A confluence of uncomfortable data points dropped in early June 2026, forcing investors to reckon with something they've been quietly avoiding: the gap between AI as a transformational technology and AI as a profitable business is wider than anyone priced in.
Start with Broadcom. The chipmaker's Q3 earnings report on June 3 projected AI chip revenue of $16 billion. Wall Street expected $17.2 billion. Broadcom shares dropped between 12% and 14%. The selloff dragged names like Micron and SK Hynix, which had earlier posted year-to-date gains exceeding 300% and 260% respectively.
Then came the Bain survey, conducted in late May 2026, which sampled 951 companies on their AI investment outcomes. Nearly 40% of respondents reported achieving only a 0-10% reduction in costs from their AI deployments. For context, 37% had originally targeted cost reductions in the 11-20% range.
On June 7, executives from major companies including Microsoft acknowledged a blunt truth: the costs associated with running cutting-edge AI models, particularly those from firms like Anthropic, have become prohibitively expensive.
Big Tech's AI capital expenditures are expected to reach hundreds of billions of dollars annually. AI infrastructure costs are rising faster than the returns those investments generate.
For crypto and digital asset investors, the AI narrative has been deeply intertwined with markets since 2023. If centralized AI deployments are underwhelming on cost savings, the value proposition for decentralized alternatives could sharpen. The key metric to track isn't AI spending-it's AI revenue per dollar spent.