AI debt worries mount as CoreWeave's bond yield tops 15%, insuring SoftBank debt gets pricier, and some Oracle debt trades near 80 to 90 cents on the dollar
THE SO WHAT
Double-digit yields on CoreWeave bonds and discounted Oracle and SoftBank paper are a warning that cheap capital for AI infra is not guaranteed. If your strategy assumes ever-expanding third-party compute, stress-test it against tighter credit, higher financing costs, and slower buildouts.
READ THE SOURCE
MORE FROM THE WIRE
Tech & InnovationThe Apple iPhone 18 Pro is (almost) free this weekend — how to claim
Carrier promos that make a flagship iPhone 18 Pro “almost free” are a reminder that ARPU and lock-in, not hardware margin, drive the model. If you build around mobile, assume the high-end device base will keep expanding even in weak consumer cycles.
Tech & InnovationPBOC Deputy Urges Private Role in Sovereign Debt Restructuring
Beijing calling for more private-sector participation in sovereign debt workouts is a signal that restructuring cycles in emerging markets will be more complex and negotiated. If you have exposure to these regions—directly or via suppliers—map counterparties and contract terms that could be hit by drawn-out restructurings.
France is aflame as peaceful protests over the cost of living devolve into chaos and looting
Sustained cost-of-living protests escalating into looting is a signal that social stability risk is now a live variable in Western consumer markets, not just emerging ones. Operators with exposure to French retail, logistics, or on-the-ground services should stress-test for intermittent shutdowns, insurance gaps, and demand volatility tied to political and economic shocks.
Tech & Innovation'This computer is going to be incredible': Nvidia CEO Jensen Huang explains why the new Surface Laptop Ultra is "really a work of love"
Four years of co-design between Microsoft and Nvidia for a single flagship laptop says more about the AI PC transition than any keynote slide—OEMs are treating local AI performance as a system-level problem, not a chip swap. If you ship desktop software, assume your users will soon have tightly integrated GPU–NPU–cloud stacks and design features that exploit that, not just generic “runs on device” claims.