Why do three US labs take ~70% of OpenRouter spend but only ~27% of tokens?
OpenRouter's Peter Walker shows three major American labs attracting about 70% of spend while accounting for only 27% of tokens. Thin reply chain; the chart is the claim. Google's weight depends on whether the metric is spend or tokens.
The read
The concentration split is the useful fact: spend and tokens are not the same market. Walker's OpenRouter aggregate — ~70% of spend vs ~27% of tokens at three major US labs — favors reading frontier APIs as a high-ASP, low-token-share slice rather than "the labs are losing share" on raw volume. What is unresolved is composition (how much Google moves either bar) and whether the gap is price, model mix, or routing. Reply chain is thin; do not overfit.
State of play
Peter Walker (OpenRouter Insights) posts a chart of spend share versus token share across three major American labs on OpenRouter. Dan Gray asks how much Google drives the result; Walker replies that it depends on whether one means spend or tokens, but "enough." Stakes: pricing power and mix at frontier labs versus open/cheaper models on a real routing venue — and whether "usage" narratives should track dollars or tokens.
Fidelity note: top-replies — essentially root + one exchange; tail unread because there is almost no tail.
The positions
Frontier labs dominate spend, not tokens. @PeterJ_Walker [named] [principal] — 'The 3 labs still attract ~70% of all spend while only seeing 27% of tokens used'
Google's contribution is metric-dependent. @credistick [named] — 'How much weight is Google pulling here?' @PeterJ_Walker [named] [principal] — 'differs whether you mean spend or tokens, but enough :)'
Weight of the room
Walker is principal on the dataset. Gray is a named VC research voice asking a composition question. No third camp with competing numbers appeared in the captured replies.
What would settle it
OpenRouter (or comparable router) publishes a time series of spend vs tokens by lab with Google broken out Knowable — next OpenRouter public insights drop — target by 2026-10-15
Lab list-price and utilization disclosures that reconcile ASP × tokens to revenue mix Knowable — lab earnings / system cards — through Q4 2026
Posts
Peter Walker (@PeterJ_Walker) — Sep 14, 2026, 5:12pm ET
Wider context on the share of spend vs the share of tokens across the 3 major American labs
This is aggregate data across OpenRouter and the split between spend and tokens is pretty stark.
The 3 labs still attract ~70% of all spend while only seeing 27% of tokens used
21 likes · 4 RTs · 1 reply · 23 bookmarks
https://x.com/PeterJ_Walker/status/2099607298025116044
Dan Gray (@credistick) — Sep 14, 2026, 5:24pm ET
@PeterJ_Walker How much weight is Google pulling here?
0 likes · 0 RTs · 1 reply
https://x.com/credistick/status/2099610223631556681
Peter Walker (@PeterJ_Walker) — Sep 14, 2026, 5:26pm ET
@credistick differs whether you mean spend or tokens, but enough :)
1 like · 0 RTs · 0 replies
https://x.com/PeterJ_Walker/status/2099610781168058726
The Open/Close · Research commentary, not investment advice. Positions may be held in securities mentioned.