arthur mensch

T3 · host / generalist

2 calls·1 name·50% bull·last heard 5 months ago·All-In Podcast

track record

no scored calls yet — needs a stated position or a categorical verdict, with a matured window vs SPY

top calls

highest conviction · one per company

1sthigh conviction

$MISTRALMistral AIposition

Mistral bets on open-source models + portable platform for enterprise customization with proprietary data

Enterprises need to customize models on their own proprietary data (manufacturing, finance, etc.) which requires open models deployable on-prem/edge; Mistral's portable platform keeps data on customer infrastructure while enabling deep customization via forward-deployed PhD engineers.

All-In Podcast2026-03→

most discussed · click a bar to filter

$MISTRAL

recurring themes

Frontier AI Models1

2 total

$MISTRAL

Mistral AI

MEDarthur mensch·All-In Podcast·5 months ago· position

Mistral: Synthetic data useful for model distillation warm-up but human expert signal essential for final quality

Synthetic data from large teacher models efficiently compresses knowledge into smaller models initially, but human expert feedback remains necessary for high-quality specialization in vertical domains.

"We use synthetic data as a way to warm up the models. It's a way to actually be quite efficient at the beginning. If you have a large model and you want to train a small model, yo…"

73:40

$MISTRAL

Mistral AI

HIGHarthur mensch·All-In Podcast·5 months ago· position

Mistral bets on open-source models + portable platform for enterprise customization with proprietary data

"We believe that eventually building an open source technology is a way to save cost, is a way to have better control because you can see the thing on every cloud that you want, on…"

69:40