Mistral AI: Europe's answer to Silicon Valley?

0:02Hello and welcome back to the Times Tech Podcast with me Katie Prescott and me Danny and this week i've been speaking danny to another very tall man tall as me taller than me how tall are you six four six six oh yeah i reckon he's probably a billionaire as well so i'm just losing on all fronts well his company is valued at billions so this is the ceo and co-founder of mistral which is france's and really europe's big hope in the accelerating race for ai supremacy the sort of only rival that we've got over here to open AI. So in a moment, we'll hear that. Yeah, and then we'll reconvene to discuss what he had to say before talking a little bit about Meta's big investment, very big investment in a bid to achieve computerized super intelligence.

0:54But first, Danny, I want to take you back to Gay Paris last week. My way, my way. Where I spent a couple of days, sadly not strolling down the Champs -Élysées, but near a conference center in the southwest corner of the city for VivaTech, which is France's kind of annual big tech get together. And it's really built up ahead of steam over the last few years. They've had Elon Musk two years in a row. This year, like London Tech Week did, NVIDIA's Jensen Huang was over there. And GTC, which I think we talked a little bit about last week, GTC, NVIDIA's Developers Conference, was hosted in the same venue.

1:30And so it was a real acknowledgement of France's advances in AI and particularly of this company Mistral, which is the only frontier AI company in Europe, which is making the sorts of advances that could rival Anthropic and OpenAI. And on stage, they had Emmanuel Macron, the French president, with Jenson Huang, the boss of NVIDIA, and this 32-year-old, very tall man called Arthur Mench, Arthur Mench, who runs Mistral and is one of the three co-founders. And it feels to me like the entire weight of Europe's AI hopes rest on his very, very tall shoulders. So before he went on stage to hang out with his very important friends, he got to hang out with me for 20 minutes.

2:18You're important too, Katie. I don't know about that. Backstage in a very busy press area. Well, what a lucky man. And obviously a wildly interesting time to speak to him. Yeah, really interesting. And I think for us on the pod, he's the last of the big LLM developers that we haven't heard from. Yeah, we've had them all. We really have, yeah, if you're kind of collecting cards on this one. But it was really interesting to hear from him, and he gave a very, very different perspective on the whole thing, not least because he really, really pushes this idea of AI sovereignty and the importance of control over AI and the risks, as you'll hear, that he sees, I mean, it's a bit of a marketing gimmick, of course, but of European companies depending on American ones for something as important as AI.

3:08What's the kind of, as they say in tech world, what's the TLDR? Too long, didn't read. What is his main argument? Why we need a European answer to OpenAI and Google DeepMind? It's really interesting. So he came out of Google DeepMind, as did his two co-founders, who are also French. Wasn't he at Meta as well? No, he wasn't at Meta, but his co-founders definitely were. Yeah. And France has a very rich history of brilliant engineers and very, very strong engineering universities. So it perhaps isn't a surprise that they are as advanced as they are in AI. But they decided that European companies need to have their own control over AI infrastructure.

3:49But also he sees a really important cultural point, which we'll hear in that if you're developing AI, which is a new form of content, he believes that actually there's a mentality behind it. And in the content you're generating, which shouldn't just be dominated by one country. And I think he's, because of that, you can imagine Europeans have jumped on board and it was very interesting to see, you know, President Macron on stage with him and the support that he's had politically, but also from the big US tech companies as well, right? So Microsoft is a big investor, as is Salesforce. But so far, the company's valued at$6 billion.

4:26dollars and if you compare that to open ai's 300 billion 80 billion anthropic at the moment and it does feel like he's got a mountain to climb he does and i think that's really interesting though that that cultural point and it you know it makes sense if you think about like if there's an analog to maybe i don't know hollywood you know hollywood has a certain type of culture and european films are different and you can argue whether it's good or bad that hollywood kind of dominates pop culture globally but it does have implications but in this case this technology is way more consequential if you're talking about as we'll talk later replacing jobs it just on a basic level seems to make sense to kind of have one that comes from a european point of view although i must say i don't really know what that means a european point of view and a linguistic point of view as well i mean you talked a lot about the um the chatbots that he's developing le chat by the way like cat in french is the yeah that's chat gpt version did he explain why they chose that name well i it was supposed to be a play on chat because that's how you spell chat cat in french i see i get it i get it.

5:50But the other thing that companies here are worried about, which I found really interesting, given what an important partner America is to Europe, is this idea that, perhaps because of the political situation in America, control of AI in, say, the West Coast means that that could be switched off. And he actually says that in the interview, you know, we cannot rely on companies that could switch off our access. Interesting. Yeah. And I noticed Brad Smith, the president of Microsoft, was over in Europe in March reassuring companies that that would not happen. Why would they even do that? Well, because I think Trump back in March put out a directive saying that if tech companies were threatened with too strict regulation, you know, there could be some retaliation.

6:37And so almost for the first time, there's a sense that actually maybe we need to look elsewhere for our services. Right. kind of like, if you don't play nice with us, we're going to turn off your internet. Yeah. Wild. So Mr. Al has recently announced its first reasoning model in a direct bid to compete with the big AI firms. But more importantly, and part of the reason that he was on stage with Jensen Huang and NVIDIA was he was announcing their push into data centers and infrastructure. And this plays into this obsession at the moment with Sovereignty and ownership. And that's really what they're pushing for with this announcement.

7:11Let's hear it. So a huge announcement from you today about the building of Mistral Compute. Could you tell me about that and why it's significant? And it's quite a deviation, isn't it, from what you've been doing before? It's an expansion from what we've been doing for the last two years. The last two years have been about first building the science brick, so building state-of-the-art models and leading in the open source front on that domain. and then from the models themselves, we also built all of the software to build AI systems, to build autonomous agents, to build the observability that you need to iterate and do the lifecycle management of AI applications.

7:53And so those are the things that we currently bring to enterprises also with our solutions team. And the natural expansion of that has been to say, we have the software that can run on private cloud, the one that we're running today. and we realized that there was a lot of demand in Europe in particular for a fully integrated player, a player that would go from GPUs, serving GPUs to building applications. And so based on that demand, we started to aggregate it and then we decided that for our own needs, for the needs of our customers, we should be building it ourselves. We should be reusing also the expertise that we gathered for the last two years in how to operate GPU clusters, large scale GPU clusters.

8:36And so we're really bringing two different expertise we already have, which is operating clusters and building applications and building models, et cetera. We're bringing them together to build a fully integrated AI cloud that is really resonating within the European ecosystem. I thought it was really striking in your press release that you talked about this upholding Europe's technological autonomy, and that it also being a significant step towards technological autonomy again. And Jensen Huang on stage earlier was talking about sovereign AI. Why do you think that is so important? AI is going to be multiple digits of the global GDP in the next couple of years.

9:16So what that means is that if you are within a certain economic space, if you're depending on another provider, that's basically trade imbalance. And by another provider, you mean a foreign provider? Yeah, a US provider. So you have an aspect that is really economical. At some point, having leverage, local leverage, when it comes to just providing technology is extremely important. So that's one aspect. There's an economical aspect to it. We want to create value in Europe, and all European enterprises have vested interest in seeing technological value being accrued in Europe. That's the first thing.

9:49Now, there's another aspect, which is really what we call sovereignty, which is about it's a technology that's going to run a lot of our lives, a lot of our enterprises, many processes that's going to have a meaningful impact. If you're relying on a provider that can basically turn it off from one day to another, then you're really at risk and you're giving them, you're giving that provider a fair amount of leverage that might be very helpful for you. And so you have a certain number of actors in certain industries that just can't afford it. So if you look at defense, if you look at public sector, they do need sovereign capacity because they don't want to be threatened by any foreign provider that could actually turn off the service from one day to another.

10:32Extraordinary, though, isn't it? Do you think that European companies are really worried about US providers, who are an ally, really, turning off capability in the way that you've described? Yes. I mean, if you look at defense companies, if you look at public sector, if you look at any industry that is running services that are critical for citizens, they're really looking at it from, I mean, it's a very serious matter. You can't run your infrastructure on a technology that can be turned off. And so that's where our last two years have been about addressing that need. So we've built a platform that can be deployed on-prem, that can be deployed on the private cloud of our customers.

11:11And so in a way we've been addressing that need for the last two years. But what we realized is that it wasn't that easy, even for our customers that were in demand for sovereign capacity, to actually operate that capacity. So being able to provide not only the software, but also the hardware underlying it in a way that gives our customers the turnkeys and that makes them comfortable. This is addressing a very big demand and that's why we're doing it. And have you started to see that reflect positively for Mistrales? Have you started to see customers come to you because you're a European provider and because they want someone who's working on AI in Europe?

11:45Yes, I mean, that's been a driver of the business in Europe for the software side. The reason why we're announcing this hardware effort is also because we have off-takers from Europe that are going to be using that technology. And not only is it useful in Europe, but it's also actually resonating in other places of the world. So in Singapore, for instance, and in all of the APAC region, sovereignty is also a very big topic because you also have public services. You also have defencing companies and you also have telcos. You also have banks. those companies also need a certain level of privacy, a certain level of confidentiality, and a certain level of control.

12:20And what we're announcing today with Mistral Compute in Europe, this is, in a way, sovereign for Europe, but what we are building is a template for sovereignty. So our platform can actually be deployed on all those infrastructures so that we can address sovereign needs in places where we are not the sovereign provider. So we're also using our platform to progressively grow onto other economic regions and just bring the software, partner with hardware managers, and then address customers that requires a local solution. It's fascinating to hear. And how much do you think this push for sovereignty is being caused by the rise of Donald Trump and the creation of instability that that's brought with it in America, which traditionally, I guess, has been the home of technology, like the stuff you're creating?

13:05Well, I think historically, Europe has been relying on the U.S. as a very friendly partner. it turns out that this has given them a lot of leverage because you have 250 billion of technology services that are flowing back to the U.S. every year and that's actually pretty meaningful for Europe. So there's been a realization that this was too much and what's interesting is that there has also been a realization that the Europe tech ecosystem was meaningful, could provide state-of-the-art solutions and could address the needs of enterprises in a way that is even faster than what the U.S. companies can provide.

13:41So there's a combination of Europe tech ecosystem, and it's not only about us, it's about other companies, Europe tech ecosystem becoming meaningful and becoming competitive, plus this realization that, yeah, this is needed. European companies are now willing to partner with European tech companies because they have vested interest in seeing it happen. How important is the cultural point here? I mean, you know, we're doing this interview in English, I'm very grateful for you doing it in English. Thank you. But clearly, there are so many other languages around the world. We're seeing so much AI coming out of the States.

14:14How important is it for AI to be built, for example, in French? So sovereignty, I think, has three pillars. And I only mentioned the first two, and you mentioned the third. The first is economic. The second is strategic, turnkeys, having a service that nobody can turn off. And the third one is cultural. This is a technology that is generating content. And in a way, you can see it as a new media. and because it generates content, it has biases, it has specific ways of speaking, it has to handle different languages and historically since this is a technology that has been built by U.S. providers, they focused on English, they poured their own biases into it and they pretend that they can actually set the biases in a universal way.

14:53But it's actually not true because there's some cultural differences across countries. Languages differ, approach to biases differ. And so by bringing a technology that can be customized to every country's needs, by making sure that we are treating every language as important, treating European languages, treating Asian languages, we've announced a model that is domain specifically good at Asian languages. We did it for Arabic as well. So doing that is also a meaningful way of giving a cultural tool to countries, to enterprises that wants to play an active role in shaping the culture of the next decade.

15:30I guess the other question is when you talk about bias is, I don't know if it's the right word to use about it, but financial bias. And how do you think about it when you are growing a business that's trying to rival OpenAI in the States and Anthropic, and you look at the size of the investment that's being poured into those businesses and the valuation that they've got. Do you really feel that Mistral can be a strong challenger to them? We do because we have a very differentiated story. We have proven to be extremely more capital efficient than U.S. providers. We've been able to always be close or at the state of the art while spending just one or two percent of their spending.

16:09So what we've shown is that by focusing on the right things, you could actually be more capital efficient. And what we are on our way to doing is to show that we can be the first company that has a meaningful business model around building AI systems.

16:31How many people do you employ at the moment? Around 250 and growing pretty quickly. Okay, cool. So what are we looking at in terms of growth in the next couple of years? We'll be north of 400 by the end of the year. And where will they be based? They will be based in multiple places. We have an office in Palo Alto, which is growing. We have an office in Singapore growing. The headquarter is obviously in Paris. London? And London, actually. We have 30 people in London. We're also expanding toward the rest of the European Union, both on the engineering side and on the commercial side. How much do you engage with your U.S.

17:03peers? How much do you talk to Sam Altman, to Dario Amadei in the States? I don't talk to them much, but we know each other. So do you know? But I mean, if you look at, I mean, the AI ecosystem is actually fairly small, and I think everybody is also working to build a meaningful technology. So it's also useful to see what everybody is doing, and we're obviously tracking, everybody's tracking each other. But what we're seeing is that we're starting to have differentiated models that perform differently in different settings, different business models as well. Because we were engaging, I think, much more deeply with enterprises than any of our U.S.

17:36counterparts. And so I believe that this is going to be a space where you will have multiple, not many, because the barrier of entrance is still very meaningful. But there will be multiple AI full-stack providers. And we are on our way to be the first in Europe and to expand globally. What did you make of the open eye acquisition of Johnny Ives company and what they might be doing there? The hardware business. I have no idea. I looked at the marketing video, but it didn't seem very concrete to me. It was quite a marketing video. That's for sure. You launched a reasoning model this week, Magistral.

18:14Yesterday, yes. Magistral. So these are models that can go through complicated tasks and follow a logical thought process. How has the take up been of that so far? Great reception. We're basically the first open source provider of a model that can reason about multimodal data. So if you put an image, you have an open source model that can take the image and then do some complex reasoning about it. Because of this multicultural approach that we've always had, we've made sure that all of the reasoning traces, which are basically the way that you interpret how the model has been thinking about the problem, we've made sure that you could have them in multiple languages.

18:50If you look at the others, they do it only in English. For us, you can do it in French, you can do it in Arabic, and it works extremely well. This is the first step. It's our first reasoning models. We have others that are actually coming very soon. We're very happy with the performance and reception. We can test it on Le Chat, which is our consumer app. And we're looking forward to how people are going to be using it and to also release models that are actually even better in the near future. How does it compare to, say, DeepSeek, for example, the Chinese one? It's comparable to the March version of DeepSeek.

19:18DeepSeek released months ago something that is even better. and so we are looking forward to our new release, which will be even growing in terms of capacity. Let's talk about some of the implications of AI. I'm fascinated to hear your views on what you're seeing. We recently had the boss of Klarna on our podcast who said that AI job losses could lead to a recession. We've heard Dario Amadei at Anthropic say half of all entry-level white-collar jobs could be wiped out by AI. Where do you see the job market going in the next few years? I think in Dario's approach to marketing has always been to spread fear of the technology it's building.

19:58So, I mean, it's not very surprising. I think it's very much of an overstatement. You won't have 50 % of the white-collar job that is going to be wiped out by AI. Effectively, what you will have is white-collar jobs being changed by AI. When it comes to doing a complex analysis, to doing a complex research, now you have tools that can actually work well, that can run research on your behalf and save you multiple hours a day so that you can focus on other things. It does shift the kind of tasks that we will be doing. I do expect that we'll have more relational tasks because that's not something you can easily replace.

20:31On the other hand, everything related to searching information and combining them and synthesizing them is going to be equipped by AI. I do think that one of the biggest challenges we'll have is to make sure that despite the powerfulness of the tool that we're deploying, we're making sure that humans stay involved in the loop and do not look at whatever output is produced by an AI system as the truth. So you want to keep people active. You want people to continue learning. And in a way, being able to synthesize information and criticize information is like a core component to learning. So one of the biggest product question, and that's really a design question, is how do make systems that are keeping the user of the system, the human user of the system as active as possible so that he learns in the process so that he can give active feedback to whatever has been produced by the mobile.

21:23I think it's the only way we can avoid having what I fear is the biggest risk of AI, which is descaling. Having people that progressively become lazy. For us, it's a very important design question. How do you make sure that whatever is being produced by the AI can be easily reviewed by humans? And how do you make sure that you're actually driving human intelligence further thanks to artificial intelligence and not the other way around? Right. So that's a real risk that we'll all just become very lazy. It's called diskilling. Yeah. I think diskilling is probably the biggest risk that we're facing.

21:53But it's a risk that you can avoid if you think of it from a design perspective. If you make sure that you have the right human input, that you keep the human active. It's funny you talked about Dario Amadei's comments as sort of marketing, the fear mongering, because it feels like in the AI world, you're a bit of an outlier in that sense. You know, when all of the others, all of your peers are talking about the push for AGI, you said in the past, you don't really think it's a worry or even something that's coming down the track. Well, I think AGI by definition is whatever you cannot do yet. And so in a way, it's a moving definition because it's, I mean, whatever you do, every improvement you're making to a model, basically AGI is again one step beyond.

22:34Because we're never going to reach a setting where you have a system that can do everything that humans can do and that can be omniscient and omnipotent. And so in a way, I think AGI, the way this is framed, is either about what we cannot do or about some form of God. We think that it's more of a product problem. Like you build a general purpose system and the one thing that matters is how do you adapt this general purpose system to enterprise processes, to the people daily life in a way that is giving them, respecting their free will, that is giving them enough privacy that is not building influence on them without them knowing about it.

23:14and we think that this is really a product problem. So for me, AGI is a product problem. How do you customize systems to make them meaningful either in your daily life or in the processes of an enterprise? And is that what you're most excited about when you look to the future of AI? What I'm most excited about is how to combine the general purpose reasoning abilities of the models that we are training to proprietary data, to the actual expertise in multiple companies that built these expertise for the last centuries. So we're working a lot with manufacturing in Europe because they have very meaningful expertise there and they are not using AI a lot.

23:54So by combining the two things, you can actually make unique differentiation and help them drive their business. So for me, this is really exciting. The combination of concrete science, concrete R &D that used to be not AI-driven, but that is still very, very meaningful and very expertise-driven, expertise-loaded, and bring the AI tooling that will just move it even further. Now, I know you talked about one of the unique selling points of the company is how efficient you are. The rumor is at the moment that you're fundraising and looking for a billion dollars. Is that true? There's been rumors about the fundraising, I think, every month for the last 25 months.

24:31So your guess is as good as mine. It's probably not, but thank you. Okay, last question. This is a little bit sidebar, but I just wonder, how are you dealing with the pressure of all this? I mean, you're being held up as Europe's great AI hope, the only company that's building generative AI on this scale here. You're still so young. How are you shouldering that and how does that feel? First of all, I'm not that young. You're a lot younger than me. We have a strong team on all of the functions of the business. We have very strong co-founders, driving science, driving the technological roadmap. Great team that is doing enterprise sales, a great communication team.

25:11so I think we've been able to we've grown very fast but we've been able to structure the company so that people have enough autonomy that the pressure can actually flow back also to the rest of the company. And you're about to go on stage with Jensen Huang from Nvidia, President Macron, I guess you know them both quite well at this stage. We interacted quite a bit indeed, you know that's exciting times and it's always enjoyable to do these kind of announcements with uh with great people well really nice to meet you thank you for taking the time i know you're great you can tell we're in a rush backstage can't you yeah i've only got 20 minutes to rush through all these questions um what did you think i love how kind of like he's like yeah dario he's his approach to marketing is fear oh that video that open air put out doesn't sound very concrete it was like very french very kind of like uh nothing to see here it's great isn't it and i love that bravado and and it's interesting when we're talking before about you know the cultural point you say you couldn't quite put your finger on it it is a hard thing to pin down but when you hear him it's a very very different point of view and philosophy to the interviews i think we've done so far it is and i do think you know silicon valley is if nothing else it's a hype machine you know it's a company building machine and they they're very good at it and so i do think when you have sim altman being like you know basically hey guys we're in the process of changing the world this could destroy everything or make everything better don't worry we got this that's marketing yeah and he and he called it out dario amade saying like yeah 50 percent of entry-level white color jobs the next five years are going to be replaced by ais guys, that's marketing.

27:02Because you're like, look, we're creating this magic new thing that's going to change everything that makes you sit up and be like, well, should I be buying that? What should I do for my business? Maybe I should start paying the money to use this service. But to call out Arthur a bit, I mean, that is also the whole, you know, America might switch off our AI is also marketing. A hundred percent. Microsoft is not going to switch off a data center. Brad Smith has come over to Europe to say that. Yeah, and Donald Trump, even if he says, hey, Microsoft, shut off their data centers, those cheese-eating guys we don't like.

27:41I mean, come on, that's not happening. Cheese-eating guys we don't like. Cheese-eating guys. I'm glad I did that interview, not you.

27:53And he did say, you know, it's an interesting philosophy about this de-skilling point. You know, jobs aren't going to be lost. I'm worried that people are going to be de-skilled. And so we're trying to build AI that drives human intelligence. Yeah. It's a kind of interesting point. And again, I do think AI is transformational. Let me get that out there. But I do think it's going to be slower, more gradual and more messy than our friend Sam and our friend Daria would like to admit. And I thought that was a good point that Artoon made about critical industries. so defense and public sector perhaps being more wary about that and also thinking more about sovereignty so yes it was a marketing point but you can really see that if you are in a company like that or if you're running an organization like that if you're running the civil service you will be more sort of concerned about you know what is going on behind the curtain um before we go the other big news this week of course was meta's very large investment in scale ai fascinating yes alexander wang's company yeah 20 something billionaire now and scale is really interesting company because they started out doing like really low level like data labeling for these ai systems and initially they were like really big in driverless cars like helping label what's a stoplight what's a pothole, what's all of this, et cetera, for these AV systems.

29:21And then as that bubble burst, then they moved into these AI systems, et cetera, et cetera. And they've kind of started to move up the chain. They're just growing like a weed because they're providing all these core data labeling services to all the big AI companies. And Mark Zuckerberg took a shine to him and has put$14 billion into his company. He's bringing Wang in as his chief, new chief AI guy because zuckerberg apparently is desperately worried that he's falling behind in the ai race so it's a very big bet and the other question is what happens to yan lecun who's one of the kind of the modern godfathers of ai he is the main ai guy at meta but now there's a new ai guy so how's that going to work um which i think is just a really interesting kind of game of thrones type aspect of it to watch yeah and it really is interesting to see how the big personalities are kind of driving the AI race in this way.

30:18So do you think the rationale for it then is to try and improve how meta uses its data in terms of powering its AI? It's a little unclear. I mean, it's like Alexander Wang, he's a university dropout. He's very young. He's extremely ambitious. So I think he's similar to Zuckerberg in that way in that he's very, very aggressive and very, very ambitious. but it's unclear how what scale AI does would actually meaningfully improve what meta does. I think the other interesting thing is they bought a 49 % stake. This is what the new big tech companies are doing. We're not going to buy you because we don't want to get sued by the government.

30:59We're just going to basically take control without taking control. Don't want the competition authorities dabbling, which they, I mean, certainly the CMA here They looked at, well, Mistral at one point and OpenAI for the Microsoft investment that they've got and Anthropic for Amazon's investment to sort of ferret out whether there were competition concerns. And so it'll be interesting to see if the regulators around the world are OK with this new model, because it's effectively kind of a take, you know, a non-takeover takeover. It's like a non-denial denial. And, you know, we get in journalism all the time.

31:32Well, so far, they've got away with it. certainly the three that they've looked at over here have passed must adjust but the cma has said that it's a real issue for them and raise raising the point that arthur from mistral did about sovereignty and control or sovereignty or sovereignty i'm gonna say it 10 times when we finish this and next week i promise i'll do better i promise we'll we'll just we'll keep continuity going from from one episode to the next anyway yes people are looking at it they're really worried they seem to have got away with it but you know everyone's got all eyes on competition concerns with big tech it's interesting though isn't it because it's quite it's quite a different sort of acquisition to say Mustafa Suleiman's business yeah being bought by Microsoft and he was a founder of Google DeepMind you can see that feels a much more natural fit.

32:28That's why I wondered if there was a particular data point. It does. And that's what's a little unclear is that Wang is kind of like the man of the moment and has been for the past, you know, 18 months as his company has just like exploded due to the demand from all these big AI model makers. But they're not an AI company per se. You know, they're not developing their own models. They're kind of selling the picks and shovels to the model makers so it's unclear what that what this will mean for you know meta's ai goals but it's interesting that he's like the latest kind of young turk to be completely lionized and is now a billionaire very very young not many years removed from dropping out of uni i feel like there's a good magazine piece to be done on the young turks of ai i like that phrase indeed but But yeah, I think it's quite cool that we've had all the major AI folks on the program now.

33:23And as you say, it is fun, especially to hear their different takes on it all. And how they're trying to market themselves and kind of grab market share. But the only one, it feels like there's open AI in terms of people actually paying for stuff. Open AI, maybe Anthropik and kind of everybody else. And Miss Jarl's got a few big deals with French companies, I've noticed. So they're playing on that sovereignty point. But can they keep up is the question. Yeah. Well, good to see you. Likewise. Any conferences lined up your end? No, thankfully no. Zero conferences on the horizon. But I mean, it's really, you've been the conference queen.

34:06I have been the conference queen. South by Southwest, then London Tech Week and Viva Tech and Founders Forum. And this week we have the Times' CEO Summit, which I'm really, really looking forward to, which we host on the 17th floor of our building and we bring in FTSE 100 and 250 chief execs. Are you going to get like a frequent flyer card or something for all your conference going? No, I'll get free breakfast though. Maybe lunch. Maybe the miniature half-stale croissant that you get at these places. Half-stale at the Times, how dare you? They'll be perfectly fresh. I've had a few. I've had a few in my day.

34:43Maybe they're stale by the time they get to Sunday. That's what's going on. That's true. It's true. Oh, for our dedicated listeners, please, if you have any questions, concerns, suggestions, whatever, send us an email at techpod at thetimes.co.uk. That is techpod at thetimes.co.uk. And we're going to be back next week with another fabulous interview and more high-level chat. Look forward to it.

35:15Thank you.