# The Permit Is the New AI Chokepoint: Texas, Data Centres and Power *URL:* https://millionaur.com/millionaur-2026-11-04-daily-analysis/ *Updated:* 2026-11-04T07:27:43+00:00 Last night two things happened within hours. AI suppliers reported demand above expectations again. Then voters elected a Texas governor who campaigned on stopping new data centre construction. Futures fell. Here is how I read it, and what the portfolio did. ## The demand side printed Advanced Micro Devices (AMD) reported revenue of 13.4 billion USD, up 45% from a year ago and above expectations. Data centre sales nearly doubled, and it guided the next quarter to about 17 billion, well above what analysts expected. Arista and Super Micro also beat and raised. Nothing in these numbers says AI demand is slowing. Every supplier this autumn has said the same thing: orders are above capacity. ## The new gate: permission to build Democrats won the Senate and are projected to hold the House. The result that matters most for my themes is Texas. The new governor promised a moratorium on new data centre construction and an end to tax breaks. Texas holds more than a quarter of planned US data centre capacity to 2030. Context the headlines skip: - Texas already paused new data centre hookups in August. Regulators are auditing every project in the queue for credible financing and a realistic power plan. That report is due on 10 December. - The queue is enormous: over 474 gigawatts of requests, about 90% data centres. Much of it was never going to be built. - Contracts already signed continue. One large example: a 20-year nuclear power deal between Vistra (VST) and Amazon in Texas, starting in 2027, which the company said the pause does not affect. My read: this is the next layer of the bottleneck story. First it was chips, then tools, then capital. Now it is the right to plug in. A freeze on new sites makes capacity that is already built, powered and contracted more valuable, not less. Compute demand does not disappear. It moves to other states, to on-site power, or it waits. ## What the portfolio did - Before the results I closed the most politically exposed AI line, a cloud data centre developer, at breakeven. - Two energy stops filled at or near their levels in profit as oil eased on Iran talks. Energy is now about 4.4% of the portfolio. - Overnight, stops on Tesla and one GE Vernova (GEV) clip filled in profit as futures dropped. - I reset three positions in loss at lower prices (copper miners, gold miners, a laser maker) to keep them at full size, and added global equity and emerging-market ETFs. - Cash is about 18%. Tech and AI are about 21% of the portfolio, well below my normal range. ## A correction Yesterday I called the 12% gap in the Marvell-Lumentum deal “wide”. Over the 18 months to closing, that is about 8% a year, only 2-3% a year above cash. That is a normal spread for a large deal under antitrust review, not a warning sign. The conclusion is unchanged; my wording was too strong. ## What I am watching - Tonight: Coherent (COHR) results. I want evidence of new customers or capacity, not just price moves, before adding. - Thursday: Applied Optoelectronics (AAOI). Friday: US jobs report and Vistra (VST) results, where I want to hear that contracted power is unaffected. - Health care: a Democratic Congress tends to support insurance subsidies. I may add to the Health Care Select Sector SPDR (XLV) on a strong weekly close. - Uranium: term prices near record highs while miners sit near yearly lows. One addition after Friday’s jobs data. - GE Vernova (GEV): rules that make data centres bring their own power favour turbine makers. ## Risks, stated openly - If moratoria spread to more states, AI hardware could stay volatile even with strong earnings. - If big AI buyers cut 2027 spending plans, the bottleneck becomes a demand problem. - Election and event nights can gap prices through stop levels, including in overnight sessions. ## The takeaway The constraint keeps moving up the stack, and each time the market sells first and sorts later. I prefer to own what is already built and already paid for, keep cash for when the politics settles, and follow the triggers I wrote down in advance. *Portfolio snapshot: 4 November 2026, 07:10 UTC.* Sources: [AMD Q3 2026 results (StockStory)](https://stockstory.org/us/stocks/nasdaq/amd/news/earnings/amd-nasdaqamd-surprises-with-q3-sales-stock-soars); [Texas governor result and data centre pledge (Yahoo Finance)](https://finance.yahoo.com/economy/article/a-democrat-won-the-texas-governors-race-after-promising-to-halt-new-data-center-construction-044536389.html); [Texas pause and ERCOT queue (MarketBeat)](https://www.marketbeat.com/articles/texas-paused-data-center-hookups2-power-stocks-are-waiting-on-the-verdict/); [Vistra on the pause and Comanche Peak (Utility Dive)](https://www.utilitydive.com/news/vistra-texas-data-center-pause-comanche-amazon-earnings/827434/). **This is not financial advice.** Past performance does not guarantee future results, and copy trading carries a real risk of loss. Please invest only what you can afford to lose. --- Human page: https://millionaur.com/millionaur-2026-11-04-daily-analysis/?vwao_human=1