AI Judge changed title after evaluation, original title: "Pleias co-founder Pierre-Carl Langlais argues Europe’s AI struggles stem from training skill shortages rather than a lack of compute"
Recent US export controls that cut European access to the latest Claude models have thrown a harsh light on the continent's deeper shortfall: years of assuming frontier progress could be imported while skipping the grind of building internal training know-how.
Doing the most responsible thing an European AI labs can do after this weekend: shipping a blogpost. Why the EU can't into AI, how it's not about compute, but actual skill issue and failing for years to build an actual training ecosystem. https://pleias.ai/blog/fable-eu
Continuous training skills like sparse MoE tuning, RL post-training loops, and models-that-train-models decay fast without daily practice, yet EU efforts have mostly funded fixed-term grants and policy papers instead of dense lab ecosystems.
Public clusters exist and some distributed-training talent has formed, but private inference-heavy setups and a thin market for data and buyers leave expertise as the scarce resource no budget line has yet fixed.
Users react to claims blaming Europe's AI lag on skill gaps and culture rather than compute, with some praising the analyses as excellent while others dismiss them as overlooking talent like DeepMind or fixating on irrelevant factors.
We especially intended to correct a major misconception: we don't have the expertise anymore. European research is fast outdated, and cannot get up to date to recent model research (let alone frontier). And worst, at decision-making level, we don't know we don't know.
Doing the most responsible thing an European AI labs can do after this weekend: shipping a blogpost. Why the EU can't into AI, how it's not about compute, but actual skill issue and failing for years to build an actual training ecosystem. https://pleias.ai/blog/fable-eu
The United States has sucked Europe dry of all the research talent, and for good reason too
We especially intended to correct a major misconception: we don't have the expertise anymore. European research is fast outdated, and cannot get up to date to recent model research (let alone frontier). And worst, at decision-making level, we don't know we don't know.
Applies equally for Canada, well written
Doing the most responsible thing an European AI labs can do after this weekend: shipping a blogpost. Why the EU can't into AI, how it's not about compute, but actual skill issue and failing for years to build an actual training ecosystem. https://pleias.ai/blog/fable-eu
I have almost nothing to add. Good, sober, unflinching post.
Doing the most responsible thing an European AI labs can do after this weekend: shipping a blogpost. Why the EU can't into AI, how it's not about compute, but actual skill issue and failing for years to build an actual training ecosystem. https://pleias.ai/blog/fable-eu
My intuition says that we are all generally under indexing on young talent and giving them the freedom to cook
Applies equally for Canada, well written
What can be done? A leapfrog? Maybe but not in any direction. As auto-research is heating up, frontier models (which the EU can't access) are bound to take the lead in architecture experiments. The harder but more promising path might instead be to target open-endedness itself…
We especially intended to correct a major misconception: we don't have the expertise anymore. European research is fast outdated, and cannot get up to date to recent model research (let alone frontier). And worst, at decision-making level, we don't know we don't know.
The reason is money. Absolutely absurd amounts of money
The United States has sucked Europe dry of all the research talent, and for good reason too
@yacineMTB Europe has guillaume lample, I believe in him
"quietly" ffs
@Dorialexander Great root cause analysis as usual
@Dorialexander It's seriously such a beyond cooked continent man
@umi33563 @Dorialexander Many things can be done till the end of this day!
@Dorialexander Excellent. Let's hope some key people will read it.
@Dorialexander there is more than one group of people being misinformed in one direction though.
@Dorialexander Great article, but would add one point: >Start with the labs P0 is compute. It should be done today, like literally on Monday 6/15/26, and it can be done today it's safest bet for any big capital allocation (worst case, rent to US labs if all fizzles, and recuperate money).
@Dorialexander It is largely about compute tho, but the reason they don’t have any compute is a skill issue
@yacineMTB At this point, its a race for Europe/Canada to escape Vinge’s slow zone
@Dorialexander
@yacineMTB dolla dolla bills, yall
@yacineMTB It's all about money, but Europeans consider this taboo and instead prefer to blather on about "muh health care premiums" as if that matters when you're pulling 400-600k TC. They will never wake up.