極AI基礎設施支出,無預算限制跡象。Google的Gemini與相關AI應用顯示改善的實用性,暗示Q3訂閱戶強勁成長。

供應鏈依賴關係清晰: - 第一層:晶片製造商(NVIDIA、AMD、TSMC)創造硬體 - 第二層:算力供應商(CoreWeave、Iris Energy)出租容量 - 第三層:應用巨頭(Google、微軟)開發消費產品

每層依賴上一層,創造支持持續AI投資的連鎖需求結構。

科技七巨頭分歧

盈餘成長差距持續擴大: - 2025年Q3科技七巨頭:年增率+14% - 標普500其他493家公司:年增率+4%

分析師持續低估此分歧達3年以上: - 2024年初:預測收斂 → 未發生 - 2025年初:再次預測收斂 → 仍在擴大 - 當前2026年估計:七巨頭從0%上修至+3%;其他493從0%下修至-3%

投資意涵:在此模式打破前,資金集中科技巨頭與AI基礎設施合理。留意兩個潛在輪動訊號: 1. 傳統產業顯示加速營收/獲利成長 2. 市場修正期間,落後股拒絕下跌並開始反彈

特斯拉的獨特定位

特斯拉以~200倍本益比脫穎而出,儘管僅35%盈餘成長預期 - 明顯的「夢想溢價」取決於: - 完全自動駕駛(FSD)訂閱營收:經常性軟體收入流 - 人形機器人(Optimus):工廠自動化與消費機器人潛力 - 策略轉向遠離電動車價格競爭:無法在成本上與比亞迪/小米競爭,專注軟體與機器人

平價電動車公告近期令人失望 - 非承諾的25,000美元新車款,而是便宜10%、簡配的Model 3且無FSD。此暗示: - 短期:有限電動車銷量成長催化劑 - 長期:策略聚焦高毛利軟體與機器人

技術訊號:甲骨文案例研究

甲骨文(ORCL)提供如何解讀強勁上升趨勢修正的教學範例:

情況:Q3財報顯示Blackwell算力租賃虧損1億美元(新業務初期毛利率低)

市場反應:初期下跌7%,造成恐慌

技術分析: - 股票先前財報強勁而跳空創新高 - 賣壓造成跳空下跌但未完全填補前次跳空 - 20日均線支撐守住 - 市場快速恢復,抹除黑K

解讀:只要跳空下跌未完全回補前次跳空上漲,且支撐守住,上升趨勢保持完整。短暫恐慌代表買入機會而非趨勢改變。

警訊觀察:跳空上漲後立即高量反轉收在低點附近,將顯示籌碼分散與潛在趨勢反轉。

聰明部位管理

主講人強調此多頭市場後期的細膩方法:

長期持有者具低成本基礎: - 維持核心部位無壓力 - 任何修正憑時間地平線與低進場價可渡過

當前水準近期進場者: - 波動期間心理壓力較高 - 維持現金儲備彈性至關重要 - 考慮逐步獲利了結:「每天賣一點點」 - 修正來臨時,將慶幸手握現金

關鍵原則:切勿基於估值恐懼完全退出強勢部位。市場非理性(昂貴)持續時間可能遠超你保持償付能力(持有現金)的時間。漸進再平衡 > 全有或全無的時機嘗試。

可能打破AI循環的因素?

主講人識別潛在轉折點:

需求面風險: - 經濟衰退減少企業/消費者AI支出 - AI模型改善邊際報酬遞減(當前模型「夠好」) - 監管限制AI開發/部署

技術訊號: - 創新高後高量反轉K線 - 多時間框架均線下破 - 廣度惡化:更少股票參與反彈

基本面訊號: - 儘管營收成長但盈餘成長減速(毛利壓縮) - 領先公司調降財測 - 大型科技減少資本支出公告

關鍵洞察:轉折可能發生時無明顯新聞催化劑。待主流媒體識別「市場為何下跌」時,跌勢多已完成。技術訊號警覺性比新聞流更重要。

更廣泛市場背景

輪動指標監控:

  1. 修正期間落後產業不跌:訊號資金開始輪動
  2. 非AI產業營收/獲利加速成長:基本面改善驅動興趣
  3. 估值正常化:隨AI股變昂貴,價值投資者輪動它處

當前狀態:輪動為時過早。AI/科技領導地位保持主導。傳統產業(消費品、紡織、服裝)困於下降趨勢。在這些產業顯示技術/基本面改善前,堅守贏家合理。

投資課程推廣背景

主講人推廣台美股配置課程(早鳥優惠2小時後結束,代碼USA88額外折扣),強調:

課程價值主張: - 結合台灣與美國市場涵蓋以分散風險 - 匯率與產業風險雙重緩解 - 識別成長股的系統化方法 - 技術與基本面分析整合 - 私密學員社群有講師參與 - 工具包含券商比較與資產配置計算機

教育哲學:無固定公式存在,但多元分析工具增加勝率。課程提供系統框架而非規範規則。

關鍵引言

“長期來看,資產真的還是要放在股市當中。如果說可能都完完全全沒有投資,真的會跟資產成長的速度是脫節的。” - 「長期而言,資產確實需要投入股市。若完全不投資,真的會落後資產成長速度。」

“每當這種形勢一出來之後,就代表說其實股市短線的漲幅真的是已經有一點太過誇張了。” - 「每當這類預測圖出現,代表短線市場漲幅確實已有些過度。」

“國安基金真的是很深準的,只要喊國安基金進場的時候,幾乎都是相對低點。” - 「國安基金極為準確 - 每當宣布進場,幾乎總是相對低點。」

“台灣的進口金額的成長率,是遠遠超過其他的國家…基本上台灣他的成長可以說是一枝獨秀。” - 「台灣的進口額成長率遠超其他國家…台灣成長基本上可說一枝獨秀。」

“既然泡沫來,即便質疑泡沫但是打不贏就加入,就是真的等到泡沫破裂我們再退場。” - 「既然泡沫來臨,即便質疑它但打不贏就加入 - 真正等到泡沫破裂再退場。」

“這個永動機就越滾越大,因為其實等於說這個錢,其實都一直以來,都是在這幾家公司換來換去。” - 「此永動機越滾越大,因實質上這些資金持續在這些公司間循環。」

“我們永遠不知道市場最瘋狂的一個點位到底會到什麼時候。” - 「我們永遠不知市場最瘋狂時點究竟何時到來。」

“你在4月對等關稅的時候你是受恐慌的時候做進場,其實你當下進場的時候,你是找不到什麼利多。” - 「四月對等關稅恐慌進場時,當下進場你找不到任何利多。」

“未來可能真的股市開始要陷入一個漲多修正的時候,你一定也不會找到什麼利空。” - 「未來市場真正開始漲多修正時,你肯定也找不到任何利空催化劑。」

“比起自己虧錢,我看見大家虧錢我是更難過。” - 「相較自己虧損,看見大家虧錢我更難過。」


主題標籤 / Tags

#AI泡沫 #AIBubble #台股 #TaiwanStocks #美股 #USStocks #NVIDIA #TSMC #台積電 #國安基金 #NationalStabilizationFund #科技七巨頭 #Magnificent7 #算力 #Compute #投資策略 #InvestmentStrategy #技術分析 #TechnicalAnalysis #市場警訊 #MarketWarnings #摩根大通 #JPMorgan #川普關稅 #TrumpTariffs #半導體 #Semiconductors #OpenAI #Oracle #AMD #CoreWeave #IrisEnergy #特斯拉 #Tesla #黃金 #Gold #資產配置 #AssetAllocation