Yeah, good morning everyone and thank you Fabian for having us. I'm happy to kick off this conference with the first panel and with two panelists most of you have heard of before. Each of them has founded one of Europe's most important and successful AI companies. And many people are hoping they can help answer one big question. How can Europe stay competitive in the AI race? So, please join me in welcoming from Finland the founder and CEO of SiloAI, Peter Zalin, and from H Highleberg, the founder and CEO of Alfa, Yonas Andrudes. Hi. All right. So, our panel is called Good News from Europe. So, to prepare, I looked through my team's reporting from the past few months.
And here's what I found. US President Trump announced Stargate, a colossal AI infrastructure project with up to $500 billion in investment. Deepseek, a littleknown Chinese startup, realized released a model set to outperform the latest version of OpenAI's GPT4, and it shook up the AI world for days. Openai closed a record $40 billion funding round, raising its valuation to $300 billion. And both Entropic and XAI also raised theirs significantly. So I thought to myself, did Fabian invite me really to moderate a panel or to launch a full-blown research project just to find some good news from Europe? So before we start uh looking for some or any good news, I might have overlooked Yunas and Peter, how did you feel about those headlines?
Do they still worry you or Yeah. have you stopped worrying and comparing us to the US and China altogether? What do you think? You want to start? Yeah, sure. It's a big question. Um and uh I mean I think overall it's good for humanity that we're seeing progress. Um uh I at the same time as uh Fabian and Daniel highlighted uh and you highlighted um implicitly it's a reality that uh we are not seeing the investments. We're not seeing all of the development in in Europe um for Europe to to compete on on all matters. Uh but at the same time I um I do think that Europe has its strengths and Europe has a possibility to compete in AI.
Um but AI is not necessarily just one thing. It's not necessarily just about uh building general purpose models. Um, and my take is that uh the investments that we're seeing now with the AI gigafactories, even though I do agree that there's no full clarity yet on where the money is coming from and how much. Um, but I do think that if that sort of infrastructure and capacity is spent well, um, it can build competitive edge and competitiveness in in Europe. How do you see it, Jonas? First of all, I agree with uh everything Peter said. Um but there's maybe two things I want to highlight. First of all, infrastructure investment is not the game. Like this is a distraction.
Um we need value creation. We need innovation. We need so like for the future of Europe, we need two things. We need value creation and value capture. And this is innovation and sovereignty. And of course once we build once we transform our industries or our governments we will need infrastructure but this is something that will follow automatically in a way right I mean once we create value we can invest this into infrastructure it the like the other way around it doesn't work so we cannot just build infrastructure and then hope that the value creation happens automatically so we're maybe in a way thinking about the second step before we've got the first really done. I think that's important and we should not be distract and I I know I understand that it's like quite appealing to build a great data center, something you can touch and I'm sure the sales people that want to sell you this data center, they have very compelling arguments why you should invest a billions into a data center.
But the most the more important thing is transforming our value chains and get getting our industry and our governments into a strong position. I think that's one aspect of that. Thanks. And and on the other thing, what were you expecting? This is the biggest technological shift in humanity. And we're talking about the world's biggest and best companies. And I mean having having worked we all have worked in one of those companies or with those companies they are smart they are fast and they are ruthless in like their strategy around like dominating the future and so with everything we see now with everything AI will bring did we expect that it's going to be easy and they'll leave the field for us and they'll scale down investments and wait for five years until everything settles down.
No, this is this is this has always been from a European perspective a mission against the trend like we've missed so many technologies um and we failed to establish industries and and like modern technologies in Europe. But I also with all the conversations we currently are having, I don't think Europe is done yet. There there is a lot of brilliant talent. There's phenomenal leadership in Europe. There's a lot of money in Europe. And I think Europe can do a lot more than than what we've done in the past. And more is to come from Europe. But certainly we have to we have to pick up the pace. cannot we cannot just relax on this on the status that we have and kind of say well we do like a little bit of a small startup here and then we have this university and they have a great paper this it's not going to be enough right before I come to the next question I have one question for the team so there's an AI or
something else behind the sofa trying to engage in our conversation can we please stop that I don't know it's come from loudspeakers behind the sofa um so yeah you want us to build up on Um, Fabian has already clarified his point to us. He's not looking for the big headlines and not for the big data centers, but for the tiny little success stories of European companies starting their AI transformation. I mean, both of you be honest. As the founders of two of Europe's biggest AI companies, can it really satisfy you to just be responsible for AI baby steps? I mean, I maybe just answered that half uh with the last answer. No, I'm I'm not happy with that.
It's not going to be enough. We need We have some of the world's best companies in Europe. We have phenomenal countries, governments, and we have great startups. And of course our expectation cannot be to build a hyperscaler that rivals in terms of capex rivals what what the what the tech giants have spent. This is like in doesn't make any sense but we should aim for big moves and those big moves doesn't need they don't need headlines but I have the ambition that every one of our partners I'm fighting for their position their strength I want all our partners to be at the forefront of transforming their industries and they should win big and and then we can win our win can only come through the win of SAP, Bosch, Schwarz Group, Dodger Bank like those companies need to win and then startups like ourselves can also win.
Mhm. Yeah. So, so I think we need to be specific about things when when we talk about sort of AI companies and digital platforms, tech platforms um and and and eventually commercial success in Europe. Um, I mean, yeah, silo um built a large private AI lab. We we've operated with some of the largest companies in the world and and and built AI wi with them for them um and and and and seen um sort of what's happening on a wide spectrum everything ranging from autonomous vehicles to smart devices to large language models um and and built our own models for for European languages and so forth. But I mean eventually I think what what Draggy said in his report last autumn even though he talked about AI eventually what he talked about was tech platforms.
Um and it's it's it's it's not about having the biggest data center or infrastructure or um even having the biggest AI lab. Eventually, Europe has to have products. Products that end users want to use products are not an AI or models. Products are you the end-to-end experience of being able to create value to an end user in a specific context. And of course, AI assistance is one part of that, but it's so much more than that. I think we live in a reality where the tech platforms of today are not coming from Europe. Um I I think there are some obvious reasons to that. This sort of what we call single market is not a single market.
It's an extremely fragmented market and for you as a tech platform, as a digital platform to scale into Europe, it's extremely difficult and we're sort of continuously hearing from entrepreneurs. Um, exactly that problem. And it's not about access to capital. It's not about access to talent and so forth. It's about access to a market where you can scale. And many European entrepreneurs eventually leave to markets where you can more easily scale. So I think you know when we are looking for this AI success story I think we should be looking at the sort of underlying structure where we operate and see whether you know that's actually something that allows us to scale. Now that's one part.
The latter part is that I do actually think that if we look at the next wave of AI, we will we are already seeing everything ranging from the Nobel Prize last year going to uh uh say biology and and life sciences and and and you know a range of different industries that will be transformed and there I think actually Europe sits in a pretty good spot with some of the largest bioarma companies uh with for instance significant players in automotive where we're seeing a transformation and robotics as well and so many other industrial domains that are not equally dependent on owning the tech platform. Right? Let us come back to that later. Uh Jonas, there has indeed been some good news from Europe.
Um at Davos Alfa introduced an innovation that could significantly improve how companies use AI Tree. what is it and why do you see it as an important step? I mean that's joint work also with like Peter's team. Um it's so there's I'm give me the secret hand sign if I'm too nerdy. Um so the the architecture that everyone is using is that called transformer. All LLMs are built on that. And what every transformer so far does is um transforming text into a numerical representation with something called a tokenizer which is a fixed set of vocabulary. And because this vocabulary is optimized on the distribution of the training data set which is English heavy and which is general purpose heavy, the tokenizer is not well suited towards different languages or the knowledge that some companies have for their internal processes.
Right? Follow us so far? Okay. So now people are basically saying, okay, we can still take an open source model and fine-tune our new language or our our enterprise data onto that model. And a lot of companies and and have tried that and and sovereign states have tried that and they all came out of it or mostly came out of that disappointed. It's underwhelming and the reason is that basically what the tokenizer does is is creates a highdimensional space of like token combinations and if you want to shift the distribution in higher dimensional place to a very different position usually it doesn't work and so what we did with with Peter's team is we um changed we we we introduced an evolution of this architecture that doesn't have a tokenizer anymore so no limited vocabulary and no optimiz optimization no purpose op like no no customization optimization of the model towards any distribution in any training data set.
So it's this model can represent all text representation equally well and we've shown that for finish which is arguably an extreme case but for finish we we achieve 400% efficiency compared to llama and we've also shown that this architecture is much better suited to fine-tune new outer distribution knowledge whether that is languages or whether that is company data into these models. So this is uh from from our perspective and and our customers seem to agree uh this is a gamecher for sovereign AI strategies because now as a company as a sovereign state you can use this architecture to master to build models that are in quality far better what what any uh other offer can give you.
So as Yonas said you indeed tested the solution with the new parent company AMD. Is it as good as he says? Yeah. So I I there was a moment of silence. No, let let me give let I mean I I think the credibility comes from the fact that we're we're we're actually proving that the models that we've done together and this let me sort of tell you this from another perspective but but but the same thing and yes um uh but but the angle is really that from our perspective that uh we're together proving that what we're doing now together is better than the models that we built before. So also also better than llama also better better than llama but but llama was never supposed to be optimized for for finish whereas our models were supposed to be optimized for finish.
Um so we've been solving this sort of low resource uh um low resource language problem for the past several years. Some of our researchers have been doing that for the past 15 years. Um so it's not a new topic. Um, and we built be best-in-class models uh for Finnish that covered also English and programming languages and then covering all the Nordic languages and then all the European languages. Um, and we've used specific approaches that that taps into a cross-lingual signal between the low resource language and the high resource language. So, Finnish and English or Nordic languages in English and and so forth. And that was a way to sort of get to performant models uh that that sound like a Finn rather than like an American speaking Finnish and so forth.
But uh the problem with with these in in most cases is that they don't come with high reasoning capability. Uh when you sort of tap into low resource languages, you basically just don't have uh the data to allow for similar reasoning capability to say Llama and and many others. Um and now what what we are doing basically now is looking at it from another perspective where we basically take the reasoning capability of an extremely powerful model and then we adapt it optimize it for a low resource language and that's giving us better models. Yes. So maybe you should be aware which uh languages you really want to translate but but see. So speaking of your parent company uh if we look back to the time since the last rise of AI conference, there was one truly major uh piece of AI news from Europe.
You sold Silo AI to US chip giant AMD for $665 million. It was one of the biggest, maybe even the biggest AI exits in Europe. Um, the question is, wasn't that just good news for you or also for Europe? I mean, the title is misleading because you write the news, right? So, so I don't invent them. So I I mean this you're not the first person to ask this and it's not the first time you're asking this either but um it's so I I believed then I still believe that we can um create more value and have a larger impact in Europe um as part of AMD. Um and that of course sort of requires that uh many things um u sort of happen.
It it implies that silo is still a European entity. We're headquartered in Helsinki in Finland. Uh we have offices across Europe. Uh we had then we continue to have we continue to grow. Um and uh of course we have a global perspective. Um but uh uh at the same time it's obvious that we are sort of operating in the vicinity of uh our offices and and where we uh live and operate and creating more value um uh to those companies that that um are close to silo. So um what basically happened was that silo got more resources. uh silo got a sort of global agenda um and uh we continue to do basically uh the same things that we did previously.
So as I said earlier, silo uh was a private AI lab basically helping worldleading uh AI initiatives succeed and we continue to do that. So you know it's it's u as part of AMD it implies that we of course have an even larger focus on some of the largest AI initiatives or LLM initiatives. Um but what we are also sort of now tapping into is these different uh verticals and different industrial work streams. So do you buy that? Yes. Um so in success I like to me this is a success for the worker silo and a success in a success as at that magnitude is a success for Europe. Um so first of all that second of all AMD I think is a is a great partner and there's no European alternative.
AMD is very collaborative. It's like an open system. They they they don't want to dominate as like a monopoly player. Um, so I think we need partners like that and to get like teaming having silo team up with AMD gives us easier access and and and strengthens the partnership and I'm sure that I I would I wouldn't have been able to start Alf Alpha without my own entrepreneurial past. So and I'm sure there's a lot of great things and some great things have already come from Peter himself and there's a lot of great things um to come from Peter. So I think this is just more resources, more success and of course it would also been phenomenal if uh like silo would have stayed independent and grew and became massively successful.
Um but this is not to say that this current version is not a good outcome for Europe. So you see a reference to your own story by selling your second company to to without my time at Apple, without the um the economic success, but also without the time I spent surrounded by brilliant people in their AI R&D leadership team. I would have not been able to start this company in 2019. Yeah. Okay. And AMD is probably Nvidia's most important competitor when it comes to AI hardware. Um so you You assured me about the new generation of GPUs. That's really good stuff. Why doesn't AMD yet have the same strong expectation as Nvidia and what can a small Finnish startup do to help change that?
Yeah, happy to give my perspective. You can then correct me. Uh so we we know everyone knows that AMD has the best hardware, right? uh and I the at the same time the reality is that there's some catch-up to be done on the software side um and uh you know if you look at the AMD's past 10 years um it's it's been an amazing journey um driven especially by Lisa Sue um first taking over the CPU market and and now taking over the GPU market and and um there is catchup to be done on the software side And that's very well known in the community. It's known by developers. It's it's known by the community widely.
Um and and that is explicitly where we are helping. Um we we were one of the first companies in the world to scale uh LLM training or or prove throughput scaling on AMD GPUs. We've trained all of our models on AMD GPUs. Um and uh basically now we are helping companies across the world do that. Um, and so I think it's an it's an obvious thing for us to to sort of help AMD succeed on the software stack. Um, and and that's where we need to do some catchup. And one thing is for sure about um the more good than bad news story um of the cyi exit. Since the sale, you've kept making headlines as the millionaire who gives every Finnish university an a AI professorship, invest in startups, and even saved the football club.
But probably the biggest news for you personally is the launch of Foundation PS. So, what's your goal with that? Yeah, I I don't write the headlines. Uh you write them. Um but and indeed they focus on on things that I I don't care too much about but um but yeah so foundation PSS is is basically foundation that I set up um last fall um with the intent to give back um and especially support world-class research in in Europe uh and and of course especially in Finland. Um I think uh Germany paved the way with the first Ellis Institute in Tubingan. Um and I I donated 13 professorships to the second illness institute in Finland. Um and and that sort of uh together with a number of other initiatives in Finland uh made it made it possible and and I think it's just extremely important that we have world-class uh AI research um in in Europe and in Finland.
I don't think it's enough. I I do really think that we have to make sure that sort of commercial success follows uh and commercial activities and world-class R&D initiatives outside of universities also follow. Uh but I do think that you know you can make a difference and and the way that I just chose was to focus on um attracting uh worldleading professors uh to Europe uh because I believe that the best people want to work with the best people and eventually that will have a significantly larger impact. Yeah. And to make this crystal clear, you've kind of become a national superstar since then. So in Germany, we often complain that startup founders don't get celebrated as heroes.
How do you see that? And do you think some Finnish Finnish teenagers now dream of becoming a founder who saves football clubs instead of just playing for them? Yeah, but that's probably not due to me, right? So I mean I think the world is changing. I think entrepreneurship is more popular than it used to be. I think it's you know more common uh to be a university student and already know that you want to start a company when you graduate if you graduate. Um and so I think you know we're that's a good thing. Um and um uh I mean then um I do think also that uh it's how should I say it's it's probably easier to be a somewhat successful entrepreneur today than it was 10 years ago.
Um and and and I think it's more more accepted in that sense. So uh even though I don't agree with all the headlines, I I I you know do feel that it's more about the millionaire part than the I do feel that there's a positive atmosphere uh around entrepreneurship and and successful entrepreneurs like uh like you rightunas for a while you were very much um in the spotlight too seen by media and politicians as one of Europe's biggest AI hopes but that also meant more people were watching closely And then came criticism like claims of an inflated valuation, weak models and so forth. So did that lead to a negative spiral, maybe even lost customers, or did you manage to turn things around since then?
Well, I mean, some things were plainly not true and um I'm I was happy to see that all the high quality media out of Germany and internationally didn't participate in that. But of course um we are we're in an era where once you you're so visible once you're so maybe overhyped then of course any criticism whether that is substantial or not will generate impressions and so will cannot drive the business model of of journalists. So I think it's part of the game. Um it's certainly not pleasant or helpful in any way, but to be expected and um in Germany, everyone in Germany and and maybe in the world um but especially maybe in Germany uh that that gets out of cover and expose themsel in some way face criticism and sometimes unfair criticism.
And I think the important thing is that this should not lead to you like giving up on Germany, on Europe, on the mission, on the team. U and of course there was some uncertainty and there were some questions on in our team and with our customers, but we could we could show that uh some of this stuff was plainly not true without any substance. Um and so yeah, um not necessarily pleasant, but I mean it's just just part of what we expected to happen. Is there any piece of advice you could share with founders going through similar challenges? Um so one of the things where I think we we did a did a good job in some way and and but but could have explained so our whole business model our positioning right so looking at it from today's perspective you and everyone here might say very smart positioning I spoke to a um duck CEO a few days ago that basically told me the same thing he thinks Our positioning is spot on.
We didn't waste money on things that are what how one investor called it the fastest depreciating asset in tech history. Right. So I think we so that's that's because you don't compete anymore in the race for the best uh AI language model. Well for general purpose models, right? So I mean we we all learned now that uh the capabilities of LLMs are don't generalize out of distribution meaning they memorize patterns in training data and um this kind of makes some of the benchmarks we're using ridiculous right so of course I can check with benchmarks like MMLU I can check whether the model knows what was the 12th US president or what was the kind of I the second Disney movie that hit more than a billion whatever.
Yeah. And the models memorize these patterns and they can repeat them. And of course if you have that specific question it's useful that the model has learned this pattern. But if you are kind of transforming um a company like Deutsche Bank or Bosch or Schwartz Group then the knowledge about the 12th US president is frankly irrelevant. Right. So um many companies and have realized that there are general purpose models and they have their use um but they are everything we measure them on is not an indication that the models are suited to tackle industry specific problems. And that's where our work comes in by saying we we extract all the general purpose knowledge but we trans translate it into a new architecture that's able to learn and learn new knowledge that is highly relevant for for enterprises and governments.
That leads us to a very important point. So what are the chances that the aging European industry can actually renew itself through AI? Do you think Peter? That's a that's a very important question. Let let me maybe say a few words still on on this u I mean of course uh without having a a good understanding of what what UNAS has gone through and and how the German market has been covering this sort of opportunity uh that lies ahead for AI. But but I do think that there is some resemblance to what I said earlier about basically um you know eventually if you are to tap into such opportunities that the market has been hyping you're not actually an AI or model company you're a product company and you should scale and I do think that you know we at the moment we're not seeing particularly many AI sort of companies or model builders coming out of Europe that that are proving scale.
Uh I think we're seeing a lot of uh extremely talented teams uh and capable companies uh in also building models out of Europe. Some of the most used open source models actually both in image uh and text come out of Europe uh and European researchers. But we don't see the scale. Uh and and I in that sense I think the the sort of path that Alfalpha has chosen um m makes a lot of sense because I think you have to also acknowledge the fact that if it's not as easy to scale in European markets then you might need to choose uh different paths than for instance the consumer market. Um but now to the question on on sort of industries um yes uh that I alluded to that already earlier in that um I think uh these sort of AI assistants general purpose AI models u that's the first chapter it's the trigger to this revolution um but it's it's sort of one use case it's an extremely important use case uh the way we
use chat GP BT or Claude or or other models out there is transforming sort of the way we operate uh and do certain tasks. Yes. But it's one use case and and with that I mean that we're we're we're sort of seeing a revolution happening in life sciences. We're seeing a revolution happening in robotics and automotive for instance. We're seeing a revolution happening in visual media, entertainment, gaming for instance. and and there's so many other industries will will see this going forward. And that's that's just why I think it's so important that we now sort of in Europe we we stop and we think about where we could actually play a significant role going forward.
And the reality is that we have some of the largest biioarma companies, we have some of the largest automotive companies. Um we actually have also some of the largest gaming companies in Europe. And we all know that these same transformer models that are being used for AI assistants and so forth, chat bots, um they are already transforming these industries. So this is where you know I think what we really need is sort of um more ambitious more more sort of courageous initiatives in transforming these traditional companies putting AI at the core and investing into large scale AI initiatives. Yeah, you repeat it again and again and I think it's worth repeating it um today. So that European companies focus too much on efficiency and productivity gains from AI.
But what does it take to come up with truly new ideas and to reinvent entire business models? Those are totally two different things. I mean the way I've said already earlier or for a long time I mean the last AI wave in 2017 2016 17 I said the same thing uh nothing has changed I mean you know if you're you're it's a buy or build question if uh you know if if you are looking for efficiency and productivity gains you buy the best products out there and you use the best tools out there and that's what every company should do that's not innovating that's just using the best tools out on the market and everyone should do that the same way we started to use PCs and and you know text processing and and so forth 25 years ago.
We should use AI powered tools today. But but that's a totally different thing compared to sort of putting AI at the core at the core of your strategy and then all the way from the strategy go to execution and and yes it requires significant budgets and it it sort of requires that you rethink um you know how how a drug discovery process looks like. you you rethink um what autonomous driving implies for you uh and you rethink what your for instance robot for a robotics company what it is actually doing and how it's operating and and it it's a significantly different uh sort of approach to looking at AI. Um and unfortunately I would agree with what you basically alluded to that a lot of companies in in Europe they're looking for efficiency gains and productivity out of AI you know the AI capability that they acquire and when we look at the business models of siloi alpha but also openai and almost trial this is no longer plug and play so like like many
um earlier software solutions from the startup world. So you both send teams into companies, work closely with them and develop custom solutions together. So to to really understand that is AI maybe not as scalable as everyone has hoped. So I mean what what Peter just said is we need to transform companies and the answer the right answer is different for every large company. Um, but this doesn't mean that there aren't there isn't a foundation to build like a scalable platform or an ecosystem. I mean, that's what we what we built with with Faria and we um just a few weeks ago at at Han Hanova Messa. We showed how we we created a foundation to think production and industrial processes um new and we've kind of already um we are kind of in flight with several customers that are doing precisely that that are of course using AI for efficiency gain but they're also thinking about how can we position ourselves in a in a postAI world like how would a an an automotive company
or a a semiconductor company um be run in a in an AI native way. Uh and I think both is both is important and and both is is uh necessary. Um and we should also when we compare ourselves to the US, we should not think of it as a zero sum game. This is like the goal is not to be bigger than the US tech giants or to win against the US. The goal is to build a great future for Europe. And that means a a that we have the our values represented. Our liberal democracy is is uh flourishing but also our companies and our governments can do the work they need to do to secure prosperity and and like a region worth living.
We are uh one of the big dangers we there's several in like giant challenges we have. The demographic change is a giant change. um the the climate change and in order to address these challenges we need innovation, we need collaboration and we need technology. I don't see any way any other way as we all so we already highlighted that um you aren't competing for the best large language general purpose model anymore but companies like Alfa and with tri from France have already burned through huge amounts of money uh trying to stay in the race that many say can't be won we haven't Not I mean it depends right. So um um where I'm coming from it's a lot of money but it's nowhere near what some other companies have burned.
Okay. So but nevertheless how can this go on? Don't you think that some companies or organization need to join forces to compete in that race? So certainly we need to join forces as Germany as Europe like we we have this um kind of the way we we approach things the way we think about things and maybe this is due to languages or to to government but even in Germany there is there is this uh kind of rivalry between the states right and of course this is not and this is not helpful right like we we need to join force ources because once we do we have a giant market we have an a phenomenal talent pool and some great companies and by the way we just said that there's no tech platform coming out of Europe I would argue there's one SAP like all the important business or like 95 whatever percent of all the important business processes touch SAP systems go through SAP systems and SAP is not as bold as some other tech
giants are, but it's a phenomenal success story for a tech platform that's crucial to operating the world and operating value chains around the world. And um I think that's that's phenomenal and SAP should um I hope SAP kind of further wins through AI and further establishes their their lead and yeah, let's build more of those. Let's build more ecosystems um founding ecosystems. Um, I think it's phenomenal what companies like Silo and ourselves and many others are doing because this is a new era of startups, a new era of companies. There's already been three phenomenal startups coming out of people leaving Alf Alpha. And of course, I'm sad to see them go, but it's the best way to see people leave.
Peter, when you look at the European landscape, is there a way to bring the different efforts together more effectively? Yeah. Um you just mentioned ecosystems of course that's a term that um not necessarily always is unambiguous um in its meaning. So but I actually think that ecosystems is the right word. uh we're extremely fragmented not only as a single market but we're extremely fragmented in um our sort of various efforts uh around AI around you know research in universities and and and so forth. Um I I do think um we actually have a good example to follow. um not sure whether that's a sort of good example to say here in Berlin but uh I think Paris is an extremely good example um where they've sort of built an extremely strong academic cluster uh they've attracted world-class R&D initiatives ranging from meta to uh open AI um Google uh and and and so forth and then eventually um we've seen many leading AI companies being founded um or that have moved to the Paris region
ranging from Mistral to to Poolside and Kyoai and and so forth. So, so I I you know I I think we we're too small to have AI initiatives spread out across Europe and every country to have their own and and so forth. I think we have to focus on on ecosystem building building strong ecosystems. Um in in Germany of course Halbrun is an example of that and and I think you know tubing and of of course as well and and what would be important is to see even further collaboration between such um and then I think eventually Europe needs to also sort of the this the public actors need to make decisions and pick winners.
we can't sort of foster growth in the same technology area in multiple different places. So what I'd like to see is that there is more focus in, you know, there's a quantum ecosystem built somewhere, there's an AI ecosystem built somewhere. Um, and eventually making sure that we have the world's best researchers, we have the resources there, we have uh world-class R&D initiatives um that allow you to eventually use that as a platform to to build your next AI company. As I got the impression that you would laugh to write the next headline by yourself, I have one last question for both of you. If there's one piece of good AI news you're still dreaming of this year, and that might actually come true, what would the headline be?
So, I'm happy to start. Um, my Christmas wish list um for for AI news is that our partners are leading the transformation of their industries. So, Schwarz, Borch, SAP, Deutsche Bank, Bora, uh, HPE that they're at the forefront of like a sovereign AI movement that is strengthening the ecosystems, strengthening those companies and making the world a fairer, balanced, more balanced world where not all the power and all the value is created with like a handful of companies. That's a beautiful story. I still wonder how that might fit into one headline. What about you? Well, the headline is your job then. Yeah, I I' I'd sort of put focus on on other things than AI to actually enable AI and and that's really to to decrease the fragmentation of the market and to strengthen the single market in in Europe.
Uh the EU Inc. initiative that was launched last year that's actually sort of now happening and we are actually seeing uh progress in how we can sort of make sure that you as one company can operate in all European countries um and scale towards all all European countries if we don't succeed with that I think it's going to be extremely difficult uh to scale consumerf facing uh or or overall customerf facing ing AI companies in Europe. See, creating headlines is an art. Our time is up. You want to Peter, thank you for being with us. Thank you audience for listening and have a wonderful conference. Thanks everybody. Thank you. Oh, thank you Larissa. Thank you Yonas.
Thank you, Peter, uh, for this great beginning of the rise of AI.