Top 1% of AI spenders direct over a third of AI budgets to raw GPU compute, Ramp data shows
rohanpaul_ai · x · 2026-10-26
Ramp's AI Index data shows the top 1% of businesses by AI spend direct more than a third of their AI budgets to raw GPU compute rather than LLM subscriptions.
Ramp counts LLM subscriptions, coding agent subscriptions, and API tokens as LLM spend, while GPU compute covers paying for hardware capacity itself — heavy users rent GPUs to train, fine-tune, or host their own models, not just call APIs.
The accompanying chart shows GPU compute's share hovering around 25%-56% from January 2025 through 2026, with LLM spend at roughly 24%-55%, indicating a remarkably steady compute commitment among the heaviest AI buyers.
Related event: Top AI spenders put over a third of budgets into raw GPU compute(2 posts)→
More from Venture
- Cursor contributed $608M revenue in ~1.5 months post-SpaceX acquisition, ~$4.6B ARR — DanielLockyer · 2026-10-26
- Gartner: 80% of neoclouds to cease independent ops by 2030, $4T at write-down risk — rohanpaul_ai · 2026-10-26
- Postiz-cheap ships with zero-restriction license: repackage and resell freely — matchaman11 · 2026-10-26
- SoftBank plans $50B AI and data center outlay in Malaysia — VectorWireAI · 2026-10-26
- Postiz-cheap: a free open-source alternative to the $99/mo social scheduler — matchaman11 · 2026-10-26
- Unlimited AI tokens, a $10,000 bill and 285,000 wasted lines: Deloitte finds 53% of CFOs only see AI costs when the invoice arrives — alex_verem · 2026-10-26