- 20 hours ago
The honeymoon phase is behind us. In 2026, "AI-powered" is no longer enough; the market is more critical of ROI. We contrast the strategies of investors to decipher where, once all is said and done, real value will lie in the AI chain. Which technologies will drive productivity in the years to come and create a real return on investment? And which business models will prove the most sustainable?
Category
🤖
TechTranscript
00:01Great. Good morning, everyone. Hope you're all doing well.
00:04So this session is going to be focusing on surfing the wave
00:09and answering the question, what is the best strategy to invest in AI?
00:14I've got an esteemed panel with me today.
00:16I've got Susanna on my right and then Maya and Hala as well.
00:21So before we get started, maybe we'll just do quick introductions,
00:24maybe a minute each, just describing what you do
00:26and your experiences for the benefit of the audience.
00:28Maybe Susanna, we'll start with you.
00:31Great. For those of you who have listened to me for an hour already this morning,
00:35I promise I will think of some new things to say.
00:38Susanna from Managing Partner at the UK's Sovereign AI Fund.
00:41We're a 500 million fund.
00:43We launched a couple of months ago, based in London.
00:46We invest in companies with a significant UK nexus
00:50that are in the AI space exclusively.
00:55Hi, everyone. So I'm not an investor, but I am managing director of France Digital,
01:01which is the biggest startups and VC association in France and also in Europe.
01:06So we work a lot with the different investors in France.
01:11And what we do is to be a big business network for them to help scale up to grow.
01:17And also we represent the interest as what we can call a lobby,
01:22to talk to policymakers both in France and also in Brussels,
01:25to create the best regulation landscape possible for startups to grow.
01:32Hello, everyone. I'm Hala Fadel.
01:35I lead the growth equity at Euraseo.
01:38We have over 8 billion invested in tech companies across Europe.
01:43And we just did the first closing of our latest vintage at around 650 million, aiming for a 1 billion
01:52fund,
01:53only focused on growth companies in tech in Europe.
01:57And of course, most of that is AI.
02:00Great. Thank you.
02:01And I forgot to introduce myself.
02:02My name is Naylin Patel, director of research at PitchBook.
02:05So let's jump straight into the questions.
02:08And Hala, maybe back straight to you.
02:11Let's address the question of the title.
02:13Thankfully, we're lucky today that the title has got the question in it.
02:16So what is the best strategy to invest in AI?
02:19And maybe we'll go to the panel after that as well.
02:23Yeah.
02:23So we were discussing this a bit backstage.
02:26I wish I had a clear answer.
02:29I will give it, you know, my best shot.
02:32I think we need to distinguish maybe a number of things within AI.
02:40So first, I mean, we've all tried the different AI tools and we know how powerful they are when we
02:47run them on any kind of data.
02:49So maybe the easiest part of the question is to say, you know, what are the business models that we
02:54should not invest in?
02:55And I would say the first category is anything around analytics.
03:01AI and analytics is something that, you know, you have the data within your company.
03:07You use any AI tool and you get lots of information out of that.
03:15So anybody selling analytics with AI or an app around that, I think, is not going very far in today's
03:23world.
03:24One thing that is clear also is that we will be using lots of tools, lots of databases, lots of
03:30AI infrastructure, cloud.
03:34So I think this is a segment that is very valuable today.
03:39Another segment that's very valuable and create defensibility is something where we have hardware involved and where you're creating your
03:47own proprietary data with the hardware that is involved in it.
03:51So take, for example, robotics or anything that involves sensors or cameras that will be, you know, detecting your own
04:01proprietary data and putting a layer of AI on top of that.
04:05Now, having said that, we are at the very beginning of this AI revolution.
04:11So I think we need to wait a bit for the sand to settle and see, have more visibility.
04:19I would give it, you know, 12 to 18 months to have clearer visibility on what will be, I think,
04:25the future business opportunities and everything around the application layer.
04:29Because as the analogy with electricity has been used many times around AI, with time, nobody pays a very high
04:41price just to get electricity.
04:43You pay a high price for all the use cases around electricity.
04:46So anything around the model, the foundational models, I don't think will be something that will bring value in the
04:56future, but more the application around that.
04:59And to know who's going to win and how they're going to win in the application layer, I would sit
05:04on the sidelines today, not necessarily invest and wait 12 to 18 months to see how things will play out.
05:11Great.
05:12Great.
05:13Maya, any other thoughts?
05:16Well, I think it depends also what is your strategy as an investor because you are talking as a growth
05:22investor.
05:22And I think when you are investing in venture capital at the more early stage, you won't see the thing
05:31the same way.
05:32I think you won't expect to see in which topic, in which vertical you have to invest.
05:37You will make a bet anyway.
05:40What I want to add is I think it's important today to really take the AI in the whole value
05:47chain, meaning the shapes and also the data centers and also the application, the models.
05:56I think that at every level of the value chain, there is some value to create.
06:03And I think it's important, especially in Europe.
06:05And I will talk with my hat of manager director of France Digital because what we want to do is
06:11also to make some European startups to grow and only see the concentration of wealth in the US or in
06:18China.
06:18And what we need today, what we can see in Europe is that we have great companies at every level,
06:24but they are not the leader on the market.
06:27And we need to help them to be a real alternative.
06:32So, as I said before, I'm not an investor, but I think it's important to really have this vision of
06:38the whole value chain and to collaborate also with the public sectors to help each level to grow also in
06:46Europe.
06:47Understood. And, Suzanne, maybe shifting towards kind of the more picks and shovels debate.
06:54We just had a question from an audience member about before we joined the stage about, you know, demand for
07:00electricity, water consumption.
07:01And, you know, there's always been that narrative around, oh, you should invest in the picks and shovels in industry.
07:05What's your take on that? And then what is your take also on the best strategy within AI?
07:10On your sort of first question, best strategy within AI, I think clearly from looking at your slides earlier, and
07:17I'm a big pitch book user and fan, you know, if we all had a time machine, we would have,
07:23you know, done early rounds of anthropic and open AI.
07:27And in the absence of that, I think we need to be clear about the areas where Europe can win
07:33and not just have strong domestic players or regional players, but categories where it is plausible that Europe builds a
07:42global player.
07:43Because, you know, if we European countries want to have a seat at the table, we need some leverage and
07:50the easiest and most straightforward way I think of having that leverage is having some really significant domestic AI companies.
07:58You know, and at the moment, largely all of those companies are US, arguably a few in China as well.
08:05And if we can help some of the names that were on your list, whether it's, you know, Wave in
08:10London, who are one of the leaders in embodied AI or 11 labs in voice to really, like, take that
08:15next step and become global leaders, I think we'll be in a much stronger position as a geography.
08:22On picks and shovels versus sort of the application layer.
08:27I think as a fund, we're looking at both.
08:31Our first three investments are across both categories.
08:37And I think we have, like, different time lags as well.
08:43Generally, the application layer companies come out of the gate more quickly.
08:47They're quicker to commercialization than some of the picks and shovels businesses that arguably have slightly more durable moats.
08:57I think the moat point is, you know, is one we can explore in more detail.
09:02It's something investors historically have been very, very focused on.
09:06I think the durability of moats in general has just come down hugely.
09:10And now what gives founders and companies an edge is really just ability to execute versus any meaningful technical moats.
09:21Understood.
09:22And, yeah, maybe touching on that valuation point.
09:26Obviously, we had SpaceX that IPO'd at an enormous valuation.
09:30I think they announced that they've acquired a UK-based AI company for about 60 billion a few days ago.
09:35We've got Anthropic, OpenAI, which are probably going to redefine VC exits and AI exits moving forward.
09:44So the valuations are very, very high.
09:46So, Hala, maybe next question to you.
09:48How do you avoid overpaying within the AI space?
09:53Yeah, I mean, the valuations we're seeing today are very typical of a nascent technology.
09:59I mean, we've seen this at the very beginning of the internet.
10:02We've seen it at the very beginning of the cloud.
10:04There is investor appetite to catch this innovation wave because they have been the creators of those large companies of
10:12the Amazon of this world, of the Google of this world.
10:16They were all created around those waves.
10:19And if I go back to the cloud wave, you had the Salesforce, the Workday.
10:25I mean, so many companies that were born around those AI waves, sorry, those technology waves and investors wanting to
10:32catch them.
10:35So as growth investors, so we're all Europeans here.
10:39First, valuations in Europe are much more reasonable than valuations in the US.
10:44So anybody wanting to catch the AI wave and the AI talent present in Europe, please feel free because I
10:52think we have a great example in our portfolio where we have Cognigy, one recent exit that we did, a
10:59German company doing AI for customer support.
11:02Sierra, their closest competitor was at the same level, if not lower ARR and was valued 10 times more.
11:10And because we entered at a reasonable price, we were able to do an exit and do within two months
11:15the 2.2 cash on cash on this investment.
11:19So Europe is a great opportunity.
11:21And then if you think the valuation is crazy and you don't have anything to build the defensibility of this
11:30company, you know, to build conviction around the defensibility, I think it's okay to stay on the sidelines.
11:38So we're investing a bit less these days.
11:42And as I said a bit earlier, just watching how things will unfold before making wiser decisions.
11:51So there's no fear of missing out and just, yeah, you're still being prudent.
11:54Exactly. This is what you need to control.
11:56I think that's a valuable lesson considering historical trends within private markets as well.
12:01There's been a lot of, you know, overheating, you know, particularly with certain industries.
12:06It reminds me of kind of the non-fungible tokens during COVID.
12:10We don't really hear much about them or delivery companies that consolidate quite a lot more post-COVID.
12:17Yes, go ahead.
12:18I have an early stage investor view on price.
12:21Yeah.
12:22And that's, you know, the power law kicks in and it's the winners that matter and you need to be
12:29in them almost irrespective of the entry price.
12:34And if you look back at a lot of the best AI deals or even in waves before, the early
12:41rounds looked exceptionally expensive.
12:43And the best companies grow into their valuations.
12:48And so I think, you know, while price is one factor when considering an investment, if you're a seed fund,
12:55pre-seed, even Series A, you are certainly paying a price ahead of today's valuation.
13:04But you've jumped ahead and taken one of my later questions, but that's good.
13:07That's good.
13:07I guess let's keep the discussion going on valuations.
13:10Any other further feelings about valuations?
13:12I'd like to agree with this at an early stage.
13:14Yeah.
13:14Like if you take the example of Mistral, when it first came out, everybody was saying, oh, this is crazy,
13:19the biggest seed round in Europe.
13:23And then it pays off.
13:24Like if you look today, the company and what they were able to build.
13:27So there is a case also for paying high price at an early stage.
13:30Yeah.
13:30We definitely think about that future growth in terms of revenues.
13:33And there's a lot of loss-making companies within private markets that obviously are pricing in future valuation growth.
13:37Any further thoughts, Maya?
13:39And how do you, you know, justify some of the valuations that are out there, kind of, you know, considering
13:44all things?
13:44What I can see on the market is that around valuation, maybe there are two big trends.
13:49One is beating on the hype companies when you see that there is a potential to exit anywhere at the
13:57short time because it's hype.
13:58Yeah.
13:59Or the other trend is to really focus on the, let's say, let's go back to basics and really try
14:05to understand where the value is added, what are the strengths of the company, on the team, on the ability
14:13to go to the market, to really define where you can save costs, where you can add some value with
14:20AI.
14:20Because what you can see, and I think we can talk about the AI ecosystem in Europe, sometimes it's very
14:25difficult to really define if it's just AI-based company for marketing purpose, because when you see today the landscape
14:33in Europe, it seems like, I think the figures is around 25% of companies today in Europe, scale-ups
14:39in Europe, startups are on AI, in AI.
14:42But it's very difficult to really feel if the real value is because of AI, or it's just because they
14:50want to raise some money, so they put some AI in the pitch deck.
14:53So what I wanted to say is when you are an investor and you want to really feel the good
14:59valuation, you have really sometimes to step aside, put AI on the side, and really see what are the other
15:06strengths regarding the relation between the startups and the potential customer.
15:12And how do you think a company can remain durable and protect their moat, and not be commoditized in like
15:1812 months' time?
15:20I mean, it's complicated to keep some advices.
15:22I think that there is a huge excitation, so a kind of real form of, nobody wants to miss this
15:30revolution, because it's a revolution.
15:32I think it's important to really consider it, and really be aware of what is going on in the markets,
15:41but you have to stick with the vision.
15:44And I think that, once again, I'm not the investor, so I'm not going to give advice to people to
15:49how they have to do their job.
15:50But I talk to a lot of VCs, and I think that at the end, whether you really understand the
15:59purpose of the company and the market, or you really feel the team, and that's it.
16:03It's not because it's AI or not AI, and they are also, and I think that we shouldn't forget that
16:09we also have some great companies, which are not just AI-based, not great fundamental models,
16:15but they do have stronger sets to find the market and to create value.
16:21Happy maybe just to put a small framework around, you know, what to look at.
16:27I think Maya was saying back to basics, and she mentioned the importance of a team.
16:33Usually this is more a venture thing, like to look very much at a team, but now we do also
16:41analyze much more teams on the growth side.
16:44I mean, this has always been important, but you need teams that are very technical.
16:48I think it's the rebirth of the technical founder and surrounding himself also by a technical team, so that's one.
16:58Then in terms of maybe this is more difficult at early venture stage, but at growth stage, you know, looking
17:05at who are the customers,
17:07how long have they been a customer for, are they renewing, because you have companies trying lots of, you know,
17:15pilots,
17:16and then, so they buy any type of product, but after six months, after one year, are they renewing, are
17:23they buying again this product,
17:24are they expanding this product from one division to several divisions, so that's another point, you know, to anchor your
17:30conviction on.
17:31Then what type of data are they using?
17:34Is there any kind of proprietary data?
17:36That does the service that this company provides becomes better over time because of what they're doing?
17:44You also have to look at how close they are to the workflows of the enterprise.
17:49Is this just a wrapper on a model, or are they deeply integrated into the workflows of a company?
17:57And this is, of course, much more defensible.
18:00And then in terms of valuation, if you think about the exit, we know today, I mean, we've seen,
18:07if you have to wait to be SpaceX to IPO, I mean, the bar to become an IPO-able company
18:15is getting higher and higher
18:17in terms of how defensible you need to are, so you need to be.
18:21So if you look at the exit in terms of a strategic exit, we know that strategics typically never pay
18:29more than $3 billion for an exit.
18:31So if you reverse engineer how much you need to pay to actually get a profitable exit,
18:39then this gives you some kind of valuation discipline as to your entry point into those companies.
18:45One last comment on early-stage valuations.
18:49I think there are two things that are pushing them up that we haven't discussed.
18:53One is that certain costs have gone up significantly for these companies,
18:58so they're spending much more on compute, which means they're raising larger seed rounds.
19:03And founders are taking roughly the same amount of dilution,
19:07so if you raise a larger round, that pushes the valuation up.
19:10I think the second point is capital in and of itself has become a moat.
19:16So when you look at sort of why does Mistral, why do Ineffable need to raise these sort of billion
19:22-dollar rounds,
19:24it is both cost of compute and also that having that capital gives you weight in the room,
19:31particularly against some of the very, very well-funded global players.
19:37And how, I guess, some of those rounds are absolutely enormous,
19:42and some of these companies have only been created a few months ago.
19:46How, I guess, how do you solve that problem?
19:49Like, how do you solve the problem of, like, a company that hasn't really got much...
19:52Well, obviously, it has probably got, you know, developing tools that we don't know about,
19:54it's in stealth mode, and then raises a round that's like a billion dollars.
19:57Because we see that in our data when we're writing research reports.
20:00We see a new round at seed level, and it's like a billion,
20:02and we're like, wow, that's incredible.
20:04But how, like, what kind of goes on behind the scenes with that?
20:07Like, how is a company being able to raise a billion without, you know, much track record
20:12or, you know, much, you know, having a long history?
20:15The answer is almost always talent.
20:17You know, you look at the profiles, the people that can raise those mega rounds.
20:22Yeah.
20:23Generally, they have been senior employees at leading global tech companies.
20:29Take Dave Silver, he was the founder of Ineffable, was arguably, like, number two, three at DeepMind,
20:36helped on AlphaGo, you know, supported on their protein folding project, which won a Nobel Prize.
20:45When exceptionally high-quality talent comes out, they tend to raise really large rounds.
20:52And the other thing they need to do with those really large rounds is attract top-quality talent themselves.
20:59And those individuals are currently very handsomely compensated at Anthropic, OpenAI.
21:07And they don't have to match compensation, but they, you know, compensation is now a significant driver
21:14of why these rounds are so big.
21:16Mm-hmm, understood.
21:18I guess, yeah, moving on to kind of sectors now.
21:23And, Maya, maybe, you know, what verticals or, you know, where do you think AI is potentially, you know,
21:29going to make more of an impact in terms of verticals?
21:31Do you think it's actually going to be horizontal platforms and LLMs that is going to grow more?
21:35Or do you think certain verticals are right for a bit more of AI disruption?
21:41What we can see today in Europe, we have more and more companies.
21:45I think we have around 10,000 companies in Europe on AI.
21:49We can see that there are some big hubs.
21:51And so you have London, for sure.
21:54This is the leader in Europe.
21:55But just behind, you have Paris.
21:58We have more of Paris, France, see?
22:00We have a bit more than 1,000 companies.
22:03What is very interesting when you look at those companies is that we are very strong on some verticals.
22:07One of them is health.
22:09I think we can create a level of value on this topic because we do have great doctors, some great
22:16researchers in this field.
22:18So I think this is where we can create all the value.
22:21But as I said before, I think that it's important to really have the control on the value chain on
22:28AI.
22:28So when you look at the different levels, the value I did is sometimes big, I think, on the LLM
22:41today.
22:41Because if I have to find some figure to compare it, but one gigawatt per hour coming from the LLM
22:54is a value I did at 35,000 euros.
22:59One gigawatt coming out from the data center is 400 euros.
23:04So there is less value at that point today.
23:07So maybe you want to invest a bit more on the LLM, but the competition will be very strong.
23:11So at the end, you have to, I say, to diversify the risk and also focus when you know that
23:19you are very strong.
23:20And as I was saying, what we can see in Europe, there are some fields, especially in the health and
23:27the defense, there are a lot of value.
23:30And we have strong people in this field, great talents.
23:34So I would say that I would start from there.
23:36Yeah, nice.
23:37And Ala, any sectors that stand out for you?
23:39Yes.
23:41So, I mean, first of all, I agree with what Maya said.
23:44I think we can make the distinction between, you know, the productivity gains, if you look at a vertical like
23:52coding, where productivity has just been multiplied by 10, if not more.
23:59And then industries that can be revolutionized because of the possibilities that AI opens.
24:07For example, I mean, she mentioned healthcare, but you can think of biology, biotechnology, and all the drug discoveries that
24:15can come from that.
24:16Defense, she mentioned defense.
24:20Having, you know, drones or any type of defense engine without humans in it is completely changing the way defense
24:34is being done.
24:34So on the productivity side, I think there are many, I mean, we mentioned coding.
24:40So what happened to coding, coding is an enabler of the productivity gains in all other verticals.
24:46So I think this, combined with agentic AI, we're at the very beginning of the productivity gains we're going to
24:53see within the enterprise.
24:55And comes with it the question of, you know, how many jobs are going to be replaced.
25:01And so I don't have the answer to that.
25:04Of course, the two sides can be right, like everything replaced, anything from every job replaced to a human AI
25:13interaction that can be much more productive.
25:17The answer is I don't know, but there will be significant productivity gains.
25:21I can see it already in my team.
25:23We used to, it used to take us two weeks to look at a data room.
25:28Now, in less than a day, it is completely done with the same level of accuracy.
25:33Of course, there is human intervention, but it's humans empowered by AI that are much more productive.
25:39And you can think of that within every function of the enterprise.
25:44And that's very exciting at the same time, freeing all this time for us to do other things.
25:51To do what?
25:51I'm not sure, but to do other things.
25:54And then on the discovery side, I mean, I mentioned healthcare.
26:00But so many new areas of discovery that will be much easier with AI.
26:06There was the CEO of Palo Alto Networks that was saying that with Mythos, so the cyber software that Entropic
26:15released, it took them six weeks to discover the equivalent of six years of work at Palo Alto Networks.
26:23Wow.
26:24So this is how powerful this tool is beyond productivity into getting us onto the next level of discoveries and
26:32innovation.
26:34That's super interesting.
26:35You know, six years into six weeks, that is quite a ROI.
26:38Susanna, what sectors are exciting you?
26:41Where do you think, you know, high growth opportunities are going to occur at the early stages?
26:45I think to a large extent, Europe can choose what sectors they want to excel in.
26:52Yeah.
26:53And if governments have the willingness to support those categories, we can build global winners.
27:00And I would point to defense tech that you've both mentioned as a really good example.
27:06Five years ago, it would have been a real struggle as a European defense tech founder to get your business
27:11funded.
27:12And Germany, through, like, weight of support and spend, has created two companies in Helsing and Stark that are now
27:22global defense tech leaders.
27:24And I think the same model can apply to other categories, too.
27:28So to start with, I don't think we just need to try and play in sectors where we think Europe
27:35currently has the right conditions to win.
27:37I think we can forcibly create conditions to win.
27:41And two, like, you know, you mentioned SpaceX and the sort of insanely huge SpaceX IPO.
27:50That company was also created on the back of, like, huge government support from DARPA, from NASA.
27:57Yeah.
27:57And I think in Europe, there's a sort of view of government supporting early stage or even growth stage companies
28:04with a certain amount of suspicion.
28:06Like, are we propping up companies that are too weak to survive on their own?
28:10But actually, in the U.S., they have been doing this for 50 years.
28:14And it's not, you know, life support for early stage companies.
28:19It's enabling the winners.
28:21And I think in Europe, we're going to have to be a little bit more comfortable on, you know, buying
28:26from domestic AI companies and really helping them win.
28:31Yeah, that's a very interesting point.
28:32And I think that's a great opportunity to probably transition to more kind of regional and policy-focused questions and
28:38focus a bit more on Europe.
28:40So, Maya, kind of next question to you, you know, what does the European ecosystem look like?
28:48Maybe touch on, you know, the EU AI Act potentially as well.
28:52I believe you guys were involved with that as well or provided, you know, feedback on that.
28:56So, it would be interesting to get your view on that and how that positions Europe moving forward.
29:03So, how is the landscape today?
29:05So, there are more and more companies in AI, sure.
29:09More investment as well.
29:09More than 10,000 in AI in Europe.
29:12It represents more than, so as I say, it's 25% of the startups today are in AI, but it's
29:17more than 50% of the investment.
29:20So, I think that there is a strong acceleration and we have a deep ecosystem of talent working on this
29:25field.
29:25And there's also the regulation part, which is part of the European landscape in AI.
29:31And I would say it's both at the same time a problem, but also a good thing to promote the
29:38European AI through the regulation point of view.
29:41Because I think it shows that, and you are talking about the AI Act, the purpose of this text at
29:47the very beginning.
29:48I mean, the philosophy around this text is really to create an AI, a trustworthy AI with a real purpose,
29:56respect the privacy, which is responsible.
30:00And I think this is how Europe is branded today on the AI field.
30:05But at the same time, the first one is regulation.
30:08And I think that when you're talking about Europe at the international level, you do feel the heavy part of
30:16the regulation.
30:16And you think it's very complicated to scale and to really build AI.
30:20So, it's both regulation is both, I think, good.
30:24And I think sometimes I feel proud of being in Europe because you have this text about AI, the past,
30:31the reading next to be responsible.
30:32And on the same time, I think people are saying, I don't know if I go there because it's too
30:37complicated to really scale in Europe because of the regulation.
30:42What I can say about it is that, for sure, it's complicated to be compliant, but the text is built
30:50today and there were the omnibus that has been released in May 7th that give a timeline and also make
31:00the text eligible only for the biggest one.
31:03So, it gives times for the smallest company, it gives some sandbox to really deploy that technology without caring about
31:12being compliant at the very beginning.
31:15Do you think it's a help or a hindrance?
31:17Do you think it's a help or a hindrance?
31:19I think it's very important to really give the space for the new challenges to grow in front of the
31:28biggest company that we do all know.
31:30All the hyperscalers are in this AI race, and it's very complicated to enter the race because it costs a
31:36lot to develop some technology and AI.
31:39You were talking about the public procurement in Europe.
31:41For sure, we do need more domestic command from customers.
31:47We do need to, as European, to have these European preferences for our own tools and AI.
31:52But we also need to really create the European offer that can compete with the extra-European tools.
32:00And it costs a lot.
32:02We do need more capital.
32:04And so, we do need to still invest a bit more.
32:08So, regulation is a good tool, I think, to really create a direction, a vision of what should be AI
32:15in Europe.
32:16But, at the same time, we do need more procurement, and we also need more capital.
32:20And I think my message today is that we hear a lot about the fact that there is too much
32:27regulation in Europe.
32:29Maybe, but it shouldn't block our companies.
32:34And I really think that we need to promote also Europe as a different thing than just a regulation landscape.
32:42Hello, any thoughts on the U.S.-Europe debate?
32:45Yeah, no, I have a very controversial view on the topic, I have to say.
32:49Excellent.
32:50Yeah, no, I don't agree.
32:52I think it's the same way it's not the right time for me to tell you what to invest in
32:57in AI,
32:57because we need to have more clarity about where things are going.
33:01It's the same approach that the regulator should take.
33:04It's no time to regulate AI now.
33:06I have lots of entrepreneurs in our portfolio that are shifting their headquarters to the U.S.,
33:12specifically around this issue of regulation.
33:15It is the time to support every single European entrepreneur,
33:18and then regulate in due time when we start seeing the problems or the side effects that AI can have.
33:28You can regulate social media, which is another topic, and maybe outside this panel,
33:34but AI, it's way too early to do it.
33:36You're scaring off entrepreneurs.
33:38And what we're seeing, and I could hear it when Suzanne was speaking,
33:46like Europe can choose which sectors to support.
33:49I can see already an ambition that is tamed, and I don't like to see this.
33:54I think we have great entrepreneurs, great talent.
33:57What is really missing is the capital piece.
33:59We need much more capital to support this talent, and we should not tame our ambition.
34:05We should be as ambitious.
34:07If we want to own our future, our sovereignty, what is going to happen with our lives,
34:13and then regulate and have the means to regulate as opposed to consuming the technology
34:18that is coming from the U.S. or from China, because this is what we're doing today.
34:23We need to support these entrepreneurs as much as we can with no regulation, a lot of capital.
34:30In fact, deregulation of the different home markets create one unified big market,
34:38and let it take off, and we'll regulate way later.
34:43Any thoughts?
34:44On the regulation point, completely agree, and I think there are a handful of areas
34:52where Europe could regulate, could create new entities that would be incredibly supportive of founders.
34:59I think the most obvious one is this movement around EU, Inc.
35:03I think still a significant proportion of top European founders, even if they put their headquarters in Europe,
35:12even if they plan to stay as founders in Europe, will have a top code.
35:17Their legal structure, the main company, will be based in Delaware.
35:22And, you know, as we've seen over the last week, that could be pretty challenging for some frontier AI companies.
35:30And I think if the EU had a Delaware-equivalent structure,
35:35and this is what the guys at EU Inc. are trying to do,
35:37so that European founders didn't have to have a German company or a French company,
35:42but could have a, like, EU company with consistent regulation around employee option schemes
35:48and other, like, key points that they need when they're deciding where to start their business,
35:55I think it would be incredibly attractive for European founders.
35:59Can I just answer, because I think it's very interesting, your point of view,
36:03and I think at the end that we all agree that we regulated too early about AI,
36:08because the text was imagined at a moment where there was no Gen AI.
36:13It was only on AI, and the use of AI, what are the risks with using AI?
36:17So I think that we agree about the fact that the text arrived maybe too soon.
36:23But what I wanted to say is that there is a way of doing some marketing around the political decision,
36:30and this is how I see EU Inc., this is how I see the AI Act.
36:33I think we should use as European those tools, as marketing tools,
36:38saying that if you come to Europe to develop your companies in AI,
36:41it's because you want it to be more responsible.
36:44If you come to Europe, we are not as fragmented as you can believe,
36:49because we do want to be more European, and this is why we are creating EU Inc.
36:53At the end, EU Inc. is not the only solution.
36:55It's more an administrative framework.
36:58But if you do a kind of strong marketing around it,
37:01saying, do not think that Europe is so fragmented.
37:05Actually, we really try to work together,
37:07and there is a way to really build your company
37:10and really to scale at the European level with some fast trucks.
37:15It's a good way of saying things.
37:17And I think that today, when I talk to people,
37:20and I think there is a kind of strategy also from the extra-European actors
37:23to say, it's too complicated in Europe, don't go.
37:26But actually, it's not that easy to go to the US also.
37:30I mean, there are different states, different rules also,
37:33but there is, for sure, a deeper market and more money.
37:37And this is what I was saying.
37:38We do need capital and procurement,
37:41and regulation is a framework, a tool that we can use to regulate.
37:45But what we do need is some strong talents, some capitals,
37:48and also some procurement.
37:49Yeah, it's an interesting point you make about that kind of EU Inc.
37:53And it kind of leads me to my next question.
37:54And Susanna, obviously, you work with the kind of UK Sovereign AI Fund,
37:59and the mission is to kind of support the best UK founders
38:03and develop, you know, AI companies in the UK.
38:06So how does that work in practice in terms of developing the UK
38:09and then the competitive landscape in Europe
38:11and then also developing European companies to compete with the US?
38:13How does that all fit together from your perspective?
38:18So, I mean, we are exclusively investing in companies with a UK nexus.
38:26Many of those companies will do business in Europe,
38:28but, like, as it stands with the current movement,
38:32we wouldn't back founders with, you know, an HQ in Paris, for example.
38:39And I actually think that over time there are really good ways for us
38:44to kind of work together.
38:46And, you know, if you're running BPI,
38:50the sort of French equivalent of both us at Sovereign AI
38:53and our sister fund, the British Business Bank,
38:56you know, I met one of the managing partners there yesterday
39:00and we had a huge amount in common.
39:01And we have a lot of the same challenges, you know.
39:04One of our challenges is a lot of early-stage British founders
39:08are attracted to Y Combinator on the West Coast
39:10and they don't accept UK topcos, French topcos, etc.
39:17So they all end up with Delaware topcos.
39:20And we both have a shared interest in changing that
39:22because ultimately these are British businesses or French businesses
39:27that, you know, by chance, because of one decision,
39:30end up being a US company.
39:34And so while as a firm we don't, you know,
39:38invest directly in European companies,
39:40we certainly need to work with our peers to, you know,
39:44lobby in areas where we have, like, joint interests.
39:48Understood. Great.
39:50We've got about three minutes left.
39:51I think that would be a good opportunity
39:52to maybe shift towards a forward-looking outlook
39:56thoughts from our panel.
39:57So maybe, Hala, to start with you,
40:01you know, how do you build a portfolio for the next 12, 24 months?
40:05What are you focusing on?
40:07What's your outlook for AI investment strategies?
40:10So you only gave me 12 to 24 months.
40:14You can extend that if you want.
40:16You can pick longer periods as well.
40:18I mean, as I mentioned at the beginning,
40:20everything around the AI infrastructure is here to last
40:24because you need to organize your data to be able to use it.
40:29Everything also around physical AI, you know, robots, autonomous driving.
40:39I mean, there are some technologies.
40:41It's too early to talk about them, like quantum.
40:44But defense, we mentioned defense.
40:47So everything involving hardware, industry, what we call industry 5.0.
40:52So revisiting how manufacturing is done.
40:54This is what we're looking at right now.
40:57We have a number of companies already in the portfolio
40:59because these are very defensible.
41:02They stand inside the workflows of enterprises
41:06and there's some physical assets involved.
41:09So the barrier to entry or exit is high.
41:15And then if you give me a bit more than 24 months, then...
41:202030. It's June 2030.
41:21What does it look like?
41:22June 2030.
41:23Then I think the same way we've seen, you know,
41:27the cloud unfold, the internet unfold,
41:29we will see the application layer on the AI infrastructure unfold.
41:35And there will be very big investment opportunities in that layer.
41:39But that layer is too fragile today to see how it's going.
41:43Even the economics of it are going to unfold
41:46with the price of tokens being where they are today.
41:49Great.
41:49It's June 2030, guys.
41:51What does the investment landscape look like in AI?
41:53Any final thoughts?
41:54I hope in Europe, our IPO market will catch up with the US
41:58and we're talking about a Lagora IPO,
42:01a lovable IPO to match the upcoming OpenAI and Anthropik IPOs.
42:09Because while we're catching up with capital at the earliest stages,
42:14still, until Europe really sorts out its exit opportunities,
42:17that makes it harder to attract growth stage capital.
42:21I am really excited about the potential
42:23of the new European scale-up fund that EQT are managing.
42:27It's going to bring much-needed capital to that later stage.
42:30So I hope that enables the next wave of AI companies to go public here.
42:35Great.
42:36Maya, any final thoughts?
42:40From a fund perspective,
42:42I think the wave is very important
42:43and so you have to have this short-term view of 12, 24, 30 months.
42:52But if you look at the ecosystem level,
42:54I think it's very important to keep the trend aside
42:58and to really focus on the industrial long-term view.
43:03And what is interesting with AI today is not a trend like an area,
43:08a field where you invest.
43:10It's an infrastructure.
43:11And what we do need in Europe is really to have this long view,
43:16long-term vision of the infrastructure,
43:18the constriction of these assets on the long term.
43:22Great.
43:22Thank you very much.
43:23We're perfectly on time.
43:25So please join me in thanking the panelists for their time today.
43:28applause
Comments