Skip to content

← All Q&A

Why are fine-tuned models on proprietary data the real moat, not another foundation model?

fine-tuningdata moatcompetitive advantageLLMs

Drawn from Lutz Finger's Forbes column, LinkedIn writing, and Cornell teaching. Sources are cited inline so you can read the originals.

The same transformer everyone has becomes a defensible asset only when trained on data nobody else holds.

Cyriac uses the same transformer that is the basis for AI generally, but he trains it on data from his own wet lab that nobody else has, to get a fine-tuned model that can answer questions nobody else can. The future right now is really fine-tuned models trained on specific data nobody else has. There is a race on data and a race on fine-tuned models, and the same underlying idea is happening in every industry across the globe.

Source: Tomorrow’s Medicine, with Cyriac Roeding (eCornell Keynote)


Have a follow-up? hello@lutzfinger.com. Or pick another question: all Q&A →