AI’s Investment Case Hinges on Whether New User Habits Can Be Monetised: Analyst

Why It Matters

If AI-driven habit changes can be monetized, they will validate heavy infrastructure spending and reshape winners across cloud providers, chipmakers and application developers; persistent compute bottlenecks and rising energy needs will also influence capital allocation and supply-chain dynamics.

Summary

Speaking at Goldman Sachs’ Communacopia & Technology conference, an analyst argued the AI investment thesis depends on whether new user and enterprise habits driven by platforms can be monetized, with computing shifts moving from infrastructure to platforms and only later to applications. The platform layer is firming up while the application layer and daily-use AI agents will determine whether initial capex yields returns; physical AI—robotics, autonomous vehicles and warehouse automation—may be an underappreciated near-term frontier. Supply constraints for compute (memory, GPUs and power) are likely to persist into at least H2 2027, even as token costs for general compute begin to decline unevenly, and competition from custom silicon (Google, AWS) will shape semiconductor winners. Nvidia’s results are being watched for signs of how the industry will close the gap between demand for compute and available capacity.

Original Description

Joining from the sidelines of the 'Goldman Sachs Asia Communicopia and Technology Conference’ the bank's Co-lead of TMT Research Eric Sheridan says AI’s real returns will come when changing user habits turn into revenues. He argues investors are too focused on chatbots, overlooking the bigger opportunity in robotics, automation and autonomous vehicles.