The math changed. According to BigGo Finance’s reporting, an index tracking fees that users pay for AI tokens, called the “Silicon Data LLM Token Expenditure Index”, has fallen nearly 20% from its May 2026 high. The Los Angeles Times covered the same pricing dynamics on July 3, framing the trend as a fragility signal for AI’s broader pricing power. Neither outlet identifies who publishes the index, what its methodology is, or how its “blended rate” reading of 1.62 is constructed, so the index itself can’t be treated as an established market benchmark. What both outlets confirm is the direction: token pricing power is eroding.
Allianz research, as cited by BigGo Finance, identifies a 46% growth gap between AI investment and sales, more severe than the 32% divergence seen during the 2001 telecom bubble. That comparison doesn’t come from the index; it comes from a named research institution and is worth separating from the index’s unverified methodology.
Why it matters
Over $700 billion in AI capital expenditure has been committed on the premise that frontier labs can sustain pricing power over enterprise customers. Analysts warn that if token prices keep falling, driven by frontier lab competition, enterprise CFO optimization, and structural migration toward open-source models, the return math on that capex becomes strained. The Allianz growth-gap figure is the sharpest articulation of that concern: investment is running nearly 50% ahead of the revenue growth needed to justify it, a gap wider than what preceded the 2001 telecom crash.
Investment-to-sales growth gap (AI vs. prior bubble cycle)
Optimists counter, as BigGo Finance notes, that while token prices have collapsed, total spending has doubled, the market is expanding even as per-unit prices fall. That’s the bull case: volume more than offsets price compression. Whether volume holds as open-source migration accelerates is the unresolved question.
Context
This story connects directly to the inference cost dynamics the Hub has been tracking. Token price compression isn’t new, it’s been building since late 2025 as frontier labs cut list prices to defend market share against open-source alternatives. What’s new is a named index with a reported magnitude (nearly 20%) and a comparison to prior bubble cycles that gives investors a historical frame. The Epoch AI analysis showing hyperscaler capex growing 3x faster than cash flow sits in the same analytical territory: spending is outpacing returns, and the market is starting to ask when that gap closes.
What to watch
The near-term trajectory of the index, if it continues declining, is the indicator to monitor. Pricing power, not chip supply, is now the critical variable for AI infrastructure stocks, according to BigGo Finance’s framing. Watch Nvidia’s next earnings call for any commentary on inference utilization rates and whether hyperscaler customers are signaling demand moderation. Watch whether any of the major frontier labs announce a price increase, that would be the clearest test of whether pricing power is recoverable or structurally impaired.
Analysis
The Silicon Data LLM Token Expenditure Index has no publicly identified publisher or methodology in available reporting. The nearly 20% decline figure is confirmed by two outlets (BigGo Finance, LA Times) but cannot be cross-checked against an independent data source. Treat as a directional signal reported by credible outlets, not as a verified market benchmark.
TJS synthesis
The bear case and the bull case are both internally coherent right now, and that’s exactly the problem for investors trying to price AI infrastructure. Falling token prices with rising total spend is a razor’s edge: it works until enterprise migration to open-source tips the volume math against the closed labs. Together AI’s $800 million raise this week, led by Aramco Ventures, is a direct capital market bet that the tipping point is closer than frontier lab valuations imply. Watch whether open-source inference platform bookings, Together AI’s claimed $1.15 billion and others, show year-over-year acceleration in Q4. That’s the number that will tell you which side of this debate the market should be on.
Sources: Bloomberg, BigGo Finance.