At Intercom we have a sizeable AI R&D presence in Dublin; you'll have to make your own mind up how far from the frontier we are, but we're doing some stuff.
It seems like one could equally ask, why are our maths departments so weak and underfunded when so much of AI, tech more generally and finance depends on having a mathematically-able workforce. We don't educate many to a *high* level (see PISA studies at secondary level); instead, we import outstanding people.
From an economic perspective, some might argue that it is unnecessary for *us* to fund advanced education and research, I don't know. It does hinder our young.
Nice post! It's unclear to some people in British (and I suspect also European) policy circles what they have to gain from buying AI chips. There's no domestic producer, so they'll have to buy from an American company. And there's no domestic frontier AI lab that knows how to use the chips, so the highest ROI move will be to rent them out to American labs. But then what's the benefit to Britain? What value are they capturing by buying chips from Nvidia and renting them to OAI?
That's what some people think, anyway. I take you to be pointing out that national HPC is the first step in a process that leads to domestic AI labs. When your universities have reliable, subsidized compute access, they can recruit better professors and grad students, and they can work closer to the frontier. And then you get positive spillovers into industry.
Very interesting piece, Conor. I’m not sure how best to explain this phenomenon, but it feels familiar from previous waves of technological change. It feels like a fear in the wider community of being seen as part of an avant garde. I wish it would change.
A genuinely excellent piece. Thank you. One minor quibble - "Ireland’s current position in AI – a place where products built elsewhere are localised and sold into the European market – is economically valuable, but it depends on continued multinational presence." - IME the products are not localised here, in any sense that particularly matters. Multilingual training for the model takes place off-island, product decisions are off-island, etc etc. The only localisation is the sales team and the currency translations.
At Intercom we have a sizeable AI R&D presence in Dublin; you'll have to make your own mind up how far from the frontier we are, but we're doing some stuff.
We're sharing some work openly, e.g.
https://arxiv.org/abs/2601.11471
https://arxiv.org/abs/2603.12133
https://arxiv.org/abs/2602.12078
https://fin.ai/research/
We'd have hundreds of H100 equivalents dedicated to R&D.
We're also good at post-training, and doing some pretraining.
I'd say we're legitimately at the cutting edge in applied AI. Fin has >$100M ARR.
It would be amazing if there was the R&D presence here in Dublin that there is in London or Zurich (per capita).
Those are tough benchmarks though; path dependencies arising from Deepmind, Google Research
I'm not sure if having plentiful compute would make a big difference; but it wouldn't hurt.
You definitely are doing stuff Fergal, but I don't think we can really say that amounts to a national strategy ;)
Oh, hello Fergal!
It seems like one could equally ask, why are our maths departments so weak and underfunded when so much of AI, tech more generally and finance depends on having a mathematically-able workforce. We don't educate many to a *high* level (see PISA studies at secondary level); instead, we import outstanding people.
From an economic perspective, some might argue that it is unnecessary for *us* to fund advanced education and research, I don't know. It does hinder our young.
Nice post! It's unclear to some people in British (and I suspect also European) policy circles what they have to gain from buying AI chips. There's no domestic producer, so they'll have to buy from an American company. And there's no domestic frontier AI lab that knows how to use the chips, so the highest ROI move will be to rent them out to American labs. But then what's the benefit to Britain? What value are they capturing by buying chips from Nvidia and renting them to OAI?
That's what some people think, anyway. I take you to be pointing out that national HPC is the first step in a process that leads to domestic AI labs. When your universities have reliable, subsidized compute access, they can recruit better professors and grad students, and they can work closer to the frontier. And then you get positive spillovers into industry.
Very interesting piece, Conor. I’m not sure how best to explain this phenomenon, but it feels familiar from previous waves of technological change. It feels like a fear in the wider community of being seen as part of an avant garde. I wish it would change.
A genuinely excellent piece. Thank you. One minor quibble - "Ireland’s current position in AI – a place where products built elsewhere are localised and sold into the European market – is economically valuable, but it depends on continued multinational presence." - IME the products are not localised here, in any sense that particularly matters. Multilingual training for the model takes place off-island, product decisions are off-island, etc etc. The only localisation is the sales team and the currency translations.