E112: Is Davos a grift? Plus: globalist mishaps, debt ceilings, TikTok's endgame & more

Jason Calacanis said the World Economic Forum recruited him about 15 years earlier as a “global future leader,” then sent him a roughly $40,000 bill; he declined. That anecdote fed the panel’s distinction between Davos’s genuinely powerful invited speakers and a broader paid status network surrounding them.

Chamath Palihapitiya said he is investing heavily in underexplored natural-resource regions such as India as deglobalization reshapes supply chains. His thesis is that governments will accept slower consumer cycles and longer paybacks for domestic jobs, while Western technology restrictions raise China’s production costs and narrow its advantage.

Calacanis proposed separating mass labor immigration from elite-talent recruitment: roughly 2–3 million legal workers annually, plus hundreds of thousands of top graduates encouraged to remain in America. The group broadly converged on skills- or points-based selection, while Sacks argued border control must come first for reform to win public support.

Palihapitiya argued TikTok’s geopolitical exposure could erase $70–80 billion from ByteDance’s valuation and advised existing investors to sell if they can. He said a forced U.S. sale could let buyers bargain brutally—illustratively valuing TikTok U.S. around $10 billion rather than paying anything close to ByteDance’s prior private-market marks.

The strongest evidence discussed against TikTok was not hypothetical surveillance but ByteDance’s own admission that employees improperly accessed U.S. user data while trying to identify journalists’ sources. The panel also noted that this proved internal misuse, not systematic Chinese-government spying, a distinction often lost in the political argument.

BioNTech’s roughly $600 million purchase of about 250-person InstaDeep highlighted how strategic buyers can value AI capability even without a conventional software product. Palihapitiya countered that pharma AI still has not clearly improved discovery hit rates, arguing the stronger model may be giving tools away for a small royalty on successful drugs.

Calacanis argued accounting is a cleaner near-term AI target than many flashy generative-AI demos, citing Americanas’s multibillion-dollar accounting failure as the kind of human error software should prevent. His proposed end state was AI producing the P&L directly, giving CEOs and CFOs a machine-checkable system rather than relying on opaque manual workflows.