Top Tweets for #VirginO2

SpaceX is about to become the most important company on Earth. And maybe off it. $30-40B raise. $1.5T valuation. Largest IPO in history. Those are the headlines. Here’s what they’re actually telling you. Musk just told investors that Starlink will generate $22-24B in 2026 revenue, up from $8B in 2024. That’s tripling in two years. At 100x forward revenue, the market is pricing Starlink like it’s the next AWS. The comparison isn’t crazy. AWS did $90B last year with 31% of cloud market share. Starlink has 8M subscribers and approximately zero real competition. Amazon Kuiper has 153 satellites. Starlink has 7,000+. OneWeb did $216M in 2024. Starlink did $8B. By the time Kuiper launches commercial service in late 2026, the race is already over. But Starlink is the boring part of this story. Musk said last month that “cost effectiveness of AI in space will be overwhelmingly better than AI on the ground” within 4-5 years. SpaceX plans to use IPO proceeds to build space-based AI data centers. Starship can deliver 100GW per year to high Earth orbit. For context, total US electricity consumption is 490GW. Average AI data center demand is projected to hit 123GW by 2035, up from 4GW today. The math: Earth cannot build power plants fast enough. Gas turbine backlog is 7 years. Nuclear takes a decade. Meanwhile Musk is talking about solar-powered AI satellites with radiative cooling in orbit. Google’s Sundar Pichai called it a “moonshot” and said Project Suncatcher could launch prototype servers by 2027. His exact words: “only possible because of SpaceX’s massive advances in launch technology.” ARK just published an open-source valuation model. Monte Carlo simulation, 17 variables, 1 million iterations. Base case: $2.5T by 2030. Bull case: $3.1T. That’s a 38% CAGR from December’s $350B round. The model assumes: - Full Starlink constellation (42,000 satellites) by 2035 - $300B annual revenue at maturity (15% of global communications spend) - Mars development funded by Starlink cash flows And here’s where it gets weird. Musk announced 5 uncrewed Starship launches to Mars in 2026. He gives it 50-50 odds. If they land intact, crewed missions begin 2029-2031. By 2033, he’s targeting 500 launches per window. The goal: 1 million people on Mars, self-sustaining city in 20 years. The timeline: - 2026: 5 test flights (10t payload each) - 2028-29: 20 launches, first humans - 2030-31: 100 launches - 2033: 500 launches ARK’s model explicitly excludes Mars revenue because “projecting cash flows from extraterrestrial settlements can be speculative.” But they note Mars infrastructure could eventually enable asteroid mining. Jensen Huang called space data centers “a dream.” Fair. Radiation shielding for Blackwell GPUs doesn’t exist. Deploying tens of thousands of square meters of thermal radiators is science fiction today. But SpaceX has a history of turning science fiction into quarterly revenue. At $1.5T, you’re paying for: 1. Starlink monopoly on LEO internet (priced in) 2. Space-based AI compute infrastructure (speculative) 3. Mars colonization (not priced, possibly unquantifiable) Here’s the bear and bull cases. Bear case 100x multiples assume flawless execution on subscriber growth, zero price compression from competition, and regulatory goodwill in 100+ countries. Musk’s timelines slip by years. Starship has had multiple test failures. Bull case SpaceX is the only company that can simultaneously solve the AI power crisis, provide global internet, and make humanity multiplanetary. The TAM is literally civilization. I don’t know if SpaceX is worth $1.5T or $3T or $500B. Neither does anyone else. The honest answer is that traditional valuation frameworks break when you’re pricing optionality on space-based computing and interplanetary colonization. But I know this… if you’re building a 20-year portfolio and you can only own one company, the argument for SpaceX has never been stronger.