AI Token Prices Fall Nearly 20% From May Peak as $700 Billion Capex Faces Return Doubts
Updated
Updated · Bloomberg · Jul 3
AI Token Prices Fall Nearly 20% From May Peak as $700 Billion Capex Faces Return Doubts
3 articles · Updated · Bloomberg · Jul 3
Summary
The Silicon Data LLM Token Expenditure Index has fallen almost 20% from its May high, signaling weaker pricing for AI usage even as demand for the technology remains under scrutiny.
That drop matters because token pricing is one of the clearest market gauges of whether the AI sector can turn heavy spending into durable revenue and growth.
The index had nearly doubled since its December launch before reversing, underscoring how quickly optimism around AI monetization has started to fade.
More broadly, the slide adds to investor unease over a $700 billion-plus AI capital-spending boom and whether those investments will generate acceptable returns.
With AI's economics failing, is the industry heading for a dot-com-style collapse?
AI promised abundance, but its scarcity-based costs are breaking business models. What comes next?
As AI's energy thirst grows, will power grids become the ultimate bottleneck for progress?
The $700 Billion AI Cost Paradox: Why Cheaper Tokens Are Driving Soaring Enterprise Bills and Infrastructure Risks
Overview
AI tokens are becoming significantly cheaper due to rapid technological and algorithmic advancements, with new, more efficient models on the horizon. However, as AI shifts from pilot projects to a core enterprise function, overall spending is paradoxically soaring. This is driven by increased adoption, employees using AI for unnecessary tasks, and the deployment of complex agentic workflows that consume far more tokens. Many companies are surprised by unexpectedly high AI bills, revealing that effective cost management must go beyond just token prices to address the broader and more complex cost structure of enterprise AI.