# MALAGA PANEL SESSION 3: NEW STRATEGIES FOR PHOTONICS TO PROFIT FROM SEMICON

https://www.youtube.com/watch?v=hkbgokIkGk0

[00:03] [music] I think opticist is
[00:05] differentiated. It's it's additive
[00:08] accreative. It's additional to the other
[00:10] societies.
[00:12] I I think [music] it's unique in the
[00:14] sense that it it's filling a niche which
[00:17] is creating this cross collaboration
[00:20] [music] and providing an opportunity and
[00:23] a forum for people to be able to be
[00:25] speaking [music] and getting
[00:26] enthusiastic not only among investors,
[00:29] innovators and business [music] people
[00:31] but also with regard to public and
[00:33] private investments. there's a big shift
[00:35] in that that's already starting and um I
[00:39] think there's an opportunity [music]
[00:41] not only for entrepreneurialism
[00:44] but to enthuse a next generation of
[00:47] talent and this is a great a [music]
[00:49] great u platform for that too. Optic is
[00:52] doing a great job.
[00:53] >> All eight keynotes from the global
[00:55] phutonics economic forum 2025
[00:59] are already published in full online.
[01:02] Now it's the turn of the seven panel
[01:04] sessions.
[01:06] Remember posts on LinkedIn only carry
[01:09] the first 15 minutes. The full session
[01:12] is on YouTube.
[01:18] Last year the panel that got that got
[01:20] the most positive feedback was the one
[01:23] on semiconductors.
[01:25] I had the chance the luck to moderate it
[01:27] and they asked me to do it again and I
[01:29] of course I couldn't be happier. I love
[01:31] doing that. We have some of the
[01:34] panelists of last year and new panelists
[01:36] of this year. There is a lot of things
[01:38] to talk about over the next over the
[01:39] next 45 50 minutes. We're going to talk
[01:43] about indium phosphate growth. I know
[01:45] that advanced materials growth. We're
[01:46] going to talk about supply chain
[01:47] localization and what we said this
[01:50] morning how difficult it is to build a
[01:51] fab. Going to talk about policy chips
[01:53] act in the US and in Europe. What is
[01:55] going on now and how we can benefit from
[01:57] that? gonna talk about the real growth
[02:00] drivers for the next three years. Is it
[02:02] only AI? We're also going to talk about
[02:05] the design for manufacturing and we also
[02:08] get to know our panelists and see what
[02:11] they can do for you and what you can do
[02:12] for them. I would like to invite now to
[02:15] the floor I am going to be so well
[02:17] surrounded now my panelist for the
[02:19] semicolon panel.
[02:26] It is not easy to introduce these
[02:29] figures from our industry. So I ask them
[02:31] to introduce themselves with a short
[02:34] round of introductions. And allow me to
[02:37] start with a personal hero, Chuck Mater.
[02:43] >> Hi, I'm Chuck Mater, the former chair
[02:46] and CEO of Coherent Corp, a market
[02:49] leader in indium phosphide technology
[02:51] and products. I'm currently the CEO of
[02:54] Avalanche Thinking, which earlier today
[02:57] announced its first investment in
[02:59] Hyperspectral, an exciting company
[03:01] that's bridging multimodal spectroscopy
[03:04] with advanced AI algorithms, enabling
[03:07] breakthroughs in medical diagnostics.
[03:10] >> Followed by Matt Estrinski.
[03:13] >> I am Matt Strashinski. Previous to this,
[03:15] I was co-founder and CEO at Inosce. Now
[03:18] what at AMD I'm responsible for fetonic
[03:20] development which includes co-ackaged
[03:22] optics as well as pathfinding.
[03:25] >> San I am Sanjay Pata Sarati. I'm the CMO
[03:29] of Coherent. Uh for those of you who
[03:31] don't know Coherent we started as 26 in
[03:34] 1971 and in 2019 we acquired Finar which
[03:38] is a datacom company and 2022 we
[03:41] acquired Coherent Inc. which was the
[03:43] world's largest we still are the world's
[03:45] largest laser manufacturer and then we
[03:47] changed our name uh a few months later
[03:50] and with the new branding of Coherent
[03:51] I've been with the company for a long
[03:53] time but as uh CMO since 2019
[03:57] >> Greg uh good afternoon everybody Greg
[04:00] Bartlett chief technology officer at
[04:01] Global Foundaries been there for a
[04:03] little over 16 years and of course we
[04:06] are a broad semiconductor manufacturer
[04:09] but uh for the purpose of today it's all
[04:11] about silicon photonics
[04:13] and Johan Huyak.
[04:16] Hi Johotonics.
[04:20] We are an Indian phosphite based uh
[04:23] foundry based in Antoven. We are making
[04:26] indium phosphite picks uh all the way
[04:29] from components to laser arrays to
[04:32] high-speed modulators but all the way to
[04:34] full integration monolithically.
[04:36] >> Needless to say these people have a lot
[04:38] to say to you but they also want to
[04:40] listen to you. If at any point you have
[04:42] a question, raise your hand and the
[04:44] technical directors of optic are here
[04:46] and they will come with a microphone to
[04:47] you. This is a panel that has to be as
[04:49] useful as possible to you. So please,
[04:51] please, please ask questions and raise
[04:53] your hand whenever you have something to
[04:55] contribute to this panel. So let me
[04:57] start. Let's cut to the chase. We had a
[04:59] lot of discussion for this panel and one
[05:01] thing that for me was a conclusion of
[05:03] last year by Greg was that photonix is
[05:07] becoming more like a weekday job and
[05:09] less like a Friday evening job. The
[05:11] volume is is happening and we can tell
[05:13] from what is going on in the room that
[05:15] the volume is happening. Uh could you
[05:17] tell us let's start with Greg a bit of
[05:19] an update what happened over the over
[05:20] the last year and there are reasons for
[05:22] us to believe that for the semiconductor
[05:24] industry photonix is now a core.
[05:28] >> Yeah. So for sure silicon photonix is
[05:30] mainstream. Um I think you won't see it
[05:33] in reported revenues from commercial
[05:35] companies until probably in significant
[05:38] volume until 2027. But to your point
[05:41] about uh is it a
[05:43] 5day a week? No, it's a 7 by 24. You
[05:46] know, as evidenced by and here's here's
[05:48] one of them. We announced a $575 million
[05:52] investment in establishing a silicon
[05:54] photonix packaging line. I think a year
[05:56] ago I was highlighting the fact that the
[05:58] packaging ecosystem for silicon
[06:00] photonics because of the blend of things
[06:03] related to optical components being
[06:05] coupled into silicon is a big gap in the
[06:08] industry right the traditional
[06:09] semiconductor OSATs uh don't have the
[06:11] capability uh folks that do know optics
[06:15] don't have the high volume manufacturing
[06:18] so um we decided that we would make a
[06:20] big bold decision and and committed to
[06:22] establishing our own manufacturing line
[06:24] so that for me is an indic indication
[06:26] that uh we are preparing for what is
[06:28] going to be significant
[06:29] commercialization in the next two years.
[06:32] >> This is big thing. So you are going to
[06:34] do offer packaging solutions which
[06:37] companies are we talking only about the
[06:39] ones that are looking for ultra large
[06:41] volumes or
[06:42] >> so what it really boiled down to is the
[06:44] traditional OSATs that the semiconductor
[06:47] has worked with um many of them had nent
[06:50] or uh uh non-existent photonic
[06:54] capabilities and that's associated with
[06:56] test methodology but you know the
[06:58] biggest challenge in industrializing
[07:01] photonics remains uh light in light out
[07:04] right fiber attach modules uh there and
[07:07] we recognized that in addition to
[07:09] working with the OSATs to bootstrap them
[07:12] we needed control of our own destiny
[07:13] because there is still a lot of customer
[07:15] diversity in the form factors and that
[07:17] we needed to have uh our own hands-on
[07:20] engagement beyond the OSATS of enabling
[07:23] our customers
[07:24] >> Johan we discussed after last year that
[07:27] it is not about silicon photonics it's
[07:29] not about Indian phosphide it's about
[07:30] the combination of those to go from chip
[07:33] to module. You are growing. You are
[07:36] increasing in wafer size, increasing in
[07:37] customers in volume. Give us an update
[07:39] what happened over the last over the
[07:40] last year.
[07:42] >> Uh yeah, I I think things are growing
[07:44] very rapidly. I think that from the
[07:46] perspective of indium phospite, I mean
[07:48] the the desire for higher speeds of
[07:51] course is driving the need for other
[07:54] materials than silicon as well.
[07:56] uh I think bringing the best of multiple
[07:58] worlds together I think is one of our
[08:00] biggest challenges of the of the
[08:01] industry right actually I mean from the
[08:03] the advanced packaging from the
[08:05] heterogeneous integration is we haven't
[08:08] got a single platform that does
[08:11] everything uh so you know and then we
[08:13] all focus on actually maturing our
[08:15] platform growing it bringing it to scale
[08:18] uh but in the end we need to bring them
[08:19] together to actually get the customer to
[08:21] be happy with the performance because
[08:22] the actually the customer doesn't
[08:24] necessarily care about what the material
[08:26] platform is. They care about the end
[08:29] performance. So we I think and you know
[08:32] we said this morning as well I think we
[08:33] need to really work together collaborate
[08:36] and bring these platforms in a fashion
[08:39] that in the end the customer benefits.
[08:42] >> Matt few days after the acquisition of
[08:45] semi by AMD I I reach out to you and I
[08:48] asked you to come here and thank you so
[08:50] much for accepting. It seems to me
[08:52] obvious now that HPC requires optics for
[08:55] the future and CPO is an intermediary
[08:58] step. What is the vision for for you now
[09:00] as in ENO semi part of AMD? Well, yeah,
[09:04] thank you for the opportunity to be
[09:05] here. So, right one one thing is
[09:08] absolutely clear. I think you mentioned
[09:10] the the physics of it the physics of uh
[09:13] of communication. I mean that we need
[09:16] optics for AI compute system. That's a
[09:19] fact and it it's maybe helpful to look
[09:21] back at the last decade of why it hasn't
[09:23] happened yet. I think co-packaged optics
[09:25] has been proclaimed as as coming soon
[09:27] for for quite a while across startups
[09:29] and uh standardization efforts and you
[09:32] know we're fighting against CMOS scaling
[09:34] and advanced packaging and materials and
[09:37] digital signal processing but now after
[09:39] you know leveraging all of these tricks
[09:41] and uh going through a few few more
[09:44] design cycles and learning cycles the
[09:45] the crossover point is is pretty clear.
[09:48] it's coming up and so the uncertainty of
[09:50] when when will this ramp happen is is
[09:52] coming down to quarters and or even
[09:54] months rather than years now and so I I
[09:57] I think you can count on it.
[10:00] Congratulations by the way once again
[10:02] for what you have achieved. The panel is
[10:04] open for questions and here comes the
[10:06] first one. One of the biggest dilemas so
[10:09] Indian fight is is now being scaled to 6
[10:12] in. That's our reality. Is this the
[10:16] inflection point we've been waiting for
[10:17] in the Indian forite community? Johan,
[10:21] >> whether it's the inflection point, it's
[10:23] certainly a inflection point. I mean,
[10:25] having looked at earlier presentations,
[10:26] I mean, this is going to be, you know,
[10:28] there's there's always the myth about
[10:30] Indian phosphite not being mature or not
[10:32] being there. You know, we've been out
[10:34] there for like years, right? I mean 30
[10:36] years, 40 years and we've been
[10:37] manufacturing these things also with
[10:39] reliability but actually driving the
[10:42] cost down by going to 6 in is going to
[10:45] be pushing it into new applications is
[10:47] going to be pushing it into areas which
[10:49] weren't available before which in the
[10:52] end again drives the volumes up and
[10:54] actually make us grow and mature as a
[10:57] full industry. So I think you know it's
[10:58] absolutely an inflection point. We are
[11:01] together with TNO in the Netherlands
[11:03] actually setting up one of these
[11:05] six-inch lines and we know that actually
[11:07] other people in the panel and other
[11:08] people in the world are doing the same.
[11:11] Uh you know we want to be at the
[11:12] forefront of bringing the cost down and
[11:15] actually making it more available to any
[11:17] applications.
[11:18] >> Sanjay it's two years more or less since
[11:21] the biggest announcement on Indian
[11:23] history for the 6 in coherent. What's
[11:25] going on? Yeah, we we announced our 6-in
[11:28] uh platform in March 2024. I mean,
[11:31] indium phosphide is a very versatile
[11:32] platform. Of course, all the excitement
[11:34] today is around AI data centers is
[11:36] great. We we we love that business, but
[11:38] indium indium phospite is a great sensor
[11:40] material. We use it in sensor
[11:41] applications. There are applications in
[11:43] in 6G there. It's a it's a pretty
[11:46] versatile platform and um I mean our our
[11:49] 6- in we announced uh publicly that we
[11:52] are in production in our from our 6-in
[11:54] platform. So it's um it's it's just it's
[11:57] all about the RAM Ram Ram Ram. That's
[11:59] that's what's going on.
[12:01] >> Jack, is this an inflection point?
[12:04] >> Well, first of all, let me thank you
[12:06] very much for inviting me. It's a
[12:08] pleasure to to be here with these
[12:10] panelists. Um I agree with Sanjay that
[12:13] um Coherence's announcement in March of
[12:16] 2024 may appear as a footnote to some. I
[12:19] believe it was a turning point for all
[12:21] of us.
[12:23] I'd like to share just a few thoughts to
[12:25] be additive to uh to what the rest of
[12:28] the experts on this panel can can talk
[12:31] about today. Let me reflect. The first
[12:34] room temperature indium phosphide laser
[12:37] was reported 50 years ago.
[12:41] For the last 40 years that I
[12:43] participated in this industry, it's been
[12:46] an empirical science that brought us to
[12:49] today.
[12:51] It's a vital agre a vital ingredient of
[12:53] the semiconductor industry's drive to1
[12:56] trillion dollars in sales by 2030.
[13:00] But since our combined technologies,
[13:03] photonics and electronics have ignited
[13:06] an avalanche in AI data centers, things
[13:10] will need to change even faster.
[13:15] I believe that to advance the next 50
[13:18] years
[13:19] and especially to mitigate what I
[13:22] believe is an unnecessary waste of
[13:25] electricity and water being projected to
[13:28] be required by 2030. The avalanche will
[13:31] extend to necessary advances in material
[13:35] science, adoption of state-of-the-art
[13:37] tools, and educating a 10,000 new
[13:42] engineering and scientist workforce who
[13:45] are bound to take it to another level.
[13:49] And I believe that these actions taken
[13:52] all together will still not be enough.
[13:56] I believe that it's time to adopt AI
[13:59] systems capable of closing the loop
[14:02] between materials, design, and processes
[14:06] that will lead to profound breakthroughs
[14:10] in half the time and with half the
[14:13] investment.
[14:15] I believe that these are worthy of
[14:17] public policy debate and action. And by
[14:21] the way, I'm not the only one apparently
[14:24] thinking this way. In fact, Jensen Wong
[14:27] recently said that if he were a
[14:30] 20-year-old today, he would focus on
[14:32] physical sciences being vital as AI
[14:36] integrates with the physical world.
[14:40] >> That's your vision.
[14:42] You know, you know what, Chuck? Last
[14:45] year after the event, five people told
[14:48] me the same thing. Talking to you for 10
[14:51] minutes was worth it coming to Malaga.
[14:53] So never again thank me for bringing you
[14:55] on a stage. You such a huge value to the
[14:57] industry.
[14:59] Greg, can I be a bit provocative? I I
[15:01] was a bit provocative last year and it
[15:03] was a big success. So allow me to do it
[15:04] again. So outside Taiwan, are we ready
[15:09] for localized end to end supply chains?
[15:14] How short do you want this answer to be?
[15:16] Because I
[15:17] >> as long as Chucky.
[15:18] >> Yeah. No, the answer is uh and I think
[15:21] it's important to state that you said
[15:23] outside of Taiwan. Uh Taiwan has done a
[15:25] remarkable job creating an ecosystem
[15:28] first and foremost around semiconductors
[15:30] itself and uh we have seen a lot of
[15:32] developments in the last 12 months of
[15:34] that extending into the photonic uh
[15:36] space. So with that as a reference
[15:38] point, you look around uh you know
[15:40] Europe and North America and you do not
[15:44] see the same uh capabilities there.
[15:47] First and foremost uh the traditional
[15:49] assembly capabilities uh are not
[15:52] non-existent in Europe and the US uh but
[15:55] they're small by comparison to Asia. You
[15:57] add to that the complexities around
[16:00] fiber attach or materials like lithium
[16:02] nabate or barerium titanate and you
[16:05] recognize that those ecosystems just
[16:07] don't exist. So as I said last year, you
[16:11] know, we're working to fill in the
[16:13] supply chain. Uh and you know, I I was
[16:15] highlighting a year ago the gap that we
[16:17] saw in advanced packaging. We voted with
[16:20] a half billion dollar investment
[16:21] ourselves to say that we're going to go
[16:23] fill in that portion of it ourselves.
[16:25] But it is a team sport. So we had an
[16:27] announcement last week at ECO with uh
[16:29] with Corning. Uh we work with a number
[16:31] of different partners is slowly building
[16:33] that ecosystem. Our customers are not
[16:36] going to have confidence until they see
[16:37] an entire supply chain that is not tied
[16:40] to Taiwan or China uh that exists for
[16:42] all the materials and the capabilities
[16:44] that are necessary. So still a work in
[16:46] progress.
[16:46] >> I love that you mentioned Corning. I I
[16:48] love that you continuously mentioned
[16:50] collaborations. In your opening speech,
[16:52] you said that you now offer packaging in
[16:55] the room. There are people from from
[16:57] fix. Can they raise their hand? The
[16:58] people from fix people from Alter
[17:01] Technologies also raising the hand from
[17:04] Alter Technologies. So I can see them at
[17:05] the back.
[17:07] How do you envision to collaborate with
[17:09] current packaging suppliers?
[17:13] >> So we we don't see it as necessarily an
[17:16] us versus them. We see it as an and
[17:18] knowing that they bring scale and cost
[17:22] effectiveness in an environment that has
[17:25] billions of dollars invested over the
[17:27] last decades there and we're literally
[17:30] transferring the technology knowhow
[17:31] we've developed internally over the last
[17:33] decade into those OSATs to establish
[17:36] that capability. That said, we also know
[17:38] that control of the supply chain and the
[17:40] technology elements requires continued
[17:43] industrialization. So we see it very
[17:45] much as complimentary on their you you
[17:47] look at the biggest guy in the industry
[17:49] that's doing co-ass today they are not
[17:51] trying to meet all of the customers
[17:53] demand themselves we think silicon
[17:55] photonics we're preparing for a world
[17:58] where their number of optical modules is
[18:00] measured in millions of units per month
[18:02] that is a team sport so we see that
[18:04] there's a perfectly good environment uh
[18:07] for equivalent is in semiconductor
[18:09] manufacturing there's IDMs and there's
[18:10] foundaries uh there's OSATs and
[18:12] foundaries and we see that coexistence
[18:14] here as well.
[18:15] >> Let's make this interactive. The people
[18:17] from fix who raised the hand. Can you
[18:19] raise it again? Helena is bringing the
[18:21] microphone.
[18:23] >> How do you feel about this? Now, Global
[18:24] Foundries is offering packaging. It's a
[18:27] great unity. It's a challenge, a room
[18:29] for collaboration.
[18:30] >> I think it's introduce yourself name.
[18:32] >> Oh, sorry. Albert Husper, CEO of Fix. So
[18:36] uh we have been in contact with uh
[18:38] global foundaries in uh so uh I I think
[18:42] it's an perfect proof point that the
[18:44] industry is getting mature
[18:47] >> that so I see it exactly the same as
[18:50] Greg. So the more the better. I did a
[18:54] little bit of a study on how difficult
[18:56] is to build a a successful packaging
[18:58] company and how much you have to charge
[19:00] to get return on investment and what
[19:03] you're doing at fix is is fascinating
[19:05] how successful you are. You deserve a
[19:06] huge credit for what you're building.
[19:08] Congratulations.
[19:09] >> Thank you.
[19:10] >> He's my friend as well Sanjay.
[19:12] [laughter]
[19:13] How is coherent ensuring resilience in
[19:16] this in this changing world described by
[19:18] by everyone? Diversification, vertical
[19:21] integration, shifting production. What
[19:23] is the strategy of coherent?
[19:24] >> Oh yeah, it's it's it's certainly the
[19:27] the need of the hour, right? I mean when
[19:29] we we engage with our largest customers,
[19:31] the hyperscalers, uh these are the
[19:34] companies that are building the the data
[19:35] networks of tomorrow and and all of
[19:37] that. The first question is do you have
[19:39] the enabling technologies? Right? We're
[19:41] not talking about assembly. We're
[19:42] talking about you know the the nextG
[19:44] laser the 400G laser the 200g vexel it's
[19:47] and then very quickly these days the
[19:50] last I would say for the past 2 3 years
[19:51] now the conversation shifts to okay
[19:54] where are you going to make it how is
[19:55] your supply chain where is your supply
[19:57] chain based out of so we've supplier
[20:00] resilience has been a very strong focus
[20:02] for us I mean it's dates back for many
[20:04] years and we believe there there maybe
[20:07] three main ingredients for that the
[20:09] first one is you need to own the
[20:10] enabling technologies if you're shopping
[20:12] for the key technologies. Okay, you're
[20:14] you're a little bit out of luck in in
[20:16] that realm. Second is you need to be
[20:18] vertically integrated because our
[20:19] customers, you know, now we are we are
[20:21] engaged with optical switching systems
[20:23] for example. These are these are, you
[20:25] know, carrier class switching level
[20:27] equipment, right? These aren't simple
[20:29] transceivers that that's got software,
[20:30] it's got and and so you have to be
[20:33] vertically integrated. The third one is
[20:35] the ability to to manufacture in diverse
[20:37] locations. I mean all of our datacon
[20:39] transceivers for example comes out of
[20:41] Malaysia. We have production in China
[20:43] but we do other things in China. So the
[20:45] ability to move to react to all the
[20:48] changes that are happening in the world
[20:50] not in real time but at least in a few
[20:51] months I think that's that's an
[20:53] important part of supplier resilience.
[20:55] >> But to what extent your customers know
[20:58] where your chips are made? Do they
[21:01] actually even have the right to know
[21:02] where the chips are made with?
[21:03] >> Oh our our customers I mean our our
[21:05] customers are super sophisticated. They
[21:07] want to know every I mean they would
[21:08] want to know where the sand for the
[21:09] silicon wafer comes from. I mean that's
[21:11] that's the level of detail that that
[21:13] they drive into. So
[21:15] >> it's a it's a complex world. Johan, you
[21:18] live in the Netherlands. I live in the
[21:19] Netherlands for many years. Wonderful
[21:21] country. Netherlands has blocked
[21:23] actively Chinese acquisitions. It's a
[21:27] difficult situation but smart has is a
[21:30] global company. You have customers all
[21:32] over the world. How do you ensure that
[21:34] your customers remain well served? They
[21:37] remain independent. They remain
[21:39] reliable. How do you ensure that
[21:41] situation?
[21:42] >> Yeah, I mean I would say I mean the
[21:45] blocking of Chinese actually we are a
[21:48] very good example of that because that's
[21:50] exactly what happened in 2020 when we
[21:52] had the opportunity to actually become
[21:56] at least partly Chinese I would say. Uh
[21:58] and in the end we we didn't. Uh and
[22:01] looking back that's been a pretty good
[22:03] decision.
[22:04] Um so you know that that actually is
[22:07] part of the start of the story that we
[22:10] don't want to be Chinese. We want to be
[22:12] an independent foundry that actually
[22:13] serves the whole world including the
[22:16] Chinese and customers by the way right
[22:18] because I made that point also very
[22:19] clear. I mean yes we do not want to be
[22:21] maybe in Chinese shareholdership but the
[22:24] world is our market and therefore the
[22:26] Chinese that includes the Chinese right
[22:30] um if we as a foundry are not actually
[22:34] independent are not actually working
[22:36] with the IP very careful with the IP
[22:39] very careful with the way we work and
[22:41] also our partnerships I would think
[22:43] we're out of business pretty soon so
[22:45] [clears throat] this is extremely
[22:46] important to our business to actually
[22:48] show that to our customers And you know,
[22:51] in the end, the best way to to build
[22:53] that trust is on a day-to-day basis,
[22:55] right? I mean, trust goes by horse and
[22:58] comes by foot. That's the way you have
[23:00] to do it. You build these relationships
[23:02] over time.
[23:04] We're about to talk about the chips act.
[23:07] Are there any questions in the room
[23:08] before we go there?
[23:11] The microphone is coming to you. Please
[23:13] remind us your name and company.
[23:19] Uh it's Litz Ashka from uh Photonix 21.
[23:22] Again, I I have a question which maybe
[23:25] turns the topic a little bit around.
[23:26] It's more like what can photonics do for
[23:28] the semicon industry? I I ran into
[23:31] several startups in the last few months
[23:33] building satellites and what they all
[23:36] complain about is radiation radiation
[23:38] hardness, single event testing and all
[23:40] that stuff in their electronic
[23:42] components for for their satellites. So
[23:45] I actually have kind of two connected
[23:48] questions. The first one is um can uh
[23:51] photonics help making these things for
[23:54] space more robust uh the electrical
[23:57] circuits. Second one is do you need
[23:59] different uh metrology for that because
[24:02] there there is a lack of testing
[24:04] facilities for these kind of things and
[24:06] would you all of a sudden need a neutron
[24:08] source in your semiconductor lab to test
[24:10] it or something like that? Would be
[24:12] interesting to hear that. The question
[24:13] is clearly reliability and current
[24:15] metrology solutions. I Sanji I I can uh
[24:19] I I can start uh we we do have products
[24:22] that that we set to the space-based uh
[24:25] deployment both for communication as
[24:27] well as for observation telescopes etc
[24:29] etc and you're absolutely right I mean
[24:31] this is uh there is no set standard set
[24:34] of criteria that you can go and test
[24:36] against but but our customers have been
[24:38] great partners they tell us exactly what
[24:41] type of because there is IP and knowhow
[24:43] in that and they they hold a lot of that
[24:46] quite close to themselves So but but
[24:47] they do share with us at the type of
[24:50] testing they would they would uh they
[24:51] would need but we do we do those kinds
[24:53] of testing. It's
[24:55] >> anybody else wants to answer on current
[24:57] reliability challenges.
[25:00] >> I have the feeling myself that we talk
[25:03] about reliability and we talk about
[25:04] harsh environments. We talk about
[25:06] automotive and in the CPO business we're
[25:09] going to be challenged by reliability
[25:11] challenges.
[25:13] Would you like to comment on that Greg?
[25:15] Would you agree disagree? Is Photonix
[25:17] still not reliable enough to survive a
[25:20] data center on CPO 247?
[25:24] >> Yeah. Um,
[25:26] first of all, the ramp hasn't started
[25:28] and you always find unusual
[25:31] circumstances when you scale, right?
[25:33] We're in the process of preparing for
[25:35] the ramp in 27, which means we have
[25:38] second, third, fourth of a kind tools
[25:40] being installed this year and next year.
[25:42] Tool matching subtleties, etc. So you
[25:44] always find those things and until you
[25:46] start ramping in millions of units on an
[25:48] annual basis, you don't get there. Now
[25:50] from an in there's there's two
[25:52] components to think about from
[25:53] reliability. Intrinsic reliability
[25:56] meaning do you have the fundamental
[25:58] physics underlying this that you know to
[26:00] be reliable that we're certain of. It's
[26:03] more the now I'm at volume. I'm got tool
[26:06] matching. I have multiple supplies of
[26:08] materials coming in. And are there
[26:10] exttrinsic factors that become
[26:12] reliability testing? And you don't know
[26:13] that until you start the ramp. So,
[26:15] anybody tells you that they've got it
[26:17] licked, they haven't been through a
[26:18] semiconductor ramp?
[26:19] >> Yeah.
[26:20] >> Marco, VGO ventures.
[26:22] >> Uh, yes. A question concerning the the
[26:25] approach to strategies that are being
[26:27] mentioned here. Also adhering to what
[26:28] what Michael mentioned in the um in the
[26:31] introduction before in the current
[26:34] situation uh that we're facing how do
[26:37] you approach new strategies or how do
[26:40] you feel that uh the businesses that
[26:42] you're doing could or should change to
[26:46] um to align with the new geopolitical
[26:48] realities and to try to build a more
[26:51] resilient and this better version uh or
[26:54] the more expected by us vision version
[26:56] of the future.
[26:59] Um well of course I mean you know as far
[27:02] as we can adapt uh being still not you
[27:05] know a large company uh I don't suddenly
[27:08] go and build a fab somewhere else uh you
[27:11] do actually look at what we uh I think
[27:13] that you know the [clears throat] way
[27:15] we've set smart up the way actually the
[27:17] photon delta ecosystem has been set up
[27:20] actually has been almost preempting the
[27:23] current situation right and of course we
[27:25] do adapt there but I mean you asked a
[27:26] question earlier about uh building a
[27:29] supply chain. Very often the supply
[27:30] chain is built around the foundry. Uh
[27:34] and and the good the good news is we
[27:36] actually have that that value chain
[27:37] already built, right? Or at least we
[27:40] have the the skunk works of it. And now
[27:42] we're actually growing it to maturity
[27:44] all all of it. Uh so being able to adapt
[27:47] to the geopolitics uh actually is is
[27:51] then becomes easy because we have it
[27:53] locally.
[27:55] uh do it on a daily basis. I mean, you
[27:57] know, honestly, when people talk about
[28:00] tariffs, 15% one day, 100% the other,
[28:05] 20% the day the next. I mean, honestly,
[28:07] I just listen and it goes that way
[28:10] because what do I do, right?
[28:14] >> Yeah. I I do want to comment um
[28:16] especially since the name of our company
[28:18] is literally Global Foundaries, right?
[28:20] uh when we started out in 2009 uh we uh
[28:24] we started acquiring companies with a
[28:26] geographically dispersed base. So today
[28:29] Singapore, Dresden, Germany and the
[28:31] United States um that was a liability a
[28:35] decade ago and it's looking to be a real
[28:37] asset for us today. Now the furtherance
[28:40] of that is we are not only establishing
[28:42] you can have this technology here. We're
[28:44] transferring technologies from Singapore
[28:47] to Dresden, from Dresden to the US to
[28:50] ensure that where we see geographic
[28:52] sensitivity to where it's produced that
[28:54] we are avail making the same
[28:57] technologies available in more than one
[28:59] geography and our customers get a
[29:00] tremendous amount of value out of that.
[29:04] >> It is chief sack time and we're
[29:06] [clears throat] talking plural because
[29:07] we had the US chips act and the European
[29:09] chips act. It's one year old. What's the
[29:12] real impact of the US chips act? What is
[29:16] your what is makes you happy? Is it
[29:18] working? And I look at Greg and Sanjay
[29:21] obviously since you're heavily involved.
[29:23] >> You want to go first?
[29:24] >> Okay.
[29:24] >> Please go ahead.
[29:25] >> So um for sure the $575 million
[29:29] commitment we made was as a consequence
[29:31] of a discussion with the CHIPS program
[29:32] office in the US. they recognized that
[29:35] this was a strategic gap for the US and
[29:37] the fact that GF was willing to stand up
[29:39] and say we're willing to put in the vast
[29:41] majority of it. Uh so from that
[29:43] perspective it's a key enabler. I do see
[29:46] some of the other investments as being
[29:48] uh strategic and then you know that was
[29:51] a $39 billion envelope uh that was made
[29:54] available and many companies have
[29:56] committed to establish a volume
[29:58] manufacturing. I think wait six years to
[30:01] see how many of those facilities have
[30:03] actually been built. We're also a
[30:05] beneficiary of the European Chips Act
[30:07] for expansion in the Dresden facilities
[30:09] as well. Uh I won't comment on the R&D
[30:12] side of things at this point which has
[30:13] been a much messier space in the last
[30:15] two years.
[30:16] >> Sanji, no I would just to add on what
[30:20] Greg said. Um I mean as a as an industry
[30:23] as a photonix industry and as an
[30:25] industry that focuses we look at a data
[30:27] networking piece. I mean for us it was
[30:29] all about networking but the last couple
[30:31] of years we're now part of the compute
[30:33] cluster.
[30:33] >> Yeah.
[30:34] >> Phutonix is not networking it's part of
[30:36] the compute cluster and soon you know we
[30:38] we are super excited by scale up. These
[30:41] are connections between the servers that
[30:44] are all electrical now. Those are those
[30:46] are starting to go optical. So photonix
[30:49] is is is really it's it's no longer a
[30:52] small part of the semiconductor
[30:54] industry. It's going to be you know
[30:55] right front and center [clears throat]
[30:56] very soon. So there is you know so any
[30:59] all the government efforts to respect of
[31:01] which government to drive it that's
[31:03] that's exciting and we've made several
[31:04] announcements uh talking about our
[31:07] interactions and engagements with the
[31:08] government from chips act in particular
[31:10] but photonix is included too so yes is
[31:13] great
[31:13] >> you in Europe the chief act one barely
[31:16] included photonics although we made a
[31:18] good deal with them on the 6 in new
[31:20] phosphite the chief act 2 maybe will
[31:23] have a different view for photonics but
[31:25] it's not until 2028 8 lots of things are
[31:28] going to happen until 2028. So what
[31:30] makes you excited? What worries you
[31:32] about the current situation of chiefs
[31:33] act in Europe?
[31:35] Yeah. So we will at least have twice
[31:37] Mulliga between now and then. Um so you
[31:42] know I'm looking jealous a little bit to
[31:44] my my neighbors on the right in terms of
[31:46] uh how you know US chips act is actually
[31:49] just bluntly supporting companies.
[31:53] Um and you know the the chips act has
[31:55] been a indeed a bit of a mixed bag. I
[31:57] mean we got finally photonix in which I
[32:00] think has also been recognized as
[32:02] actually literally by the European
[32:04] Commission saying you know with the
[32:05] photonic pilot line we have now
[32:07] completed our strategic picture for the
[32:10] future and therefore we're going to fund
[32:11] on the basis of this this strategic
[32:14] picture. So you know having the Pix
[32:17] Europe line there as a beach head for uh
[32:20] chipsack 2 is very very important
[32:22] because it's going to contain phutonics
[32:25] but we need to wait for 3 years and I
[32:27] think by that time we also better fix
[32:30] the problem of funding and state aid
[32:33] because uh I don't think state aid
[32:35] actually exists in Chinese probably
[32:39] uh and you know we talk about the level
[32:40] playing field and then we look at the US
[32:43] and China to sort play to our game and
[32:46] play to our rules and honestly that's
[32:48] never going to happen. We need to change
[32:50] our rules in order to make that that
[32:53] level playing field. We need to be able
[32:55] to willing to actually fund companies
[32:58] directly from things like the chip act.
[33:02] So we better get that right by 2028 and
[33:05] and in the meantime I am worried about
[33:07] the next two years of us not doing
[33:09] anything.
[33:10] >> Correct. We have some questions in the
[33:12] room. Please remind us your name and
[33:13] company.
[33:14] >> Hi, this is Ramos Nikolai Square Cloud.
[33:17] Um, just a followup to Greg's comment.
[33:20] Are we going to have 45 SPCLO in
[33:22] Dresston? [clears throat]
[33:24] [laughter]
[33:26] >> Sorry, I don't mean to put you on the
[33:27] spot or in Singapore.
[33:28] >> Dresdston would be more interesting.
[33:30] >> Well, you did. [laughter]
[33:32] Um, so actually we are evaluating our
[33:36] footprint in Singapore. We've actually
[33:38] had a 200 millimeter photonic footprint
[33:40] in Singapore there. And um uh we know
[33:43] that the probability of being able to
[33:45] address the China photonix market where
[33:47] there is a keen interest in the CLO
[33:49] platform is unlikely to be supplied out
[33:52] of upstate New York. So uh if if we were
[33:54] to take any step, it would probably be
[33:56] bringing in 300 millimeter into our
[33:58] Singapore facility.
[34:01] >> Point Cloud is a is a lighter company.
[34:03] So now we should talk about other growth
[34:05] drivers beyond pluggables and CPO. But
[34:07] before we do that, my colleague Tamia,
[34:09] there are many questions in the room.
[34:11] Please remind us your name and company.
[34:13] >> Yes. So Robert Harrison um from a lawyer
[34:16] as I mentioned. I really want to pick up
[34:18] on Yan's comment about state a because I
[34:20] think it's a really important thing and
[34:22] we've seen that the new administration
[34:23] in the United States has taken a direct
[34:26] equity stake in Intel in exchange for
[34:28] funds. Um, do we think that this is a
[34:31] good idea that government should
[34:33] actually start taking direct equity
[34:35] stakes in high-tech companies? Is that
[34:38] what we're proposing? Or is this going
[34:40] down a very slippery road towards uh
[34:43] state socialism?
[34:46] >> Best question so far at the global
[34:48] photonist economic forum, including both
[34:50] editions.
[34:52] >> Who's brave enough to answer that?
[34:56] Uh I think the slope is somewhat
[34:58] slippery. Um I mean there is you know
[35:02] there are actually programs in Europe
[35:04] where people I mean like the step
[35:06] program where there is actually equity
[35:08] being pushed into uh startup scaleup uh
[35:11] places which I think is
[35:15] uh solid enough not slippery uh to do it
[35:18] like the way it's done in the US I would
[35:20] say is quite a quite a step. I'm not
[35:22] sure we should be doing that kind of
[35:24] stuff but I think funding companies
[35:28] directly
[35:29] and uh you know doing it because I mean
[35:32] we of course we we should learn from
[35:34] what the US as an example is I mean the
[35:36] US just simply looks at how good a
[35:38] proposal is if the proposal is better
[35:40] than the other one it's perfectly
[35:42] defendable you fund this one and you
[35:44] don't fund that one and I think that's
[35:46] the right approach actually that's very
[35:48] fair and that actually has nothing to do
[35:50] with state aid Any
[35:54] further views?
[35:58] >> It was a really good question. I'm going
[36:00] to tell you my views in private over a
[36:02] beer tonight.
[36:05] Let's go to the growth drivers now. Uh
[36:07] Matt or in general everyone is it all
[36:10] data centers? Is it all pluggables and
[36:13] CPOS? Do you see parallel engines, new
[36:15] growth applications? Matt, go ahead.
[36:18] Yeah, I think for sure the nearest term,
[36:22] you know, order of magnitude increase in
[36:24] the fetonic space certainly is
[36:25] co-ackaged optics. It's pretty clear
[36:27] that more compute it leads to better AI
[36:31] models and uh phetonics is a pretty key
[36:34] enabling piece of technology to get
[36:36] there. And so um yes, 100% I think I
[36:39] think it's obvious that that's that's a
[36:41] near-term growth driver. But uh you know
[36:44] there's a whole host of interesting
[36:46] technologies that then get to leverage
[36:48] all of the learnings that that we'll go
[36:50] through to to to bring about that that
[36:53] order of magnitude increase. And so you
[36:55] know with better AI compute you you know
[36:57] more sensors means uh more data that
[37:00] that you could uh you know process
[37:01] whether for classification or for uh
[37:03] generative AI uh whether there you know
[37:07] uh and fetonics has had a huge role
[37:09] historically speaking for display and
[37:11] and sensing you look at the huge success
[37:13] of pixels and face ID and so on. So, you
[37:16] know, I think those those follow-on uh
[37:18] applications are are certainly coming,
[37:20] but uh definitely the the nearest term
[37:22] thing as far as I could see is CPO.
[37:25] Sanjay, we had a great OFC this year.
[37:28] Really great. So, is it looks like
[37:30] inside data center is Christmas, but
[37:32] it's an economic forum. Please don't
[37:34] tell me that photonics has many
[37:35] applications. Are there real
[37:37] economically valued growth markets
[37:40] beyond photonics for data centers today?
[37:42] >> Oh, completely. I mean um here you know
[37:45] three um matches talked about u um you
[37:49] know biometrics on the on the phone I
[37:51] mean that used to be the rage four or
[37:53] five years ago um right so I I think the
[37:57] what is really cool about photonix it's
[37:59] cool about photonix is that it's all the
[38:01] different market verticals that
[38:03] addresses um and and sensors great I
[38:06] mean uh sensors are the eyes and ears of
[38:08] AI right with no and and a lot of those
[38:11] sensors are for photonics whether you're
[38:13] sensing people in front of you or you're
[38:15] sensing what's going on inside your body
[38:17] through these devices like this. There
[38:19] is a there's a lot of excitement.
[38:20] There's a lot to come. Um I mean we're
[38:23] talking then if you look at energy
[38:25] applications, lasers for fusion, I mean
[38:28] computing the laser requirements for
[38:30] fusion applications
[38:32] >> is uh
[38:33] >> is is just um it's it's beyond belief. I
[38:36] mean when we do the calculations say
[38:38] really that many lasers? I mean there
[38:40] there's a lot going on. There's a lot to
[38:42] come. The best is yet to come.
[38:44] >> The best is yet to come. Greg, do you
[38:45] agree? Do we go to all the applications?
[38:49] >> Yeah, we need uh to be honest with you,
[38:51] we have so many adjacent engagements to
[38:55] uh to data center that we have to turn
[38:57] down, right? In particular, so a big
[39:00] part of our normal semiconductor
[39:02] business is working directly with
[39:04] universities. We have a university
[39:06] partnership program with more than 100
[39:08] universities that do prototyping. We
[39:10] always see that as an avenue for you
[39:13] know company or universities doing
[39:16] research on 6G telecommunications long
[39:18] before the carriers are are engaged in
[39:21] it. So uh we do see that adoption in our
[39:25] silicon photonix platform for a lot of
[39:27] applications as bio sensors uh you know
[39:29] LAR for sure is an example where we
[39:32] actually do see you know a few years
[39:34] beyond where we're headed with data
[39:35] center and I would highlight the fact
[39:37] that this is public knowledge that we're
[39:39] working with quantum on the quantum
[39:41] computing side of things as well. So
[39:43] yeah we do see plenty of other markets
[39:45] none of them at the scale that CPO is
[39:48] going to drive
[39:49] >> Johan. Yeah, very similar picture I
[39:53] would say. I mean I think that uh you
[39:55] know I would say data centers is not
[39:57] necessarily equal CPO there could be
[40:00] other other stuff happening there as
[40:02] well which is also large volume so it's
[40:04] not just uh synonymous
[40:06] um you know consumer electronics I mean
[40:08] what Sanjay was saying biometrics I
[40:10] think let me add one more there I mean
[40:12] spectroscopy to actually add uh that to
[40:16] your consumer electronics to be able to
[40:18] look around and and get and feed
[40:20] information you know we're we're we want
[40:23] all want to see data. We all want to see
[40:24] information. And I think actually
[40:26] spectroscopy is another interesting way
[40:27] of looking at the world. Getting that
[40:30] into consumer electronics, I think, is
[40:31] going to be a big thing as well. It'll
[40:33] take a while, but it'll get there.
[40:36] >> Any further questions in the room?
[40:39] We keep talking when we compare for the
[40:43] microphone is coming to you. Please
[40:44] remind us your name and company. By the
[40:46] way, I know who he is and his company.
[40:48] Really interesting one. Hi, this is
[40:50] Steven Tan from Rapid Photonics. I have
[40:53] a question for um for for
[40:55] [clears throat] Greg but also for for
[40:57] Matt. Gre in your last uh time we were
[40:59] here in Malaga, you said that one of the
[41:02] biggest challenges for photonics is
[41:03] reliability
[41:05] and uh we see now CPO really hitting off
[41:09] uh and their reliability is extremely
[41:11] important because it's very difficult to
[41:13] swap a CPO device. So has things changed
[41:17] all of a sudden in a year?
[41:21] So a couple things first and foremost
[41:23] from uh fundamentals of the physics of
[41:25] the optical devices the intrinsic
[41:28] reliability is good. I I haven't seen
[41:30] the talk yet but I understand that Meta
[41:32] actually presented a CPO versus
[41:34] pluggables uh uh uh lifetime analysis
[41:38] and reliability data at ECO last week uh
[41:41] that actually indicated CPO had superior
[41:44] system level reliability relative to uh
[41:47] uh to pluggables on there.
[41:48] Unfortunately, they didn't go into the
[41:50] parados.
[41:51] >> Interesting. Very interesting. Yeah.
[41:52] Okay.
[41:52] >> Yeah. So, look, CPO is still quite
[41:55] immature and the fiber coupling piece is
[41:57] probably the uh area of the most intense
[42:00] focus now on ensuring uh you know CPO
[42:03] requires detachability of the fiber
[42:05] connector on there. Uh how many times
[42:08] can you do that? Uh you know, under what
[42:10] circumstances, what level of automation
[42:12] is required. So, there's there's things
[42:13] to be sorted out still. So by no means
[42:15] am I declaring that uh uh that all
[42:18] reliability issues especially the
[42:20] exttrinsic ones from volume
[42:22] manufacturing are sorted out but uh
[42:24] intrinsic uh highly confident in the
[42:26] technology
[42:28] >> right and maybe also just to build off
[42:30] of that and I think there's an earlier
[42:32] question about space in prior lives I've
[42:34] looked at building picks for space
[42:35] applications I've looked at building uh
[42:37] siliconics for outside plant for inside
[42:40] plant for for CPO and I could tell you
[42:43] that uh you know we the the amount of
[42:45] engineering effort that goes into
[42:46] cooling and and the overall system
[42:49] design for for CPO it just makes it you
[42:53] know several orders of magnitude easier
[42:55] than say putting chips out and for
[42:57] outside plant for telecom and so uh yes
[43:00] reliability is a concern but uh it's
[43:03] being designed for and uh there's pretty
[43:05] clear pathways to achieve that that
[43:07] system level reliability and uptime that
[43:09] that's needed to make this successful.
[43:12] Thanks very much.
[43:14] >> Any further questions?
[43:16] We always talk about semiconductors
[43:19] being like 20 years ahead of photonics.
[43:21] We've seen the the clear ecosystems
[43:24] around OIP FD accelerators created
[43:27] designs ecosystems that that really are
[43:31] using
[43:33] tools that designers can quickly quickly
[43:35] go from a design for an idea to an
[43:37] design. Infotonics we have PDKs, we have
[43:40] IP libraries. Now we have also the
[43:42] integration. Um
[43:46] what are the the the unmet needs the
[43:49] necessary steps that we need to make so
[43:51] we are considered as soon as possible
[43:53] and then equal playfield maturity with
[43:55] electronics and allow me to go through
[43:58] all of you because you have a lot to say
[43:59] on that. I will start with Matt.
[44:02] >> Sure. So uh it's I think there's an
[44:05] earlier uh presentation about the how
[44:07] volume drives quality and I think that
[44:09] with the rise of silicon phatonics to
[44:11] where it is today that's certainly been
[44:13] true. I I know a decade or two ago uh
[44:16] you had to start from scratch and every
[44:19] uh facet of of of this design of PD
[44:22] standing up your own uh software
[44:24] ecosystem. And so teams of PhD
[44:26] scientists built writing a ton of code
[44:28] to rebuilding an entire Foundry PDK
[44:31] every time you wanted to go uh build a a
[44:34] new chip. But you know nowadays that's
[44:37] just not true anymore. It's it's night
[44:39] and day. And so uh you know the
[44:41] foundaries have learned a lot. The EDA
[44:42] ecosystem has learned a lot. And so, you
[44:45] know, there's a whole lot that you can
[44:47] now leverage out of these PDKs, out of
[44:49] all those learning cycles that now
[44:52] enable getting a design from idea to to,
[44:55] you know, to beta and a one or two tape
[44:57] outs at most.
[44:59] >> Greg, is there an IP market yet? Do we
[45:03] have a critical mass compared to
[45:05] electronics? Of course.
[45:07] >> Well, there's a lot of folks in this
[45:08] room that believe there's an IP market.
[45:10] And the one area that I would highlight
[45:12] and it's you know it's not current
[45:14] generation CPO solutions but as you move
[45:17] into advanced materials for modulators
[45:20] uh BTO lithium nabate indium phosphate
[45:22] etc. One of the unique things is who the
[45:26] who the supplier and who the customer
[45:28] is, right? Foundaries uh responsibility
[45:30] is to integrate the full solution end to
[45:32] end. But there's many companies that
[45:34] have access and capabilities for novel
[45:36] materials that wind up getting
[45:38] integrated into a pick and it's like
[45:40] well where do who's the customer and
[45:42] who's the supplier in that relationship.
[45:44] So I I I I fully expect that that is one
[45:48] of the things that'll mature in the
[45:50] industry in the coming years either
[45:53] through sorting out business alignments
[45:56] or through M&A activities that uh bring
[45:58] all of those capabilities into a uh into
[46:01] a single house. Uh I think one of the
[46:03] areas that is an unmet need is still in
[46:06] the optical test domain that there's
[46:08] still tremendous amount to be learned of
[46:11] a highly complicated CPO with EIC and
[46:14] PIC and in the form factor that goes
[46:16] into an optical module being able to
[46:18] handle the optical test component at the
[46:20] cost that is required to uh to achieve
[46:22] crossover with electrical.
[46:24] >> We have a question in the room. Please
[46:25] remind us your name and company.
[46:28] >> My name [clears throat] is Stafar. I am
[46:29] from Scandinics.
[46:31] So we are here at the global photonics
[46:34] economic forum. So my question is about
[46:36] e economy. Um,
[46:39] photonix uh never has been considered as
[46:42] a cheap technology and uh today we see
[46:46] that uh look into the uh quick uh ramp
[46:51] up in the uh
[46:54] in the data centers and the huge demand
[46:57] from the hyperscalers
[47:00] and especially from companies who are
[47:02] vertically integrated. So uh the uh
[47:05] volumes
[47:07] increased significantly
[47:09] and uh do you see here the chance that
[47:12] the cost uh will continuously uh drop
[47:15] down for the wafers for the manu for the
[47:19] manufacturing processes for the
[47:21] materials because um um if we will look
[47:24] at around of um uh data centers there's
[47:28] a lot of applications and still what we
[47:31] can say still those applications are
[47:34] hardly affordable from the cost price
[47:37] because it's not high volume. So what do
[47:39] what do you say about the cost and the
[47:41] further projection on the cost um uh
[47:44] decrease uh maybe some percentages in
[47:46] some couple of years? What do you say?
[47:50] >> Yes, go ahead.
[47:51] >> Yeah, I don't know which mic is working.
[47:52] Is the mic working?
[47:53] >> Both I think are working right now.
[47:55] >> I'll double myself. So I mean we've been
[47:58] in the dataccom business for many
[48:00] decades, right? And uh the one thing is
[48:02] for sure the cost per bit goes down. But
[48:05] when the cost per bit goes down let's
[48:06] say we go from 800G to 1.6
[48:09] the cost per bit goes down but the 1.6
[48:12] transceiver is a much is a more
[48:14] expensive transceiver and we make more
[48:16] money at the nextG technology. So the
[48:19] companies I think photonics as an
[48:20] industry we've been getting smarter and
[48:22] smarter. We we're making more money than
[48:24] we did in the past. So cuz I I think the
[48:27] value that we bring to the customers is
[48:29] is now starting to get appreciated a lot
[48:31] more by them which is a good thing.
[48:34] >> We have couple I mean you know just
[48:38] repeating what what we what talked about
[48:40] already you know there's a reason why we
[48:42] go to 6 in right and and it probably
[48:44] doesn't stop there. So uh you know going
[48:46] to larger V portfolios we can we have
[48:49] learned and that's why we are 20 30
[48:50] years behind but we can learn from the
[48:53] infrastructure that has been set up from
[48:54] the way it has been set up and from the
[48:56] way that the cost has been driven down
[48:58] there and do the same thing over here
[49:00] right we're not there yet but we'll get
[49:02] there
[49:04] company
[49:05] >> yeah so I'm Philip from Keystone
[49:06] Phetonics I have a question uh uh
[49:09] concerning the uh CPO situation and
[49:13] reliability I mean part of The story
[49:15] about increased reliability in CPO is
[49:17] that the active components have been
[49:19] removed.
[49:21] So, so, so, so, so coherence lasers and
[49:24] smartphanic components, they've been
[49:26] removed and they're now pluggable. So,
[49:28] it's a well lower cost component that
[49:31] that you can just have anywhere and you
[49:33] can easily exchange it. So, so, so
[49:35] what's the strategy to fight way back
[49:37] into this CPO?
[49:40] Sanjay
[49:41] >> so um so first of all CPO is not there's
[49:45] no oneizefits-all there are multiple
[49:47] platforms that CPO is being explored for
[49:50] different technology options in fact we
[49:52] some of us just came back from uh the
[49:54] European conference on optical
[49:56] communications where we spoke a lot on
[49:58] CPO in fact Jose was moderating a panel
[50:02] like this at at ECO and it was a CPO
[50:05] panel and I was on it as well but so it
[50:08] it's it's not one thing because there
[50:10] are vexel based approaches where the
[50:12] vexels are right there. It's not like an
[50:14] external laser bank like the
[50:16] traditionally done with silicon
[50:17] photonics. Um so it is it it there are
[50:21] considerations there is you know it's
[50:24] the the the complication or the beauty
[50:26] of CPO is it's not a planer device like
[50:28] a semiconductor device you got light
[50:30] coming in and out and you got coupling
[50:31] optics you got fibers that you need.
[50:34] There is a lot of stuff if if you look
[50:36] at the content in CPO everybody talks
[50:39] about silicon photonics and the external
[50:40] laser there's a whole bunch of other
[50:42] optics that is needed otherwise you
[50:44] can't make CPO work right all of those
[50:46] need to be scaled up those platforms
[50:48] need to be demon you know the
[50:49] reliability needs to be demonstrated so
[50:51] there's a lot of work to be done um and
[50:54] and that is why I think Matt Refford
[50:56] talked about this earlier is that it's
[50:58] always it's been you know CPO is coming
[51:00] it's coming yes it's come but there's
[51:02] still a lot of work to be
[51:04] This has been a great panel and time
[51:06] flies when you have fun. I would like to
[51:07] go to you to look at you Chuck. Is there
[51:10] any closing remark? Anything that you
[51:12] would like to add to everything that has
[51:13] been discussed and what is your future
[51:15] your vision for the next two years? What
[51:17] would you have? What would you teach us?
[51:19] >> Uh thank you Jose. I think if I look
[51:21] back the last 50 years to look forward
[51:23] the next 50 years, I think what served
[51:26] us best was to imagine, invest, invent,
[51:31] and to incubate a workforce including
[51:34] for the future and stay with problems
[51:36] longer. I think if we do, we'll find out
[51:39] five or 10 years from now that new
[51:42] strategies of semicon will have
[51:45] benefited from photonics. I would like
[51:48] to thank you and thank all the panelists
[51:50] for a great great session. Thank you
[51:51] very much. [applause]
[51:57] That's where we conclude coverage of
[51:59] this panel on YouTube. There are another
[52:02] six panel sessions in this series. And
[52:05] don't forget to take advantage of early
[52:08] bird deals and exhibition space for
[52:10] [music] your company during the third
[52:13] edition of the Global Phutonics Economic
[52:16] Forum. [music]
[52:17] It will be held in Malaga September 24th
[52:20] and 25th, 2026
[52:24] in the same venue as Eco
[52:34] [music]
[52:50] >> [music]
[53:06] >> Heat. Heat. [music]