- 26 minutes ago
Europe's economy is often viewed through two lenses: high-growth startups and global corporations. Yet much of its strength lies in the businesses in between—companies that evolve, adapt, and grow over decades rather than years. And in a business environment defined by constant disruption, longevity is becoming an achievement in its own right: many of Europe's most successful organizations have thrived by continuously adapting to changing markets, technologies, and customer expectations. As AI reshapes industries and economic uncertainty persists, what separates companies that successfully reinvent themselves from those that fall behind? In a world increasingly focused on disruption, how do you build not just the next success story, but the conditions for long-term prosperity? Bringing together perspectives from venture capital and banking, this conversation explores what it takes to create enduring value in Europe today.
Category
🤖
TechTranscript
00:18Hello. Hi, everyone. My name is Yajo. I'm a reporter at Bloomberg covering venture capital
00:25and startups. And I'm very thrilled to be joined by Philippe and Yannick, representing two very
00:32important asset allocators at various stages of a company's timeline. Topic of the day is ensuring
00:41business longevity, but I thought I would start with the news of the week, which is the U.S.
00:46government kind of kneecapping Anthropic and blocking its latest models. Obviously, maybe I'll
00:55start with you, Yannick. Sure. What does one do when a foreign government can just cut off a very
01:02important utility like that? Well, this is, to a large extent, it's pretty unprecedented. Oops.
01:08There's a bit of a Larson effect.
01:14Okay. I think we're good. All right. That's much better. So it's a bit unprecedented to tell you
01:19the truth, Yajo. It's the first time that the U.S. is cutting access to technology to customers around
01:29the world, not to be in Europe. So I think for a lot of European companies, it's a bit of
01:34a further
01:34wake-up call on the fact that there's a pretty high level of dependency that's been built over the years.
01:43Philippe, obviously, Anthropic is one of Excel's portfolio companies. I'm sure that your firm might
01:50be expecting a large windfall. But on the heels of the news over the weekend, what has your discussion
01:58been like with your own portfolio companies? And what do you think? Do you think sovereignty right now
02:05makes for a good business case? Well, I mean, to be honest, I would say right now that, you know,
02:14there is already a lot that can be done with models that are not best in class, right? And so
02:22you don't
02:22need the best in class model to do everything you need to do with AI today. And I think a
02:28lot of
02:29companies right now, and we see it in our portfolio, they're actively experimenting with different models
02:34and trying to find the best model for a given task. Because, you know, there is no point in overpaying,
02:41and it's all about getting the right ROI. And ROI meaning that you want the cheapest cost for the
02:48results that you want to achieve. So overall, it feels like, you know, you look at the models today,
02:56and even the model that are, you know, a couple generations away can generate, like,
03:01amazing ROI for, for a lot of tasks. Talk a little bit about the models that you're using. Are these
03:07the blend of Chinese, European, American, typically by your portfolio companies? So which model are they
03:15using? Yeah. Yeah. Well, I think they're experimenting with all of them. Okay. Obviously,
03:22some models are better at certain tasks. I think open source models, I mean, we have some great
03:32French open source models. Sure. But I mean, talking about Chinese open source model, then I think it
03:38creates an issue in some industries. So clearly, some regulated industries are not very keen on relying on
03:47Chinese models. I think open source is part of the toolkit. It has always been part of the toolkit.
03:53And if you look at software, software has always been divided between, you know, closed source company
04:00and open source company. And obviously, with, with SaaS and cloud, I think that gave, you know,
04:07more edge to the closed source than to open source, like we saw with perpetual licenses, but still a lot
04:15of
04:15open source going on right now. And I would expect it to be the same for the model. And for
04:20some,
04:20Sure. Some use cases and some companies, I'll prefer to, to use, you know, private models. And for
04:27other tasks, I think companies are going to be fine with, with open source. So I don't think it's
04:31one or the other. I think it's going to be a combination of, of both. And now if you ask
04:37me what
04:37the market share of each is going to be, I think I'm going to pass on this one. I think
04:41we'll see.
04:41But, uh, sure. Uh, we, uh, Bloomberg actually broke the news last week that, uh,
04:48Mistral is looking to fundraise, uh, I think around three billion euro at, uh, upward north of 20
04:53billion valuation. I'm sure that figure is, uh, increasing, uh, as we speak. Uh, but, um, I'm
04:59curious, like what is the, the adoption and adoption at BNP Paribot? Um, well, it, it, it, it very
05:04much echoes what Philip was saying. Um, it's, um, it's a case of a lot of experimenting. It's a
05:11case of, uh, applying different, uh, levels of sophistication. Indeed, there are some, uh,
05:17models that are relatively simple to use and can provide a, a lot of, uh, added value without
05:22moving into the frontier model day one. Uh, so there's a lot of, uh, things that are happening.
05:27Typically we're using it a lot in the coding space. Uh, we're using AI also increasingly in
05:33client interactions to have AI enabled or AI powered type of interactions with customers.
05:38And we're starting to apply AI also to, uh, uh, uh, the most industrial labor intensive and data
05:45intensive parts of our, uh, industrial set up. Okay. So that's, uh, I would say it's a multi-year
05:50journey, right? It's no cliff effect. Uh, it calls for a lot of preparation and calls for a lot of
05:55planning and, uh, yeah, well, sometimes trial and error. I think Mr. Jeff Bezos is speaking
06:01upstairs. So I'm obligated to bring up the space X IPO, um, $2.4 trillion company. Uh, the IPO really
06:08debut really doesn't disappoint. Um, so I think, uh, the calculation right now is the eight largest
06:14U S companies are worth a total of 24 trillion. Um, which, uh, you know, back to the question,
06:21are we in a stage of rational valuations? Uh, well, um, I think spacex today has become the
06:30fourth largest market cap in the U S if I'm not mistaken. It's worth huge multiples of its actual
06:36sales. Uh, it has an amazing business model with a fantastic prospects, but personally as a banker,
06:44I'm the guy with the ties from the banker. Philip is the investor. He's the one without the time. Um,
06:49but as a banker, I find it difficult to justify that level of valuation. Uh, I was a tech banker
06:56in 2000
06:57and 2001 at the time of, uh, what was now referred to as the internet bubble. And there's a little
07:04bit
07:04of feel of, uh, deja vu, uh, when you see some of the very, very hot stories. And I appreciate
07:09the massive, uh, disruption, uh, that is at work, uh, for, by the anthropics or the open airs of this
07:16world or, uh, uh, by spacex, they're, uh, they're, uh, opening entire new fields that are going to generate
07:2110 years to come. But putting a price on that, uh, is very tricky. Okay. Well, markets are to some
07:31extent always right. So, uh, there's a price for spacex. It's very high. Philippe, is this, uh, a
07:39fundamental driven IPO? Is it a storytelling driven IPO? And does that, does this sort of debut, um,
07:46apply to other companies as well? Well, I mean, I, I think what, what's, what has been very interesting
07:53to observe in, uh, I would say the past 15 years is that the private market are becoming the new
08:00public
08:01market and the public market are starting to behave like private market. Yep. And you, to give you an
08:07example, I mean, in 2010, you know, we could have a software company in a portfolio which would reach,
08:13you know, a couple hundred million dollars in revenue. They would go public at, you know,
08:18probably a couple of billion. They would raise an IPO of 150 million. And that was how that was
08:25success in, in, in venture at the time. That was 15 years ago. Today, a round of 150 million,
08:33that's 100% of private route. Sure. Right. And, and you see now private companies where,
08:39you know, when you start reaching, you know, 10 billion, 15 billion and up in market cap,
08:45there is liquidity. So to some extent, they're also behaving like public markets because you can
08:51actually buy and there's, you know, share sell in secondary, not just primary. Um, so that's
08:56basically the dynamic we've seen in the private market. And then in the, the public market, I mean,
09:02yes, the valuations, you know, they look high today, but if you look at where the valuations are coming
09:08from, they're coming from the terminal value of an asset. And when we invest in venture,
09:16the way we, I mean, there, there is no revenue, so there is no multiple, but basically the way we
09:21think about how about the valuation is really thinking, what can this company be in five,
09:27seven, 10 years from now? And then if it becomes what we think it can become, then what is the
09:33value?
09:33And then we back that in terms of risk adjusted, what we're wanting to pay today to get to this
09:39outcome. And when you look at the public market today in technology, it is very much the same.
09:44It is all about the terminal value. It is all about how you think, what do you think this company
09:49can
09:50do in five to 10 years from now? And if you look at where SpaceX was 10 years ago and
09:57where they are
09:58today, and you project that in the future, then how much is that worth, right? Okay, so you're a believer
10:04in the...
10:05I don't think I'm a believer, but I think that that is now the way that people are looking at
10:11these assets.
10:11And it's very different from looking at a company and saying, well, you know, they're growing 20% per
10:17year. If I project, you know, is this growth sustainable for another five, seven years, and then
10:23profitability is going to increase slightly. So I do my cash flow evaluation, and this is how much the
10:28company's worse, right? This is a totally different way to look at the public markets.
10:34And if I may, yeah, Jean, I think it's really a case today for a very patient capital,
10:42because of the pace at which technology is accelerating. It calls for investors to hang around
10:49for probably longer for the company to reach a certain scale and size. And when I started as a
10:55tech banker in the 1990s, I think the average duration before a company going public in the US
11:00when it was a tech startup was about seven to eight years. Now, I think the average in 2025 was
11:06around 15.
11:07So it's a longer hold investment. And over that period of time, that gives the company the capacity
11:14to continue to expand and grow. And in the meantime, there's an increasing amount of capital
11:19that is going into the private capital space. And if you look at the amounts that are being managed
11:25by venture capital in the US, I think it was about $1.3 trillion last year, which is in itself
11:34already a gigantic figure. And this is forecasted to go to nearly $1.8 or $2 trillion within the next
11:39five, six years. So there's going to be a lot of capital, private capital available that enable
11:44those companies to continue to grow, develop in private hands without having to go public.
11:50So we'll see more gigantic IPOs of that nature as the company mature.
11:54If Philippe, if you're one of your portfolio companies exploring a dual track approach,
11:59either going public or acquisition or buyout, what are the pros and cons? How do you walk them
12:06through the different options and where? I would say that in venture, you just don't run a dual track.
12:16At some point, you decide that it is good for the company to be public, and then you go for
12:21it.
12:21I do it. You don't sell a technology company. It gets both. So it's possible that during this IPO
12:29process, a strategic will come and say, hey, we're ready to make an offer on the table. And then we'll
12:34decide if we want to take that offer or go public. But you don't go and try to say, okay,
12:41I'm going to try and sell a company. Great company get both. They don't get sold. So if we decide
12:46to go
12:47for an IPO, we go for an IPO. And if something comes along, then it comes. But you don't proactively
12:54go and talk to strategic buyer while you pursue. And sometimes it can happen. But I would say that
13:00for a great company, you don't shop them like someone is going to come and try to take them.
13:07Last question on SpaceX, which is, I think there's been a fair amount of criticism over how
13:13private companies have grown to such a scale. They're being valued at nearly a trillion in the
13:19private markets. So when retail investors finally get a hold of them, they're not really getting their
13:24money's worth. So question for the banker here. Are retail investors sort of being shortchanged in
13:31this dynamic where public and private markets are converging? No, look, I don't think so in the sense
13:37that at the end of the day, investors, they make educated investment decisions. So if you look at one
13:44of the early backers of SpaceX or a very big one, take Google. So Google was an early investor in
13:50SpaceX.
13:50I don't know whether they were in the very, very early arms, but they invested over a billion dollars.
13:56Their billion dollars has become a hundred billion dollars. I'm pretty confident that anybody private
14:02individual or investor who has bought into the IPO is not going to get a return of 100 times their
14:08original investment. That boat has sailed, right? So I think what you need as a banker to be worried is
14:15to make sure that you always pay attention to what we call suitability, meaning the investment proposal
14:22that we are bringing to a private individual has got to be commensurate in terms of risk profile and in
14:27terms of magnitude with the ability of that investor to take a hit. Right now, the SpaceX stock price is
14:35going through the roof, right? There might be negative headlines further down the road. That means it
14:40might drop below the stock price. And we need to pay a lot of attention to investors' ability to stomach
14:47that. Sure. Especially retail, private individuals. Yes, indeed. Moving on from disruptive technologies
14:58such as bringing the human inhabitants to Mars and the moon. Let's just talk about AI for a second here.
15:05Philippe, there's been so much criticism of some of the metrics that startups use such as ARR. Can you talk
15:13a little bit, just walk me through your thinking on some of these KPIs that startups tout? Is there much
15:20science behind it? Is there a more reasonable way of examining if there's less churn or there's enough
15:30the profit margin? What's the best way to do it? Well, I don't think, you know, on our side,
15:37we don't think that the metrics are any different today than they were 15 years ago when they were
15:43designed. I think ARR is just a recognition of a recurring revenue that a company is making. Now,
15:50I think some of the questions that people are asking is like, you know, if the models are turning to
15:55a
15:56consumption model, how recurring is that revenue, right? Because historically, a recurring revenue
16:01was a recurring contract where you're paying, you know, 100,000 a year and that's what you're paying for
16:07the year. And then it's renewed for, you know, 100,000 plus potentially some sum of sales and price
16:13increase. Today, when it's basically based on consumption, you can say, well, how is my consumption
16:19going to, you know, change? Is it going to go up, down? So there is probably a bit more variability
16:24in
16:24it. But so you have to take this into account. But I think for me, like this core metrics are
16:32still,
16:33you know, their basic microeconomics are still working. The key question is, in a world where
16:39manpower, you know, is complemented with token, then how do you think about productivity metrics?
16:47And how do you integrate the token, the cost of the token and the usage of the token to understand
16:55how a company is performing? Because, you know, if the company, you know, as uses a lot of token,
17:04you know, to deliver their products with a low gross margin, then, you know, obviously you're saying,
17:10well, there's just a limited value add that this company is providing on top of a model. So the
17:16question on how valuable this company is. Now, if the company doesn't deliver, that doesn't use any
17:21token to deliver its product, you would say, well, is it really a modern product or it's more legacy
17:27products that's going to be disrupted with AI? So I think what companies are trying to figure out
17:32right now is kind of what is the balance in terms of the optimal usage of token in their product
17:37that enables them to have software gross margin in the 70, 80%. And that's, we don't have an answer yet,
17:44because, like, everybody's trying to figure it out, right? And you have the SaaS companies that are
17:49putting more AI in their products. So they're coming off, you know, less token, more software. You have the
17:54next generation AI companies that are coming with more token, less software. And, you know, my guess is that,
18:01you know, at some point, the next two, three, four years, they're going to converge towards an optimum.
18:07Yannick, you know, software, which was the darling for the private equity and private credit
18:13industry really brought the entire industry down to its knees a few months ago. And I'm just curious,
18:20you know, for the institutional lenders, how are you guys thinking about this changing chapter
18:25industry for this technology? There's been a lot of talk about, like, this SaaS-pocalypse,
18:32the end of the year of software. And, you know, we at BNP Parima, we are, for all intents and
18:37purposes,
18:38in Europe, we're the biggest leverage finance house, right? So any software company that is borrowing
18:46money in Europe, chances are they either bank with us or we've had to look at it. So we're running
18:53very
18:55tight, closed assessment of their ability to pay down the debt and their ability to generate cash
19:02flows. And so far, the analysis we have conducted, and again, this is based on decades of experience
19:10as a lender, doesn't give rise to any specific concerns, right? Okay. There's a handful of borrowers
19:17that might be a little bit too leveled. They might have too much debt, but that's a fact of life.
19:22It
19:22happens across lots of industries. So I'm not worried about the actual underlying credit risk,
19:30at least for what sits in our book as a lender and as one of the biggest lenders in Europe.
19:35Mm-hmm. Philippe, let's talk about your book. So I'm sure every venture capital firm has had some
19:42traditional SaaS companies and a lot of them has gone through what would have had to go through
19:48the process of retrofitting themselves in this AI age. Could you talk a little bit about, like, what is
19:54that transformation typically like? And I'm sure that there are winners and losers. So how are you
20:00underwriting that risk right now? Yeah, well, I would say that, I mean, AI is a great opportunity,
20:05has been a great opportunity for all our software portfolio because it basically gives them a way to
20:11be more productive and especially accelerate their roadmap. So things that were taking 18 months now
20:17can be done in 10 months. So you can develop a lot more products with the same head counts. And
20:25two,
20:26now you can add new functionality, more intelligent functionality in your software. I mean, if you take
20:31a company like Dr. Leap, to take a French example, I mean, for the past two, three years,
20:37they have been super active at incorporating AI into their product to basically help the doctors
20:45prepare, you know, the meeting with their patients to basically inform the doctor about, you know,
20:53what our medication would be helpful in, you know, in a case. And so basically, it makes the doctor a
20:59lot more productive, it helps the doctor in the diagnostic. And so it made their platform a lot more
21:05powerful, like they just launched a new app for parents to basically give guidance. If you just
21:11have a newborn, like, and your parent, you're anxious, and now you have an AI that can actually
21:18help you. And obviously, in case it becomes critical, there are also doctors on the line who can take over
21:24the AI, etc, because they're like, they're not giving medical advice. But I think it has been a real
21:29boost overall for the SaaS company. Now, are all going to make it? Probably not. Some, you know,
21:36are going to be left behind, the ones that are not able to take that turn. But the ones who
21:40are able
21:40to take that turn, they are going to be a lot more valuable than they were before.
21:44We have about two minutes left. I really want to touch on defense. Staying with you, Philippe,
21:51XL has recently gone into defense investing. You're the backer of Cambridge Aerospace, a drone
21:57interceptor. Can you maybe just, you know, in your mind, where do you think we are in this defense
22:04cycle, investment cycle right now? Because I can think of at least 10 drone companies that's
22:09fundraising, hundreds of millions, and clearly, there's a lot of froth in the industry. When
22:14do you think this will settle out? Well, I think there are two fronts on the defense side. There
22:20is more the hardware, and then there is cyber. I think on the hardware side, I mean, clearly,
22:28the defense budget in Europe have increased from, you know, one percent of GDP to two, three,
22:35four percent of GDP. And Europe is realizing that it cannot rely on capabilities from other countries.
22:41So we have to build our own capabilities. I mean, France is the only country in Europe
22:46who can manufacture rockets, right? I didn't know that. Yeah, but I think in a world where
22:54actually ballistic missiles now becomes, you know, a very important weapon on the battlefield, like,
23:00you know, Europe needs to build these capabilities. And so we're in this cycle,
23:03and I think it's going to run for some time. I think on the cyber side, it has always been
23:10very
23:10active. Now it's becoming even more urgent to do this transformation. And so, you know,
23:17we have companies like Sayera in our portfolio who focus on data and AI security, and we see, like,
23:22huge momentum because every company is looking at Mythos and what the vulnerabilities that Mythos can
23:28uncover and say, wow, we know what's coming. We need to be ready for it. And we don't have two
23:33years.
23:34Like, we need to do it now. So there is very strong momentum on the cyber side as well.
23:39Yannick, how are banks adjusting to the rearmen of Europe?
23:43Well, we're playing our part, you know, BNP Paribas being the biggest bank in the Eurozone.
23:49We're also the stalwart protector of European sovereignty. And so today, we are extending over
23:5526 billion euros of loans to the aerospace and the defense industry. We're managing over 8 billion
24:03euros of a direct equity investment on behalf of our client base. And we have close to 300 million
24:09euros of direct investment from our own money into startups and mid caps across the space. So
24:18we are firm supporters of that industry. And this is to protect the European way of life.
24:23You know, defense is about defending yourself. And we're very strong supporters of that industry.
24:31So I guess sovereignty is the flavor of a day. Yannick, Philly, thank you for your time.
24:37Thank you, Yejean.
24:37Thank you. Thanks all.
24:39Thanks.
Comments