AI Evolution: Merchants of Compute
THE SO WHAT
The rise of CoreWeave, Nebius, and peers is fragmenting the GPU market into specialized “neo clouds” optimized for AI. If you’re planning large training runs, multi-cloud now means mixing hyperscalers with these merchants to arbitrage price, availability, and flexibility.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechGM is giving away free home chargers, and paying drivers to charge off-peak
Managed charging is turning EVs into grid resources—GM and PG&E are effectively buying flexible demand for $1,999 upfront plus $15/month. If you operate large fleets or multi-site facilities, treat charging hardware decisions as an energy-market integration choice, not just a perk.
Deep & Emerging TechMost Americans now oppose data centers in some way as tide continues to turn on AI technology
Local opposition is becoming a real siting constraint for AI infrastructure—data centers are now politically salient, not invisible utilities. If your roadmap depends on new capacity, factor in permitting, water, and community relations as first-order risks, not afterthoughts.
Deep & Emerging TechHackers Hack the FBI’s Hacking Unit: Report
If a specialized FBI cyber unit can have personnel records exposed, assume your own offensive and security teams are high-value targets too. Treat red-team and sensitive ops data—names, tools, infrastructure—as crown jewels with separate controls and monitoring.
Deep & Emerging TechBentley’s first electric car arrives, and its next one is not due until after 2030
Bentley pushing its all-EV target from 2030 to 2035 and spacing models beyond 2030 shows even luxury OEMs are rethinking straight-line EV roadmaps. If you’re tied to auto timelines—charging, materials, software—plan for staggered, segment-specific adoption rather than uniform EV curves.