# Rui Zhang of HeyGen shares GPU optimization and scheduling lessons for lower AI inference costs

> Original title: Lower serving costs start with the GPU.

- Source: SGLang
- Published: 2026-11-09T02:26:00.000Z
- Bad Influence: https://badinfluence.news/items/ra4hgg18yuo9zanb0nj81qzve
- Original: https://x.com/sgl_project/status/2119617472802459769

## Summary

HeyGen's Rui Zhang discusses GPU performance optimization and intelligent scheduling, drawing on lessons from running production AI inference. The post frames the GPU as the starting point for lowering serving costs.

## Post

Lower serving costs start with the GPU.

Rui Zhang from @HeyGen on GPU performance optimization and intelligent scheduling, with lessons from running production AI inference.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWphIXuaEAAaC2S.jpg&mode=thumb&exp=1794528000&sig=cce87ebd9768c1a4)