In a ‘Breaking Analysis’ column, Dave Vellante examines whether the AI industry is in a capital bubble, arguing that scarcity in semiconductors, high-bandwidth memory, and power has so far delayed the market-clearing moment that would expose weak underlying demand. He uses the Oracle-OpenAI-Stargate arrangement as a case study of capital commitments preceding revenue-generating deployment, and lays out three possible outcomes: a soft landing as demand catches up, a delayed reckoning with persistent supply constraints, or a bubble break once supply normalizes faster than profitable use materializes. The piece estimates this dynamic could play out through roughly 2029.
