The reason I had only 10.2GW at the end of 2027 was because of an expectation that we'd have a blue wave (But this also means rev/GW will be higher) The reason I had CapEx/GW going down in mid 2027 is because of an expectation that Neoclouds will be forced to sell assets at a discount The more complex the terrestrial compute game... the more it benefits those that can deal with complexity And the more it benefits space compute *Again, this is not investment advice. I'm trying to help people think through a very complex space*

News: Alphabet $GOOGL increased its SpaceX $SPCX exposure last quarter. Its Q3 13F, filed today, shows 555.0 million shares versus 551.2 million at the end of Q2 — up roughly 3.85 million. SpaceX is still the dominant position at about $83.7 billion, nearly 95% of the $88.3 billion total portfolio. Reuters framed it the same way: a genuine increase, not a mechanical unlock effect, because 13Fs track beneficial ownership and lockup expirations don’t move share counts. This is also the first clean quarter-over-quarter read we’ve had, since SpaceX only IPO’d in June. Alphabet added Qualcomm too — nearly 3 million shares, about $548 million — while trimming Revolution Medicines, Freshworks, and LIFE.

There were already bottlenecks — memory and others. This round of tighter regulation on ground data center construction actually gives SpaceX’s ground data centers smoother progress and lower-cost builds. So going forward, investors in AI-related industries need to think carefully. If you just chase the bottleneck, you might not get the returns you originally projected. One more point: from here on, all the neoclouds will basically be marginalized. It’s not just about their bonds — it’s that their capital, technology, and talent simply can’t compete with the big players. So this type of investment, I’ve never recommended from the start.

I believe most people still don’t realize SpaceX $SPCX is the biggest beneficiary of this round of tighter regulation on ground-based data centers. I’m talking mid to long term, because the ultimate solution is space. Too many people don’t get it — society as a whole can no longer function without AI. After this midterm election, I believe Musk will accelerate Starship development and pull the timeline forward. My point is that with Democrats winning, it actually (force to) speeds this up, because they’ll regulate ground data centers even harder, making space the only real option left.

Today Morgan Stanley estimates that levered ETF rebalancing will create 52 billion dollars of forced selling in US equities before the close, with semiconductors taking about twenty-one billion and the Nasdaq 100 another fifteen billion. It’s the second-largest single-day since June fifth, which saw 55 billion. My take: this kind of forced selling is the cleanest signal in the market. The sellers aren’t exiting because fundamentals got worse — leverage formulas are pushing them out. The portion of price that’s been knocked down is mostly pure noise, not a repricing of value. But today there’s one key difference: inexperienced new investors can’t tell. All they see is red screens and panic headlines. They don’t ask who’s selling, or why. Forced selling and genuine fundamental deterioration look identical on a chart. Same decline — some treat it as mechanical noise and step in, others treat it as doomsday and run. Retail behavior itself amplifies the volatility, creating a second layer of stampede. So the real opportunity isn’t the fact of forced selling. It’s whether you can keep that distinction alive inside the chaos.

News : Texas Gov-elect Gina Hinojosa (CNN interview): “We can have a strong working relationship with Musk and his companies ( $SPCX & $TSLA )— on our terms.” “We will have laws to protect Texans, and we need to stop what we are doing with data centers until we have those protections.” “It is not that I hate data centers. I hate the way they are taking advantage of Texas.” “It cannot be just running over the people of Texas, and that’s what changes now.” Post-election shift to negotiated relationship after campaign criticism of incentives and Abbott-Musk ties. Existing permitted projects face lower risk; new expansions under more conditions. Source: CNN via Reuters, Nov 5.

SpaceX-linked $SPCX : 6 GW supercompute campus filed in Cochise County, Arizona. The applicant is Tombstone Resources LLC — a Wyoming entity formed in July, headquartered in Atlanta. SpaceX hosted the Nov 4 town hall; the county confirmed it was privately organized by SpaceX. No official SpaceX statement yet. The numbers, straight from the filing: • ⁠747 acres along US 191, adjoining AEPCO’s Apache Generating Station • ⁠Up to 9 computing buildings, ~493,000 sq ft each, across 4 phases • ⁠Full build-out: 6 GW. Phase 1: 1.5 GW in 3 stages • ⁠All grid power via SSVEC and AEPCO. No on-site generation. 90 MW of natural-gas standby for outages and testing only • ⁠24 GWh battery storage • ⁠Dry-cooled. No cooling towers, no evaporative spray. 7–10 acre-feet/year at full build-out — mostly staff domestic use, roughly 20–30 Arizona homes. No new Willcox basin well, no industrial groundwater • ⁠$3.6B in applicant-funded transmission and substation upgrades • ⁠Claimed 10–40% bill reduction for all SSVEC members for the project’s life, plus $13–53M/year to the SSVEC Foundation • ⁠~1,000 permanent jobs, ~5,500 at peak construction • ⁠$34–103M/year in property tax Context that matters: SpaceX’s Q3 call (Oct 20) raised the 2027 compute target to 13 GW or more, from 1.7 GW operating at Sept 30. SemiAnalysis counts 3.6 GW under construction and 9+ GW identified. This Arizona site sits inside that pipeline — and inside Musk’s separate 15–20 GW power-and-cooling target for YE27. It is not Brownsville. Different state, different permit chain, grid-served instead of behind-the-meter turbines. The Texas freeze does not reach it. But three things are unfinished: no signed Energy Services Agreement, no ACC approval, and the source of generating capacity is not yet identified in the co-ops’ responses. An opposition petition has passed 1,900 signatures. The Planning and Zoning Commission hears it Dec 9 at 4 PM in Bisbee.

Breaking News: SpaceX $SPCX and Qualcomm just announced a multi-year deal to build Starlink Mobile’s nationwide terrestrial 5G network in the U.S., with rollout starting in 2027. Qualcomm’s Dragonwing QRU100 and QDU100 platforms will power tens of thousands of cell sites, using 64T64R Massive MIMO for better coverage, capacity, and spectrum efficiency. Power-efficient processing improves deployment economics at scale, and the architecture supports multiple bands with a path to AI-native 6G. The goal: the first fully integrated satellite-plus-terrestrial mobile network in the country. Satellites overhead, 5G on the ground, one seamless experience. Qualcomm calls it its biggest 5G deal to date. Financial terms weren’t disclosed. This puts SpaceX on a direct collision course with AT&T, Verizon, and T-Mobile — and it’s another data point that SPCX is building infrastructure, not just rockets.

News: SpaceX $SPCX is adding aluminum furnaces at its Bastrop Starlink factory. TCEQ Permit 182733: technical review done, draft permit issued, preliminary approval because it meets all regulations. This supports on-site melting for high-pressure die casting of Starlink terminal parts. Continues the “raw aluminum in → finished kits out” vertical integration at a site already running ~15k kits/day and scaling further. Not primary smelting. Separate from the turbine foundry.

News: Morgan Stanley just nearly doubled $SPCX SpaceX’s 2027 revenue to $175B, up 93%. FCF positive pulled forward four years to 2031, not 2035. About $280B less debt needed. At $140, SPCX trades at 17.6x 2028 earnings — cheaper than the S&P’s 17.7x, while growing earnings at roughly 150% CAGR from 2026 to 2030. My own model says even that is conservative — 2027 revenue can easily above $200B