Pretraining fundamentally does not make sense anymore for anyone other than frontier labs. Although there are a lot of p…

SemiAnalysis Twitter · SemiAnalysis (@SemiAnalysis_) · 2026-06-11

SemiAnalysis argues that pretraining large language models no longer makes economic sense for any organization except frontier AI labs, attributing continued enterprise pretraining to career-signaling incentives rather than genuine ROI.

Open original ↗

Appears in

SemiAnalysis: Local LLMs Are the 'Great Leap Forward' of Inference — Structurally Doomed by Scale Economics

Extraction

Topics: pretrainingllm-trainingenterprise-aiml-strategy

Claims

Pretraining large language models is only economically justified for frontier AI labs.
Many enterprises and startups pursue pretraining primarily for internal career advancement and impact signaling, not genuine business ROI.
Higher-ROI alternatives to pretraining exist for non-frontier organizations.

Key quotes

Pretraining fundamentally does not make sense anymore for anyone other than frontier labs.

There are a lot of people at enterprises & startups who have 'Pretrainitis' to show 'impact' and get promotions, fundamentally, it doesn't make sense.