Like all good stories, the tale of EurIPS begins in the kitchen…

A few years ago, Serge invited Aasa and me for dinner and a playdate for the kids. While Serge was preparing a delicious gnocchi dish (note: if you get the chance, you should let Serge cook for you!), we were discussing the pros and cons of European versus American academia. The reason for this topic was that Serge had recently relocated from Cornell to the University of Copenhagen.

One of our shared frustrations was that many top conferences (CVPR, NeurIPS, …) are predominantly held in the US, making attendance more difficult for European researchers. I remarked that the community should experiment more with distributed conferences, and jokingly added that we should start with NeurIPS because the naming would be easier: EurIPS.

We had a good laugh (and a delicious meal), and otherwise enjoyed the rest of the evening. But from time to time, the discussion resurfaced in various forms.

We fast forward to the beginning of 2025. At that point, it was becoming increasingly clear that there was a demand for a European venue for NeurIPS, and our gnocchi-induced discussions restarted. We, therefore, invited members of the ELLIS unit in Copenhagen to meet in the dungeon basement of the Pioneer Center for AI in February 2025. About 20 people joined. Perhaps unsurprising, the result was around 20 different viewpoints on the best course of action 🙂

Not having a clear direction, Serge and I decided that it was perhaps best to get some numbers. In late March, we decided to run a short online poll (shared on LinkedIn and Bluesky) asking if people would be interested in a European venue for NeurIPS. I’m not a social media person, so I was rather overwhelmed when more than. 2000 people replied positively in just a few days. Clearly, there is an interest in the community.

With those numbers, we decided to take action. Serge approached the ELLIS Board to get their opinion, and Bernhard stepped up. With his guidance, we wrote a two-page proposal to the NeurIPS board suggesting that they should support some sort of European branch of NeurIPS.

And here the long wait begins…

We were, honestly, not expecting much of a (if any) reply. We did feel a bit like ‘crazy internet people’, so our expectations were low. We, thus, spent time discussing various alternatives, including starting a new conference called ‘Eurainus’, which arguably is the best name ever for a European AI event 💩

Many things happened in parallel in this period, but a key point was when the organizers of ‘NeurIPS in Paris’ reached out. Clearly, EurIPS needed to pick up a lesson or two from these folks (a good bulk of them are today heavily involved in the EurIPS organization). They also pointed to the beauty of the virtual author registration, which, at the time, solved many logistical challenges.

Mid-April, I travelled to DALI in Sorrento – a wonderful conference in a wonderful town. I was mostly expecting to talk about Bayesian neural networks, but already on the first day, Bernhard and I had extensive EurIPS discussions over breakfast. I ended up spending the bulk of the conference discussing EurIPS with other participants, because rumours of the idea spread like wildfire. During the conference, ELLIS had its annual assembly, where EurIPS was also discussed. The interest was overwhelming!

And that’s when the reply came. The NeurIPS board was interested in discussing EurIPS as an experiment. Here I was thinking that we were being treated like crazy internet people, but instead, the NeurIPS board was discussing the proposal in depth.

In all honesty, our initial proposal to the NeurIPS board wasn’t all that great. Practically all relevant details were missing, so the NeurIPS board asked us to give them some more details to consider. At this point, the NeurIPS paper submission deadline was about a month away, so we, realistically, needed to produce a fleshed-out conference proposal in a matter of days.

Impossible, you think? Let’s go back one month in time…

At the beginning of April, Aasa was asking the basic question: What happens if we actually have to organize EurIPS? She reached out to a contact in Wonderful Copenhagen (a local tourist organization) to ask for advice on hosting a conference on short notice. They told us that such a project was `crazy, but not impossible.’ That was good enough for us. Anna – our contact at Wonderful Copenhagen – did a lot of legwork on our behalf. She found us a venue, made an initial budget, and even drafted a fancy-looking conference proposal.

So, when the NeurIPS board asked for a fleshed-out proposal, we were ready. A few days after their request, we were able to dazzle them (well, that’s at least what we tell ourselves…) with a professional proposal that was, indeed, quite fleshed out.

And the NeurIPS board decided to support EurIPS in 2025 🎉

Now things started to move really fast. We got connected with the NeurIPS organizing committee, who greeted us with open arms and cheered us along. Fundraising started (big thanks to DDSA and NNF for early yet strong backing). The team assembled. And so forth. We had as much drama as one can expect (well, more…), but those stories will be told another day.

End note: Of course, many more people contributed and much more happened than the short story above indicates. I apologize for cutting things short to keep the reader engaged.