Big Tech may need to rethink AI spending plans amid adoption concerns

The AI gold rush has a timing problem. Companies are pouring hundreds of billions into data centers, chips, and energy infrastructure that takes years to build out, while the software those investments are meant to support evolves on a timeline measured in months. Former IBM CEO Sam Palmisano is now publicly flagging that disconnect, urging Big Tech to develop contingency plans for the very real possibility that their massive capital commitments don’t pay off on schedule. The spending spree under the microscope Microsoft, Alphabet, Meta, and Amazon are collectively projected to spend hundreds of billions annually on AI-related infrastructure. But the picture gets more uncomfortable when you factor in off-balance-sheet commitments, which some analyses suggest could push total future obligations toward $3 trillion. Palmisano’s core argument is straightforward: the AI boom rests on long-cycle bets in physical infrastructure, but AI software is changing at breakneck speed. The hardware you commit to building today might not be optimally suited for the models and applications that dominate two or three years from now. That asymmetry creates a peculiar kind of risk, one where the investm...