Political Power Current
Musk in the Lincoln Bedroom, AI Valuations Under Scrutiny, and the Limits of the Bull Case
A new book by Maggie Haberman and Jonathan Swan, titled 'Regime Change,' reports that Melania Trump repeatedly objected to Elon Musk's stays in the Lincoln Bedroom and was overruled. When the First Lady raises a concern about a tech billionaire's access to the White House residence and the objection is dismissed, it communicates something concrete about whose concerns carry weight in the internal hierarchy of this administration. A separate book revelation — that Musk warned Trump about a potential Taiwan chip crisis — carries more strategic significance: Musk's financial exposure to Taiwan through Tesla's supply chain and SpaceX's semiconductor dependencies gives him a direct economic interest in how U.S. policy treats Taiwan's security, blurring the line between policy advice and advocacy for personal financial outcomes.
The arc of Mark Zuckerberg's relationship with Trump, as described in the same reporting, is instructive about power dynamics between political and tech leadership. Zuckerberg spent years resisting political entanglement, then cultivated what the book describes as 'full-blown camaraderie' with Trump — while Trump privately mocked Zuckerberg's outreach. The asymmetry, Zuckerberg earnest and Trump dismissive behind closed doors, reflects which party in that relationship holds the structural leverage.
The Social Security tax cap proposal from Senators Moreno and Warren — a rare cross-party alliance to remove the payroll tax ceiling currently set at roughly one hundred and sixty-eight thousand dollars — carries enormous fiscal implications. Removing the cap would extend the six-point-two percent employee tax to all wages, significantly extending Social Security's solvency horizon while representing a substantial tax increase on high earners. Passage in the current Congress is considered unlikely, but the unusual alliance signals some political movement on entitlement financing.
A federal judge barred ICE from making arrests at immigration courts nationwide, ruling that arresting people who have voluntarily appeared to comply with legal process undermines the court system's ability to function and creates a deterrent against legal compliance. If the ruling holds on appeal, people with pending immigration cases can attend court appearances without arrest risk. The administration is expected to appeal, and the Supreme Court's recent green card ruling may constitute relevant precedent depending on how that challenge is framed.
The most consequential analytical question of the day may be the one least amenable to certainty: whether AI valuations are justified. The bull case, articulated most forcefully by Son, rests on three assumptions — that AI capabilities will continue scaling at recent rates, that infrastructure builders will capture the economic value rather than the businesses using AI tools, and that a small number of foundational model providers will maintain pricing power. Each assumption has serious vulnerabilities. Scaling returns appear to be diminishing per dollar of compute. Business software history suggests productivity gains flow to users, not vendors. And open-source models from Meta, Mistral, and others are advancing toward near-frontier capability without the training costs of commercial providers. Three signals will clarify the picture: enterprise AI contract renewal rates at current pricing, the revenue-per-unit-of-compute ratio at leading AI providers, and whether open-source models close the capability gap with commercial frontier models by late 2027. Both the bull and bear cases have genuine empirical support. The gap between the ambition embedded in a sixty-billion-dollar Cursor valuation and demonstrated revenue remains large enough to warrant real skepticism alongside real enthusiasm.