Paris is quickly becoming a major European hub for AI startups, and now another deal in the works could improve its position.
Poolside.ai, a Paris-based generative AI company that makes tools to speed up software development, is raising at least $400 million, at a valuation of $2 billion, sources told TechCrunch.
Bain Capital Ventures and DST are in talks to co-lead the round at the moment, sources told us. BCV is a previous backer of the company, and DST is a new investor. Bain’s participation has been previously reported, with PitchBook listing this as a roughly $450 million round.
Poolside made a splash (ahem) last August when it wasn’t another AI startup in town to raise a big seed round. It took $126 million from backers that included, in addition to BCV, early-stage specialists like London’s Air Street, Abstraction and Scribble Ventures, and New Wave and Frst from France. Bpifrance, Felicis, Point Nine, and Redpoint are also in the round. None of the investors we contacted would comment for this story. Poolside’s CEO did not respond to our request for comment.
Mistral and H, two basic model companies, are among those that have also raised 9-figure seed rounds ($113 million and $220 million, respectively) out of the city. City of Light should probably be named AI City at this level.
Looking at the bigger picture, it feels like the market is heating up, really fast, for AI startups, which have collectively raised billions of dollars when you include Anthropic and OpenAI. Do we need another basic AI company, you might ask?
There are several reasons why Poolside is getting this level of funding to take its own big swing at the generative AI opportunity:
– The strength of the founding team and its relationship to the company’s premise. Both are in the world of developer tools and DevOps. One of the founders, CEO Jason Warner, is GitHub’s CTO and leads engineering for Heroku and Canonical. The other, CTO Eiso Kant, previously founded Athenian, which has created a series of tools for developers to help them optimize the way they create and work. (The company was bought by the Linux Foundation for an undisclosed sum.) Warner also previously served as a VC at Redpoint and understands the value and language of interaction between VCs and founders.
– Problem solved. Unlike the basic LLM model that is more common in its approach, Poolside (for now at least) sees one specific use case: helping developers work faster. This will resonate with investors, and like the memorable essay written by Paul Graham many years ago about startup ideas, building tools that you know you will need, which by default means founders and possibly other technologists should.
Although people like Mistral also focus on developers and developer tools, but also Microsoft’s adaptation of OpenAI for GitHub (actually OpenAI is all about APIs that developers also use), there are some important examples of how the code works. one of the blind spots more problematic (again, for now at least) for LLMs more generally. Now, there is an opportunity to build on this, and build better than the more common approach.
That’s not to say that Poolside doesn’t think big either. In the three-step plan described on the website, it is stated that in the end it hopes to work, after the developer, with anyone who wants to write code and software; then, after that, “Generalize this capability beyond software to all other fields.” NBD!
– Early signs of what has been built. I have not seen whether the company has released a product not yet publicly available, but there is some evidence that they are used and many (and perhaps running in private beta?). Computing supplier IREN in April reported that Poolside had finalized a cloud services deal with the company.
– Last but not least, there is monetization. As one source close to the company told me, there are many examples of AI targeting different areas of the market, but not many with clear indications – much less evidence – of monetization potential. (Yes, the big issue that OpenAI has flagged with PwC could open the door to other companies’ business, but it may be a years-long endeavor; in the meantime, it remains to be seen what can be done at any company.)
However, with Poolside, when you consider the opportunity and scale of creating a co-pilot tool for developers, it’s probably one of the best and easiest places to implement AI, if done right. This is an obvious and huge need and all computer programming has a syntax, so it’s not nearly as open as many other areas where AI is applied. In addition, it has constraints, and performance indicators, and benchmarks – a winning combination for investors and startups looking to build a business.
We’ll update this post as we learn more.