425 inference API providers now exist worldwide, mostly vLLM wrappers in a losing price war
vaibhavbetter · x · 2026-11-22
After an investor asked "aren't there too many inference providers?", Akashi203 scraped every catalog available — models dev, OpenRouter, LiteLLM, Hugging Face, etc. — and counted 425 inference API providers worldwide:
- 58 listed on models dev in all of 2025; 232 so far in 2026
- 34 added in August alone; 94 in November, with the month not yet over
- 32 are routers or resellers by name
His verdict: 99% just took vLLM, put an API key in front of it, and shipped. The only lever left is price — and a price war on rented compute loses more money the more you scale. Reposter vaibhavbetter compares it to web hosting in 1999.
(The thread includes promotional links to YC-backed inference providers like RightNow AI.)
More from Infra
- Shorting AI-buildout IPOs on day 3 averaged ~37% returns this year — vaibhavbetter · 2026-11-22
- Qualcomm nears $2B quarterly auto chip revenue, tipped to lead by 2027 — BenBajarin · 2026-11-22
- Gavin Baker: $2T AI CapEx has produced only ~$120B in revenue so far — josh_bickett · 2026-11-22
- Why a $200 subscription can burn $100k of tokens: it's all about infra utilization — RexDouglass · 2026-11-22
- Inkling 2 Mini hits 200 tok/s on RTX 5090 via vLLM with native NVFP4 — VoidAsuka · 2026-11-22
- Redditor predicts cheap GPUs will return, 'only a matter of time' — ridhwanss · 2026-11-22