Did Anthropic's pacing call already cost frontier enterprise spend share?

Opening

Ramp's spend split favors reading Anthropic's pacing call as already expensive for frontier share — with an open sample caveat. Ara Kharazian (Ramp) posts Astra at 13% of enterprise AI spend vs Fable at 8% this week, and argues Anthropic's frontier model has already fallen behind on adoption after the pace call; OpenAI growth is shifts from Sol and some Anthropic models plus net-new. Visible replies push back that Ramp may miss Global-2000 / regulated pipelines and that OpenRouter shows a parallel OpenAI spend lead. Tail unread beyond top replies (top-replies).

Sides

FOR — pacing already shows up as lost frontier spend

• @arakharazian [named] [principal] — Astra 13% vs Fable 8% on Ramp; pace call was a "big risk"; OpenAI gaining from Sol + some Anthropic shifts and net-new
• @firstadopter [named] — relays the Ramp framing that Anthropic's frontier model has already fallen behind on adoption
• @trmcdonald [anon] — five-point spend gap turns pacing from principle into strategy

AGAINST — wrong sample / wrong buyer

• @bmlascolea [named] — ETR-linked claim Anthropic still gains among Global 2000 even as it slips among smaller orgs
• @reachmaheshkmb [anon] — pacing is aimed at 12-month procurement / audit buyers; this week's spend share is the developer market
• @patrickdonohoe [anon] — Anthropic still dominates highest-margin Fable spend; OpenAI owns the rest of the Pareto; compute rationing may explain Ant prioritization

THIRD — same spend signal, different venue

• @PeterJ_Walker [named] [principal] — OpenRouter: users spent more on OpenAI than Anthropic last week for first time in >2.5 years; Astra top by spend, Luna top by tokens
• @victoria_neiman [anon] — Ramp = enterprise, OpenRouter = developer; Claude Code still #2 app by volume while routes go elsewhere

Room: Ramp principal on card-spend data; OpenRouter principal on router spend; Global-2000 counterclaim is secondhand via reply quote.

Quotes

OpenAI is winning enterprise spend at the frontier. As of this week, Astra takes 13% of enterprise AI spend vs. Fable (8%) per Ramp data. — Ara Kharazian (@arakharazian)

Anthropic took a big risk in its recent call to pace the frontier. It's frontier model has already fallen behind on adoption. — Ara Kharazian (@arakharazian)

OpenAI's growth is primarily coming from shifts from Sol and some Anthropic models, net-new usage too. — Ara Kharazian (@arakharazian)

OpenRouter users spent more on OpenAI models than on Anthropic models last week. This hasn't happened for more than 2.5 years. — tae kim (@firstadopter)

Astra = top model by spend last week / Luna = top model by tokens (by a LOT) — Peter Walker (@PeterJ_Walker)

Interesting data - we're finding that Anthropic continues to gain among the largest organizations (Global 2000 shown below). — Brad LaScolea (@bmlascolea)

What would settle it

Ramp AI Index (or comparable card-spend panel) publishes a multi-week series of frontier model spend shares with Anthropic vs OpenAI broken out Knowable — next Ramp public drops — through 2026-10-15

OpenRouter weekly spend share OpenAI vs Anthropic stays inverted for ≥4 consecutive weeks (or reverts) Knowable — OpenRouter insights — through 2026-10-15

ETR / Global-2000 (or similar enterprise survey) prints Anthropic vs OpenAI share among largest orgs for the same window as Ramp Knowable — next ETR / survey release — through 2026-10-31

Connects to

Pacing as mix, not cliff — Baker's mechanism that pacing reallocates compute rather than kills spend (discussion).

Spend vs tokens — OpenRouter's prior ~70% spend / ~27% tokens split at three US labs (discussion).

Astra as product — overnight loop-transformer claim on GPT-6 Astra (discussion).