AI는 거품 속에 있는가?

Is AI in a Bubble? THE DAILY --- H.E.A.T. Torching Wall Street’s Obsolete Playbook. Delivered straight to your inbox 5 days a week. Subscribe Home Posts Is AI in a Bubble? Is AI in a Bubble? Oct 1, 2025 Wall Street’s 60/40 formula was born in 1952 — the same year as the first credit card. A lot has changed since. That’s why we created a new approach — The --- H.E.A.T. Formula — to empower investors to spot opportunities, think independently, make smarter (often contrarian) moves, and build real wealth. — # (# ) Table of Contents 🔥 Here’s What’s Happening Now 🧠 Is AI in a Bubble? The punchline What just changed (and why this week mattered) Five bubble signals I’m watching Four reasons it’s not 1999 What could break (and what to watch) Positioning: How I’d play it now (HEAT lens) Risk dashboard My view 📈 Stock Corner 📬 --- In Case You Missed It 🤝 Before You Go Some Ways I Can Help 🔥 Here’s What’s Happening Now Market selling off a bit this morning, probably on the government shut down. I think you buy this dip. Meanwhile, both crypto and precious metals moving up. As I continue to say, it’s not crypto OR precious metals, it’s crypto AND precious metals…. Rareview 2X Bull Cryptocurrency & Precious Metals ETF The Rareview Total Return Bond ETF (the “Fund”) seeks to provide a total return through the combination of current income and capital appreciation. rareviewcapital.com/2x-bull-cryptocurrency-precious-metals-etf I continue to say NVDA is a must own as the obvious winner of the AI trade and it broke out a bit yesterday…. I also continue to say that you need to broaden out your AI exposure (more on AI below) as the easy money has probably been made in NVDA. This is a topic I will probably cover at some point….. How long can Nvidia stay ahead of Chinese competition? The country has been a critical market for the AI chipmaker but the company is entangled in geopolitical tensions and is watching the rise of new rivals www.ft.com/content/c24a9b6c-1664-4e46-affb-0c0dc16e3c4a Typically, I like to buy China when nobody wants it, but this move up in Chinese stocks feels different (I know, famous last words). Do think you own the US AND the Chinese AI names here….. Drone stocks continued their run up yesterday…. Russia named a 'permanent threat' to Europe's security as leaders discuss a drone wall EU leaders are set to debate new measures, including the idea of a "drone wall" in the eastern part of Europe, close to the Russian border. www.cnbc.com/2025/10/01/russia-has-become-a-permanent-security-threat-to-europe-luxembourg-pm-says.html KTOS was our stock of the day on Monday…. Positioning yourself in the top themes is not that hard :) On the flipside, Argentina is not doing well. I was optimistic this could turn into a Trump trade, and it still might, but this is such a target rich environment no reason to waste resources in something in a clear downtrend….. Argentine assets hit by fresh selling amid worries over US bailout Peso, stocks and bonds all fall as investors fret over lack of details from Washington on rescue plan www.ft.com/content/6333a667-bdde-4e74-9dec-37afcb4db89b I use a moving average crossing system on my thematic basket to keep me out of downtrends. I used to be 100% systematic, I am not anymore, what I do is too complex for any system or set of systems. I do have an RSI sleeve that is purely systematic, but my moving average crossing system is more a guideline, I will override it from time to time when I think I know more than the market. Usually, that does not end well. 🧠 Is AI in a Bubble? I continue to think that as long as AI Capex stays at these levels you want to be buying stocks all along the AI spectrum. At some point though, we will hit a 2000 type event, that could be next week, next year, or in 5 years. That’s the problem with trying to predict bubbles….. The punchline We’re not in a uniform, 1999-style bubble. We are seeing pockets of speculative excess—especially where AI capex is being debt-financed , where revenues lag build-outs , and where business models depend on circular flows (GPU vendor → AI lab → GPU cloud/lessor → back to vendor). Expect rotations, sharp drawdowns, and failures at the edges—but not a systemic collapse of the core. Think selective bubbles inside a durable, decade-long compute/power super-cycle. What just changed (and why this week mattered) Oracle sold $18B of bonds to fund AI data centers—an “unusual” jumbo deal that marks a shift from cash-flow-funded capex at the hyperscalers to debt-funded arms race dynamics. That’s the inflection reviving “bubble” talk. Bloomberg+1 Analysts flagged the risk of “dark fiber 2.0” —overbuilding capacity ahead of monetization; plus power constraints, model-improvement plateaus, and circular financing. MarketWatch Another MarketWatch take: the AI trade increasingly hinges on OpenAI . If the keystone stumbles, the tentacles (vendors, infra, industrials riding the build-out) feel it—especially with more capex leaning on debt. MarketWatch Five bubble signals I’m watching Financing mix flips to debt: Oracle’s $18B deal isn’t fatal—but it’s a regime shift . The more capex is debt-driven, the lower the tolerance for ROI delays. Bloomberg Breadth & concentration: Since late 2022, ~41 AI-linked stocks drove ~75% of the S&P’s advance —healthy while earnings follow, dangerous if they don’t. Morningstar Circular revenues: Hyperscalers are Nvidia’s biggest customers; Nvidia in turn is a giant customer for GPU clouds (e.g., CoreWeave) and server assemblers. When the same dollars loop, one break in the chain bites everyone. MarketWatch Power constraints: “Time-to-power” (grid hookups) becomes the gating KPI. If hookups slip, you can get stranded data-center capacity —the essence of the “dark fiber” analogy. MarketWatch Momentum extremes & stall-outs: AIQ’s RSI >80 earlier this week and NVDA going sideways ~2 months are classic “froth meets digestion” tells. MarketWatch Four reasons it’s not 1999 Cash flow at the core: MSFT/GOOGL/AMZN/META still produce extraordinary FCF (even if pressured by capex). Many AI build-outs are anchored by users with real budgets. MarketWatch Real productivity curves: Unlike 1999 eyeballs, inference/training spend is attached to measurable outputs (code gen, copilots, search ads lift, industrial automation). Capex has option value: Even if near-term utilization lags, power-adjacent assets (land + substations + transformers) retain value. Scenario math from the bears isn’t apocalyptic: Barclays estimates a 20% drop in data-center capex would hit S&P EPS by only ~3-4%—a correction, not a crash. Barron's What could break (and what to watch) Keystone risk: A stumble or slowdown at OpenAI (funding, model cadence, legal/policy) would ripple through vendors and “AI beta” equities. Track deal flow, debt access, and model releases . MarketWatch China substitution risk: A Huawei/SMIC breakthrough reduces NVDA’s pricing power and U.S. export leverage—one of JPM’s top macro risks tied to AI concentration. MarketWatch Power pricing & politics: Rising utility bills create consumer/political blowback —a margin headwind and project-delay risk. MarketWatch Positioning: How I’d play it now (HEAT lens) Tactical index hedges into capex headlines / policy weeks (CPI/Fed): puts or put spreads on AI-heavy baskets (e.g., AIQ , semis) sized to offset 25–35% of your AI beta. MarketWatch Pair trades : short weaker-moat GPU lessors / server assemblers vs. long power enablers (below). (Name specifics offline if you want tighter ideas.) Power & grid bottlenecks: Long the “time-to-power” winners— GEV (GE Vernova), ETN , HUBB , select utilities with available capacity and constructive rate cases; WMB/KMI for nat-gas throughput; uranium/SMR option-ality (e.g., CCJ , selective SMR exposure). AI application tolls with monetization line-of-sight: hyperscalers ( MSFT, AMZN, GOOGL ) and platforms turning copilots into ARR, but fade melt-ups where revenue lags capex. Event-driven 0DTE overlays on data-center/macro prints monetize intraday vol without sleeping with it. Selective distressed optionality : if we get a “dark fiber” scare, quality data-center REITs ( EQIX/DLR ) with power access on sale. Persistent AI-power super-cycle (your JUCE theme), not a straight line up. Position to own the bottlenecks , rent the story stocks. Risk dashboard Debt-funded AI capex share (rising = riskier). Bloomberg Data-center utilization vs. capacity additions (watch for widening gaps). Grid interconnection queue times in key nodes (Mid-Atlantic, Texas, Southeast). Unit economics : gross margin lift from AI products at MSFT/GOOGL/AMZN vs. capex growth. Breadth : % of S&P returns from top 10 AI names; NVDA trend vs. revenue cadence. Morningstar My view Core AI is real and durable. The compute/power build-out is the next great capex cycle. Edges are frothy. Expect boom-bust dynamics among GPU lessors, server assemblers, and anyone reliant on circular revenue. Portfolio stance: Barbell quality cash-flow AI platforms + power/grid enablers , with disciplined hedges and a trader’s mindset on speculative AI beta. MarketWatch (Adinolfi): AI bubble talk, Oracle debt, circular financing & “dark fiber” concerns. MarketWatch MarketWatch (Nguyen): AI trade hinging on OpenAI; debt-driven capex & power-grid “time-to-power” risk. MarketWatch Bloomberg/WSJ: Oracle’s $18B bond sale details and demand. Bloomberg+1 Morningstar reprint: 41 AI stocks → 75% of S&P’s advance since ChatGPT launch. Morningstar Barron’s on Barclays: 20% DCAPEX drop ≈ 3–4% S&P EPS hit. Barron's 📈 Stock Corner Today’s stock is Tempus AI (TEM)…. Jensen Huang and Marc Andreesson have both said that they think healthcare will see the biggest impact from AI. Tempus has been my favorite name, and Nancy Pelosi’s also. Trump’s announcement yesterday about using AI to cure cancer could be another positive for this stock. From Chat GPT….. My best guess: impact magnitude & timeline If I were modeling this: Short-term (weeks to months): modest positive bump in TEM’s valuation sentiment. Some speculative money will flow in. Medium term (6-24 months): real opportunity, but only if TEM lands one or more of the new grants / partnerships / trials tied to pediatric oncology or AI oncology infrastructure. Long term (>2 years): meaningful upside if TEM can use this as a lever to accelerate its growth into pediatric cancer diagnostics and AI oncology platforms, improve margins, and scale revenue. But execution is key. In other words: this announcement does not guarantee success for TEM, but it increases the optionality. The wind is at its sails, but it must steer well. 📬 --- In Case You Missed It Not sure why they had to do a voice over on me but great podcast anyway…. 🤝 Before You Go Some Ways I Can Help ETFs : The Antidote to Wall Street Inside HEAT: Our Monthly Live Call on What Wall Street Doesn’t Want You To Know Financial HEAT Podcast https://www.youtube.com/@TuttleCap Freedom from the Wall Street Hypocrisy Tuttle Wealth Management: Your Wealth Unleashed Advanced HEAT Insights: Matt’s Inner Circle, Your Financial Edge All rights reserved. Keep Reading No posts found View more caret-right 0