[D] What do you think of Mistral.ai's value proposition and open source strategy?

Mistral.ai, a 4 week old startup founded by star researchers from DeepMind and FAIR recently raised a whopping €105m in their seed round just 4 weeks after set up.

Their leaked pitch "deck" made several arguments that I find quite interesting and debatable (not that I think they are totally wrong) given recent progress in research and open source.

• They seem to believe that a moat can be built around better models, or at least the engineering know-how required to pursue such endeavours

• That closed sourced models served behind an API poses integration challenges and safety issues that will become a bottleneck for adoption

• That open sourcing the majority of their work, puts them in a position to win over the very finite research talent necessary to keep them ahead of competition

• Open source models and tools accelerate the development of ecosystems and aid in dominance through standard setting. (personally I think this is their strongest argument). Very similar to what Yann LeCun cited as reasons for FAIR's commitment to open sourcing.

Curious what the industry practitioners and academics here think about this?

https://sifted.eu/articles/pitch-deck-mistral