AI의 전력 문제: AI가 에너지 인프라가 현금화할 수 없는 수표를 작성하고 있습니까?

AI’s Power Problem: Is AI Writing Checks That Energy Infrastructure Can’t Cash? THE DAILY --- H.E.A.T. Torching Wall Street’s Obsolete Playbook. Delivered straight to your inbox 5 days a week. Subscribe Home Posts AI’s Power Problem: Is AI Writing Checks That Energy Infrastructure Can’t Cash? AI’s Power Problem: Is AI Writing Checks That Energy Infrastructure Can’t Cash? Sep 25, 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 ⚡️ AI’s Power Problem: Is AI Writing Checks That E … Heading 2 The demand reality The bottlenecks (what actually breaks first) Is this a 2000–2002 risk if power slows capex? The capacity playbook (how the grid does cash the … Winners & losers (12–36 months; ratings 1– • 10) 🏆 First-order winners — the power stack ⚙️ Second-order winners — compute remains king 🧪 Third-order winners — platforms that monetize w … ❌ Likely losers / “show me” Positioning 📈 Stock Corner 📬 --- In Case You Missed It 🤝 Before You Go Some Ways I Can Help 🔥 Here’s What’s Happening Now Maybe we should have sold Rosh Hashanah :) Two down days in a row, but at this point we still see this as a healthy pullback in an uptrend. So far this morning China is still looking strong, as is precious metals, crypto looks to be taking a hit. Could be another opportunity to buy the dip on crypto….. BITK: Tuttle Capital IBIT 0DTE Covered Call ETF | Tuttle Capital Income Blast ETFs The primary investment objective of the Tuttle Capital IBIT 0DTE Covered Call ETF (the “Fund”) is to seek current income. The Fund’s secondary investment objective is to provide capital appreciation. www.incomeblastetfs.com/etf/bitk Continued to see a rotation into energy yesterday, hence our discussion below. I decided to play it with EQT and TPL as I talked about the past couple of days. I don’t think I’ve talked about coal for a bit, but that’s another Trump trade. ❝ “I have a little standing order in the White House. Never use the word ‘coal.’ Only use the words ‘clean, beautiful coal.’ Sounds much better, doesn’t it?” Donald Trump I took profits in BTU as it looks like an AI stock….. And I bought CNR, which now also looks like an AI stock…… I was about to blow Argentina out of our ETF model portfolio, but then it became a Trump trade as well. Really like ARGT here…… Bessent Says U.S. Is in Talks With Argentina for $20 Billion in Aid Latin America is a region where President Trump is picking winners and losers www.wsj.com/world/americas/bessent-says-u-s-talking-with-argentina-about-20-billion-in-aid-0a821d79?mod=finance_lead_pos1 Be interesting to see if we get a real rotation out of tech and into real assets and energy, or just a head fake. Nasdaq in the red again so far this morning, but only slightly. ⚡️ AI’s Power Problem: Is AI Writing Checks That Energy Infrastructure Can’t Cash? Higher Electricity Prices Aren’t All Good for Utilities Utility bills are catching the attention of politicians. That should concern utility investors. www.wsj.com/finance/investing/higher-electricity-prices-arent-all-good-for-utilities-23113b91?mod=finance_lead_story Short answer: locally, yes—for now; nationally, not yet . The bottleneck isn’t just “enough electrons,” it’s the where/when/how : siting, interconnection queues , transmission , transformers , thermal management , and firm, carbon-free 24/7 power . In the near term, AI is outrunning node-level infrastructure (NoVa, Bay Area, Phoenix, Columbus, Dallas–Fort Worth), forcing data centers to bid up scarce megawatts and scramble for long-dated PPAs. But with the right mix— restarts/uprates of existing nuclear , targeted SMR pilots, gas-plus-storage , geothermal , aggressive HV/MV build-outs , and HVDC transmission —the U.S. can cash the check over a 3–7 year horizon. The risk isn’t “we run out of power”; the risk is capex cadence slows because power-at-the-right-node lags deployments. The demand reality Data-center load is going parabolic : credible ranges call for +150–200% U.S. DC power by ~2030, with hyperscalers signing multi-GW PPAs and siting gigawatt-scale campuses . Training vs. inference split : training wants massive, contiguous, high-reliability power; inference wants latency-proximity to users—so even if there’s “enough” power nationally, congested metros still break. Thermal density : advanced GPU clusters require liquid cooling and substation upgrades; many legacy sites can’t be retrofitted fast enough. The bottlenecks (what actually breaks first) Interconnection : grid queues in constrained RTOs (PJM, CAISO, ERCOT pockets) are 3–5 years ; substation builds and protection schemes add months to years. Transmission : moving bulk power from where it’s cheap (nuclear, hydro, wind belts) to where the DCs are is multi-year HV/HVDC work with permitting/NIMBY friction. Equipment lead times : large power transformers (LPTs) , switchgear, high-voltage cable, chillers, and liquid-cooling skids—lead times 24–36 months in tight markets. Firm, carbon-free supply : wind/solar are essential, but intermittency + capacity credit limitations mean you still need nuclear, geothermal, hydro, gas-with-storage/CCS for 24/7 SLAs. Water & heat : water rights and discharge constraints are real; immersion or direct-to-chip liquid cooling eases this but requires purpose-built plants. Is this a 2000–2002 risk if power slows capex? Not at the core —training is a real bottleneck and application value is emerging—but the periphery is exposed . If power procurement or utilization stumbles, expect a mini-2000 in neo-clouds , second-tier data-center builders, and thin-margin “AI apps.” The winners remain the bottlenecks (compute, HBM, optics) and the power stack that removes constraints. The capacity playbook (how the grid does cash the check) Now (2025–2030): Fission restarts/uprates (e.g., Three Mile Island, Palisades; uprates at existing nuclear fleets). Utility-scale gas with battery peakers for ramping; geothermal pilots where resource allows. Long-dated PPAs with utilities; behind-the-meter generation for mega-campuses. Liquid cooling standardization; substation & transformer build-outs; HV/MV upgrades. Select HVDC projects to move bulk power; co-location near generation or industrial load-pockets. Next (2030–2035): First SMR deployments (Natrium/advanced PWRs) at utility sites; maturing enhanced geothermal (EGS) . Expanded HVDC backbones and grid-forming inverters for stability with high inverter penetration. Beyond (2035+): Fusion optionality if demos hit steady-state; broader SMR adoption; grid becomes compute-aware (orchestration across regions). Winners & losers (12–36 months; ratings 1– • 10) 🏆 First-order winners — the power stack Nuclear operators: Constellation (CEG) 9/10 , NextEra (NEE) 8/10 — restarting/optimizing fleets; hyperscaler PPAs de-risk cash flows. Fuel cycle: Centrus (LEU) 9/10 , Cameco (CCJ) 8/10 — enrichment & uranium are the new chokepoints if nuclear scales. Grid EPC & transmission: Quanta (PWR) 8.5/10 , MasTec (MTZ) 7.5/10 — interconnects, lines, substations. Electrical gear & transformers: Eaton (ETN) 8.5/10 , Schneider Electric (SU FP) 8.5/10 , GE Vernova (GEV) 8/10 , Powell (POWL) 8/10 , Hammond Power (HPS.A) 7.5/10 — switchgear, LPTs, MV gear. Thermal/liquid cooling & DC infra: Vertiv (VRT) 9/10 — the must-have for high-density clusters; Boyd (MOD) 7.5/10 . Geothermal / baseload renewables: Ormat (ORA) 7.5/10 — firm power where resource supports. ⚙️ Second-order winners — compute remains king Compute bottlenecks: NVIDIA (NVDA) 9/10 (training choke point), HBM : SK hynix/Micron (MU) 8–8.5/10 ; network/optics: Broadcom (AVGO) , Arista (ANET) , Credo (CRDO) 8–8.5/10 . Semicap tools: AMAT/LRCX/KLAC 8–8.5/10 — advanced packaging & capacity adds. 🧪 Third-order winners — platforms that monetize work Data/ops/workflow: ServiceNow (NOW) , Snowflake (SNOW) , Datadog (DDOG) 7.5–8/10 — if AI moves from sizzle to workflows (agents, copilots, RAG), they capture consumption + price/mix . ❌ Likely losers / “show me” Generic DC REITs without high-density liquid cooling 5.5–6/10 — stranded as hyperscalers demand >100 MW blocks with immersion. On-prem IT vendors (low AI attach) 5–6/10 — spend is consolidating into hyperscale. Pure-play intermittent renewables without storage 5/10 — still grow, but AI PPAs increasingly demand firm profiles. Cash-hungry AI infra startups 5–6/10 — sensitive to power procurement, utilization, and credit spreads. Positioning Own cyber (PANW/CRWD/ZS/CYBR) and nuclear operators (CEG/NEE) to hedge the two real constraints: data/security and energy . Overweight the power stack (nuclear/fuel cycle/EPC/transformers/liquid cooling) where scarcity pricing is strongest; pair with NVDA + HBM + optics . Select application platforms only where there’s measurable ROI (ticket deflection, dev productivity, sales lift) — not just model announcements. Themes to track: Interconnection/queue milestones and transformer lead times (call the pace of DC adds). PPA pricing & tenor (firm, carbon-free vs system mix). Utilization at hyperscalers/neo-clouds; RPO→billings→revenue cadence; credit spreads on AI-linked DC debt. AI isn’t a Ponzi; it’s a real computational revolution tangled in very physical constraints . In hotspots, AI is indeed writing checks today’s local infrastructure can’t cash—yet. But the U.S. has the toolkit to clear the backlog: restart nuclear, accelerate HV/MV builds, standardize liquid cooling, and wire the country with targeted HVDC . The investable alpha sits with the players that remove those constraints . Own the bottlenecks and the power stack , and treat everything downstream as a trade—funded by real, meter-read cash flows, not a prayer. The Investment Strategy Wall Street Hopes You Never Discover Tue, Sep 30, 2025 2:00 PM - 3:00 PM EDT -Why the 60/40 strategy is dead and what to do instead - How to use AI to uncover today and tomorrow's hottest themes - 4 unknown edges that still exist in today's market - How to set up your portfolio for asymmetrical returns - Little-known asset class that has limited risk and potentially unlimited returns - 4 ways to hedge your portfolio that don't include bonds Click Below to Register The Investment Strategy Wall Street Hopes You Never Discover register.gotowebinar.com/register/2237592374108794972 📈 Stock Corner Today’s stock is Freeport McMoran (FCX)…. I’ve been out of this for a while, but thinking about buying this dip, especially if we get an undercut and rally around $38 or at the 200 day. Risky play here….. ❝ On September 24, 2025, Freeport-McMoRan declared force majeure on contracted copper supplies from its Grasberg mine in Indonesia—the world’s second-largest source of the metal. This follows a fatal accident and ongoing search for missing workers, with production halted and Q3 sales guidance cut by 4%. Grasberg’s status as one of the largest and lowest-cost copper mines globally amplifies the market impact. Jefferies maintains a BUY rating on Freeport , trimming the price target from $48 to $46, and sees upside in copper miners broadly— Glencore, Lundin, Teck, First Quantum, and Anglo are all expected to outperform as supply tightens. Jefferies 📬 --- In Case You Missed It — # (# ) 🤝 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 Unshackled Advanced HEAT Insights: Matt’s Inner Circle, Your Financial Edge All rights reserved. Keep Reading No posts found View more caret-right 0