Mistral AI Raised $110m Funding As A 4-Week-Old Startup

Our team has analyzed the investment memo that Mistral AI used to get their funding of over $110 million when they were only four weeks old based on ideation. Now this whole sentence sounds utopian. It’s not easily achieved unless they have a founder who is Elon Musk or Adam Neumann, for example, who already had a very successful startup. Yet, the founders worked in Google’s DeepMind and Meta and not in such big positions that would make this make sense.

Breaking Down Mistral AI Funding Memo

The overall feel of Mistral AI funding investment memo is that it is quite realistic. It’s not saying that they’re building something out of the blue. On the contrary, they start the first section of the investment memo saying that generative AI is a transformative technology.

“The last year has seen its spectacular acceleration and generative AI systems able to generate text, image conditions on text image. Those systems can help humans produce content, read, interact with the world.”

This first section shows that they’re already stating facts. There is no argument there. This, I would say, is a very useless introduction if anyone is starting an innovative AI startup. But Mistral is not.

Mistral is just stating a fact and acknowledging it, showing that they’re aiming to build a new airplane in an airplane industry that has already been established.

Mistral AI Funding Strategy

Then in the next section, they start talking about how there is an oligopoly. They indicate that they believe while this narrative is true, which is not to be argued, that they believe the strongest AI platforms are going to be based on the models themselves – the training sets and the training data. Anyone can build a new generative AI platform. I can do that in the next few days. But where will I get the tremendous amounts of training sets and generated models that would fit this narrative?

For any logical investor reading this document, it is persuasive a little bit. So it is possible that what sways this generative AI market are models, similar to what swayed the automotive industry was how strong the engine is, not the invention of the car itself.

European Market Focus in Mistral AI Funding

The next section they started to talk about is that the current generative AI did not meet market constraints. Now they say that specifically because they’re addressing European investors where there are a lot of market constraints. GPT, Claude, Gemini, all of those AI tools process quite a huge amount of personal data.

They don’t have the same regulations that fit the European market.

That’s also a fact, and until this moment, there is not a single European investor who would disagree on the investment memo of their funding, because they’re all facts.

The next section they started to talk about disrupting the market from Europe and talking about how they could actually have strong technology with strong models. Their solution was mostly patience to take more of an open approach when it comes to model development, as well as enhancing the quality of data.

Team and Execution

This is exactly similar to how they’re saying “we’re building a new car, but we’re focusing on an engine that is actually better and have very good energy sources, like electricity.” Then they start talking about their executional skills. This is quite important because in my opinion, this is the weakest point that this investment memo has. If you follow this investment memo by saying that Tim Cook, the CEO of Apple, is leaving Apple to work on this or one of OpenAI’s chief scientists is going to work on this, then it would make sense that they would actually raise and reach a valuation of $6 billion.

They start talking about their team, which is an interesting mix and does have people who worked in DeepMind as well as Meta.

But they were not really a surprising team- it wasn’t the CEO of DeepMind, for example.

Then they started to dig into the European models and their data sources to say how exactly they intend to get the information, as well as the clients they’re focusing on, followed by talking about their investment roadmap in details rather than just say “oh, we’re going to develop the model.”

The Future Plan

It’s more of a technical document, which I personally like.

But it’s not so technical because it’s quite short, making it a good investment memo document that would raise questions. I would imagine the first question any investor would ask after reading this memo is that they would want to see an MVP and they would want to definitely see a financial model.

After their roadmap, they talk about their next stages just as a call to action to say how they intend to raise $200 million for their fundraising rounds in order to train models that exceed the capabilities of GPT-4.

Of course that was a while ago. Now GPT-4 is a bit outdated. There are different submodels inside it like GPT-4 that are faster or better than it. Claude is also in the mix with Anthropic and they’re doing quite good work in this sense. So yes, the AI market is growing. This is how Mistral AI got the funding of over $110 million when they were only four weeks old.

Let’s watch how they perform in the upcoming few years.