US tech stocks see sharp falls over AI cost fears

The discussion largely rejected the headline’s simple story that markets are recoiling from AI infrastructure bills. Several commenters argued the sell-off instead reflects AI disrupting the businesses expected to finance the buildout: answer engines keep users from clicking search ads and publisher links, while referral-dependent sites lose traffic. Others noted that expensive datacenter expansion is still an AI-cost story because it has tightened supplies of memory and compute hardware, raising PC prices and creating a broader capital-spending hangover. The most concrete AI debate centered on Google’s search summaries. Critics described a product that confidently delivers stale or false information and appears rushed into search to keep pace with the model race; even cited sources allegedly fail to support many answers. Defenders said the system has improved sharply, now resolves ordinary questions accurately enough to avoid opening reference pages, while retaining caution on niche subjects. That split fed a wider anxiety: AI demand is plainly strong and deeply embedded in some developers’ work, but cheap access may not survive if providers cannot turn massive compute spending into durable revenue. There was also frustration that consumer GPUs and local-model hardware remain scarce while capacity is concentrated in datacenters. Some see on-device inference as the practical escape hatch, while others stress that fabs, advanced lithography, memory, and skilled labor cannot expand on command. A minority welcomed a bubble correction as overdue discipline for inflated AI claims, despite the economic damage it could cause.