Five architects of the AI economy explain where the wheels are coming off

The significance here isn't that problems exist, it's that the people building the infrastructure are now saying so publicly at a major investor conference, which is a different kind of signal than analyst commentary or academic critique. When supply-chain insiders surface foundational architectural doubts at Milken, the audience is capital allocators, and that shapes what happens to the $725 billion already committed.

This story sits directly alongside our May 1 coverage of 'Big tech's AI spending balloons to $725 billion this year,' which framed infrastructure depth as the primary competitive lever. That piece assumed the architecture being scaled was sound. The Milken discussion complicates that assumption considerably. Our earlier 'AI Demand Is Outpacing the Scaffolding to Support It' piece identified operational bottlenecks, but this goes further, questioning whether the underlying technical architecture is the right one to be scaling at all. Those two threads together suggest the capital commitments may be locking in approaches that insiders already doubt.

Watch whether any of the named supply-chain stakeholders from this panel follow up with public technical proposals or defections toward alternative architectures within the next two quarters. If capital allocation patterns among the major hyperscalers shift away from current datacenter topologies in 2026 earnings guidance, that confirms the Milken skepticism is actionable, not just rhetorical.