# Optica Executive Forum 2026 Complete CEO Panel. Recorded March 16th 2026

https://www.youtube.com/watch?v=w3oqbYHaHVY

[00:00] Highlights from the Optica Executive
[00:02] Forum at OFC. Thanks to these sponsors.
[00:08] >> But the data center's ging deep inside.
[00:11] Copper running hot. Hear the signals
[00:13] getting f and cold package light is how
[00:15] we win.
[00:20] >> We're releasing seven sessions from
[00:22] Optica's executive forum because the
[00:25] world is watching. Feedback encouraged.
[00:29] Now enjoy.
[00:37] >> That was an impressive panel. We had
[00:38] just one more to go and many of you have
[00:41] been waiting for this the entire day. We
[00:44] have the traditional CEO panel with five
[00:48] of the most impactful companies in the
[00:50] history of photonix. and to moderate the
[00:53] CEO panel. Ladies and gentlemen, it is
[00:56] my honor to welcome to the stage,
[00:58] Michael Levy.
[01:03] >> Let's see, the microphone is working.
[01:04] Great. Let's hide this cup. I'd like to
[01:09] uh bring the uh the panelists to the
[01:12] stage. Uh Matt Murphy, chairman and CEO
[01:15] of Marvel.
[01:16] Uh Bill Gartner, senior vice president
[01:18] and GM of Cisco Optics.
[01:22] uh Jim Anderson, CEO of Coherent,
[01:25] uh David Herd, president of Nokia
[01:27] Infrastructure,
[01:29] and I don't know if Michael Hursten is
[01:32] here yet,
[01:34] but I've asked him as soon as he arrives
[01:36] to come straight onto stage.
[01:40] >> So, a couple of our panelists were uh
[01:43] we've heard Jensen's name like at least
[01:46] 30 times in the last hour or two hours
[01:48] today, right? So, so a couple of the
[01:50] panelists were actually at GTC this
[01:52] morning. Jim was one of them who's flown
[01:55] down to make the panel. So, thank you,
[01:57] Jim. Michael, unfortunately, is a little
[02:00] late. He decided to have lunch with
[02:02] Jensen. So, as soon as he arrives, we're
[02:04] going to ask him what what happened. So,
[02:06] he's not going to get off scot-free.
[02:09] Um, and like I normally do with the CEO
[02:13] panels, I always start off with a
[02:17] background question. I mean, who are
[02:19] you? Because if you think about yourself
[02:22] as the audience, these are folks who
[02:24] basically run our industry and you never
[02:27] get to meet them that much because
[02:29] they're pretty busy and they're flying
[02:30] everywhere. And so that will be the
[02:32] first question. But the first thing I
[02:34] want to say is that we call in this the
[02:37] CEO panel, but we have two folks that
[02:39] are not CEOs. So Bill's not a CEO,
[02:42] right? And and David used to be CEO of
[02:44] Infinera, but is now president. But as a
[02:47] committee, we really wanted to have the
[02:50] folks who influence and make big
[02:52] decisions in our industry who run big
[02:54] businesses. So we felt it was
[02:56] appropriate to have five folks that you
[02:59] know we look up to because their
[03:01] decisions are really important for all
[03:03] of us. And uh for that reason we've got
[03:06] a mixture this year. Um I'm going to
[03:09] start with a background question. at the
[03:11] uh the global phetonics event in Malaga
[03:13] last September. Um I did the CEO panel
[03:17] there and in the in the background
[03:19] question I said I collect uh old and
[03:22] antique spectacles even design my own
[03:23] spectacles but that's something uh that
[03:26] is personal to me and sometimes you see
[03:28] me in different pairs of specs but today
[03:30] for a personal question I also collect
[03:33] stamps you know that's those things you
[03:35] put on envelopes and you mail that you
[03:37] don't do that anymore so I collect
[03:39] British stamps and in doing collecting
[03:41] British stamps I buy most of them
[03:43] because I live in San Francisco on
[03:45] auction
[03:46] And when you go on auction, you know, I
[03:48] generally win about 95% of the bids I
[03:51] make because I try and calculate what
[03:54] the right value of the stamp is. And
[03:56] then if I'm betting against another
[03:59] collector where they go and the stamp
[04:02] price goes really up, then I sort of
[04:04] like I leave it alone. And so why is
[04:08] this and why am I bringing this up?
[04:09] Because I'm acquiring stamps just like
[04:12] these folks acquire companies. And
[04:14] sometimes they acquire companies at the
[04:16] right value and sometimes they're
[04:18] bidding against other other folks and it
[04:20] ratchets up. So maybe we'll explore that
[04:22] a little later. But let's let's start
[04:24] off with Matt. Um background. Who are
[04:27] you? Give us a story.
[04:28] >> Hey Michael, good to see you. And I
[04:30] think none of our stories are going to
[04:32] be as interesting as yours as a stamp
[04:34] collector, but uh but but but I'll try.
[04:37] Okay, I'll try. So, so, uh, yeah, Matt
[04:39] Murphy, uh, Marll, great to see
[04:41] everybody, and I appreciate being on the
[04:42] panel. Uh, first of all, the guy in the
[04:45] photo is me. It's my high school
[04:47] graduation photo. That was from 10 years
[04:50] ago. Uh, so I've been CEO for 10 years.
[04:52] It's been a it's been quite a run. Um,
[04:55] outside of work, I guess I, uh, I got
[04:57] involved a few years back in the, uh, US
[05:01] Olympic and Parolympic Committee. I'm a
[05:03] board trustee there. Uh, cool. Believe
[05:05] in the Olympic movement, the Olympic
[05:07] spirit.
[05:08] And so I spend some time on that. I'm on
[05:10] the uh this uh performance and
[05:12] innovation committee that's part of the
[05:13] USOC which is very cool to see all the
[05:15] technology that the athletes are using.
[05:18] You'd be surprised at how far behind the
[05:21] US is on this because uh little known
[05:24] fact, but the US Olympic athletes have
[05:27] no funding from the US government.
[05:28] Actually, zero. We're the only major
[05:31] developed country in the world that uh
[05:33] doesn't provide any funding for its
[05:35] athletes at all. So, it's all from
[05:36] private philanthropy and donations. So,
[05:39] I've been involved in that and uh I went
[05:41] to the Winter Olympics in Milan and
[05:42] Cortina and saw the gold medal hockey
[05:45] game which was pretty cool.
[05:46] >> Oh, excellent.
[05:47] >> So, that's what I've been up to. Yeah.
[05:48] Thanks.
[05:51] >> Excellent. Well, mine isn't going to be
[05:53] as exciting as stamps nor as
[05:55] philanthropic as uh as Special Olympics,
[05:58] but uh my name is David Herd. Uh I'm now
[06:01] uh the president of Noki Infrastructure.
[06:03] uh you know I started at Bell Labs in uh
[06:06] 1991
[06:08] and I uh was there for 10 years and then
[06:10] took a very uh different path along
[06:13] different networking technologies and uh
[06:16] was at Infinera in the optical business
[06:18] was at JDSU before uh Infinera by the
[06:21] way I had swore I would never go into
[06:23] optics because it was a tough business
[06:26] uh I will tell you you know personally
[06:28] uh something of interest you may not
[06:30] know I box and that's probably pretty
[06:33] appropriate for this industry because
[06:34] I've gotten the kicked out of me uh
[06:36] for at least the last 20 to 30 years. Um
[06:39] but it does teach you resilience uh how
[06:42] to take a punch in in many cases uh that
[06:45] you don't win without understanding uh
[06:48] what your strengths and weaknesses are.
[06:50] So uh super happy to be here. Super
[06:52] happy to see the uh crowd out here as
[06:55] fired up and optical being in the the
[06:57] main domain of what's required to allow
[07:00] both our WAN network to grow and inside
[07:03] the uh data center network uh to
[07:06] flourish. So good to be here.
[07:08] >> Great Bill.
[07:09] >> Thanks David. Resilience is certainly a
[07:11] good attribute for this industry. My
[07:13] name is Bill Gartner. I have
[07:14] responsibility for optical systems and
[07:16] optics at Cisco. Um I started my career
[07:20] at Bell Labs as well and ran the DWDM
[07:23] business at what became Lucent until
[07:26] 2000. I left Lucent to join a startup
[07:29] probably just after the peak of the uh
[07:32] the internet boom and um joined Furus
[07:36] which was a first generation RODM
[07:38] company. Um the the market collapsed
[07:41] after about four years and we shut the
[07:44] company down and sold what was left of
[07:46] it to another startup Mahi Networks. Uh
[07:50] after two years the Mahi board decided
[07:52] it was time to sell and we sold it again
[07:54] to Maritan Networks. And after two years
[07:57] that board decided it was time to sell
[07:59] and we sold it again to Extera. Nobody
[08:02] made any money on any of those
[08:03] transactions.
[08:05] And I left at that point and I made
[08:07] myself two promises. One, whatever I did
[08:10] next was going to be with a small
[08:11] company because despite the fact that we
[08:14] were on fumes for maybe half of that
[08:16] time, I really liked a small company
[08:18] environment. And number two, whatever I
[08:21] did next was not going to be in the
[08:22] optical industry because that was such a
[08:24] shitty industry. And six months later, I
[08:27] took a job with Cisco running the
[08:29] optical business and I'm still here.
[08:33] >> All right. Uh Jim Anderson, CEO of
[08:35] Coherent. It's been almost two years
[08:38] now. Although, as you guys said, in the
[08:40] optics industry, it's kind of like dog
[08:42] years. So, it feels like 20 years, 14
[08:44] years. Um but, uh let's see, outside of
[08:48] work, you know, in our jobs, we spend at
[08:51] least me, I spend most of the time
[08:52] indoors, most of the time sitting in one
[08:55] place, a conference room or something
[08:56] like this. So, on the weekends, I want
[08:59] to be doing the exact opposite of that.
[09:01] So outdoors, moving as fast as I can. So
[09:04] it's usually something that involves an
[09:06] engine of some kind. So motorcycles,
[09:09] off-road vehicles, uh race cars. Latest
[09:12] is um working on my pilot's license. So
[09:16] are there any pilots in the audience? Do
[09:18] we have anybody? I'm the I'm the only
[09:20] one. You needed one today. I'm not
[09:23] qualified yet. So there there are any
[09:25] pilots in the audience. So I'm I'm
[09:27] working on my uh license. The only part
[09:30] that I'm having a hard time with is the
[09:32] landings, which um apparently is a
[09:36] important that's an important part of
[09:38] flying. So, if there's any pilots out
[09:40] there, I could use some advice on the
[09:41] landings, but
[09:44] >> is this live?
[09:45] >> Yeah, welcome, Michael. It is.
[09:47] >> We're just doing a background question,
[09:48] as you could probably figure out.
[09:49] >> I'm I'm figuring it out. I apologize for
[09:52] being late. I was uh at GTC. I was
[09:54] called to the principal's office by uh a
[09:57] certain character and so I had to make a
[10:00] cameo appearance and uh as normal he
[10:03] likes to run on a little bit, right? You
[10:06] know, so it lasted a little bit longer
[10:07] than we thought and and then we were
[10:09] missing a pilot to fly us here. So now I
[10:12] know where to go, Jim. Right. I know
[10:14] where to go. Next time I'm going to hit
[10:16] a ride with Murphy. Murphy's got a much
[10:17] nicer plane. So uh we we missed all
[10:21] that. Anyway, I'm here. Uh, what do I
[10:24] like to do outside of work? Um, I just
[10:28] just like Jim, I uh I I can't stand
[10:31] being indoors, so I do everything I can
[10:33] to be outside, running, biking, whatever
[10:36] it might be. I'm not good in any of
[10:37] these things, but you know, I have a lot
[10:40] of fun trying. So, there we go.
[10:42] >> Great. Thank you. Well, that's that's
[10:43] very interesting backgrounds from all of
[10:45] you. So, the next the next question is
[10:48] what I'm termed the $5 billion question.
[10:51] So, $5 billion. Well, Nvidia has given
[10:55] well there's some type of deal that
[10:56] you've got $2 billion to Jim and $2
[10:59] billion to Michael and and David got a
[11:02] billion dollars in October last year.
[11:04] So, three of our panelists have been
[11:06] supported by Nvidia. Before I get into
[11:08] the ins and the outs of that, I think
[11:11] everybody here is trying to figure out
[11:14] what happened at GTC because GTC has
[11:16] been talked about all day. So, Jim,
[11:20] Michael, wherever you want to go first.
[11:22] So, I know you both were there. You both
[11:23] listened to Jensen. I mean, any nuggets?
[11:25] >> Yeah, there was just one phrase that
[11:28] stuck out for me. It was CPO. That That
[11:31] was it. That's the one thing that I
[11:33] heard out of all the
[11:35] >> Does that some character in Star Wars or
[11:37] Oh, that's CO.
[11:38] >> Has that been discussed at all today
[11:39] during the forum?
[11:41] >> It hasn't come up in any panel.
[11:42] >> Hasn't come up in any panel.
[11:44] >> No, I mean it was uh and Michael, you
[11:47] were there too. You should comment too,
[11:48] right? But um you know, look, Jensen's
[11:51] phenomenal, right? I mean, he's the
[11:53] guy's amazing and the company he's built
[11:55] is incredible. Um what's pretty amazing
[11:58] that he showed was just the scaling of
[12:02] the if you look at like tokens per watt,
[12:05] right, or tokens per dollar. And he was
[12:08] showing the road map of how they're
[12:10] going to progress in terms of scaling
[12:12] the total compute capacity in a tokens
[12:15] per dollar, tokens per watt um metric.
[12:18] And it's just it's incredible. It's 35x,
[12:21] 40x, 50x each generation. It that's what
[12:24] really struck me and that was that was
[12:26] really amazing what he's what he's got
[12:28] planned for, you know, and this is not
[12:31] not 10 years from now. This is over the
[12:33] next few years, right? So, it's pretty
[12:35] incredible.
[12:37] >> Yeah. I I I Kathy T, our investor
[12:40] relations person, was showing us a graph
[12:43] of our stock price and it was dropping
[12:47] dropping because there was lots of
[12:48] opportunity to mention CPO, but it
[12:51] wasn't mentioned until 1259. Okay? And
[12:54] then it we saw our stock go boom like
[12:57] that in the one minute, right, that it
[13:00] was discussed. Uh look, this is the
[13:03] industry's single biggest opportunity.
[13:05] There's no question. And you know uh I
[13:07] think between Jim and I we probably
[13:10] can't supply enough based on the demand
[13:11] that we see and Jensen was talking last
[13:14] year was all about scale out right
[13:17] connecting out to the switch bank. This
[13:19] year for the first time scale up was
[13:21] mentioned optical scale up. So optical
[13:24] scale up to the extent that you believe
[13:27] that's going to happen will drive just a
[13:30] tremendous amount of indium phosphide
[13:32] capacity. And I think that was the
[13:34] essence of the agreements that were done
[13:36] with both coherent and momentum was
[13:39] around trying to secure as much of that
[13:41] indium phospide capacity as possible.
[13:44] And we think you know scale up is
[13:46] obviously going to hit faster than
[13:47] people think. Uh it will drive a lot
[13:50] more units than people think and it's uh
[13:54] it really is it's a it's an industry
[13:56] changing event CPO.
[13:57] >> Thank you. Well, let's um let's go to
[13:59] the other three panelists. I mean, we've
[14:02] seen the investment and David, you can
[14:03] talk about the billion dollars in a
[14:05] second, but let's go to Bill. And one of
[14:07] the things that I'm trying to look at
[14:09] here, I mean, if we do a cross-section
[14:12] of the audience, there's probably a
[14:14] third of the audience there are senior
[14:15] execs and small companies doing really
[14:18] cool technology trying to get in, trying
[14:21] to grow the company. And so when I hear
[14:24] that, you know, $5 billion is being
[14:26] invested by one of the giants to secure
[14:30] supply, the first thing comes to my mind
[14:32] is the word allocation.
[14:34] So when you the last time we really got
[14:37] into trouble with allocation other than
[14:40] COVID was the last bubble back in 99200
[14:44] when there was the Nortell and the
[14:46] Lucent and the small companies were
[14:48] trying to get allocation of lasers and
[14:51] all sorts of components and things. So
[14:53] Bill, I mean is this something that from
[14:56] your perspective worries you as you go
[14:57] forward? Well, five billion dollars into
[15:00] these guys worries me a little bit,
[15:02] frankly. But, um, no, I think of course
[15:07] we worry about allocation. It's
[15:08] something that our supply chain team
[15:10] works on every day to make sure that
[15:13] we've got sufficient diversity in our
[15:15] supply chain and that we're securing
[15:18] supply in one way or another. We're not
[15:19] writing two billion dollar checks, but
[15:21] we do secure supply in other ways. Um
[15:24] all of these guys are our suppliers and
[15:27] in some cases our customers and um so
[15:30] yes we do diversify our supply chain.
[15:33] That doesn't mean that we don't have
[15:35] worries. We certainly have areas that
[15:37] are tight right now. I feel like we've
[15:40] got those under control um assuming we
[15:42] don't have a major major glitch in the
[15:45] network somewhere where somebody has a
[15:47] major supply chain problem. But um our
[15:50] supply chain team is basically working
[15:53] with these guys and others to make sure
[15:55] that we've got sufficient supply going
[15:57] forward. The other thing I would say is
[15:59] that our customers um are just
[16:03] notoriously bad at forecasting. And what
[16:07] was a very predictable
[16:10] forecast for Cisco with enterprise
[16:12] customers historically where one
[16:14] enterprise customer or one service
[16:16] provider customer would be off on
[16:18] forecast would generally be absorbed in
[16:21] any any uh buffer that we had allocated.
[16:25] When a hyperscaler is off, it's it's a
[16:28] major major challenge in the forecast.
[16:30] And so we have adopted a much more
[16:32] aggressive stance about material flow,
[16:36] about securing capacity, manufacturing
[16:39] capacity for ourselves and for our
[16:41] suppliers than we ever have in the past.
[16:44] >> Thank you, David. I mean, you've had an
[16:46] investment and is this good? I mean, do
[16:49] you have
[16:50] >> Well, money and investment is always
[16:51] good.
[16:52] >> All right. All right. You know, one of
[16:53] the one of the reasons Infinera had to
[16:55] come, you know, we came together with
[16:57] Nokia for that scale. And you know a
[17:00] couple years ago you know one of the
[17:02] most you know underknown things about
[17:05] both Infiner and now Nokia is we you
[17:07] know we actually do have you know two
[17:09] indium phospide fabs uh silicon photonix
[17:13] uh capability uh DSP capab we just are
[17:17] announcing four DSPs uh this week which
[17:20] if you think back a couple years ago
[17:23] when we would talk DSPs you'd talk about
[17:24] hey we're going to do a new DSP every
[17:26] couple of years. um the cycle has
[17:29] changed and I think the good news is the
[17:32] the people on stage here you know many
[17:34] of us compete but we're also partners
[17:36] and customers of one another and so
[17:39] getting super creative partnering that
[17:42] core fab capability whether it's you
[17:45] know indium phosphide substrates or
[17:48] whether it's you know packaging you know
[17:51] capabilities or materials we're having
[17:53] to go into kind of the secondary
[17:56] tertiary supply chain and make sure that
[17:58] we're provisioned. I think the good news
[18:00] that that we've seen over the last
[18:02] couple years though are just the sheer
[18:04] numbers. So we've been dealing a lot
[18:06] with you know DCI and scale across
[18:09] architectures. Now you know the numbers
[18:12] that we see in the uh in the scale up
[18:16] architectures and scale across
[18:19] they say all of us need every bit of
[18:22] investment we can and need to move at
[18:24] that same pace of research and
[18:26] development. not having but actually
[18:29] really moving the clock speed to kind of
[18:31] year year and a half splits to be able
[18:33] to keep the pace with technology.
[18:36] >> So I'm understanding from what you're
[18:37] saying that a billion dollars is
[18:39] scratching the surface. You want to see
[18:41] more investment? Yeah, we're putting $2
[18:42] billion a year in research and
[18:44] development across that same footprint
[18:47] along with all of the capital that we're
[18:49] talking about putting into these fabs.
[18:51] And it's still not enough when you look
[18:54] at the I think Andy had like 10
[18:56] different solutions to to scale out the
[18:59] architecture. you know the people at
[19:01] this you know across the stage here
[19:04] there's plenty of opportunity uh you
[19:07] know for us all uh which is a change
[19:10] from you know 10 years or 15 years of
[19:12] the past
[19:14] >> yeah maybe I'll add just uh agree with
[19:16] everything that was said but I think
[19:17] from a industry perspective standpoint
[19:19] we've we've we've been in a constrained
[19:22] environment really since the pandemic
[19:24] fundamentally across the supply chain
[19:26] and I I think so I think it's kind of an
[19:29] and meaning you've got these constraints
[19:32] out there that we've been going through.
[19:34] I think some of the investments you're
[19:35] seeing
[19:37] are are notable due to the magnitude,
[19:40] but they're not really surprising. So
[19:42] they're getting headlines and they're
[19:43] they're important and obviously the
[19:45] these guys are building a ton of
[19:46] capacity. But like I look back five
[19:49] years ago at Marll and that was the
[19:53] first 5 years I had been at the company
[19:54] we basically doubled. But we were able
[19:56] to do that on our own without any real
[19:58] major capex investments. So sitting
[20:00] there around 2021 after we closed Infi,
[20:03] we were about 4.3 billion in revenue.
[20:06] we're in the pandemic and we we had to
[20:08] put in then for the first time
[20:10] significant capex investments actually
[20:12] in our backend our supply chain. Uh we
[20:15] signed long-term substrate contracts out
[20:18] through 2030 did a whole bunch of things
[20:20] spent hundreds of millions of dollars uh
[20:22] you know over a couple year period and
[20:25] that's what's actually enabled us like
[20:27] we've guided we've guided the company
[20:28] now to do 11 billion this year five
[20:30] years later. So that would be, you know,
[20:32] over double over the past five years.
[20:35] But that's we were able to do that
[20:36] because of what we did 5 years ago.
[20:38] >> So when I look at what's happening, I
[20:40] think for Nvidia as an example, I mean
[20:43] they're doing what you need to do if
[20:45] you're running an infrastructure
[20:46] business. You actually have to think in
[20:48] not twoyear cycles, but like four, five,
[20:51] six years and even beyond. So I think I
[20:54] think it's actually a part of how the
[20:56] industry's evolved. But it is requiring
[20:57] everybody to put in more capital, more
[21:00] skin in the game, a lot more partnership
[21:02] oriented to go get it done. But um but I
[21:06] think that's that's just the world we
[21:07] live in and it's just at a different a
[21:09] different scale now. Um
[21:11] >> I I look at this and I see like it's the
[21:14] AI related optical bottleneck that
[21:18] Nvidia is is is helping out in terms of
[21:21] we've heard scale scale up scale out
[21:24] scale across but this is scaling in
[21:27] manufacturing capacity and I think what
[21:30] we're seeing here I mean I'm I'm getting
[21:31] some nods on the panel here I mean
[21:34] the whole industry needs to scale up in
[21:37] capacity and the last time I saw this
[21:39] was I go back to the bubble when Lucen
[21:41] and Nortell were were talking about the
[21:43] same story on the same panels and the
[21:45] executive forum at the same time. I
[21:47] think it was in Baltimore way back when
[21:49] and it sort of worked for a couple of
[21:51] years and then we sort of got ourselves
[21:53] ready and it sort of all inflated and
[21:56] the next question is the question I ask
[21:58] the panel every year. Are we in a GI
[22:01] bubble or not? And every year the
[22:03] answers change and so it relates to that
[22:07] question.
[22:09] I don't know if we're in a bubble, so
[22:10] I'm not really going to ask that
[22:11] question. But what I'm going to ask is,
[22:14] are we moving towards a cyclical
[22:17] um
[22:19] effect? Just like the semiconductor
[22:21] industry where you have rises and lows,
[22:24] you have magnitudes. I mean, what's our
[22:27] magnitudes going to be? Are we making
[22:28] too much product now that we're going to
[22:30] go into some type of dip at some time
[22:32] frame? Or is the potential for this
[22:35] extra capacity just going to mean we're
[22:37] not going to go into a cyclical action?
[22:39] Michael, why don't you Oh, do you want
[22:40] to try that one?
[22:41] >> Can I lead it off and then we'll hand it
[22:42] over to Michael?
[22:43] >> Are you stealing, Matt?
[22:44] >> Yeah. Are you stealing from me? I just
[22:46] want to be one step ahead of you.
[22:49] >> As long as I go ahead of Mike,
[22:50] >> you can go next. You go next.
[22:51] >> We'll let you go next. You can follow.
[22:53] Um, no,
[22:54] >> I'm always following you, Matt. Okay. I
[22:56] I think just the way I'd tee it up
[22:57] though, so I think this this bubble
[22:59] concern is out there. It's been out
[23:01] there um on and it happens on every high
[23:03] growth new market we've seen over the
[23:04] last 30 years. Um I I think that there's
[23:07] three factors though that make me feel
[23:10] better about it net. And let's just
[23:12] assume for a minute that the demand is
[23:13] there right now. We're at the beginning
[23:15] of Agentic. We're at the beginning of
[23:16] AGI. There's there's a lot of runway I
[23:18] think on on the applications and usage.
[23:21] And by the way, the state of the
[23:22] technology, I mean, if you just looked
[23:23] at the last couple of panels, there's a
[23:26] lot of work, right, to get this whole
[23:27] industry to even a even a you know, even
[23:30] a
[23:32] sort of a median level of where we need
[23:34] to be. So, but the three big constraints
[23:36] I think that are going to probably help,
[23:39] I think, meter this out a little bit.
[23:41] The first is there's a constraint on
[23:42] advanced node wafers. I think everyone
[23:44] who's making trying to buy CMOS wafers
[23:46] at 5 nanometer and below knows that
[23:49] constraint. It's a big one. TSMC is
[23:52] going to spend like $200 billion in
[23:54] capex
[23:55] and build 12 or 14 fabs and that's going
[23:58] to layer in over the next few years, but
[24:00] that's going to take them years. I think
[24:02] the second is memory, right? And there's
[24:04] a there's definitely a a big memory
[24:06] shortage and that's got profound impact
[24:08] across the industry, but that's not
[24:10] abundant. And then I think what's been
[24:12] relegated to third is sort of the grid
[24:14] and just the the ability to get power.
[24:16] But that was the big concern like a year
[24:18] ago. And I think it's actually flipped.
[24:20] So when I look at it, I I think there's
[24:23] some pretty macrolevel constraints in
[24:25] the system that net net probably help us
[24:28] from completely overdoing it. And okay.
[24:32] >> And so that's that's just a perspective.
[24:34] I I think we all have to be careful in
[24:36] our own businesses and and because it's
[24:38] moving so fast, you can still get upside
[24:40] down very quickly on a technology
[24:43] transition within the transition. But I
[24:45] just think those three issues are not
[24:47] getting solved anytime soon. and
[24:49] probably uh put a natural cap on how
[24:51] fast people could go, which is a little
[24:54] different than at least I you know I've
[24:56] been in the industry since 1994. So I
[24:58] saw the 95 semiconductor cycle and 98
[25:02] and 2000 all those
[25:04] >> those were not really that constrained
[25:06] per se in meaning you could you could
[25:08] build the product pretty quickly. This
[25:10] is going to take a while. That's that's
[25:12] my perspective.
[25:12] >> That's good observation, Michael.
[25:15] >> Yeah, I I mean I think your question was
[25:16] was spot on. Um the famous were last
[25:19] words. This time it's different. Okay,
[25:22] this time it's different. Um you know
[25:25] what we see obviously is just incredible
[25:29] demand out along horizons that allow us
[25:32] to invest in fabs which take a long time
[25:37] to bring up. It allows us to invest in
[25:40] manufacturing capacity which takes us a
[25:43] long time to bring up. But it seems like
[25:46] all the signs are there that this is
[25:49] going to be a good run. It's going to be
[25:51] a multi-year run. And I think if you
[25:53] think about optics now, inside that run,
[25:56] there are growth drivers that are sort
[25:58] of not dependent on a massive data
[26:01] center boom. It's just built on some of
[26:03] the things that Matt talked about.
[26:05] Increasing speeds, increasing bandwidth.
[26:08] Assuming that's there, you're going to
[26:10] have to have optics drive a big piece of
[26:14] the traffic, right? Which is a growth
[26:16] driver over and above whatever you might
[26:20] assume for AI. So, uh I think we're in
[26:23] the middle of a long and extended cycle
[26:26] and um you know all these things do have
[26:29] cycles and we have to worry about it.
[26:30] Jim has to worry about it. It is a
[26:33] concern and you know my my CFO Waj and I
[26:36] speak all every day about continuing to
[26:38] spend but we feel really good about
[26:41] where this is setting up and we're
[26:43] certainly willing to invest thinking
[26:45] that this is going to be a protracted
[26:47] cycle.
[26:47] >> Okay, Jim.
[26:48] >> Yeah, you know, I think um and I think
[26:50] you touched on this uh the great thing
[26:53] for the optics industry is we've got two
[26:55] big strong secular growth uh drivers.
[26:58] The one the obvious one is data center
[27:00] growth, right? a lot of companies
[27:01] benefiting from that, right? And
[27:03] there'll be peaks and valleys and that
[27:04] sort of thing. But the the other the
[27:07] second big secular growth which is
[27:08] unique to the optical industry is this
[27:12] sort of conversion of more of the data
[27:13] center connections from electrical to
[27:16] optical. Right? So this is the adoption
[27:18] of CPO integrated optics in the scaleup
[27:22] network. If you look at the number of
[27:24] connections in the scaleup network, it's
[27:27] an order of magnitude more than the rest
[27:29] of the network, right? And so as we
[27:31] start to see adoption of optical
[27:33] connections in the scaleup network, that
[27:36] growth driver is unique to this
[27:38] industry. And so we have two big secular
[27:40] growth engines underneath us, which I
[27:43] look, I just I'm seeing the same thing.
[27:45] I mean, demandwise,
[27:47] uh, the forecasts that we're seeing from
[27:49] customers, the long-term agreements that
[27:51] we're seeing from customers, I mean,
[27:53] they extend out um to the end of the
[27:55] decade. So they certainly have
[27:56] conviction in the demand.
[27:58] >> Wow. Bill.
[27:59] >> Yeah. I think we I would agree with
[28:01] what's been said and I think we see this
[28:03] uh this is not a select bubble about to
[28:06] burst. Um
[28:07] >> good
[28:08] >> the business models are very very
[28:10] different and much more fundamental. The
[28:12] funding for the buildouts is very
[28:15] different and I think much more
[28:16] fundamental. The the other thing I would
[28:18] say is that I don't know whether this
[28:20] this rate is sustainable for more than a
[28:23] few years but we are we are bullish that
[28:26] it will be in part because we have not
[28:28] yet seen the enterprise take of of AI
[28:32] for inference and we believe that that
[28:35] will be a huge huge expansion of the
[28:38] market for optics and GPUs and uh and
[28:41] general infrastructure. We've seen very
[28:43] very little of that to date in
[28:46] concentrated sectors like financial
[28:48] sector and healthcare. But we know in
[28:51] talking with our enterprise customers
[28:53] there's huge appetite for this as their
[28:55] applications are developed. We believe
[28:57] that that's sort of the next wave that
[28:59] we'll see beyond the the training models
[29:02] that are being developed right now
[29:03] primarily by hyperscalers.
[29:05] >> Okay. David, any thoughts?
[29:07] >> Yeah. No, I think these are all great
[29:08] points. I mean I think you know again I
[29:10] I find it funny we talk about optical
[29:13] not being cyclical but it's been one of
[29:15] the most cyclical industries over the
[29:17] last 25 years uh to see I think that
[29:21] what's what is different this time to me
[29:23] are three foundational things which are
[29:25] one power um there's no doubt that the
[29:29] compute you know uh veracity and
[29:31] appetite for compute and the advances
[29:34] that have made uh in comput I saw
[29:36] something where compute over the last 20
[29:37] years is advanced by 60,000 times and
[29:40] when you look at the connectivity piece
[29:42] it's 30 times. So we have a lot of
[29:44] catchup to go to catch up to compute and
[29:47] with the grid being outstripped there's
[29:49] no doubt it's outstripped and we play a
[29:51] foundational role in changing that when
[29:53] you're talking about driving power down
[29:55] by 50% or 70%. I once heard a customer
[29:58] in the last year say I don't care what
[30:00] the price is uh if you can take power
[30:03] down 70%. But then we got to PO stage
[30:05] and things changed. Um but to be clear,
[30:09] power is number one, math is number two,
[30:12] MedCap's law, the sheer number of
[30:14] connections is just if you start to do
[30:18] Medcap's law inside the data center and
[30:21] in between huge numbers and that brings
[30:24] up the third, which is you know as
[30:26] somebody who also supplies multi-rail
[30:29] line systems and huge optical systems
[30:31] around the world, don't forget that when
[30:34] inferencing hits the network, And you go
[30:37] from a 12 to one, you know, downstream
[30:40] to upstream that out on the edge of the
[30:43] network where you have radios and
[30:44] broadband access and where optics are
[30:47] driving deeper and deeper into the
[30:49] network that that inferencing has a
[30:53] profound and foundational impact on the
[30:55] architectures. And we all serve in a
[30:57] privileged seat to be able to mop up
[31:00] those three areas of both foundational
[31:02] demand and underlying need that the
[31:04] world has for what we do.
[31:07] >> Okay. Thank you. I the next question is
[31:09] going to be completely different and if
[31:12] we think about the audience today we're
[31:15] and the focus of this session is more
[31:18] towards the what's the impact of AI on
[31:20] our business but we're all hardware
[31:22] people for the most part in this room
[31:24] maybe even 99% including the folks on
[31:27] the panel so the next question and this
[31:30] is not a case of whether it's good or
[31:32] bad but I I just want to ask the
[31:34] question are you the the main companies
[31:37] that are doing AI hardware optics
[31:39] electronics related. I mean what is the
[31:43] have you thought about AI governance in
[31:45] your companies and you don't have to
[31:48] answer the question maybe you haven't
[31:50] but as as we go forward with AI and we
[31:53] start implementing it everywhere and we
[31:56] are sort of the key people that to help
[31:58] to implement it from the hardware
[31:59] standpoint what's your thoughts about
[32:01] this is your company been thinking about
[32:03] this and if not that's okay but I
[32:06] thought I'd ask a question Matt do you
[32:08] want to start off
[32:09] >> sure I may I have two takes on that.
[32:11] Mike, is it okay if I go first?
[32:12] >> I was going to steal. I was I was
[32:14] thinking about it.
[32:15] >> Steal. Yeah. Yeah. I I Yeah. I I think
[32:18] there's there's two angles to it. I
[32:19] think one is
[32:21] >> I think rest of the panelists can answer
[32:23] for them, but you know, there's there's
[32:25] sort of standard governance you're going
[32:26] to put in place from a security
[32:28] standpoint, data privacy. I think that's
[32:30] all
[32:31] >> table stakes. I think like you and I
[32:33] were chatting out there earlier about
[32:36] what's the role of the suppliers into
[32:40] the application layer that are
[32:41] ultimately going to have lots of bigger
[32:43] governance issues around safety and risk
[32:46] and Skynet and you know like is
[32:49] something bad going to happen and who's
[32:50] in charge of that and today it doesn't
[32:53] appear to me that there's any guard
[32:55] rails on any of this technology. What
[32:57] what about being able to be agile and to
[33:00] move quickly when you see, you know,
[33:02] there's weak areas or holes to fill? I
[33:04] mean, is that something you think about?
[33:06] >> Well, yeah, I think so. I think one is
[33:08] Yeah. So, I think there's there's a
[33:09] couple things. I think one is so that's
[33:11] a different question maybe a little bit
[33:13] in in being agile around how to make
[33:16] sure that you're maximizing the use of
[33:18] the technology while while making it
[33:20] safe. The other part of it is just
[33:23] in general in this environment we're in,
[33:26] I think the companies that are the most
[33:29] agile and open-minded and willing to
[33:31] change their mind
[33:32] >> when something changes quickly and have
[33:36] actually the confidence and fortitude
[33:38] and kind of leadership team that's
[33:40] willing to make those changes and not be
[33:43] embarrassed and say, "Well, I guess I
[33:44] made a mistake. We need to change our
[33:46] mind." You better be able to do that
[33:47] quickly in this environment or you're
[33:49] going to get absolutely crushed. And I
[33:51] think that in through this cycle is
[33:53] actually going to be a big definer of
[33:55] which companies are really successful
[33:58] ultimately is those that can adapt and I
[34:02] think reverse their positions or adapt
[34:04] to changing dynamics quickly versus
[34:07] getting hung up on their own ego that
[34:08] they already made a decision and they
[34:10] can't take it back or you know the
[34:13] development's down the road and all the
[34:14] engineers are working on it and so how
[34:16] are we going to tell them that it it all
[34:17] changed? These are things you never
[34:19] really hear people talk about, but
[34:21] actually it's what makes the difference
[34:22] over the long term in terms of your
[34:25] ability to uh to execute and really
[34:27] deploy your R&D effectively. So, that's
[34:29] another thing at least I I think about a
[34:32] lot and and and not trying to get too
[34:34] cute on we've got some vision for the
[34:36] world and that's how it's going to be.
[34:38] Guys, it's it's going to change next
[34:39] week and I think if you decide you're
[34:41] going to be hung up on your old
[34:42] strategy, you're going to get killed.
[34:44] So, I gave you a couple different things
[34:45] to think about there from governance
[34:47] down to agility, but that's that's how
[34:49] that's how we're looking at the world.
[34:50] But, um,
[34:51] >> thank you.
[34:51] >> Let the other panelists answer.
[34:53] >> Michael, you got to go next.
[34:54] >> Okay. I'm always as usual behind Murf.
[34:58] Right. There you go. Uh, you know, a
[35:00] couple of things. My chair and I uh were
[35:03] asked to come to Rome actually and and
[35:06] meet with the Pope and talk. I'm not
[35:08] kidding. Right. On
[35:10] >> telling Pope jokes now.
[35:11] >> Yeah.
[35:12] I'll let you go first on that one. I'll
[35:15] let you go first. Okay. All right. Uh
[35:18] Murphy's a good Irish Catholic name,
[35:19] isn't it? Okay. I got to be careful now.
[35:21] All right.
[35:21] >> Let's keep focused. Let's keep focused.
[35:23] >> Okay. So, uh you know, I think the guard
[35:27] rails on on AI are something that that
[35:29] needs to be discussed. I I don't think
[35:31] that they're there. And uh as an
[35:34] industry, and Matt spoke to it really
[35:36] well, I think it's up to us to sort of
[35:38] figure out how we get some guard rails
[35:40] on it. So, it is an issue. Um, and I
[35:43] think it's it's one that we don't talk
[35:46] about enough. Uh, I think it's one that
[35:49] deserves a lot more attention than it's
[35:51] getting.
[35:52] >> All right, that that's fine. It's not
[35:53] expected to be a long question, but
[35:55] let's let's go with Bill. You look like
[35:56] you're ready.
[35:57] >> So, um, Cisco has put a lot of effort
[36:00] into defining a a framework for AI and
[36:04] uh has established six principles. Those
[36:07] principles are transparency, fairness,
[36:09] accountability,
[36:11] um, security, privacy, and reliability.
[36:14] And everything we do from from designing
[36:17] products to selecting tools for AI to
[36:20] deploying products that that are land in
[36:23] our customers networks, including
[36:25] security products that are guarding
[36:27] their networks. Um we're thinking about
[36:29] the impact of AI and this includes
[36:32] things like you know tool selection what
[36:35] are the development processes that we
[36:37] are we're going to adopt and how do we
[36:39] safeguard those how do we safeguard our
[36:42] information and how do we safeguard our
[36:44] customers information when we're using
[36:46] for instance customers network
[36:48] information to help train an AI model.
[36:51] Um, so this gets a lot of attention
[36:53] within within Cisco. The fact that we're
[36:55] deploying things like security
[36:58] infrastructure for our customers makes
[37:00] it really, you know, mission critical
[37:02] for us to make sure we've got a good and
[37:04] uh, and solid framework around how AI is
[37:07] going to be used.
[37:08] >> Thank you, Jim. Any thoughts?
[37:10] >> The only I think these guys covered it
[37:12] well, but the only other thing I'd add
[37:14] is, you know, the other thing that comes
[37:15] up about AI is is job job losses, right?
[37:19] is people being replaced by AI,
[37:22] you know, in the, you know, speaking at
[37:24] least from coherence perspective, I I
[37:28] don't see it that way. I uh the way
[37:30] we're using AI is just massive
[37:34] productivity and time to market
[37:36] improvements. We're looking at it as a
[37:37] massive human accelerator, right? So,
[37:40] I'm not looking at, you know, how is AI
[37:42] going to help me cut some heads, right?
[37:44] I we're looking at how is AI going to
[37:47] help us accelerate our time to market by
[37:50] six months, by a year. How's it going to
[37:52] help us double the number of products we
[37:54] can get to market? That is way more
[37:56] valuable to us than cutting a few heads,
[37:59] right? So, that's the other concern I
[38:02] hear come up with AI. That's definitely
[38:03] not the way we're thinking about it.
[38:05] It's all about acceleration,
[38:07] better productivity, a augmenting the
[38:10] human and getting to market faster.
[38:11] >> Thank you. I mean, David's not in. I
[38:13] mean I I guess you agree.
[38:14] >> Yeah. No, I I think when you when we
[38:16] talked about how you scale, we we didn't
[38:18] get into, you know, angstrom level yield
[38:22] improvement from station to station and
[38:25] how you can use a manufacturing data
[38:27] lake to be able to improve that and
[38:29] again get the through throughput that
[38:31] we're all trying to you know catch up to
[38:33] in the industry. I think that to me the
[38:35] the really really difficult balance is
[38:38] in the wide area network keeping trust
[38:40] with your clients making sure you you
[38:43] know we also have at Bell Labs you know
[38:45] kind of the governance board to be able
[38:47] to make sure that that happens and make
[38:49] sure all use cases are passed through
[38:51] that you're collaborating very closely
[38:54] with you know thousands of customers and
[38:56] making sure that active learning is
[38:58] always coming on guess how you do that
[39:00] you got to use AI tools so while you
[39:02] talk governance and and trust and and
[39:06] making sure that that resiliency is
[39:08] there. Getting people that have been in
[39:10] the industry for 10, 20, 30, 35 years to
[39:15] use new tools and not be afraid of those
[39:18] tools and push them to the limits cuz in
[39:21] order to deliver what's in front of us
[39:23] for the next 5 to 10 years, we have to
[39:27] leverage those tools.
[39:28] >> Okay. So that I'm going to I'm looking
[39:30] at the time here. I've got one more
[39:32] question for all of you and then we'll
[39:34] we'll have a few questions from the
[39:35] audience. Um, new technologies. So,
[39:39] we've seen AI have a huge impact on our
[39:42] industry. We're getting big investments
[39:44] by big companies. We've got to increase
[39:46] our manufacturing scale. But at some
[39:49] point, we're going to have to bring in
[39:50] some new technologies.
[39:52] And every time we see a data center type
[39:55] talk or presentation, they always talk
[39:58] about reliability. always comes up in
[40:00] every presentation. And so the question
[40:02] in my mind is is how do we bring in new
[40:05] technologies down the road, maybe even
[40:08] it's tomorrow, and get them qualified in
[40:11] half the time that we're used to, maybe
[40:13] even a quarter of the time we're used
[40:14] to. Because if you look at those 10-year
[40:17] road maps folks are talking about today,
[40:19] I mean, how do we fulfill them? Now, we
[40:22] could have our own incumbent technology
[40:25] be okay for the next 10 years, but the
[40:27] chances are we're going to have to bring
[40:28] in some new technologies to to help fill
[40:30] it out. I mean, one way of doing it is
[40:33] acquisition and stuff or things like
[40:35] that. But whatever it is, whether it's a
[40:38] bolt-on or you do it internally, you got
[40:41] to get products out fast and you've got
[40:43] to solve the reliability problem. Matt,
[40:45] do you have any thoughts on that one?
[40:47] Yeah, I think um I I think this is where
[40:50] scale really matters because ultimately
[40:53] you have to have enough R&D spend and
[40:56] opex and and sort of wherewithal to do a
[41:00] lot of things in parallel you're not
[41:02] used to because ultimately what happens
[41:04] is you have some some technology we're
[41:06] all pushing and then there's always the
[41:08] next thing and the next thing's always
[41:10] got a smaller team that's not quite
[41:12] funded and then the existing business
[41:15] just sort of takes over and so
[41:18] and then M&A does help but then of
[41:20] course you got to integrate that
[41:21] acquisition you got to bring the people
[41:22] in so I think this is where maybe an
[41:25] unhealthy byproduct of this you know for
[41:28] at least the broader industry is like
[41:30] you could see in this a scenario where
[41:32] if you really have this accelerated
[41:33] demanding pace the big guys just kind of
[41:36] keep getting bigger and maybe you're
[41:38] seeing that with Nvidia to some extent I
[41:39] mean they're able to just put out so
[41:42] much technology ahead of everybody but
[41:44] they have just these enor enormous teams
[41:46] and parallel teams. So I think that's
[41:49] one dynamic that companies need to look
[41:51] at. Um, you know, we just in our own
[41:53] case, we we we um I mean, we did some
[41:55] acquisitions just in December, but for
[41:57] the prior five years, actually since we
[41:59] acquired Infi and Anovium, like we just
[42:02] doubled down on our organic investment.
[42:04] And so we had to fire up the CTO office.
[42:06] We had to fire up advanced teams. We had
[42:08] to invent things ourselves in parallel
[42:11] and then drive test chips and PC's and
[42:13] get them qualified so they're ready. Got
[42:16] it. But it just doesn't happen if you're
[42:18] just trying to make it as the nextG part
[42:21] because to your point
[42:23] it it's just never going to be ready in
[42:25] time. And so that there in there in lies
[42:27] the rub. But I think companies need to
[42:29] think about how they and AI can help
[42:31] like people are talking about more
[42:32] efficiency and getting things out
[42:33] faster. But those are some of the
[42:34] dynamics otherwise you just never never
[42:37] get there.
[42:38] >> Okay, we'll go we'll go to Michael again
[42:40] because that's the order. But you've had
[42:42] an investment in phosphite. You're
[42:44] scaling the manufacturing capacity just
[42:46] like Jim. Fine. But what about other
[42:48] technologies? There's other technologies
[42:50] we're going to need.
[42:51] >> Look, I mean, I I think and you know,
[42:53] David and Bill have been around this
[42:55] industry a lot longer than than
[42:57] certainly I have. What we're finding,
[42:59] Michael, more often than not, is
[43:02] technology
[43:03] like the semiconductor business. It's
[43:05] more revolutionary than evolutionary.
[43:08] Meaning some of the products that we are
[43:11] building today that have the most
[43:13] excitement and most traction are coming
[43:14] from our old telecom investments. Matt
[43:17] was talking about it. You got to invest
[43:19] 5 years ago to be where you are today.
[43:22] So OCS gets a lot of attention. The
[43:26] basis of an OCS is a telecom WSS. Same
[43:30] thing that's been used for 10 years. The
[43:32] basis for CPO, co-ackaged optics, is a
[43:36] pump laser that's been used for 15 or 20
[43:39] years.
[43:40] >> What was CPO again?
[43:41] >> The co Oh, that co-acked Optus thing.
[43:43] Yeah, that thing. That thing. So, what
[43:46] what we find more often than not, and
[43:48] this is where you have to stay on that
[43:49] treadmill. You don't try to reinvent the
[43:51] wheel. You try to just iterate and
[43:54] iterate along the curve and axis that we
[43:56] see our customers and the industry is
[43:57] going. And that seems to be proving I
[44:00] think the most successful.
[44:01] >> And Jim, are you do you have similar
[44:03] thoughts on those lines?
[44:04] >> Yeah, I the way we look at it is uh look
[44:07] we we see ourselves as a photonix
[44:09] company, right? Um everything that we do
[44:13] is about generating a photon,
[44:15] transmitting a photon, detecting,
[44:16] amplifying, it's all around photons,
[44:18] right? So, if you break that down into
[44:21] its elements, what we're trying to do is
[44:23] make sure that we're ahead of the rest
[44:25] of the industry on every element of
[44:27] that, right? Um, and if we're ahead of
[44:30] the industry, we got we got to make sure
[44:31] we stay ahead and if we're not, then we
[44:34] got to make sure we get in the lead. And
[44:36] that's a combination of organic
[44:38] investment and then if necessary,
[44:39] inorganic. And so, we're constantly
[44:42] looking at the landscape. Um we're
[44:44] engaged with lots of different startups
[44:46] to look at, hey, if we need to add any
[44:49] key techn any key photonic technologies
[44:52] to the portfolio, we'll absolutely do
[44:54] that, right? Um so that's kind of the
[44:56] way we think about it.
[44:57] >> Yeah. In terms of Bill and Cisco, I mean
[45:00] since the Oacia acquisition, I mean,
[45:03] what are your thoughts about next
[45:05] generation optics technologies as you?
[45:07] >> Well, I would say in general, I think
[45:09] big companies uh fall prey to innovators
[45:12] dilemma. It's just very natural. We have
[45:14] large customer bases. We have legacy
[45:17] technology that has to be supported for
[45:20] many years. And so it's very easy just
[45:21] to say let's do the next incremental
[45:24] thing. I think acacia is a great example
[45:27] where we made a conscious decision we
[45:28] were going to cannibalize a portion of
[45:31] the portfolio and say we're going to
[45:33] disrupt with a pluggable coherent optic
[45:35] in that case. And that is is as much a
[45:39] cultural change as it is a technical
[45:41] issue. The technical issues are actually
[45:43] I think solvable
[45:46] in rapid time. The cultural issues that
[45:48] that big companies have to overcome and
[45:51] our customers as well have to overcome
[45:53] as we look at very innovative disruptive
[45:56] technologies is really the thing that we
[45:58] have to attack when we're looking at how
[46:00] can we accelerate bringing in new
[46:02] technology to our customers.
[46:04] >> Okay. And David
[46:06] >> as I said I started with Bell Labs and
[46:08] now I'm back with Bell Labs. So you know
[46:10] there is no lack of PhDs and brilliance
[46:13] um investing ahead in key enabling
[46:15] technologies that might not be relevant
[46:17] now in terms of packaging heterogeneous
[46:20] material science um you've got to be
[46:24] diligent and continue to do that and
[46:26] you're right with creative destruction
[46:28] in in mind but I think the number one
[46:31] thing for me is the way to innovate has
[46:33] changed it used to be when I when I
[46:35] joined Bell Labs you'd go to a customer
[46:38] and they you'd take binders, like
[46:40] literally stacks and stacks of binders
[46:42] back of specifications, and then you'd
[46:44] fill them out and comply, not comply.
[46:46] Maybe we comply later. Um, and guess
[46:49] what? That took time.
[46:52] >> Now, when you're sitting with
[46:53] hyperscalers and you're sitting with
[46:54] your client base and even some of the
[46:56] partners here, we sit in rooms. We're on
[46:59] whiteboards doing things in real time,
[47:01] leveraging AI, putting together test
[47:03] picks, you know, layering that out,
[47:06] running algorithms to see what will
[47:08] work. That's kind of the new pace of
[47:11] innovation that's required, but you need
[47:13] all the elements uh to come together.
[47:15] And you're right, it is a cultural
[47:17] shift.
[47:18] >> Oh, that's that's a good answer. Any uh
[47:21] questions from the audience? Yeah, it
[47:23] wasn't that long ago when uh there were
[47:26] there were four hyperscalers that
[47:27] dominated the dataccom market and then
[47:30] became one Nvidia dominating the
[47:32] dataccom market and not a pretty
[47:35] challenging situation and now we have
[47:38] three three of you on the panel who are
[47:41] received funding and with the promise of
[47:44] one or more of you getting additional
[47:45] funding. I'm just curious is does this
[47:48] have a problem with too much conformity
[47:51] maybe uh top executives feeling the need
[47:54] to mention CPO an order of magnitude
[47:57] more than they used to? Uh is there
[48:00] anything really problematic about that?
[48:01] Thank you very much.
[48:02] >> Did we do we mention CPO again?
[48:04] >> I thought we weren't going to mention
[48:05] CPO. Yeah, I mean what I would say is um
[48:11] first of all in that press release that
[48:13] Nvidia and Coherent did um I think a lot
[48:17] of people missed it but it was important
[48:19] that for us that it was in there and it
[48:21] said non-exclusive right and that was
[48:24] really important for us to have that in
[48:26] the press release because we love Nvidia
[48:28] great customer they've been a customer
[48:30] of ours for 20 years but they're not our
[48:32] only customer we have a lot of other
[48:34] very important customers
[48:36] some of them. Exactly. I was going to
[48:38] say they're here on stage
[48:40] >> and we are 1,000% committed to make sure
[48:43] that we're supporting our other
[48:45] customers as well. And as soon as that
[48:48] press release went out, we went and
[48:51] talked to all those customers to
[48:52] reassure them that, hey, we're here for
[48:54] you for the long haul, right? And
[48:56] whether there's an investment or not,
[48:58] we've been, you know, we've been around
[49:00] for 50 years, right? We were founded in
[49:03] 1971 and we plan to be around for
[49:06] another 50 years supporting you. So
[49:08] that's been our message to customers and
[49:10] uh no so we'll be behind all of our
[49:12] customers. Y
[49:13] >> let's go to the next question Jose.
[49:16] >> Hi uh Sterling Parin with Omdia. My
[49:20] question is really about um I guess the
[49:22] elephant that's not in the room which is
[49:24] the telos. Um on the stage today there
[49:27] there were none. I don't think even in
[49:30] the audience um if there's any there's
[49:32] there's very few certainly compared to
[49:33] years past. Um the the question for
[49:36] whoever wants to take it or or many of
[49:38] you is um are are they still relevant in
[49:41] this world going forward and if they are
[49:43] relevant if you can kind of give me some
[49:46] some sort of specifics about what what
[49:48] makes them relevant.
[49:50] >> The answer is yes absolutely yes they
[49:52] are relevant and they're relevant for
[49:54] the hyperscalers as well. Many of the
[49:56] the telos's lumen is a good example are
[49:59] building out capacity in support of
[50:02] hyperscaler
[50:03] uh capacity in scale across where the
[50:06] hyperscalers may not have that the
[50:08] access to those those networks and
[50:10] that's I think one example that's a
[50:12] global phenomenon. The other thing is I
[50:15] think the service providers will play a
[50:17] very significant role in hosting
[50:20] infrastructure for enterprise customers
[50:22] as enterprise customers begin to adopt
[50:25] AI for inference applications. I think
[50:27] the service providers have a very
[50:28] significant role to play there. So yes,
[50:31] the uh the hyperscalers are by far the
[50:34] volume generators for the industry, but
[50:36] there's no question service providers
[50:38] are near and dear to our heart. And I
[50:41] think if you if you think about all
[50:42] those bits going out and back to
[50:43] Medcap's law and all the homes that are
[50:45] being passed and all the broadband
[50:47] that's going uh into those homes both on
[50:49] a wired basis in terms of broadband and
[50:52] a wireless basis as you see inferencing
[50:55] hit and you see creative models of grid
[50:58] compute be able to hit the industry.
[51:00] These become critical cogs. And I think
[51:03] the the challenge has always been the
[51:05] hyperscalers started off with le no
[51:07] legacy, right? Where an AT&T or Verizon
[51:10] or someone might have had 268 legacy
[51:13] services that by the way uh they were uh
[51:17] obliged to support you know that legally
[51:21] required to support. And so I think
[51:22] you're seeing the wide area network
[51:25] benefiting from some of the scale that's
[51:26] happening with the hyperscalers in terms
[51:28] of the infrastructure is going to be
[51:30] very similar. The wide area network is
[51:32] going to be a giant distributed data
[51:34] center.
[51:35] >> Okay, let's take one more question and
[51:37] then we'll go for a wrap-up.
[51:38] >> Yeah, thank you. Uh Nikil Balram Mojo
[51:41] Vision. Uh I was wondering if any of you
[51:43] might want to offer an opinion on
[51:45] orbital data centers, data centers in
[51:48] space.
[51:50] Anybody want to take that one?
[51:52] >> I mean it is actually there there are a
[51:55] number of companies right now that are
[51:57] focused on building data or focused on
[51:59] at least exploring data centers in
[52:00] space. There are many issues that need
[52:04] to be tackled there. We do actually
[52:07] support optics in satellite applications
[52:09] today. Um there are many advantages to
[52:13] having data centers in space especially
[52:14] related to power. That's the primary
[52:17] advantage. So I think it's something we
[52:19] have our eyes on. We're focused on that.
[52:22] Um there's a lot of issues relative to
[52:24] the launch that have to be solved. The
[52:26] cost of the launch before that becomes
[52:28] practical.
[52:30] >> I would say we will sell you a laser, a
[52:32] transceiver, no matter where you want to
[52:34] put that data center, right?
[52:36] >> You want to put that photons, Jim.
[52:38] >> Yeah, exactly. Photon, you want to put
[52:41] that thing in your garage, in your
[52:42] living room, go right ahead. Right.
[52:44] Well, it seems like having massively
[52:47] parallel arrays of microLED might be the
[52:49] way to go if you can't send a repair
[52:52] person up to space.
[52:54] >> Oh, I think you're getting very
[52:55] technical here with these guys.
[52:58] >> Let's um let's just finish and let's go.
[53:02] We'll start off with Michael at the end.
[53:04] I mean, in 30 seconds, I mean, what are
[53:06] the the takeaways we should we should
[53:09] take away from today and your thoughts
[53:12] given that we're in such an exciting
[53:14] time for the industry? I mean, in 30
[53:17] seconds for all of you, what should we
[53:20] take away from your your comments today?
[53:23] >> Yeah, look, uh, optics is cool again, as
[53:26] Gary said last year, right? Optics is
[53:28] cool again.
[53:30] uh this is an incredible time and I
[53:32] think the drivers are over and above
[53:35] just the hyperscaler growth right so the
[53:37] big takeaway here is you have a
[53:40] transition that's going on between
[53:41] copper and optics we're in the middle of
[53:44] it and I think it's going to drive real
[53:47] volume into this industry for for the
[53:49] foreseeable future
[53:52] I I think Michael said it well I don't
[53:53] think there's ever been a better time to
[53:55] be in the photonix industry and I'm
[53:58] really glad I joined in this industry
[54:00] two years ago. So
[54:02] >> nice.
[54:03] >> I'm glad I stuck with this industry. And
[54:06] uh I think we actually have socks at our
[54:08] booth that say optics is sexy again. But
[54:11] um I I think it's really spanning all of
[54:14] the domains of optics whether we have as
[54:17] we we're announcing multi-rail optical
[54:18] systems for long haul applications, new
[54:22] transponders that are supported for our
[54:24] customers that are deploying in high
[54:25] performance applications, data center
[54:28] optics that are, you know, 1.6T today
[54:30] and and going beyond that and then
[54:32] coherent optics that are being massively
[54:35] deployed at a 400 gig and 800 gig and
[54:37] scale across. It's just a super exciting
[54:40] time to be in this industry. I'm really
[54:41] happy to be here,
[54:43] >> David.
[54:43] >> Yeah, I agree. We're not getting
[54:45] punched. So, this is a really good time.
[54:47] Uh I'd say sees the day. Um there has
[54:50] never been a more exciting time uh in
[54:53] our industry. I think the way we engage
[54:55] with clients, the way we engage with
[54:57] partners uh that we're going to have
[54:59] co-opetition going on, we have to be
[55:02] very open to continue to keep the pace
[55:04] of innovation in the industry. And it's
[55:06] an exciting time to be here. Carpad DM.
[55:08] Yeah, I like it, Matt.
[55:10] >> Yeah. So, maybe three quick ones. First,
[55:12] plus one on the Carp ADM and the the the
[55:15] time and space we're in. It we're we're
[55:17] blessed to be here. So, every day you
[55:18] wake up and you're having a bad day,
[55:20] like it could be a lot worse. Okay, this
[55:22] is a first class problem we have in
[55:24] terms of what we're we're out to go
[55:26] solve.
[55:27] Um,
[55:30] yeah. And then I think the the second
[55:31] one and I said this two years ago I was
[55:33] on stage was I think there there's just
[55:36] this opportunity for uh really intense
[55:38] collaboration and cooperation in the
[55:40] room. I I called for that a few years
[55:42] back. You know this is an industry that
[55:44] always had these feuds and you know
[55:46] different sort of people taking stakes
[55:48] in the ground and I think it's very
[55:50] clear the ecosystem is critical. So, you
[55:52] know, I I just think continued sort of
[55:55] partnership even where you're competing,
[55:57] I think you need to figure out how to
[55:59] how to all work together to make this
[56:01] market opportunity
[56:03] um really happen. And the final point
[56:05] would be as especially as we head into
[56:06] all the sessions for the next few days,
[56:09] let's all take a deep breath and
[56:10] remember nothing is binary here, right?
[56:14] Okay, you know, scale out CPO, is it
[56:18] going to happen? Yes or no? one zero
[56:22] is scale up copper gonna stay forever or
[56:24] is that gonna go sepia is you can name
[56:26] all these different and there's these
[56:27] debates that that sort of say well
[56:29] either this technology is going to going
[56:31] to happen or not and I I just think it's
[56:33] this very intricate web and mosaic of
[56:37] all kinds of interconnect technologies
[56:39] across the board that need to all come
[56:42] together. So almost every time you get
[56:45] into one of these debates, it's yes and
[56:48] it's an and. It's an and almost always.
[56:51] It's not an or.
[56:52] >> I like it.
[56:53] >> And so that's the thing I'd say with for
[56:55] everybody here is we can get you spend a
[56:57] lot of time in these debates about
[56:59] deciding if something's going to go to
[57:00] zero.
[57:02] >> Like I don't you know like you know some
[57:05] new technology pops up well it's going
[57:06] to go to zero next year. I mean, we
[57:07] we've heard this a number of times
[57:09] across all kinds of technologies and I
[57:11] think almost every time it's an ant. So,
[57:14] that would be the thing I'd leave for.
[57:16] >> That's a great note to finish on. Let's
[57:17] thank the speakers and the panel.
[57:26] >> And this brings us to an end of a
[57:28] fantastic edition of the executive
[57:31] forum. How was it? Are you satisfied?
[57:37] I would very much like for one last time
[57:39] to thank our sponsors today. Go Photon,
[57:42] Nokia, Avisa, Icy Link, Obsidia, Moyo,
[57:45] Marvel, Resolite, New Photonics, and
[57:47] MRSI. Since it's impossible to remember
[57:50] everyone that we need to thank, please
[57:53] put up the rolling credits. And thank
[57:54] you very much. Safe travels.
[58:10] Everybody talking about scaling AI,
[58:15] but the data center's choking deep
[58:18] within.
[58:21] Copper running hot. Yeah, the signal's
[58:25] getting thin.
[58:27] So we flip the switch now.
[58:31] Optics is in.
[58:33] Bandwidth climbing fast.
[58:36] Racks are running red.
[58:39] Cloud demand exploding overhead.
[58:45] Pluggables fading as the limits closing.
[58:52] Cold package light is how we win.
[58:57] It's photoics,
[58:59] baby. 2026
[59:03] riding that light way.
[59:06] Doing new tricks
[59:09] from the fiber in the ground to the chip
[59:11] in my hand. We make that sunshine jump
[59:14] on command. Get a food on me, baby.
[59:21] 2026. See?