Back Research Notes Fed Policy Mistake vs. AI Disruptive Boom: Focus on the Forest, Not the Trees Published on September 21, 2025 By Jordi Visser In this week’s video , following the expected Fed rate cut, all four major U.S. equity benchmarks hit fresh highs together on Thursday for only the 26th time this century. Bond volatility is falling and credit spreads are at all-time lows, and futures imply multiple Fed cuts ahead. I dismiss “policy mistake” chatter as stubbornly trying to be right rather than focusing on making money with headline inflation ~3% and a K-shaped economy with a weak labor market. Retail sales and older-age consumption remain firm, global breadth is strong, and earnings-call indicators of “uncertainty” and inflation mentions are falling. Despite all of this positive news, institutional sentiment remains oddly bearish (AAII, U Michigan), which serves as contrarian fuel despite continued signs of retail euphoria. The core structural driver is AI. Oracle’s results underscore an inference demand shock while the NVIDIA–Intel and Tesla–Samsung tie-ups signal a transition toward edge/embodied AI where NPUs (on-device “brains”) matter. This will lift PMIs via a data-center construction boom but also colliding with physical bottlenecks within power availability, turbines, generators, transformers, cooling, DRAM/HDD lead times. Gas turbines may satisfy only a portion of high-case future AI power needs, pushing hybrid solutions (backup generation, batteries, grid upgrades; new nuclear mostly post-2030) to a tide that lifts many boats. I highlight generator makers (CAT, Cummins, Generac) and battery/storage names gaining on backlogs, expect a commodities upcycle led by copper, which should help select EMs (Brazil/Chile/Mexico). I maintain a structurally bullish view on Bitcoin into yearend amid low institutional allocation. Timestamps: (00:00–03:22) Markets: All four major U.S. equity benchmarks hit fresh highs together for the first time since Nov 2021; bond volatility and credit spreads are low: Fed cuts + strong economy = bullish risk setup. (03:56–06:52) Policy & inflation: Ignore “policy mistake” chatter; market pricing implies more cuts (high odds for Oct & Dec). Headline inflation ~3%; risk to 4% would mainly come from higher gasoline, otherwise cuts are justified to address a weakening jobs backdrop (K-shaped economy). (06:52–08:39) AI macro: Administration is AI-focused; AI brings future labor deflation. Retail sales and older-age consumption are strong; commercial construction momentum is soaring, especially data centers lifting PMIs despite manufacturing skepticism. (08:58–11:38) Housing & easing cycle: Mortgage spreads compressing; cutting near highs historically bullish. Global breadth strong (ACWI highs; revisions rising). Earnings call “uncertainty” and inflation mentions are falling; tariffs alone aren’t pushing CPI up. (11:38–13:31) Valuation & sentiment: U.S. “overvaluation” claims miss margins/ROE. Sentiment remains unusually bearish despite all-time highs (AAII bears elevated; Michigan sentiment low) = contrarian positive. (14:13–17:23) AI demand shock: Media frames “$3T AI bet” as cautious; meanwhile, Oracle results highlight inference demand. Author’s May 15 “Inference” basket up ~30% since vs. S&P ~13%—argues to follow demand signals, not headlines. (17:51–21:11) Next phase of AI: NVIDIA–Intel and Tesla–Samsung deals point to preparation for edge/embodiment era. Shift from GPU-heavy training → heterogeneous inference → on-device decision-making. NPUs (neural processing units) become critical “brains” for phones, robots, and robotaxis. (21:34–25:31) Supply chain bottlenecks: ASML (EUV) key for advanced chips; widespread shortages/bottlenecks across power and compute (DRAM, HDDs, transformers, turbines, cooling). Power availability now the #1 data-center siting constraint. (26:13–32:24) Power solutions: Gas turbines backlogs stretch years; likely meet only ~½ of high-case AI power needs, rest to be filled by hybrid solutions (generators, storage, grid upgrades; new nuclear post-2030). Generators surging (CAT, Cummins); Generac leaning into large-MW data-center backup. Batteries become essential for both inference scaling and embodiment. (32:24–41:39) Battery & commodities upcycle: China accelerating grid-scale storage; U.S. battery plays (e.g., NRGV, EOS) gaining on backlogs and new platforms. Expect broad commodities strength (copper, etc.); favors Brazil/Chile/Mexico equity exposure. Bitcoin remains structurally bullish amid low institutional allocation. Watch here