Well you really can’t build an AI lab in Europe.
Pleias CTO Pierre-Carl Langlais argues building AI labs in Europe is unfeasible, prompting pushback on European research contributions
Story Overview
Pleias CTO Pierre-Carl Langlais stated that building an AI lab in Europe is effectively impossible due to unspecified structural or regulatory barriers, igniting a thread where Hugging Face engineer Merve countered that Europe drives most foundational research milestones while US firms focus on turning them into products.
Research origins often trace to European labs
Merve highlighted examples like DeepMind's origins in the UK and its later acquisition by Google, noting that long-tail research rarely happens inside US companies focused on quick commercialization.
Exact obstacles to European labs stay unspecified
Langlais offered no further details on the claimed barriers, leaving open whether they involve funding, regulation, talent, or infrastructure.
Some users express hope that community efforts could overcome Europe's barriers to AI labs despite regulations, while many others see investor reluctance and bureaucracy as insurmountable and suggest migrating or giving up.
No Digg Deeper questions have been answered for this story yet.
Most Activity
Let’s just do business and move I guess.
Well you really can’t build an AI lab in Europe.
@mervenoyann @Shekswess Not really the case anymore: EU research is undercapitalized at the root, fast outdated. I'm reading US/Chinese outputs all day.
@Dorialexander @Shekswess I have the feeling that most research milestones are made in Europe, research can't happen in US (research has long tail and not to be made a product) US makes into product and sells (e.g. DeepMind, former Brain etc)
yes but we don't know which AI labs are purely in US, e.g. big closed AI companies have research offices in Zürich for sure Gemma was trained in Europe for instance and they were clear about it
US companies make efforts to build labs here and it's annoying to have export controls subjected to EU because what if part of model advancements were made here
I just think we have no talent shortage but EU is non functional which makes the whole situation meh
@mervenoyann @Shekswess Not really the case anymore: EU research is undercapitalized at the root, fast outdated. I'm reading US/Chinese outputs all day.
@Dorialexander @Shekswess I have the feeling that most research milestones are made in Europe, research can't happen in US (research has long tail and not to be made a product) US makes into product and sells (e.g. DeepMind, former Brain etc)
Well you really can’t build an AI lab in Europe.

@Dorialexander ctx? :)

@StijnSmits just one more thing that won’t happen

@PLouisV You can build business, but no way to get ressources needed for actual research (what makes an AI lab)

@Dorialexander You can’t build in Europe or can’t build with European capital?

@POTFES you need even more capital and soft corruption

@Dorialexander Did Lego broke you 😭?

@s3nhxx @Dorialexander @Shekswess I think it's also capital shortage, we have no centralisation unlike US, on top of all the big companies keeping their money there (due to taxes) and hot investors too although mistral, h company, bfl & others did raise huge rounds the rest is due to bureaucracy

@Dorialexander maybe pivot to AI infrastructure instead

@Dorialexander So we all migrate somewhere else? 😔 Maybe to USA.

@tugot17 Well it was bleak dark humor on top.

@mervenoyann @Dorialexander @Shekswess From my perspective the regulations are killing the easiest of deployment genai solution which move the centrr of distribution of task to bureaucracy, formalism etc
@Dorialexander @Shekswess the issue is being fully European ofc and having sole sovereignty and not here and there
@Dorialexander @Shekswess I have the feeling that most research milestones are made in Europe, research can't happen in US (research has long tail and not to be made a product) US makes into product and sells (e.g. DeepMind, former Brain etc)

@Dorialexander But... but... but he said... he said... I thought... 👉👈
@mervenoyann @Shekswess My preoccupation really: we have the base talent, but no way to really practice on current models/training approach. And this poured into bad policy choices that further increase the lag.
yes but we don't know which AI labs are purely in US, e.g. big closed AI companies have research offices in Zürich for sure Gemma was trained in Europe for instance and they were clear about it
US companies make efforts to build labs here and it's annoying to have export controls subjected to EU because what if part of model advancements were made here
I just think we have no talent shortage but EU is non functional which makes the whole situation meh

@Dorialexander yeah we cooked

@Dorialexander Maybe if we all unite, yes, the community can do more than an entire nation, or in this case Europe :)