Silicon Data Token Index Drops 20% From May, Bloomberg Warns

A single line on a Bloomberg terminal has become the cleanest read anyone has on whether the AI trade still makes sense, and it just gave up ground. Silicon Data's LLM Token Expenditure Index, which measures what the whole market pays per million LLM tokens, is down almost 20% from a May high after nearly doubling since its inception in December, Bloomberg reported. For stock investors, the column argues, that could be flashing a warning that AI companies are losing pricing power with increasingly cost-sensitive customers, and that expectations for an eventual AI bonanza could prove misplaced.

Why the signal matters is the size of what sits on top of it. Bloomberg calls the gauge the cleanest read anyone has on the $700 billion-plus capex boom that has done the sector's heavy lifting, and frames the setup bluntly: it's a pricing-power story, not a silicon story, that is funding the march toward $1 trillion of capex in 2027. Macro strategist Andreas Steno Larsen called it the chart everyone should be watching, warning that sustained weakness in token pricing would end the memory, hardware and data-center trades for this cycle.

The bull read is that cheaper tokens have expanded the market. Prices have collapsed more than 90% since 2023 while total spend has roughly doubled since last year, so an index pause is simply digestion and demand is real. The bear read, cited in the same column, is that sustained weakness could end the trade that saw nearly the entire AI cohort rally hard this cycle, because it's token spending that justifies the next capex order and the bill is already looking stretched. Allianz Research adds a comparison point that stings: the growth gap between AI investment and sales is running around 46%, exceeding the 32% divergence observed during the 2001 telecom excess cycle. Allianz also notes the parallel is imperfect because this round is being funded from the strongest corporate balance sheets in history, not the debt that ended in widespread capital destruction last time.

The honest caveat is that one index moving 20% in two months is not a rerating of AI, and Bloomberg leaves both interpretations on the table. The measure is expenditure-weighted, so it can shift because customers migrate to cheaper models rather than because underlying demand actually falls. What the reporting doesn't give you is whether hyperscaler internal usage is masking real sales, or how frontier model pricing behaves if cost-per-token keeps grinding down.

What is worth watching is the read-through if the drift continues. Inference-optimized silicon, open-weight hosting, and application-layer buyers of compute would be the beneficiaries of a world where the AI bill actually falls. Frontier labs and the balance sheets funding the path to $1 trillion of 2027 capex would be the ones with the most to explain.