The massive AI spending from hyperscalers and enterprises isn't justified by current profits or clear ROI. Instead, it's a defensive, game-theoretic move driven by the fear of being technologically outmaneuvered if competitors achieve a breakthrough first.
The "picks and shovels" play of investing in semiconductor companies is maturing. A better bet may now be hyperscalers, who could outperform either if enterprises start profiting from AI or if they simply moderate their own capex spending to improve free cash flow.
Enterprise AI is not a simple software upgrade. Its adoption is inherently slow because it's a paradigm shift to probabilistic systems, requiring a new technology stack and, crucially, entirely new control planes to manage the technology responsibly and compliantly.
Surveys reveal a significant gap between executives' optimistic expectations for AI's impact and the actual productivity benefits reported by employees. This disconnect highlights implementation challenges, like poor data infrastructure, and differing incentives between management and staff.
Unlike typical tech cycles where suppliers and customers thrive together, the current AI boom sees semiconductor companies capturing value while their customers (hyperscalers, model builders) incur massive losses. This unsustainable dynamic suggests a future market correction.
The scale of investment in AI (est. $7-8 trillion) is too vast to be recouped by simply capturing market share in existing industries. For a positive ROI, AI must act as a platform for generating entirely new economic activities and markets (TAMs), a much higher bar than simple disruption.