The alpha, twice a week
That was 2 min.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
MTS · 2 min · Oct 30, 2026
Sovereignty means portability and model customization
Marjorie Janiewicz defines AI sovereignty as choosing where models run and adapting them to a company's data and IP.
The episode’s best minutes
Full analysis- 5:20Sovereignty means portability and model customization2 min
- 11:40Efficient, task-specific models can compete on less funding2 min
- 13:36Open models give companies ownership of AI2 min
- One more moment in the full episode
“Model customization is really starting to be absolutely existential for companies for multiple reasons.”
Takeaways
- Portability lets companies deploy models where they choose.
- Customization can improve efficiency and performance.
- Models tuned to company IP can outperform closed alternatives.
About this segment
Marjorie Janiewicz defines AI sovereignty as choosing where models run and adapting them to a company's data and IP. That can give companies more control and a way to differentiate their models.
Marjorie Janiewicz is chief revenue officer at Mistral.
Theo Jaffee, Host · Co-founder & host, MTS
One of the first members of a16z's new media team; earlier hosted the Emergent Ventures-backed Theo Jaffee Podcast.
Sophia Dew, Host · Host, MTS
From “You Don't Need Billions to Build Frontier AI | Marjorie Janiewicz”, Oct 30, 2026. Watch the full episode
Thanks. We'll look at and fix what's off.
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 5:24
Yeah. I mean, enterprises and public sector alike want to own their AI. They want to control the AI. They want to make it their own. So there's two very important concepts of what sovereignty means. The first topic is I want to be able to really install those models wherever I want, whether it means on premise, on my own GPUs, or on any upper scalers of my choice. They don't want to depend on the third party that can disconnect those models for their business continuity.
- 5:59
So it's really the concept of portability that's absolutely essential. The second concept that really is tightly linked to the concept of sovereignty is the ability to customize those models. Model customization is really starting to be absolutely existential for companies for multiple reasons. One, a model customization can mean you can customize those models to be more efficient, more performant.
- 6:27
You can customize those models for your IP and your data. And you know what? When you customize a Mistral Lodge 4 to the customer's IP or the customer's data, it outperforms really close-source models. It means that suddenly companies, instead of using a model that's being used by a lot of other companies, they can build their own models. It's creating new opportunities for innovation, and it's also creating new opportunities for those businesses to differentiate themselves.
- 6:57
So really, for everybody, really, I mean, it's not only in Europe, it's in the Middle East, it's in Asia Pacific, it's in Canada. It's also in North America for companies that are really supporting global markets. Everybody wants to make sure they own this AI, meaning they can really deploy it anywhere they wish, and that they can customize this AI to their IP and their data. You cannot do that with closed source alternatives.
- 7:24
And now, if those models are frontier, the way Mr. Large 4 is, you do not have to choose between frontier capabilities, best of breed models, and the ability to deploy them anywhere, control them, and customize them. That's why this is a very important moment for the open world community, but most importantly, for enterprises that want to derive true value out of those models.