# 企业数据主权运动将撞上微调小模型成本难题

- 来源：Ethan Mollick (@emollick)
- 发布时间：2026-09-14 09:34
- AIHOT 分数：40
- AIHOT 链接：https://aihot.news/items/cmu0lh8t50tiyroryk4y282eo
- 原文链接：https://x.com/emollick/status/2099310692532699608

## AI 摘要

Ethan Mollick 认为，企业数据主权运动将直面一个挑战：只要 AI 能力持续指数级增长，持续微调小模型并让其稳定击败更大的通用模型既昂贵又困难。他引用一则 2026 年 9 月的设想场景：一边用 GPT-6 Astra 跑多州研究项目，一边花重金买 GPU 去微调 NVIDIA Nemotron 3。

## 正文

I feel like the data sovereignty movement for firms is going to run headlong into the challenge that maintaining fine-tuned small models in a way that continually beats larger general-purpose models is expensive and difficult as long as the AI exponential ability growth continues

### 引用推文

> prinz：September 2026: - I use GPT-6 Astra to run multi-state research projects that would have otherwise required dozens of hours of human labor - Latham spends $[red...