Source: YouTube URL: https://www.youtube.com/watch?v=TSllcwvKXP4 Date: 2025-10-11 Duration: 52:10
Despite market euphoria with Taiwan stocks approaching 30,000 points and tech stocks hitting new highs, JP Morgan’s CEO warns of a potential bubble burst within 6-24 months. The AI sector shows classic bubble characteristics with 60% of VC funding flowing exclusively to AI companies. However, Taiwan’s economy benefits massively from AI chip exports (井噴式成長), with GDP growth nearing 6% led by semiconductor giants. Key investment themes remain AI infrastructure (NVDA, TSM), cloud computing power providers (NBIS, CRWF, IREN), and application giants (MSFT, GOOGL, META). Technical caution advised: watch for high-volume bearish engulfing candles as exit signals.
“既然泡沫來，即便質疑泡沫但是打不贏就加入 - 等到泡沫破裂我們再退場”
Translation: “Since the bubble is here, even if you doubt it, if you can’t beat it, join it - exit when it actually bursts”
Rationale: - Impossible to predict exact tops (see dot-com 1998 vs. 2000) - Participate in wealth creation during euphoria - Preserve capital by exiting when clear reversal occurs - Re-enter at lower levels after washout
Execution: 1. Maintain exposure to strong momentum stocks (AI leaders) 2. Gradually raise cash as euphoria intensifies 3. Watch for technical breakdown signals (high-volume reversals) 4. Exit quickly when warnings appear 5. Wait patiently for panic bottoms to reload
Despite bubble warnings, several structural factors support continued upside:
1. Earnings Growth Divergence - Magnificent 7 tech giants: 14% earnings growth (Q3 2025) - S&P 493 (rest of index): 4% earnings growth - Analysts consistently underestimate Mag 7, overestimate others - pattern likely continues - Gap widening rather than narrowing contrary to annual predictions
2. AI Infrastructure Build-out - Companies cannot afford to lag in AI capabilities - existential competition - Compute power constraints forcing continued aggressive capex - Long-term theme spanning multiple years, not quarters - NVIDIA’s ecosystem dominance provides pricing power and visibility
3. Taiwan’s Structural Advantage - TSM monopoly on advanced nodes (3nm, 2nm) ensures continued dominance - Geography provides strategic value - US needs Taiwan semiconductor capacity - AI chip demand far exceeds supply - pricing power intact - Export data shows accelerating rather than decelerating growth
4. Valuation Not Extended (Except Tesla) - NVDA P/E 50x on 40% growth = PEG 1.25 (cheap by growth stock standards) - MSFT P/E 35x on 20% growth = PEG 1.75 (reasonable) - TSLA P/E 200x on 35% growth = PEG 5.7 (dreams priced in, needs robotaxi/Optimus success) - Compared to historical bubbles, valuations less extreme outside select names
1. Concentration Risk - 60% of VC funding to AI = capital misallocation inevitability - When sector rotates, violent unwind likely given crowding - No Plan B for investors - everyone owns same stocks
2. Circular Financing Concerns - OpenAI ↔︎ Oracle ↔︎ NVIDIA ↔︎ AMD investment loop creates phantom demand - Similar to 2000 dot-com telecom equipment circular purchasing - When one link breaks, entire chain vulnerable
3. Profit Realization Unclear - Computing power providers showing revenue but margins uncertain - Oracle lost $100M renting NVIDIA Blackwell chips - early adopter losses - Path to profitability for many AI startups questionable
4. Policy/Macro Risks - Trump policy uncertainty (關稅朝令夕改) creates volatility - Taiwan tariff negotiations ongoing - negative outcome possible - Fed policy still restrictive - higher-for-longer rates stress valuations - China competition in AI intensifying (DeepSeek, other models)
Historical Patterns - Past 6 months: S&P up 35%+ from lows - Only 6 instances in history with such rapid rebounds (1975, 1983, 1996, 2025) - Following year typically gains 43% average - However, rapid gains often lead to consolidation/pullback before continuing
Implications: - Strong probability of continued uptrend into 2026 - Near-term (weeks-months): increased volatility and potential 10-15% correction - Don’t expect same blistering pace to continue linearly - Healthy consolidation would create better entry points
Currently Strong: - AI chips (NVDA, AMD, TSM, ASML) - Cloud/AI applications (MSFT, GOOGL, AMZN, META) - Computing power rental (NBIS, CRWF, IREN) - Defense/aerospace (select names, though Kratos showing weakness)
Currently Weak: - Traditional industries unconnected to AI - European stocks (near-zero growth economies) - China exposure plays (geopolitical concerns) - Real estate (higher-for-longer rates)
Rotation Signals to Watch: 1. Weak stocks stop going down during market selloffs - accumulation sign 2. Revenue/earnings inflection positive - fundamental improvement 3. Relative strength vs. market improves - capital flows changing 4. Currently premature - stay with momentum until clear signals
Speaker’s Approach (值得學習): - Maintains 10-20% cash minimum for psychological comfort and opportunity - Uses both fundamental (valuation apps, revenue growth) and technical (moving averages, volume patterns) analysis - Prefers buying during panic vs. euphoria - higher probability setups - Willing to switch stocks/sectors freely vs. job constraints (投資的優勢) - Long-term holdings: NVDA, TSM, Ferrari (RACE), other quality names - Recent positioning: Adding AI computing power providers, maintaining chip manufacturer core
Risk Management Emphasis: - “比起自己虧錢，我看見大家虧錢我是更難過” - cares more about followers losing money than personal losses - Warns against chasing at current elevated levels for new entrants - Advocates gradual profit-taking during euphoria rather than trying to ride to exact top - Stresses importance of having cash when real opportunities emerge
Disappointment: Affordable vehicle strategy shift - Originally promised $25,000 true budget model - Now offering 10% discount on Model 3 with features removed (閹割版) - Self-driving features eliminated in budget version - Unlikely to significantly boost sales volume near-term
Long-term Thesis Depends On: - Full Self-Driving (FSD) subscription revenue - Robotaxi deployment and scaling - Optimus humanoid robot commercialization - Without these, current 200x P/E unsustainable
Market Reaction: - Recent new highs followed by pullback - Reflects mixed sentiment on affordable car strategy - Long-term believers vs. valuation skeptics debate continues
“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 prediction charts appear, it means the short-term rally has become somewhat excessive”
“國安基金喊進場的時候，幾乎都是相對低點…反而是國安基金真的出場之後，往往都代表說台股之後還有一段漲幅”
“When the National Stabilization Fund announces entry, it’s almost always a relative low point… conversely, when they actually exit, it often means Taiwan stocks have another leg up ahead”
“你怎麼可能這個佔比直接到達100%？總是會有物極必反的一天”
“How could this proportion reach 100%? There will always be a day when extremes reverse”
“既然泡沫來，即便質疑泡沫但是打不贏就加入，等到泡沫真的破裂我們再退場”
“Since the bubble is here, even if you doubt it, if you can’t beat it join it - exit when it actually bursts”
“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 will reach its most insane point”
“當股市修正的時候，你一定也不會找到什麼利空…通常等到你發現消息的時候，可能股市都已經跌完”
“When the market corrects, you won’t find any obvious negative news… usually by the time you discover the reason, the market has already finished falling”
“比起自己虧錢，我看見大家虧錢我是更難過”
“More than my own losses, seeing everyone else lose money makes me sadder”
“買不起法拉利你就先買法拉利的股票，買不起愛馬仕你就先買愛馬仕的股票”
“If you can’t afford a Ferrari, buy Ferrari stock first; if you can’t afford Hermès, buy Hermès stock first”
儘管市場氣氛熱絡，台股逼近 30,000 點、科技股屢創新高，摩根大通執行長警告 6-24 個月內可能出現泡沫破裂。AI 產業呈現典型泡沫特徵，60% 創投資金獨鍾 AI 公司。然而台灣經濟因 AI 晶片出口井噴式成長而大幅受益，GDP 成長率接近 6%，由半導體巨頭領軍。關鍵投資主題仍為 AI 基礎建設（輝達、台積電）、算力供應商（NBIS、CRWF、IREN）、應用巨頭（微軟、谷歌、Meta）。技術面警示：留意高量黑K開高走低作為退場訊號。
“既然泡沫來，即便質疑泡沫但是打不贏就加入 - 等到泡沫破裂我們再退場”
理由： - 無法預測確切高點（參見網路泡沫 1998 vs. 2000） - 在狂熱期參與財富創造 - 出現明確反轉時保全資本 - 洗盤後於更低水位重新進場
執行方式： 1. 維持強勢股曝險（AI 領導股） 2. 狂熱加劇時逐漸提高現金 3. 留意技術破位訊號（高量反轉） 4. 警訊出現時迅速退場 5. 耐心等待恐慌底部重新布局
儘管有泡沫警告，幾項結構性因素支持持續上漲：
1. 獲利成長分歧 - 科技七巨頭：14% 獲利成長（2025 Q3） - 標普 493（指數其餘成分）：4% 獲利成長 - 分析師持續低估七巨頭、高估其他股票，型態可能延續 - 差距擴大而非縮小，有違年度預測
2. AI 基礎建設建置 - 企業不能在 AI 能力上落後，屬生存競爭 - 算力限制迫使持續積極資本支出 - 長期主題橫跨多年而非數季 - 輝達生態系統主導地位提供定價權及能見度
3. 台灣結構優勢 - 台積電壟斷先進製程（3nm、2nm）確保持續主導 - 地理戰略價值，美國需要台灣半導體產能 - AI 晶片需求遠超供給，定價權完整 - 出口數據顯示加速而非減速成長
4. 估值未過度延伸（特斯拉除外） - 輝達本益比 50 倍、成長 40% = PEG 1.25（以成長股標準屬便宜） - 微軟本益比 35 倍、成長 20% = PEG 1.75（合理） - 特斯拉本益比 200 倍、成長 35% = PEG 5.7（夢想已計價，需 robotaxi/Optimus 成功） - 對比歷史泡沫，除特定個股外估值不算極端
1. 集中風險 - 60% 創投資金投向 AI = 資本錯置不可避免 - 板塊輪動時，過度擁擠恐暴力回吐 - 投資人無 B 計畫，人人持有相同股票
2. 循環融資疑慮 - OpenAI ↔︎ Oracle ↔︎ 輝達 ↔︎ AMD 投資迴圈創造虛幻需求 - 類似 2000 年網路泡沫電信設備循環採購 - 任一環節斷裂，整條鏈脆弱
3. 獲利實現不明 - 算力供應商營收顯現但毛利不確定 - 甲骨文租輝達 Blackwell 晶片虧損 1 億美元，早期採用者虧損 - 許多 AI 新創獲利路徑存疑
4. 政策/總經風險 - 川普政策不確定性（關稅朝令夕改）創造波動 - 台灣關稅談判進行中，負面結果可能 - Fed 政策仍緊縮，利率長期維持高檔壓力估值 - 中國 AI 競爭加劇（DeepSeek 等模型）
歷史型態 - 過去 6 個月：標普從低點反彈超過 35% - 歷史僅 6 次出現如此快速反彈（1975、1983、1996、2025） - 隔年通常平均上漲 43% - 然而快速上漲常導致盤整/回檔後才繼續
啟示： - 延續至 2026 年上漲趨勢機率高 - 近期（數週至數月）：波動加劇、可能 10-15% 修正 - 別預期相同飆速線性延續 - 健康盤整將創造更佳進場點
目前強勢： - AI 晶片（輝達、AMD、台積電、艾司摩爾） - 雲端/AI 應用（微軟、谷歌、亞馬遜、Meta） - 算力租賃（NBIS、CRWF、IREN） - 國防/航太（部分個股，但 Kratos 顯示弱勢）
目前弱勢： - 與 AI 無關傳統產業 - 歐洲股票（接近零成長經濟體） - 中國曝險標的（地緣政治疑慮） - 房地產（利率長期維持高檔）
輪動訊號留意： 1. 弱勢股大盤回檔時不再下跌 - 吸籌跡象 2. 營收/獲利出現正向轉折 - 基本面改善 3. 相對大盤強度改善 - 資金流向改變 4. 目前尚早，維持動能至明確訊號
講者作法（值得學習）： - 維持最低 10-20% 現金以保心理舒適度及機會 - 同時使用基本面（估值 APP、營收成長）及技術面（均線、量能型態）分析 - 偏好恐慌時買入勝過狂熱時，勝率設定較高 - 願意自由切換股票/類股相對工作限制（投資的優勢） - 長期持股：輝達、台積電、法拉利（RACE）及其他優質標的 - 近期配置：增加 AI 算力供應商，維持晶片製造商核心
風險管理強調： - “比起自己虧錢，我看見大家虧錢我是更難過” - 關心追隨者損失勝過個人損失 - 警告新進投資人於目前高位追逐 - 倡導狂熱時漸進獲利了結而非試圖騎到確切高點 - 強調真正機會出現時手中有現金的重要性
失望：平價車策略轉變 - 原本承諾 25,000 美元真正預算車款 - 現提供 Model 3 九折、移除功能版（閹割版） - 預算版取消自駕功能 - 近期不太可能大幅提升銷量
長期論點取決於： - 完全自動駕駛（FSD）訂閱收入 - Robotaxi 部署及規模化 - Optimus 人形機器人商業化 - 若無上述，目前 200 倍本益比難以支撐
市場反應： - 近期創新高後拉回 - 反映平價車策略混合情緒 - 長期信仰者與估值懷疑者辯論持續
“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”
“每當預測圖表出現，代表短期漲勢已變得有些過度”
“國安基金喊進場的時候，幾乎都是相對低點…反而是國安基金真的出場之後，往往都代表說台股之後還有一段漲幅”
“國安基金宣布進場時幾乎總是相對低點…相反地，當他們實際退場時，常代表台股後續還有一段漲幅”
“你怎麼可能這個佔比直接到達100%？總是會有物極必反的一天”
“這個比例怎麼可能達到 100%？總會有極端反轉的一天”
“既然泡沫來，即便質疑泡沫但是打不贏就加入，等到泡沫真的破裂我們再退場”
“既然泡沫來了，即使質疑泡沫但打不贏就加入，等到泡沫實際破裂再退場”
“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”
“我們永遠不知道市場會在何時達到最瘋狂的點位”
“當股市修正的時候，你一定也不會找到什麼利空…通常等到你發現消息的時候，可能股市都已經跌完”
“當市場修正時，你不會找到任何明顯負面消息…通常當你發現原因時，市場已經跌完了”
“比起自己虧錢，我看見大家虧錢我是更難過”
“相較於自己虧損，看到大家虧損讓我更難過”
“買不起法拉利你就先買法拉利的股票，買不起愛馬仕你就先買愛馬仕的股票”
“如果買不起法拉利就先買法拉利股票，如果買不起愛馬仕就先買愛馬仕股票”
#AI泡沫 #美股投資 #台股分析 #輝達 #台積電 #算力供應商 #投資策略 #技術分析 #國安基金 #泡沫警告 #AIBubble #USStocks #TaiwanStocks #NVIDIA #TSMC #InvestmentStrategy #TechnicalAnalysis #BubbleWarning #ComputingPower #MarketAnalysis