Thread

1. simianwords · 1790175588 · focus · HN ↗

   We have people suggesting that ai is so costly to run that all labs are secretly subsidising tokens and we can expect a reprice soon.

   Then we have these articles that say tokens will get so cheap that labs won’t know how to make profit.

   Who is correct?

   1. matteotom · 1790187220 · focus · HN ↗

      I find it difficult to believe the inference only providers (Baseten, Fireworks, Digitalocean, etc) are all selling tokens at a loss.

      Asking Claude for a rough estimate based on publicly available throughput and cost data for open weight models on modern GPUs suggests serverless, pay-as-you-go inference is profitable on owned GPUs with reasonable utilization (30-50%).

Unofficial Hacker News client; not affiliated with Y Combinator.