Bain's seventh global technology report frames the central question of the cycle as value, not capacity. Hyperscaler capex could hit $780B in 2026, and annual AI infrastructure spending could reach $1.5T by 2031. If capex runs at about 25% of revenue, the industry needs an AI market approaching $6T a year. Bain sees consumer and enterprise AI covering only $1.2T to $1.8T of that, leaving roughly $4.2T to come from new categories such as advertising in chat products, autonomous vehicles and machines, and AI-driven science. The rest of the 81 pages map where that new value could be built. Highlights include a $100B SaaS opportunity in automating cross-system coordination work, hardware and semiconductors outgrowing software for the first time in years, a flattening org pyramid as individual contributors direct agents, and a practical chapter on governing token spend. The report argues that absorption speed, not model access, is now the competitive variable, which favors small AI-native teams over slower incumbents.