In the fight for dominance in AI infrastructure, big tech companies are announcing Capex bigger than the GDP of most smaller countries, with a pivot toward AI infrastructure that includes job cuts in the tens-of-thousands. Amazon’s aggressive pivot toward “industrial-scale” AI came with its $200 billion announcement, which exceeded analysts’ predictions by $60 billion.This followed Amazon’s decision to put approximately 16,000 jobs on the chopping block, redirecting that money toward Trainium and Graviton custom chips, as well as robotics, Kuiper satellites, and data center builds — all of which mean a 60% increase in AI spending.  Amazon’s shares fell ~10%, as investors perhaps became spooked by the capex battle brewing among Amazon, Microsoft, Meta, and Oracle, all of which plan to spend approximately $700 billion in capex for 2026. Included in that figure are Alphabet’s recent announcement for $185 billion in Capex and Meta for $135 billion, as well as Oracle’s plans for $45 to $50 billion of gross cash proceeds through a combo of debt and equity financing. The era of industrial-scale AI has truly begun, with some comparing it to the19th-century railroad boom’s investment–crash–digestion cycle. The difference is that railroads had a long lifecycle, whereas GPUs are obsolete within a few years. That engenders an environment of constant, expensive reinvestment into what could be our most transformative era yet, or a “bust” that eventually corrects and then leads to transformation.