Back Research Notes Why AI’s Power Problem Will Rewrite Portfolio Construction for Traditional Investors Published on October 5, 2025 By Jordi Visser In this week’s video , I break down why September marked the S&P’s best in 15 years, why falling rate/FX/credit volatility plus narrowing mortgage spreads combined with more imminent Fed cuts (outside a recession, with double-digit earnings growth) remain equity-friendly. Beneath the surface, institutional sentiment remains cautious and worried about an “AI bubble” even as retail keeps buying, while labor indicators cool and the largest U.S. employers (Walmart, Amazon, Accenture) signal every job is being redesigned by AI. On the product side, Anthropic 4.5, Sora 2, and OpenAI Pulse mark the shift to proactive, agentic assistants, while Shopify/Etsy checkout inside ChatGPT telegraphs future revenue-share pressure on search and marketplace businesses. All of this AI pressure will intensify the inequality trend and force the government in an election year to continue focusing on growth. The market focus will not shift: the AI build-out’s binding constraint is now power , not GPUs. With gas-turbine and transformer shortages, and limited grid interconnects combined with the start of the massive build, alpha is shifting away from the mega-cap-dominated data-center build-out and toward utilities, grid equipment, storage, and flexible demand. I highlight how wind + solar + batteries and Bitcoin miners (as “virtual batteries”) can stabilize intermittency and improve project economics and why this favors small caps as hardware capex diffuses through the supply chain. For traditional investors anchored in large-cap growth, this emerging AI-power-capex cycle demands a new portfolio lens—one that reintroduces energy, infrastructure, and industrial exposure as key sources of alpha in the next market regime. Timestamps (00:00–02:10) Markets: September was the best in 15 years for the S&P 500; volatility falling (rates, FX, credit), mortgage spreads easing, housing gets a boost. Odds of Fed cuts are high for October and December, and rate cuts outside recessions have historically supported equities. (02:24–05:40) Sentiment & flows: Despite strong YTD winners, institutional skepticism persists (concerns about an “AI bubble”), while retail keeps buying broad equity ETFs. Micron is used as a case study of how top-down AI demand can outrun bottom-up analyst calls (e.g., 2026 HBM oversupply claims vs. multi-vertical exponential demand). (06:31–09:10) Labor cooling & AI transition: Multiple indicators (Conference Board job differential, quits rate, temp hiring, ADP, Challenger plans) point to weaker hiring. Walmart, Accenture, and Amazon say AI will change every job, major task redesign rather than immediate layoffs reinforcing the urgency for workers and firms to build an AI plan. (10:00–12:38) Education & skills: Confidence in the value of college continues to slide; argument that time is better spent learning AI tools. Frontier-model benchmarks show AI already matches/exceeds skilled human experts in ~50% of tasks, implying near-term disruption in knowledge work. (13:12–16:22) “Is this a bubble?”: Tech spending is going parabolic, but today’s dynamic differs from dot-com, fewer IPOs, faster revenue payback, and huge supply constraints (especially power). The “AI bubble” narrative misses the true bottleneck: energy and infrastructure, not speculation. (18:05–21:59) Product momentum to agents: Anthropic 4.5, OpenAI’s Sora 2, and Pulse signal the shift to proactive, agentic assistants. Shopify/Etsy checkout inside ChatGPT previews “agentic commerce” that could pressure Google, Amazon, and Meta as AI handles shopping directly. (27:06–31:59) Rotations & bottlenecks: Expect small-cap catch-up as AI build-out broadens beyond mega-caps. With GPUs less constrained, power becomes the choke point—shifting alpha to utilities, grid equipment, and energy infrastructure. (32:19–36:28) Flexibility > raw MWs: Fastest-to-deploy solutions (wind/solar + battery storage) and flexible demand are critical. Bitcoin miners act as “virtual batteries,” monetizing off-peak power and curtailing on spikes—improving renewable project economics and grid stability (ERCOT cited). (37:10–39:44) Security & reshoring: U.S.–China supply-chain realities and war-game ammo shortfalls imply accelerated U.S. manufacturing and PMI upswing. Government capital is flowing into critical minerals (lithium, rare earths), supporting the view that utilities + energy market caps could overtake consumer staples. (40:26–49:37) Investment takeaways: Rising global rate-cut cycle + improving PMIs favor small caps. Bitcoin ties the AI–energy nexus: power scarcity restrains Big Tech revenue timing while miners stabilize grids. Network effects from agents, stablecoins, and tokenization could propel digital assets amid seasonal tailwinds (Oct–Nov). Watch here