[Régis Hamelin]

Jean, before we go through your analysis of the semiconductor industry, can you introduce us to Jolt Capital in a few words?

[Jean Schmitt]

Yes, of course. Thank you everybody for being there and listening at that conference. I hope it’s going to be interactive. In a nutshell, I created Jolt 15 years ago and the idea hasn’t changed since then. The whole idea is to fund companies which are growth stage technology companies in Europe. So our main objective is to find those companies which are 10 to 15 million euros in revenue. EBITDA is another debate, but it can be negative, of course, because at that size anyway, EBITDA doesn’t matter so much, but we are looking at those mid-sized companies and to fund their growth, so taking them from, you know, 15, 20 million to 100 million. So we are currently deploying around 150 to 200 million a year. Our upcoming fund, which is going to be invested, we are going to start investing that fund in the end of 2025, is close to a billion euro fund. And so we have sizable amounts of money to invest in deep tech companies. And within that spectrum, semiconductor represents the largest portion of the money. So we do invest in medical device and AI. We do invest in materials which are not always semiconductor materials. We do invest in technical software which are not always technical semiconductor software, we invest in mobility and energy management but semiconductor represents anywhere between 30 to 40 percent of that money. In our previous funds it was the same proportion and we always had been bullish about semiconductor in Europe because there is a lot to do here because we never lost any money on semiconductor so we like it a lot. We always made money there and I think there is a kind of way to invest which works. So we are a team of 42 people. Again, we manage around one and a half billion. We are entirely focused on Europe. We have businesses, we have offices all over the world. We have people in the Bay Area, in Boston, in Montreal, in Seoul, in Tokyo, in Singapore and across in Europe as well to support our companies and support their growth.

[Régis Hamelin]

Yes in Semiconductor are you looking at more specifically Fabless companies, Photonics or equipment or design or anything works?

[Jean Schmitt]

We are in Europe so we need to go where we can operate successfully. So we are quite excited about IP companies which are licensing IP. We like a lot, and we can talk about that later, capital equipment. This is actually a business we keep liking if there is significant IP and a thoughtful strategy. Photonics of course, because photonics is in a way the semiconductor of the poor, it’s a little bit less expensive to have a fab in photonics than a two nanometers fab. Fabless, we are careful on fabless because we don’t like this dependency on two nanometers fab and spending a billion to do a new AI chip is dangerous when you don’t own the fab and when you’re not co-located with the fab.

[Régis Hamelin]

Okay, co-location is very important, that’s because of the geopolitical tension, right?

[Jean Schmitt]

Yes, and you mentioned a key word which was “pilot”. With your network across Europe, those pilot lines are critical. If you don’t see a pilot line or if we cannot establish ourselves a pilot line, there is no way to be here having an R&D center in Europe and believing that the other guys will be stupid enough to just do the manufacturing.

[Régis Hamelin]

So, and can you mention a few companies that were some success that you said that made only money, so you must have great track records? Can you give us a few examples of successful companies you have invested in?

[Jean Schmitt]

Of course, so briefly, recently, we sold the company UnitySC, which was a metrology and inspection tool company. So we were inspecting all sorts of things. What was innovative in the company was a sensor and a software. And this fusion sensor we put together and the software were great. All that was packaged in a machine, which was, of course, a big, big robotized machine. It was a million dollars a unit, roughly, and we started selling that like everybody, initially as a pilot tool, R&D tools, and then we went into lines, and then we went into full production, controlling, for example, any single NVIDIA chip at CSMC. So we sold quite a bit of machines. It was an interesting venture. The technology was very innovative. The capabilities of the company were really miserable. So we had to acquire a company with serious capabilities in Germany, another one in France, which gave us these clean homes, the processes to build the proper machines, build a partnership with a good subcontractor in the Netherlands to do the final assembly and the quality control, and all that together formed an ecosystem for that company, which was meaningful and allowed us to sell the company, by the way, to a European player, because we sold the company to Merck, which is a big pharma, but also a significant player in the semi-space. I can give you another example, which is fun, which is a company called Heptagon. And that’s a company which was, it’s a perfect exercise in strategy. Those guys were selling optical elements, which then were packaged with some semiconductor sensors in order to do optical modules for iPhones or for cell phones. And at a point when we invested, we said, guys, we need to stop that. We need to not be providing our great optics to people packaging us, but you are going to become an advanced packaging company. And we started to build some very advanced packaging, especially things like the facial recognition of the iPhone, which is a very complex package. It’s a big cell going through a diffractive element with a lens and a CMOS sensor, all that in a teeny, teeny package, which of course doesn’t have any crosstalk if possible, and which is very safe because you don’t want to beam a laser beam into the eyes of your customers. Turning your customers blind is not good for business. So we produced those for the iPhones, and that was a multi-billion contract. That company went from 100 people when we invested to 10,000 people when we sold the company and it scaled quite a bit. That was also a very good deal for us. But you have many others. I’m not going to talk about any of them.

[Régis Hamelin]

You mentioned capital equipment and also advanced packaging. We’ll see later in your slides that those specific two are especially maybe less invested, but the important point is that you mentioned advanced packaging, but it’s not a company which is centered or centric on advanced packaging, it’s more a kind of verticalization of the company with advanced packaging broad value.

[Jean Schmitt]

Yes, building a generic company in advanced packaging is very exciting but much more complicated, expensive, and one vertical application looks like nothing, but the factory itself was a $400 million factory. So it was not cheap to put that thing together. Of course, we were doing much more than facial recognition. We were doing all sorts of infrared sensors and the presence sensors and colorimetry sensors and so on. But that said, you’re absolutely right. It’s a verticalized packaging company we had.

[Régis Hamelin]

Okay, so thank you very much for this first insight. So we’re going to talk a little bit about your process in investment because you have an internal process which is powered by a specific tool that you have developed in-house. So can you tell us a little bit more about this tool?

[Jean Schmitt]

Yes, I think it’s going to be helpful for everybody to understand where the data I’m going to talk about and comment on is coming from. At the very beginning when we came and explained we are going to invest in late-stage deep tech companies across Europe, people were laughing at us telling there are not enough, the market does not exist. So we designed this tool which is called Ninja and Ninja is collecting data from the internet and trying to detect companies. So today the platform became very large and Ninja is discovering one new company every single minute. And during that minute, this same minute we are updating around a thousand business plans or business cases. So it became a very large database. It’s a large graph describing business, including patents, people, relationship between companies, funding, activities, whatever you name it, all over the world. So today we have 5 million technology companies in our database. This is actually the largest database in the world. We don’t sell it, but we use it for finding deals, for doing M&A, combination of companies, build-ups and so on. And of course, we added all sorts of AI gimmicks in it, virtual analysts. We don’t have many juniors in Jolt because this platform does the job. But also, since we have a lot of data, we can now comment on the size of the markets and how to target the quality of IP and so on. And this is what you’re going to talk about. We are organizing our own strategy on Ninja, but many others as well. So you would have Temasek in Singapore, you would have people like EIC in Europe or the government or many other funds are using Ninja today. We are not jealous of this data. I don’t believe in data you keep for yourself and you just use it for your own little business. I’m a strong believer in the network effect and of sharing that data in order to form a community of investors. I think we have 25 funds across Europe using it today. By the end of 2026 we would have a hundred funds. And I think it’s good for everybody, investors, entrepreneurs, everybody will benefit of having a more transparent market.

[Régis Hamelin]

So for our audience, from now on, the slides that I’m going to share are coming from an internal study from Jolt Capital. The first stop will be an analysis you did on the global semiconductor investments, can you elaborate and explain what we are currently seeing, because we see a huge investment leap from 2020 to 2021, yes.

[Jean Schmitt]

Yes, what you are looking at here is the investment, so is the venture capital and growth investment in semiconductor and here it’s a graph which is worldwide so you see that there is an uptick of investment starting in 2020 and AI is a part of that investment but not only and the VC and growth investment had been quite sustained during the last few years so this is boosting innovation, this is boosting the number of companies we will see that later. You also see on the middle graph, you see the evolution of the share between the different regions in the world of that money. So Asia is capturing a bigger and bigger share. Europe is very modest and the US is capturing a lower share. And you will see that over and over. Asia is taking over in terms of semiconductor now, in terms of money, and you will see in other domains, Europe is, I would say, supporting its own industry, but not developing it. So, we are mediocre, but at least we keep as medium-sized as we are, so we don’t shrink, which is a performance, because the US is shrinking significantly, as you can see that. So we talk about very large amounts of money. We will talk also about the reason why Europe is not developing. There is a very simple reason why, but today we keep operating nevertheless, so we are not in a desperate point. We don’t develop our business as we should.

[Régis Hamelin]

But it’s 25 percent of the deals in volume. The number of deals, one quarter of all the deals are made in Europe, but for only 10 percent of the money. So it means that the size of the rounds are smaller, right?

[Jean Schmitt]

So the size of the rounds are smaller if we compare ourselves with the US in terms of number of deals, and it’s true for Semi, it’s true also for any other domain. You would see that Europe did catch up in venture capital in a number of deals, and if you divide the amounts of money by GDP, Europe and the US are doing similar efforts now in venture capital. Of course, in absolute dollars, the GDP of the US being much higher now than ours, they are spending more dollars. On the other hand, US is much more expensive as well. So I think dividing by GDP is kind of creating a better comparison. So in venture capital, we are kind of comparable in terms of efforts. Where we have an issue is on gross capital, but we will cover that later. But yes, we do a lot of deals, you’re right. And we are creating a lot of companies. What do we do then is the question.

[Régis Hamelin]

Then the next topic is for the European semiconductor investment universe. You did a study on 3000 companies which are mentioning AI, right?

[Jean Schmitt]

Yes, exactly. AI is currently the buzzword for many applications, particularly in the B2C sector, but for the semiconductor community, it’s a critical concern that offers enormous opportunities. These opportunities are everywhere: in lower geometries, power management, and in creating chips that are hundreds or even thousands of times less power hungry than current technologies. These are truly game-changing semiconductor chips. This makes AI integration critical across all segments, including capital equipment, IP, power management, the chips themselves, and Fabless design. However, when we analyze where people are actually discussing AI, the data is a little disappointing, especially in terms of marketing and positioning. The mention rate is quite low across the board. For materials, manufacturers, and backend, the low percentage isn’t surprising. But in capital equipment, as geometries shrink, new tools are essential. We’ve seen companies like Near Field raise significant money, and anything sub-2 nanometers must mention AI to justify its existence. Capital equipment vendors are increasingly integrating AI, not just for tool development, but also for their own internal capabilities, as we expect more innovation using AI within the fabs. Today’s fabs consist of many disconnected silos. To achieve better results, these must be connected, with AI placed on top to manage the entire process. While capital equipment isn’t perfectly interconnected yet, we see a growing trend in Design and Fabless, and a correlated trend in EDA. Still, at only 12 to 14 percent, the adoption in marketing is very low. I would certainly expect more AI-focused marketing from startups, but that doesn’t seem to be the case. So, is this enough? Should we be seeing more? We also need to consider the potential for an AI bubble coming.

[Jean Schmitt]

Between you and me, I think European semiconductor companies are terrible at marketing. They aren’t using the right buzzwords as they should be. When I see a company developing capital equipment for metrology or inspection, and they don’t talk about AI? Give me a break.

Sometimes I see companies using old terminology. I’m an AI guy; I had to hide when I was young because, in the 90s, you couldn’t mention AI, it was a bad word. So you talked about statistics or data mining instead. Now is prime time for AI. Not using these buzzwords is directly decreasing a company’s valuation and making it less attractive. Companies need to think carefully about how to promote what they do, build their case, and even publish scientific papers explaining how AI boosts their technology. They don’t have to claim they are an AI company, but they must showcase how they use and implement AI.

The AI bubble we’re currently witnessing is largely driven by consumer usage and adoption (B2C). AI is here and is evolving. Are Large Language Models (LLMs) going to take over the world? Absolutely not. But are they a great milestone as foundational models? Absolutely yes.

Now, we could argue whether a company like OpenAI, which loses $2 billion per month, is a great company. Of course, the semiconductor industry thanks them for the business, but they’re still losing $2 billion a month. This suggests the bubble might burst on the B2C side. However, in the B2B space, it’s not a bubble; it’s a solid trend. If there is a valuation uptick right now, let’s just enjoy it. The key is to split our focus between B2C and B2B.

[Régis Hamelin]

So I suggest we move on now with the next stop with how fast companies are raising funds according first from the date of inception and from their geolocation, where they are located.

[Jean Schmitt]

Yes, it looks complicated, but it’s very simple. Here the idea is that we took all the companies and we said, how much money do they raise year one, year two, year three, year four? And we put different curves which are showing what happened to them depending on their date of birth. So you see the solid curve in blue, which is the 2022 vintage. So the companies which were born in 2022, and you see their investment trends. So that’s the latest one. You see that in the US, you have the other two curves which are starting in the same way. So in year one, year two, they raised 20 million very fast, and then the 2022 cohort raised much more after year three because the AI bubble is pushing them up. But all in all, if you compare Europe and if you compare the US, year after year, you have the same message. The US companies are raising much more money, much faster than the European guys. We don’t want to drive too quick conclusions, but I will comment on that. But what you see is that after year four, in average in the US, the guys raised around 60 million, while the European, in average, went to 20. So we are in a tough spot. Let’s not drive too quick conclusions here. Having tons of money is good, but let’s not forget that our U.S. companies waste a lot of money as well. They pay people three times the salaries and so it’s not always well invested money. So not everything is related to money. That said, it is very clear that on year four or year five, we didn’t put year five because it gets off-charts almost at a point. But what is surprising here is that growth capital in the US is pushing a lot on semiconductor, while you see that gross capital in Europe does not intervene before year seven and year eight. And actually, we see that as a result, many of the companies which we look at are seven, 10, 20 years old sometimes. So it took a while for them to get to a reasonable level of revenue, while in the US it goes much faster. So capital in the US-

[Régis Hamelin]

Is it because they are local?

[Jean Schmitt]

I think that the corporates are much better corporate citizens. So, you know, in Europe, our corporates are not desperate, but they are focusing on themselves. So the STMicro, the Infineon, the NXP are around, they do work with their ecosystem, but not that much. While, when you are in the US and you are next to the TI’s, the Intel’s and so on of the world, they are investing massive amounts and they are co-investing massively, not to mention things like DARPA and so on, which are pushing companies to grow much faster. So not to put pressure on corporates, DARPA here, we didn’t put the contribution of DARPA in those graphs, but it is very significant. In Europe, we give grants. DARPA is not giving grants. DARPA is giving contracts. So you get real revenue out of those contracts. And that’s a game changer for companies. They learn how to use that money to achieve something as a supplier, not as a grant to do research. So those differences are very much in those numbers. Would I see companies doing 20 million based on research contracts and so on as an investor? I would consider that as true revenue, not as grants. So it’s a different animal they are forming, thanks to their own way of organizing their business. Europe is underfunded, really, yes.

[Régis Hamelin]

So four more years for a European startup to reach roughly the same level as a US company. Yes, and you see that the AI bubble did hit the US semiconductor sector because the increase of amounts invested grew quite a lot in 2024-25, while in Europe the AI bubbles did not hit funding. Right, that’s true. Yes. All right. So there’s now a concept that you call the hidden universe. So can you elaborate about what is this hidden universe and is it worth of investment?

[Jean Schmitt]

Yes, it’s the idea. We are cold-blooded investors. So my job is to invest at a great price in companies and sell them at a much higher price. So when I look at the market, the one thing nobody likes is competition. So there is a place in Europe, which is what we call the hidden universe, which is like our Shangri-La. It’s the place where no VC, no private equity ever invested. So you have plenty of companies in Europe which made it without any private equity money. Those are the best ones. Those are the most interesting ones, because those are the guys who could survive probably doing uninteresting deals in order to fund some interesting technology or they find ways. And those are great survivors. So, per domain, if you look at the little graph on the right, per domain, you see that 55% of the AI-related companies are funded. So 45 are not. When you go on devices, 48. Manufacturers, 44. Design and Fabless, 44. which are backed by VC. So already here you have a gigantic universe of companies which are doing design and fabless which are not funded and if you go down you will see that semiconductor materials are less funded. When you go to EDA, almost nobody, 36% of the companies only are funded and when you go to capital equipment as you mentioned it before, 29% of the companies on capital equipment backend are funded by VC. So in other words, if you are a capital equipment company, you had better count on yourself because there is little money available. For us as investors, I like a lot those graphs because it tells me where I should go to find companies which are good at surviving and navigating turbulent waters. If I’m a capital equipment company, I put a little bit of AI in it and I survive for seven or eight years, I can look at Jolt’s door. Absolutely. In fact, if you come to me with significant IP, so good patents, a set of families and a decent business, so if you want the recipe to be loved as a capital equipment company, the recipe is the following. You have 20 families of patents, you do 12 million, you have a niche product and your niche product is starting to get into production and is already out of R&D. So, of course, if you sell one machine to IMech, one machine to Astar, one machine to Letty, you’re not yet in business. You’re just a potential business. If you sell two or three R&D machines at large fabs, it’s still R&D machines. We know what it is. You’re going to lose money forever doing customs. What you want is someone telling you, hey, I’m going to buy four of those machines, even if it’s a small production, then we love you. You’re in business. Then we are going to take great care of you because we believe there is something to scale. And the other recipe is that don’t try to kill the big guys. You kill nobody. In the semiconductor world, but here in front of a lot of people who know this sector much better than I, you kill nobody. You disrupt nobody. So the reason why actually we like semiconductor as well, you contribute. So your job here is not to go to Lam and tell Lam, I’m going to kill you. That’s going to make them laugh a little bit. It’s like Lam, you know, you don’t look at that…