Summary
Quick summary about these 3 days of congress. First of all, the laser German industry is not in a great shape, China is a difficult competitor. There was a lot of talk about Laser Inertial Confinement Fusion, still in its infancy and I don't know how it can get anywhere without a proper public support from the government. The other big topic was PCB manufacturing of Through Glass Vias (TGV) that is a big market for laser machine makers right now. All (95%) of the clients for TGV machines are in Asia. I know, it is a "very" big surprise.... AI in photonics is progressing it seems, although it is mostly in the field of Machine Learning. China is using AI combined with lasers all the time. In the West are a little more scared about using AI. Somehow a lot less talks about laser micro-structuring this year.
I am going to try to write now about certain topics, I will try to make these paragraph independent to help people read only what matters to them
European Laser Industry outlook
The first day was mostly the technology day where the morning was focusing on the laser industry in general and the afternoon on several topics that we had to chose. Between Medical, Quantum, Electronics, etc. I chose electronic for the afternoon session mostly because this is not a field I know very much and I was curious. The speakers from the general morning session were there to share their views on the state of the global photonics market and the laser market worldwide, in Europe, US and Asia. Not a surprise but the last few years have been hard on the German economy, and on the German Laser Machine market. According to the numbers, and to the speaker, China is becoming "really difficult to compete with", and Germany has difficulties finding products that it can sell in China as China can serve it's own domestic market very well now. This goes for lasers under 1kW and 10kW, while there is still a small market for lasers of much higher power, that China is not doing yet, but I would say that they are on the way to also serve that domestic demand very soon. Strangely enough, Italy's demand for laser and laser machines is growing this year. And China is starting to export more and more lasers and laser machines... Ouch. The other oddball in the numbers that I saw was the sudden demand of laser machines in Vietnam and Thailand. Apparently, this comes from the Tariffs Trump imposed on China last year that forced them to de-localize some production to neighboring countries with less tariffs. China who is really swallowing everything. According to the speaker on the subject, there is 5 mega-trends right now in China:
- Defense - Big spending in defense applications
- Hot applications continue to emerge. Less about fiber lasers, more power in ultrashort pulses, prices going down. LIDAR is huge because of car industry and robotics. Laser display industry, semiconductor, laser cleaning, etc...
- Replacement of import. Use of internal companies to serve domestic market
- Domestic companies go global (BYD, Xiaomi, etc...)
- Financial capital is pouring like there is no tomorrow
Final note from the speaker about the Chinese market is about silicon photonics.... and how the time has come it seems? Lots of investments in silicon photonics for ultra high speed interconnects where electronics is the bottleneck because of skin effect at very high frequencies. Photonics is the only solution right now, mostly for data centers.
My point of view on the subject is simple and it really comes down to one article from the Financial Times. "China is making trade impossible". Basically, as one speaker weakly suggested, we need to find what to sell to China. Problem is simple: China can do everything and is doing everything. Agriculture product, metal products, complex engineering products. Everything. Domestic market in China is supplying all its needs and there is ZERO incentive for Chinese companies to buy Western products. At all. ZERO needs. And even with the talk about how we are the only ones to make good ultra high power laser, how long before they can make it themselves? Do you really think that a bunch of patents will prevent them from copying and improving your design before they sell a competitor product for 1/3 of the price? This is a complex situation that I am less than qualified to find an answer to. Tariffs "a la Trump" is a stupid proposal as they will simply de-localize production, as they are doing right now and which is hurting the US more than it is hurting China as a whole. I think that the best solution would be for Chinese government to realize that they are strangling the World's economy doing what they do and that to have a commercial relationship, the best would be to allow some exchange to happen. As usually, someone needs to regulate what is happening. Some people might argue that we could simply stop buying Chinese products. While it could be a solution it is also going to stress the international relationship between the countries. Not to mention that decades of unregulated tech transfer from the West to the East depleted the West capacity to replace China. Decades of wanting to increase profits by de-localizing to poorer countries to produce cheaper products lead to that situation where we can't restart our own fabrication lines as simply as flicking a switch. I actually found a blog article a few days ago called "The west forgot how to make things. Now it's forgetting how to code", basically saying the same thing, but extrapolating to Software and AI. As it is written in this article: "Rebuilding takes years. Always." And listen, I am also part of the problem, earlier this year, I wanted to do some custom PCB. Simple PCB to convert a high frequency signal into another signal to drive a laser pulse picker. I asked a quote from a Chinese company (won't say which one, it is one of the famous company that has loads of ads on youtube), and a quote from a Swiss company that, according to their website, makes PCBs for prototyping and also for large quantities.... Great. I thought to myself that the Chinese manufacturer would be 100 CHF and the Swiss should be 300 CHF probably. The PCB I asked was a simple PCB with simple components at a total of 10 CHF price for all the components on the board. Basically, the connectors were more expensive than the chips in it. Well, the quotes came back a few days later. China was 100 CHF for 5 boards, international shipping included. 20 CHF a prototype board. Swiss quote was 1200 CHF for 3 boards. 400 CHF a prototype board. Same components, same everything. It was not a factor of 3, it was a factor 20... AND the experience was terrible on the Swiss company website to upload the Gerber files and select the components. I ordered the Chinese PCB, obviously, despite me wanting to support the local economy. Why? Because it is a prototype board that costs 15 CHF in raw material! How can they explain such a ridiculous price? And this is the problem really. In PCB making, if there are no companies willing to make cheap PCB in Europe, how can we be any good in electronics? If I decide to make an electronic equipment, I need 2 or 3 rev of a board till I get everything as I want. With the Swiss company that would cost me 5k. with the Chinese company, 300! This is a no brainer. Why is there no cheap pcb manufacturing in Europe. Hell, why is there no cheap manufacturing of anything in Europe? Because all prototyping is de-localized. Because prototyping is not "rewarding enough" for companies in Europe? But this prototyping is the stepping stone of everything else. I do the prototype at your company, I buy the final run of 10k units at your company! How can these "industrialists" be so short sighted?
So China is playing the game well. Some might say a little too well. And it is winning and crushing us with an ease that is only explained by how short sighted and risk avoidant we have become. Well played China. But things need to change, for the good of everybody involved in the game.
Through Glass Vias
This came as a surprise. Almost half of the talks I assisted to this years mentioned at some point or another these Through Glass Vias. Basically, the idea is to use Glass as a substrate for very demanding applications in data-centers. The idea is to create those tiny holes, between 20 and 4 microns of diameter over almost 1mm depth using Selective Laser Etching (SLE). That is, first the glass is shot with a femtosecond laser that creates glass modification on the path, micro-cracks. Then there is some wet etching that melts the glass in KOH or HF depending on the geometry of the channel that we want. This is not a new method, but I didn't hear about this application the last time I was here. Now it is the talk of every laser machine manufacturer, from TRUMPF to smaller integrators that supply ASML. Last time actually there was a lot of talk I assisted that mentioned SLE for plenty of other applications, and I don't think I heard Through Glass Via in any of those previous talks.
Laser ICF
Laser Inertial Confinement Fusion was a big portion of the talks during the second day. To be fair, the second day morning was actually about high power and high energy for novel applications, so ICF was the obvious target application that the organizing team had in mind here. Last time I was here, in 2024, ICF was already a big subject, with a lot of hope being placed there by the speaker at the time from Fraunhofer. That same speaker was there this year to continue the hype. Some ICF presentations were done the afternoon and some live presenters had ICF themed booths during the laboratory visits. What is new since 2024 is that some startups have emerged in the field, and are trying to get money to finance huge projects, such as turning the old Biblis nuclear power plant into a laser based ICF experimental plant in the future (2035) coordinated by a startup called Focused Energy Gmbh. I would say that the project is colossal. Way to big for a startup, especially if they need to find so much money as to adapt an existing nuclear power plant to house gigantic laser complex to make tests of laser ignition. In my mind, this kind of project is an absolute necessity for Europe. Vital even. And it can only be done with an oversight from the government with money from the government, not from investors in a startup environment. I would say with 100% certainty that if they do not manage to change their investment model and find support from several countries directly (Germany, France, Switzerland, Italy, Spain, UK, Belgium, Netherland and others to name a few), create an international consortium, create an ecosystem of startups and companies in the field, this will fail. And in my opinion, it is already failing in this way. They were proud to announce that ICF was getting investors, and that they were able to get 5bn€ for a period of 5 years for this project. It might seem like a lot, but compared to the recent AI investments of several 100's of Billions of dollars in data-centers, this is nothing. The money is there in the pockets of a few and they are not investing in ICF. The public hype around laser ICF faded rather fast, dethroned by gen AI. And at Biblis, the numbers are not small. They need a grand total of 1160 super high power laser. 69GW of diode peak power, 500MW of cooling capacity, 1GW of laser input. Just a small reminder of ITER: the project, which is based on existing tech, so no "hardcore R&D required as in Laser ICF", cost initially 6 bn€. Actually, it is now estimated between 18 and 22 bn€ for some, 45 to 65 bn€ for others. Again, ITER is maybe the project closer to what is being planned at Biblis, except that ITER is founded by several parties, including China, EU, India, Japan, Russia, South Korea and the US and other countries (Switzerland also, but with an interruption between 2021 and 2026). And this is part of the problem I think. ITER is also about Fusion, using magnetic confinement instead of laser confinement. But there is no startup leading the project. It is an international government led scientific collaboration that is spearheading the project. And I don't see how it could be done otherwise. And I would say that it must be the same for Laser ICF. Instead of a single startup doing this, we need government involved in the project, if only for regulations, but also for investments without any expectations. Oh, and did I mention that ITER has been founded in the 80's and is only planned to start in maybe 2035, "If all goes well"?
Because one of the funny remark came after this whole Thursday session. In the afternoon, after the panel session of the morning, one of the speaker was explaining how we need to move in Europe, how we need to act fast. And concerning ICF, the speakers of the panel this morning were all very enthusiastic in front of the audience, but behind the scene, they were apparently very reluctant to make the first move toward ICF, for fear of losing money. And this is where the limits of startup investments (and big corpo also) in this kind of project go. As long as there is no secure funding from a government led agency, no one will make the move for fear of loosing money. Because we all know that people involved in this project will lose money at first. It is an open secret. Laser ICF has been "demonstrated in research" by one actor, the National Ignition Facility, but there is no real path to commercialization before at least 50 years.... Who in their right mind would say: "50 before return on investments? Count me in." Nobody can say that. Except governments.
So my word there is : "Open your fucking eyes EU commission, don't mess this up again, we missed the go for internet, solar panels, EV, robotics, etc... in Europe. Please don't mess this one up also. Because China is ready to tackle this once they finish with EV and they will also beat us if we wait too long." Heck, even if we go now full speed, it is not sure that China will not beat us on the finishing line and sell Laser ICF reactors at 10 times less that we would. But that doesn't mean we need to stop trying now.
It actually reminds me of something else. I remember reading something very interesting somewhere said by the president of the EU Chamber of Commerce in China: Jörg Wuttke. He said this: China's view of startup innovation is that of a gym. Plenty of people get to the gym weak in order to grow muscles and become strong. 100's of startup, many fail, but out of the bunch, some manage to grow become strong and eventually win domestic market. Once they win the domestic Chinese market, they are allowed to go oversea and they destroy western companies. In the west, it is totally different. We only allow the gym to pre-aproved gym rats, who are already shredded and therefore, with a very low risk of failure. This drives innovation down significantly, as we don't allow innovative startup to even try. It boils down to very low risk appetite from western investors. Basically, if these western investors are given a choice between an investment with 3% return at 80% certainty and 80% return at 3% certainty, they will always take the 3% return. I understand that, but we need to act bold sometime. And so, who are those investors in China that are so willing to take a risk? Once again, it is the government that is willing to take the risk, with a lot of money. And this is again something that I heard during the talks: If you have a PhD and an idea in China, you can easily start your company. Fail most likely, but at least you can try, and if it succeed, the losses are more than compensated!
AI in photonics
Last and not least is AI in photonics. One speaker told us : "China is using AI everywhere in photonics. Everywhere." I was expecting a lot of talk about LLMs, robotics, automation but no. The AI that was mentioned 99% of the time is actually dumb old Machine Learning.
And it kind of bugged me that they almost always used only the term AI when everything that was shown is clearly just ML, classification, identification, decision trees. I understand that technically, ML is located in the field of AI. I also understand that most of the audience doesn't know much about AI to make the difference, but please let's call a cat a cat. I guess it looks a little more "marketable" to have a laser machine with "AI", rather than the boring machine learning. The buyer can imagine some AGI type super-intellect at work in his super expensive machine when all it is doing is some basic classification using a scikit-learn algorithm in Python.
There was one speaker with a little more ease on scene talking about AI that suddenly spoke about Agents. And with it, the dream of a automated R&D lab. Why not. I am on the fence with this, mostly because I don't think an Agent can really do this kind of work, but hey, maybe I am simply fighting for my own job. It is still far fetched and the applications are still very niche, who knows where we will be in 2 years?
I did not hear more about the possibility of machines put into a network sharing data from one customer to another to generate a pool of knowledge. Basically the work of some small company buying one of those machine would benefit the big fish anonymously, with no regard to one's data ownership. Nothing surprising there. Actually, no, the surprising thing is that someone in TRUMPF thought this was a good idea.
Overall, to me, AI, and by that, I mean ML, is crucial to understand in today's world of precision machining. It is indeed the best for process control and process adjustments. AI in the form of Agents is still not very clear to me, but maybe I only fear for the loss of jobs that this generates.
Micro-structuring and the Awards
I was surprised to see so little talks about micro-structuring during the conference. I understand that 3 days go fast and there was a lot to talk about in other fields. To me, this is one of the main reason I went there. Even during the Fraunhofer ILT visit, there were only 2 or 3 actual booth targeting micro-structuring. Anyway, the winner of the awards this year is 4Jet, which is a company that does micro-structuring on the wings of planes, the famous shark like structures. So it is still an important topic. I was surprised to see them doing these structures with a CO2 laser and a DLIP method. I guess it goes faster, especially because they decided to structure the planes directly instead of going the way of using a plastic film structured like what is done nowadays. To be fair, this is strange to me, not only is it a hassle to get the plane to be structured by a laser, you have to build a whole crane/gantry specially around the wings and body of the plane, but compared to a simple Roll to roll process where the structures are transferred to some polymer coating, then glued on the plane, well... This is way more difficult. And I say that not just because I work on the subject, in the Roll to Roll manufacture, but because it is obvious the best way... Not to mention that the price will be bigger, and once the planes are repainted, or if they are hit by hail, the micro-structures are gone and needs to be redone. When you only need to change the sticker in the Roll to Roll process, the direct engraving is more tedious. So, yeah... kinda surprised at the winner really. The contender for the price was Cailabs with their phase plate (MPLC) beam shaper that they got to act as a passive mode cleaner for high energy free space coupling into hollow core photonic fibers of femtosecond laser (among other applications). It is pretty nice indeed. It is a neat trick. They deserved the win more than 4Jet in my opinion. While I am at it, the third company selected for the awards was Coherent about laser transfer of microleds. While the idea is also neat, I am sorry but what is Coherent doing in this kind of award? They are leaders in laser market. This is exactly what I was talking about earlier, with the idea from China's Startup Gym compared to Europe's vision of innovation. I understand that Coherent must find new source of money, but couldn't this idea have been done by a startup somewhere else? Can't Coherent be fine making laser systems? This is the reason why China is beating us and why we have very little innovation in Europe. The only people that we trust with innovation are the big players, the ones that already have money. If you are a lone PhD and you have the idea of making the same microled lift process? No. No money for you. You need 1M to invest personally (or from the bank, good luck with that), before being able to do some research, make a MVP, before some investors are ready to give you a little bit, before being able to compete in this kind of events. Nothing personal against the team at Coherent. Again, they are playing the game and the way it is design right now. This is the only way. I would like to make Qbits in Silicon Photonics Cristals using optically trapped Rydgberg atoms... Well, no can do. I am no part of a University. I don't have founds to build PIC chips. I don't have founds to book clean room access. I can go die with my idea in Europe while in China, maybe I would have had money invested in me because I have a PhD in Silicon Photonics and optical trapping. Maybe if I was working at ASML I could found this idea. But I am not. So yeah. I hate where we are going in Europe. What are we doing?
A small aside now. I wanted this blog to act as a showcase of what I can do in optics and photonics. Build some MVPs in different optical related fields. Typically in laser communication, in optical systems. Then I realized how much it would cost me, using regular components sourced from Thorlabs, so not even that expensive. Given the fact that I don't have access to a lab, or anything at all, being in the industry, I need to start from scratch. It took a while for me to get the basic electronics running to program small controllers, to get my hand in robotics and programming. Mostly because it is not too expensive. As soon as I touch optics, prices fly over the window. A dielectric mirror? 100€. An optomechanical mount, 2 axes, 1"? 150€. A laser diode? Ah! don't get me started. If I want one with power control, the diode itself is not too expensive. 50€. The driver on the other hand. 200, 400, 1000€ depending. A simple project like a laser communication between two terminals cost easily 5000€ depending on how fast I want to modulate the signal. It can go a lot higher if I want to modulate at MHz speed. Truth is, I am bound to simple projects because I don't have money. I don't have money because I can't showcase my projects, because money is tight everywhere here, because no one wants to take a risk. So I realize that a lot of my projects now are low cost software stuff, now low cost embedded software and easy robotics, because it is the only thing I can afford right now. There we are. This is what is happening and this is what is making me furious. Even more furious when I learn that google is investing 40bn$ in Anthropic for AI stuff every other day. It happened last week again.
