The panel argues that SpaceX's valuation is justified by Starlink's projected revenue growth and its nascent but powerful AI compute business, dubbed 'Elon Web Services.' They highlight Elon Musk's ability to rapidly deploy GPU clusters, enabling SpaceX to resell compute at high premiums and potentially become a major hyperscaler. The discussion also explores the economics of orbital data centers, which could be significantly cheaper than terrestrial ones due to Starship's reusability. On the AI front, the group debates the frontier model race, noting that new models like Fable 5 and Mythos are extending their lead despite difficulties in evaluation. They argue that frontier models capture 90% of revenue, driven by long-running agent tasks that consume massive tokens. The panel defends massive AI CapEx, asserting that revenue estimates are too low and that demand far outstrips supply, with less than 0.2% of people using AI agentically. They conclude that AI companies are on track to add a trillion dollars in revenue in half the time it took the Magnificent 7.