# EPIC Online Technology Meeting on New Pluggable Transceivers (in cooperation with COBO)

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

[00:00] we are live it is 4 p.m central european time here in the beautiful netherlands.
[00:06] we are all over the world today with companies the united states of america and asia and of course in europe.
[00:13] welcome to epic welcome to the online technology meeting on plugable transceivers.
[00:20] in the previous online technology meetings we studied the growth in data communication demands for a new connected world with the big players in the industry.
[00:28] we have identified business opportunities in the supply chain and we've realized that the big one is developing the right transceivers at exactly the right time.
[00:39] for that companies like intel luxtera nokia effect photonics sequoia scaling optics have developed fantastic solutions by combining integrated photonics with electronics.
[00:52] they satisfy the form factor demanded by equipment vendors so what challenges lie ahead.
[00:58] well a paradigm shift in data networks.
[01:01] is coming.
[01:02] and we saw in our online meetings 800g may not satisfy upcoming demands in the medium term.
[01:09] onboard optics and even co-package optics may only be an intermediary step in the future integration of motherboard and optical components.
[01:20] so let's do what we do best at epic.
[01:23] let's bring together the beyond 400g equipment vendors the transceiver manufacturers and their supply chain that includes photonic internet circuits micro optics pixels electronics assembly specialists.
[01:37] and let's answer the epic question so switch on and plug into the epic online technology meeting on plugable transceivers.
[01:44] let's do datacom business.
[01:46] let's do datacom business this beautiful beautiful afternoon.
[01:53] i recorded this video just before my holidays.
[01:55] i made a couple of statements on the video.
[01:57] you cannot imagine how many people sent me emails during the holiday study are you
[02:03] sure you are saying this are you sure?
[02:04] you're allowed to say this.
[02:06] a couple people actually told me do you have the guts to say that?
[02:08] that's actually fascinating so i love that and thank you all of you for the great feedback and of course i like to to have fun on my job so i can make any claim i want at the end of the day i don't sell transceivers.
[02:17] i'm sorry i also don't buy transceivers.
[02:20] but let's talk about what we have in our hands today.
[02:23] we have a huge opportunity for making business.
[02:24] the epic technology team has worked extremely hard on bringing the who is who in the whole entire supply chain of photonic internet circuits pixels and electronics all the way to the huge system integrators.
[02:37] first of all i would like to thank the epic members.
[02:41] we are now 600 600 companies.
[02:44] we are now 600 companies in epic and without your support you couldn't do anything.
[02:50] this is the second chapter of the season two.
[02:52] this is like a netflix show now season two chapter two of the epic online technology meetings.
[02:58] today we talk about transceivers.
[03:00] we have already planned until christmas.
[03:02] if you see any topic
[03:05] that interests you in the upcoming online technology meeting so you have to register online or simply send us an email and we'll be happy to engage you on that particular meeting.
[03:12] i would like once again to say that i am part of a dream team.
[03:16] this is the dream of a lifetime to have people that i really really love working with.
[03:23] this is my team half of them have phds.
[03:25] the other half know the industry and the marketing extremely well and we are here to support the 600 members organize events.
[03:33] we provide access to a network please tell us which companies you want us to introduce you.
[03:36] we'll do that.
[03:38] we help companies raise capital and we have the biggest website in the world.
[03:42] i'm going to repeat that.
[03:43] the biggest website in the world in finding the job in photonics.
[03:50] www.jobsimpphotonics.com.
[03:51] and we also all of you know this we have a lot a lot of market reports in our database.
[03:56] we give for free to all the members but today we are not here to talk about epic.
[03:58] we are here to talk about plugable transceivers.
[04:03] i would like to say thank
[04:05] you so much to kobo to the consortium for onboard optics.
[04:09] without you we wouldn't be able to bring all the entire supply chain all the way to the users.
[04:15] brad booth and all the cover family you are epic vips for life thank you so much for this.
[04:21] i also would like to thank the sponsors today without you this meeting wouldn't be possible.
[04:24] savik the company the huge company in the petrochemical industry you are looking for extend lenses in harsh environments this company can provide you the material for making the optics for harsh environments from consumer all the way all the way to automotive.
[04:42] effect photonics one with the success stories of the european ecosystem indium phosphide-based modules from transceivers to self-tuning modules this company has made success out of a great technology.
[04:55] congratulations effect on all what you are doing for the european industry.
[04:59] broccoli photonics who doesn't know rocky photonics thank you very much for your support.
[05:03] any company that is looking for
[05:05] solutions in silicon photonic integration all the way to supplying in the telecom data combat also in the sensing industry.
[05:14] troglophoton is making a huge difference.
[05:16] etc is when you are looking at micro optics and optical systems based on micro optics etc has great technology to support you.
[05:24] and finally ficontec the company has positioned itself in the center of the supply chain of photonic integration providing assembly solutions assembly equipment for packaging and assembly optical instruments and now i'm gonna give the floor to my good colleague and better friend anna gonzalez anna thank you very much for being with us.
[05:45] what's in the menu today thank you jose i really i couldn't wait for this moment to arrive and finally here we go and here you can see the list of a companies that are going to give a talk and this will be very useful to trigger the discussion between the entire supply chain that you can see here i hope you agree with me that this
[06:06] is one of the most amazing list of companies put together in any kind of a virtual or physical event.
[06:13] so we have at the service providers we have a transceiver the company is doing transceiver modules a vertical integrated networking solutions.
[06:23] we have all this part a great part of the supply chain for photonic integrated circuits also for packaging assembly and testing.
[06:29] big cells semiconductor laces iot connectivity software and also uh uh well just a last mention about the final line.
[06:36] which we we are collaborating jpeg spine online for uh going to ramp up the production for indian phosphate chips also picks up for packaging and assembly photonic integrated circuits.
[06:49] a and a passion a for that is not a pilot line but it's an innovative project that is developing a metro network solutions.
[06:56] uh so that's it let's start the jose thank you very much anna.
[07:02] it's so great to have anna in epic anna loves photonic
[07:06] integration even more than me and now we are gonna say hello to the people in youtube.
[07:11] this meeting is the live stream in youtube so hello youtubers in the world.
[07:16] if you want to get in touch with any of the companies in the zoom room today all you have to do is send me an email hosted.assoc.com.
[07:24] i'll be happy to make that introduction.
[07:26] this of course is so valid for the people in the room.
[07:28] if you're going to get in touch with each other all you have to do is of course use the internal chat provided by zoom or simply send me an email saying the company or the person you would like to be introduced and so happy to make that introduction.
[07:41] almost as happy as i am to introduce the next speaker today who is also a very good friend of mine and also a person who is not afraid of making claims like i am not either mark all the way from the beautiful u.s of a.
[07:54] thank you very much being with us this beautiful afternoon here in the netherlands this beautiful morning for you the floor and the attention of everyone in the whole world goes to your living room the floor is yours.
[08:04] thank you so much jose.
[08:06] and uh i think anna's getting my slides up.
[08:09] uh there we go.
[08:11] and uh um.
[08:14] you know maybe i'm becoming an old foggy.
[08:18] i don't know.
[08:20] uh there's a i would say in the world in general a reckless disregard for the truth.
[08:25] and i think you know given that we're in technology and science and we want to be careful about that uh we ought to be careful sticking the facts not make science political science.
[08:39] and they're just too many narratives in our industry that just stick around for too long.
[08:47] can we go to the next page please.
[08:59] oh somebody has control over the next page.
[09:02] there we go you're saying the wrong screen anna.
[09:12] so can everybody see that we're gonna solve it in one second mark as i say yes.
[09:16] it's true that this industry is getting getting full with marketing statements.
[09:21] we are here to find out what we can do for each other and i always love where mark takes the floor in the beginning and place everybody in the in the right in the right mindset.
[09:29] let's try to find each other to make us each other a little bit uncomfortable because this afternoon we don't want to fall asleep.
[09:35] but also let's try to make each other work together that's what we want to do here.
[09:39] anna can you change the screen that you're sharing because you're sharing the wrong one otherwise just unplug the second.
[09:44] yes i think we are in business now.
[09:48] not yet let me see.
[09:51] before i'm going to meet myself because i couldn't before.
[09:53] okay share screen.
[09:55] yes and we are going to solve this in a second everybody is smiling.
[09:59] gloria is there from infinite my friend gloria i love you gloria thank you very much joining today.
[10:05] i think we're in business taiwana mark the floor is yours.
[10:09] thank you we go to the next slide please.
[10:11] i hope
[10:18] No, thank you very much.
[10:21] So, um, you know, some, I mean, some of these narratives, you know, never go away.
[10:25] Uh, you know, I was saying since 2010 that Google Fiber shouldn't have been taken seriously.
[10:30] That it was just about getting other operators to do likewise to bring more bandwidth to the table.
[10:37] Uh, I also think, you know, there were other reasons, you know, that there was select.
[10:42] So in case Verizon or AT&T decide to, uh, kill the internet for whatever reason.
[10:48] You know, the million and one chance they do that, they could say we, we became a select in good faith and they shouldn't do that.
[10:54] Uh, Verizon in particular, uh, you know, I don't play favorites with any company, but, um, for whatever reason, you know, it was crazy with F.
[11:05] With files apparently, um, and, and I agree at the time they were doing it, the cost was extremely high and, uh, they deserve to get some grief.
[11:14] But because, you know, they're regulated, they couldn't afford to tell everybody.
[11:18] that they were passing every fortune 2000 customer along the way to pay for it.
[11:23] and uh rise verizon gets uh hit a lot.
[11:27] you know they obviously make mistakes like every other company.
[11:30] even when it comes to millimeter wave on 5g.
[11:33] uh you know t-mobile supposedly has nationwide uh 5g service in the country uh but um supposedly verizon's doing the wrong thing and uh by by stressing millimeter wave right now.
[11:47] and i i think that's uh that's just totally wrong that that point of view and actually we'll be putting out a report uh you know giving a uh a legitimate rationale you know for y verizon is focusing so much on millimeter wave.
[12:02] but i guarantee i'm 99 sure that won't change the narrative but i'll do the best i can.
[12:06] can we go to the next slide please.
[12:13] thank you.
[12:16] um i mean we've been talking about 400 gigabit.
[12:19] ethernet for a while now maybe a good seven years.
[12:24] uh there's uh there's a market research firm who will remain nameless who said there would be a million 400 gigabit ethernet modules in uh this year actually uh so.
[12:36] um you know it's it's been it's been taking a while uh and you know we need we need to keep these things to perspective.
[12:43] uh you know and putting aside the fact that uh there's all this talk about uh you know going to 800 gig or beyond we're just in the beginning stages of 40 gigabit ethernet.
[12:55] and i think uh if you saw uh if you remember jose's presentation he's pretty much saying that we're making a transition to that.
[13:05] broadcom is just ramping up uh you know that that is nothing negative about broadcom and uh uh.
[13:11] so uh you know that that's gonna take a while but uh the main point is that uh for several years the main concentration in this market for the hyperscalers.
[13:21] it's going to be for 100 gigabit ethernet and 200 gigabit ethernet big uh 200 gigabit ethernet in particular.
[13:27] because you know the cost is getting so low it's becoming competitive.
[13:31] uh you know close to the cost of a hundred gigabit ethernet.
[13:36] and so at the end of the day you know these uh hyperscalers talk a good game about needing all this extra capacity.
[13:43] but uh in the final analysis they you know they stay at lower rates and commoditize it and do whatever they can to to lower the cost.
[13:52] and and that sort of thing we've also put out a report on forge at zr if i may say so uh it's it's an excellent report.
[14:01] we were accused of the beginning of being uh of our numbers being too small.
[14:05] uh and even um one of the one of the vendors i believe um inno light uh not yet neophotonics excuse me.
[14:14] is saying that well the market size could be 500 million 100 million.
[14:18] we took about 45 of the
[14:21] uh of the 500 mil 500 million and
[14:24] uh so we've been getting some grief over that
[14:26] uh or at least initially for one vendor in particular did the discretion especially like that
[14:33] but what had happened is uh even before the market's really taken off in a big way you have a chinese vendor
[14:39] uh that's come in on pricing that's just way lower than the initial expectations
[14:45] and so uh you know we actually might wind up being high on our numbers
[14:50] uh which uh is uh not not uncommon you know uh in the final analysis these research firms could talk about having these models and all that all this other stuff it really comes down to gathering a lot of good intelligence
[15:03] it really comes down to a gut level feeling the zr plus
[15:07] may take a while it's it's kind of a swiss army knife
[15:11] it will have longer distances and of course we're talking about the line side here with forged cr
[15:18] dci and metro types of applications so zr plus may take a while you know
[15:25] I don't think enough is made with uh.
[15:27] Even even that 400 gig uh just starting to beat this discomfort level uh with power 800 gig.
[15:33] Uh the power problem comes um, you know.
[15:37] You know really relatively large.
[15:40] Um and so you know that has to be taken into consideration.
[15:43] Going to these higher capacities if whatever reason and going beyond hrg.
[15:48] Whether it's the polymer guys or or whoever.
[15:52] You know for some reason the Shannon limit doesn't come up but uh whatever.
[15:55] Um you know and the whip scalers.
[15:59] They they have this tendency uh to want to scale in the same way as servers and storage and optics is just different.
[16:12] And I think that's just unreasonable.
[16:13] I think it's just unreasonable to to expect that to happen.
[16:18] Uh you know even that uh and I don't think most of us thought it would happen.
[16:21] I mean they did get to a dollar per gig 100.
[16:26] gig and uh you know even then that that.
[16:30] that supposedly wasn't sufficient so uh.
[16:32] uh you know it's it's very interesting.
[16:35] about that.
[16:36] so uh you know if we're talking about the standard ieee stuff uh.
[16:41] i'm saying four to one million the first million in 2026.
[16:45] um you know that that maybe that may sound overly pessimistic.
[16:50] i just think it's overly complex and uh it'll take a while uh and at the same time.
[16:57] the arguments made that uh the devices will be simpler.
[17:00] despite the complexity so that that's a that's a little confusing.
[17:04] uh but uh um you know we'll see what happens there uh but.
[17:09] there's no doubt uh that the next step is to double the 200 gig uh and get to get the 400 gig.
[17:15] and and just like what happened with uh of course with 100 gig it was also the fact that you had the uh traditional carriers uh you know paying.
[17:24] um you know getting the getting down the.
[17:27] cost but it's really the chinese vendors
[17:30] that that that have lowered the cost substantially
[17:32] on the 200 gig a bit ethernet and we'll do so on 40 gigabytes of that we go to the next page please
[17:40] so um you know given that uh this is all going to take uh take a while you know
[17:47] the ford should give a bit ethernet to take off uh given that uh we're just the initial stages
[17:55] uh broadcom uh um you know in a big way or if you're just starting to ramp up
[18:03] um the fact of the matter is you know and this is this is a um this is a point that was made in a uh an earlier presentation or an earlier virtual webcast with uh with epic um you know that um
[18:19] that cpo is going to be the intermediate step
[18:23] that we need to have true integration
[18:27] and and there's no doubt there's been uh problems with uh cpo that that it is this hybrid cluj and um and that uh you know new developments will will be need to be made you know.
[18:41] but even even when we get the hundred terabits um boy i um i don't know 50 50 plugable i'm still betting on the plugables there.
[18:49] um you know if you know let alone the 50 terabit switch.
[18:54] and you know the problem is if you look at a microsoft which is still whatever reason the no offense i mean they are a sponsor but uh i mean once you get the cpo i don't know why uh you necessarily need to uh have uh kobo anymore but uh uh but i think uh vendors are going to feel obliged to at least give to to uh have some allegiance to kobo uh if in order to uh ensure that they get business so what you have which have in effect uh are three.
[19:29] solutions that need to be developed.
[19:31] uh with the vendors assuming uh the total risk.
[19:34] and in the final analysis uh you know the hyperscalers might wind up doing uh all of this anyway as uh they're they're moving to switch chips and i mean they have the resources to do it.
[19:46] so it's a risky proposition.
[19:48] we go to the next slide please.
[19:53] so i mean this is hazardous i mean uh um yes it's it's exciting uh you know don't get me wrong i mean the the four big hyper scalers they are tremendous companies uh uh i'm proud that they are american companies uh uh they are uh helping uh to bring the economy back uh during uh this this covent period uh but but i always get the sense there there are bean counters at the top and then you have the network planners and the b counters are just kind of looking at it and saying well you know we're just using such we have such enormous costs and such.
[20:29] and gradually such um you know more
[20:32] power being used
[20:34] and uh you know we need to do something
[20:36] about that
[20:37] uh especially again in comparison to
[20:40] storage and servers
[20:41] and i just i just think that's
[20:43] unreasonable and i also think it's
[20:45] ironic because
[20:46] uh you know some of these uh
[20:49] hyperscalers
[20:50] uh you know their track records are
[20:52] mixed uh
[20:53] you know some are better than others uh
[20:56] i don't think they do adequate
[20:57] benchmarking when they
[20:59] uh they're looking at new solutions you
[21:01] know that that
[21:02] that gets eliminated in the total
[21:04] numbers uh so nobody looks at that but
[21:06] when you're staring at the cost of
[21:08] a fiber or the cost of optics and the
[21:11] power
[21:12] that's right in your face and you need
[21:14] to do that so um
[21:16] so essentially um you know the network
[21:18] operate the network uh
[21:19] network operators or the network
[21:21] planners and and the architect
[21:23] architects that are involved with these
[21:25] optical uh networks they're between a
[21:27] rock and a hard place
[21:29] uh they can get fired now uh for uh
[21:33] uh you know not not listening to the
[21:35] bean counters or they'd get fired
[21:37] later uh when um when
[21:40] possibly um what happens is uh
[21:44] you um you stick a you stick a pluggable
[21:48] into a white box and you use
[21:51] uh your own sourcing power uh you know
[21:54] maybe there's a reliability issue
[21:56] there uh and so uh you know the
[21:59] uh it could there could be a time when
[22:02] we have an extensive outage by one of
[22:04] these cloud guys
[22:06] if not a security type of thing and uh
[22:09] maybe we see a return to companies doing
[22:12] their own data centers
[22:14] and that sort of thing so we'll have to
[22:16] see if that happens
[22:18] but my general point is be very careful
[22:21] about the hyperscale getting into the
[22:23] hyperscale vortex
[22:24] these companies you know are very
[22:27] attractive
[22:28] in terms of their size and their
[22:30] reputation
[22:32] but uh uh if you can all if you could
[22:35] possibly
[22:36] uh you know avoid it and you're not a
[22:39] chinese
[22:40] transceiver provider i recommend you
[22:42] avoid it
[22:43] and that's it for me that was what was
[22:46] the best way
[22:47] if you are not a chinese transceiver
[22:48] provider that was something to quote you
[22:50] on but thank you very much mark for a
[22:52] fantastic
[22:53] introduction to what's gonna be an epic
[22:56] world of discussions
[22:57] that anybody has a question statement or
[23:00] something to comment
[23:02] on mark's presentation
[23:05] there is a person in the room who i know
[23:07] has something to comment because
[23:08] a person has followed the trends over
[23:10] the last 30 years of packaging on
[23:12] transceivers
[23:13] and a person i strongly look up to is
[23:15] one of the key people
[23:16] in the epic network bob musk thank you
[23:19] very much for joining this beautiful
[23:20] meeting today bob how are you doing
[23:22] today
[23:23] very well thank you what is the picture
[23:25] that you have behind you
[23:27] that's where i live in fact over that
[23:29] hill is where i am right now
[23:31] wow it looks fantastic you have
[23:33] something to tell me about
[23:35] the packaging of transceivers right yeah
[23:37] what i'd like to do is just show you a
[23:39] little bit of the history it's been
[23:41] going on for about 30 years now
[23:43] 30 years and what have we accomplished
[23:46] well
[23:46] let me just um share the screen and i'll
[23:49] show you what's been going on
[23:53] bob is going with the people who we have
[23:55] been interacting
[23:56] over the whole online technology
[23:58] meetings over the last months and he's
[23:59] been extremely helpful on
[24:01] telling us what has been the transition
[24:03] from the previous years until now
[24:05] but we can see your screen perfectly
[24:07] clear right well
[24:09] jose just gave me one minute to do this
[24:11] i'll be as quick as i can
[24:12] so this is just looking at the
[24:14] background history of the plugable
[24:16] optical transceivers it goes back to
[24:18] 1989 when btnd developed
[24:22] transceivers for the dupont natural
[24:25] backplane system
[24:26] and i believe these are probably the
[24:27] very first plugable
[24:29] optical transceivers then in 1995 the
[24:33] first commercial
[24:35] pluggable transceivers with the gbic
[24:37] followed shortly by
[24:38] zenpay then if we look at the small form
[24:41] sector msas this is going back to 1997.
[24:45] original devices were pinned that caused
[24:48] a lot of the um customers problems with
[24:51] trying to mount the boards and solder
[24:53] them in place
[24:54] so the request was can we have surface
[24:56] mount please
[24:58] and all the msa members looked at it but
[25:00] unfortunately ibm's optics at the time
[25:02] wouldn't survive
[25:04] surface mount temperatures so we had no
[25:07] option to look an alternative
[25:08] to through and that was to use a
[25:10] plugable device
[25:12] that way we got over the problems for
[25:13] ibm but
[25:15] the plugable device then was chosen as
[25:17] an alternative to smt
[25:19] not because of its plugability etc which
[25:22] i think was an interesting twist
[25:24] and if we look at some of the very first
[25:26] plugable sfp proposals
[25:28] it's just two of them here one from ibm
[25:30] and one from them and the one from
[25:31] ibm is based on a slide connector
[25:36] rather than the pushing connectors such
[25:37] that am proposed
[25:40] and then if we just look at all the
[25:42] pluggable transceiver uh module formats
[25:45] there's a whole wide range and this is
[25:46] just the summary it's not all of them
[25:48] it just gives you some idea of the
[25:50] variants that are out there
[25:51] and that's it i've done it i think i've
[25:54] done your minute for you
[25:55] thank you very much it was a big
[25:57] challenge because you sent me out of a
[25:58] slice and i told you can you can you do
[26:00] it in one minute and you managed to do
[26:02] i think we have set up the ground to get
[26:04] the meeting going strong brad what do
[26:07] you think we have
[26:08] already told you already what the what
[26:09] the landscape looks like
[26:11] what the history look like and now i'm
[26:13] going to the key company in the
[26:15] in the onboard optics alliance and also
[26:18] the key company in the epicard microsoft
[26:19] brat
[26:20] thank you very much for being with us in
[26:21] this beautiful morning for you beautiful
[26:23] afternoon for
[26:24] us what's on your mind
[26:27] yeah hi jose and uh good morning good
[26:30] afternoon to everyone who's
[26:32] participating you know i think
[26:35] mark and bob have raised some
[26:37] interesting aspects when we talk about
[26:39] pluggables and there's a
[26:41] this industry is very different in the
[26:42] fact of what we used to think
[26:44] as as optics and how we're using optics
[26:47] uh
[26:48] you know bob highlighted how we moved to
[26:50] plug bowls
[26:51] you know if you look at it originally
[26:52] most of the optics actually were onboard
[26:54] optics
[26:55] uh and i think you know when
[26:58] they say like everything old is new
[27:00] again i think that's one of the
[27:02] aspects of you know what it really is a
[27:04] pluggable you know is it something that
[27:06] i'm inserting
[27:07] is it something that i'm embedding
[27:09] inside a system or is it what we're
[27:11] going to be looking at
[27:12] in the future here where you know i
[27:14] might be doing some type of co-packaging
[27:16] and it's pluggable on
[27:17] on an mcm i i think the aspect to
[27:20] understand
[27:21] for a lot of this is and interestingly
[27:25] bob did touch on this was
[27:26] power we do have very different
[27:30] thermal environments depending upon how
[27:32] far someone wants to drive something
[27:35] if you're driving within a rack or
[27:36] within a row the requirement
[27:39] for power there is significantly less
[27:41] than if you're trying to drive
[27:42] metro or sub c links and so something
[27:46] that's doing say a co-packaging
[27:48] that that device is not likely to be
[27:50] driving a sub-ceiling
[27:52] but it could be driving uh you know a
[27:54] disaggregated link
[27:56] and and that's a critical element here
[27:57] because what's happening is
[28:00] when you look at building data centers
[28:03] the power we pour into them is a tax
[28:07] and we don't want to necessarily have to
[28:11] build another data center and so it the
[28:14] way i can keep that power down while
[28:17] still providing the end users the
[28:18] applications and everything else that
[28:20] they need
[28:21] helps me you know not have to move and
[28:23] build a new data center
[28:25] um which you know they're they're a
[28:27] significant investment you know it's
[28:29] billions of dollars
[28:30] it's like building i remember when i was
[28:32] at intel and they used to talk about
[28:33] building fad facilities and how they
[28:35] were
[28:35] billions of dollars to build a
[28:37] fabrication facility
[28:38] the same thing applies to data centers
[28:40] there are billions
[28:41] billions of dollars worth of investment
[28:44] and so anything that we can save
[28:46] and optimize to make sure we're not
[28:47] building new data centers
[28:49] uh one helps with the environmental
[28:51] impact but it also helps us
[28:53] in being able to maintain what we're
[28:56] doing
[28:56] but i think as we go forward some of the
[28:59] requirements are going to start
[29:00] changing in optics and i think this is
[29:02] one of the aspects that
[29:03] people are kind of missing uh you know
[29:06] why were faceplate plugable optics so
[29:08] popular well
[29:09] you know originally as bob was alluding
[29:11] to you know there was a power problem
[29:13] uh but it started becoming an
[29:15] operational thing right oh the optics
[29:17] have failed
[29:17] i pull it out i throw a new one in oh
[29:19] awesome off i go and i'm up and running
[29:21] again
[29:22] um but that's actually a bad mode to be
[29:26] in because what happens is it's allowed
[29:28] us to say okay
[29:29] we're not going to care so much about
[29:31] the reliability of that because i know i
[29:33] can replace it
[29:34] but that's going to change because
[29:35] what's happening now is
[29:37] the ability to get in there and change
[29:39] some of these things whether it's
[29:41] you know embedded or whether it's
[29:42] co-packaged that changes
[29:44] the ability to get to it but the other
[29:47] uh change is the environment
[29:49] so what's happening now is if you look
[29:51] at some of these
[29:52] as people may be aware microsoft built a
[29:55] data center that we submerged in the
[29:57] ocean uh
[29:58] it's not so easy to send a guy down
[30:00] there to to you know replace plugable
[30:03] optics
[30:03] so one of the things that we're going to
[30:05] be looking at is how do i maintain that
[30:08] i gain better reliability out of my
[30:10] optics and if i start getting better
[30:12] reliability
[30:14] does that need for face plate pluggable
[30:16] still
[30:17] sit there um you know can
[30:20] can that enable things like co-packaging
[30:23] or onboard optics
[30:24] to to to take place because now i don't
[30:26] need to have to service that
[30:28] uh the other aspect is we are so looking
[30:31] at things like liquid immersion
[30:33] and people are talking about you know
[30:35] cryogenic
[30:36] cooling all these environments change
[30:38] how you use these things and i think
[30:40] that's one of the areas as we look going
[30:43] forward
[30:43] is how do i pull the heat out of these
[30:45] systems how do i build this
[30:47] so i can maintain a lower power
[30:50] node and and drive that down but
[30:53] you know as mark was alluding to it and
[30:56] it's tough these are
[30:57] these are kind of research and and while
[31:00] still trying to push the ability of the
[31:02] technology edge to
[31:04] you know make it so that we're not
[31:06] having to build new data centers and so
[31:08] that we can provide
[31:09] those computational resources closer to
[31:12] where people need them
[31:15] so it comes down to you know do i want
[31:18] to try to
[31:18] put everything into a plugable form
[31:20] factor and make it a swiss army knife
[31:22] for everything and i think
[31:23] the industry has done that for a long
[31:25] time we did that with qsrp28
[31:27] i mean it really was a swiss army knife
[31:29] we could do everything from
[31:30] our metro dci right down to a dac cable
[31:34] and and that was awesome we're not
[31:36] seeing that same thing
[31:38] you know we're seeing a little bit of
[31:39] qsfpd and osap
[31:41] and you know it gets a little bit
[31:43] tougher when we get
[31:45] into those higher bandwidths and marco
[31:47] is alluding to you know
[31:48] 800 gig and beyond and those
[31:51] are going to come we all know that
[31:52] that's eventually going to come
[31:54] uh whether it's in the 51.2 t or the 100
[31:57] t or 200 t
[31:58] generations of switch silicon when that
[32:00] starts to happen
[32:03] that swiss army knife is not going to
[32:04] necessarily be the best thing and that's
[32:06] where
[32:07] as you know equipment purchasers and
[32:10] stuff we like to have a toolkit we like
[32:12] to be able to say here's another way i
[32:14] can do it here's how i can
[32:16] uh remove the power here's how i can
[32:18] deal with a different thermal
[32:19] environment and those become critical
[32:21] elements
[32:22] as we move forward so i'm a big
[32:24] supporter of having a toolkit i think we
[32:26] have that today we have pluggables
[32:28] we have onboard optics and now we're
[32:30] looking at things like co-package optics
[32:32] but these aren't going to be used
[32:34] everywhere you know it's not like
[32:36] one's going to replace the other there
[32:37] will be use cases for some of these and
[32:39] we know that
[32:41] but it is it is necessary to have that
[32:43] toolkit because i don't think
[32:45] one solution is going to satisfy the
[32:47] whole market
[32:51] and that's what i had to say about that
[32:52] thanks jose
[32:54] okay thank you very much brad super
[32:56] interesting presentation
[32:57] as always so please now the epic
[33:00] question er
[33:01] what can you do for them and what can
[33:03] they do for you
[33:07] well i think the that's always the good
[33:09] question i think one of the keys is
[33:11] is let's um as an industry look at
[33:15] what's going to really happen i mean we
[33:17] know the trends we know how things are
[33:19] built
[33:19] uh you know even with co-package
[33:23] you know that is we're hearing two
[33:25] different
[33:26] flavors there right now you know we're
[33:28] saying people are saying hey well we can
[33:30] use xsr and others are saying we can use
[33:32] a different type of electrical interface
[33:35] direct drive or
[33:36] lr plus or something you know we need to
[33:38] have that discussion to figure this out
[33:40] you know we're focusing a lot on how do
[33:43] we eliminate power
[33:44] what i'm looking at is eventually we're
[33:48] going to have a problem where
[33:49] driving copper cables is just not going
[33:51] to work if i get to 200 gig
[33:53] electrical interfaces copper cables
[33:55] literally have to start disappearing in
[33:57] my data centers
[33:58] and so for that transition to happen how
[34:00] do i make optics
[34:02] fit in there and and become that uh
[34:05] copper replacement uh you know i look at
[34:08] it from
[34:08] also from personally from an
[34:10] environmental point of view if you've
[34:11] ever seen a copper mine they're not very
[34:13] pretty
[34:13] uh so how do we get to you know doing
[34:16] something that's more photonics based
[34:17] whether it's driving across a pcb or
[34:20] whether it's
[34:21] driving across a back plane or across
[34:23] the ocean
[34:26] okay thank you very much brad so we have
[34:27] a question from jean
[34:29] from g-a-e-g genuine jim
[34:32] would thank you yes um more of a comment
[34:36] than a question so i
[34:38] i don't think optics be became pluggable
[34:40] because of reliability concern of the
[34:42] need to change them out
[34:43] i think it was a matter of the swiss
[34:45] army knife that brad showed so the
[34:47] ability to plug anything into a port
[34:50] and what has changed moving forward is
[34:52] as the switch matrixes got very large
[34:55] and became large close equivalents
[34:57] there's always one stage it's a fan out
[34:59] where everything connects to everything
[35:02] it's a very common consistent connection
[35:04] so
[35:05] there you no longer need the swiss army
[35:07] knife you could make something dedicated
[35:09] and that's permanently connected because
[35:11] you know it will never be changed
[35:13] it has one requirement and it's a
[35:15] permanent connection so
[35:17] there now that the hyper scales of these
[35:19] large fan
[35:20] outs in the middle of their switch you
[35:23] don't need a swiss army knife like brad
[35:25] was saying
[35:26] thank you
[35:30] thank you very much so any additional
[35:32] comment
[35:35] good to see you jim for brad for jimmy
[35:40] okay so if not let's go ahead with the
[35:42] program and it's my great pleasure to
[35:44] introduce you a
[35:45] juan pedro from telefonica that will
[35:48] give a the next
[35:49] presentation so well you know telefonica
[35:51] is the the biggest
[35:53] service provider in spain so i am super
[35:56] happy competitor that you accepted this
[35:57] invitation
[35:58] and yes if you want to share your screen
[36:02] thank you thank you anna yes
[36:07] okay
[36:11] can you see my screen
[36:14] yes yes very well excellent so thank you
[36:17] again thank you jose
[36:18] thank you anna for this opportunity to
[36:20] present you
[36:21] some uh these application scenarios for
[36:25] the
[36:25] new optical pluggables for 5g
[36:29] transport here i'll present you
[36:33] some application externals where i think
[36:35] the new plugables could help an operator
[36:38] as
[36:39] to optimize and the the course and the
[36:42] investments
[36:43] to face with the traffic increase coming
[36:46] from
[36:47] 5g application and other in
[36:50] bandwidth intensive applications uh
[36:54] telefonica i think as other operators
[36:56] are experiencing
[36:58] a traffic explosion at the edge of a
[37:00] network and
[37:02] the first question is where is the edge
[37:04] of the network because this is very
[37:05] important to
[37:06] to understand what are the also the
[37:08] requirements for the optical transport
[37:10] uh the edge is uh is not
[37:14] as close as many people can can think
[37:17] uh the edge i think that the point where
[37:20] the traffic injection
[37:22] is expected for gaming for video
[37:24] transmission and so on
[37:25] and this fulfilling the the latency
[37:28] requirements
[37:28] is what we in the level that we told the
[37:31] xl2
[37:32] this means in data centers and
[37:36] located with routers in the biggest
[37:39] cities
[37:39] in in in each country for example cities
[37:43] with a population
[37:44] above a half million this means that
[37:47] for example in telefonica network in
[37:50] countries like spain
[37:51] in the netherlands germany in uk or in
[37:54] brazil
[37:55] this means some tense of of nodes
[37:59] tens of nodes where we're expecting or
[38:02] where we are already deploying
[38:05] the uh service gaming
[38:09] video platforms and so on and
[38:12] uh this is fulfilling the requirements
[38:15] for
[38:15] for latency for example then most of the
[38:19] traffic or most of the traffic explosion
[38:22] is impacting in the regional
[38:25] and the access let's say the access or
[38:28] metro network and for regional we mean
[38:33] networks with uh typical distances or
[38:37] paths no longer than
[38:38] one thousand kilometer so most of the
[38:40] traffic
[38:41] or most of the investments will be
[38:44] in for channels for paths with
[38:48] distances lower than 1 000 kilometer
[38:51] even considering restoration paths
[38:55] important in this networks in the metro
[38:58] and in the regional
[38:59] we are deploying photonic message with
[39:01] raw dams restoration capabilities
[39:03] dynamic restoration is key in telefonica
[39:06] using gmpls nowadays and
[39:08] evolving to to sdn and for this
[39:12] having a network with photonic mesh with
[39:14] multiple paths
[39:15] the optical spectrum is not a problem so
[39:19] for example the migration to for example
[39:21] 200 gig
[39:23] or 400 gig is mainly motivated by the
[39:26] cost efficiency because the cost per bit
[39:28] is lowered for example than for 100 gig
[39:30] it's not a the evolution towards higher
[39:34] speeds 200 400 or or or more
[39:37] it's not motivated by a spectrum
[39:40] scarcity is motivated by cost efficiency
[39:44] and in order to er in order to me
[39:47] also optimize this cost one of the
[39:50] the levels or one of the technical
[39:52] initiatives we are promoting in
[39:53] telefonic is the optical disaggregation
[39:56] and i think also the new optical
[39:57] transceivers have lots of
[39:59] lots to say or there are a lot of
[40:02] motivations to have
[40:04] this optical disaggregation the idea is
[40:05] to take
[40:07] solutions that are coming from the data
[40:09] center world for data center
[40:10] interconnection
[40:11] for example single geo boxes
[40:14] doing optical and packet aggregation
[40:18] and deploy in the backbone but okay
[40:21] keeping
[40:22] the same functionalities in terms of
[40:24] performance fault management and
[40:26] surveillability resiliency
[40:28] that we have in a typical in a normal
[40:29] network uh from akiben vendor and we
[40:32] manage
[40:32] the idea we can do it using sdn using a
[40:36] standard
[40:36] sdn architecture the right interfaces
[40:40] towards the transceivers
[40:41] we can for example select the the the
[40:44] power
[40:45] the the wavelength and so on so one of
[40:47] the applications is in the metro core
[40:49] distance is lower than
[40:51] thousand kilometer typical rates 200 yet
[40:54] 408
[40:55] 400 gig we do not foresee a higher
[40:58] speeds at least in the
[40:59] in the midterm in a time frame of three
[41:01] years i think 200 and 400
[41:03] would be more than enough but very
[41:05] important to have these restoration
[41:07] capabilities
[41:08] automatic provisioning and so on and the
[41:10] optical desegregation would be key in
[41:13] order to minimize the cost
[41:15] other application scenarios are in the
[41:17] access let's say
[41:18] in the for example when in order to
[41:21] access directly to a 5g
[41:22] site we typically have fiber rings
[41:26] and for the new sites because in 5g we
[41:28] will need an intensive
[41:30] deployment of of new sites we will need
[41:33] more fiber investment
[41:35] this probably is one of the major costs
[41:37] in the
[41:38] in the deployment or for 5g evolution
[41:42] and how how can we optimize this cost
[41:45] and how the new optical transceiver can
[41:47] help us to
[41:48] optimize this cost first of all by
[41:50] migrating from
[41:51] rings for example typically currently
[41:53] for the 4g
[41:55] we have fiber rings with
[41:58] drain interfaces interconnected at
[42:00] different sites and sharing
[42:02] the bandwidth with the 5g we can
[42:04] allocate a dedicated wavelength parasite
[42:07] using passive wdm and
[42:11] optical transceivers color transceivers
[42:14] that are
[42:14] installed in both the 5d side for
[42:17] example in the bbu
[42:18] and in the router that is doing the
[42:21] aggregation in in our
[42:23] naming the dates r4 so that we can have
[42:26] a dedicated or allocated wavelength or
[42:28] capacity per
[42:30] 5g site according to that in this case
[42:33] we are
[42:33] forcing the most in the
[42:36] short medium term most of the of the
[42:39] transceiver would be just
[42:41] 10 gig dedicated 10 gig per
[42:44] 5g side and migrating to 100 gig
[42:48] important is the dsp integration in this
[42:51] in distance zipper in order to enable
[42:54] for example the
[42:54] the to install a plugable from one
[42:57] vendor
[42:58] or given border in the for example the
[43:00] radio side from
[43:01] any run radio access network vendor or
[43:04] in the
[43:05] on the router platform the typical
[43:07] distances are up to 30 kilometers and
[43:11] up to 10 hops as a reference for
[43:14] the application plasma telephonica we
[43:16] have already started
[43:17] some deployments in in this in this
[43:20] sense
[43:21] so key is the evolution from the logical
[43:24] uh
[43:25] from the tree to the to the star
[43:28] topology
[43:29] from sorry from the ring to the to the
[43:31] star topology
[43:33] another benefit that this is providing
[43:36] to the telco operator telefonica and
[43:38] others
[43:38] is to promote the network sharing
[43:42] for example telefonica can deploy a
[43:45] new fiber in a given region vodafone can
[43:48] do
[43:49] for example a deployment in a complement
[43:52] a complementary region
[43:54] and we can serve wavelengths among us
[43:56] for example
[43:57] we can do agreements sharing the
[44:01] the capacity in terms of wavelengths so
[44:03] that each one can deploy their own
[44:05] ip network or their own technology and
[44:09] using wdm we can serve cost this is
[44:12] something for example
[44:13] we can we are currently doing for the 4g
[44:16] in different countries this
[44:17] this model of network server inducing
[44:19] wavelengths and for the 5g would be
[44:22] s will be really interesting
[44:26] finally optical and packet integration
[44:30] the optical and packet integration as
[44:32] many years ago we started to talk about
[44:34] the integration of call or interfaces in
[44:37] the ip routers
[44:38] that never succeed and one of the
[44:41] reasons
[44:42] telefonica and other operators i think
[44:45] do not
[44:46] deploy this optical and packet
[44:47] integration was because
[44:49] two main drawbacks the first one was
[44:51] that
[44:52] the integration of the optics in the
[44:54] router was impacting
[44:55] in the port scalability we cannot have
[44:59] we couldn't have a high density port
[45:03] cards in hyper routers when we
[45:05] integrated
[45:06] the color the the the optics in the
[45:08] router
[45:09] and the second one was the multipenter
[45:11] interoperability
[45:12] if we wanted optical and packet
[45:15] integration we had to buy
[45:17] both ip and optical to the same to the
[45:19] same vendor
[45:20] now hopefully both problems are could be
[45:24] solved
[45:24] and mainly
[45:27] thanks to the new optical pluggables it
[45:31] is okay with new optical pluggables also
[45:33] integrating
[45:34] the for example the the dsp
[45:38] we can have we can install the plugables
[45:42] in high density cards in the ip routers
[45:46] and avoiding this not losing
[45:49] scalability in the routers and on the
[45:51] other hand
[45:52] we can configure these uh this we can
[45:55] configure this wavelength or the optical
[45:58] parameters using sdn interfaces
[46:02] in the case of telefonica we are using
[46:04] netconjunct and openconfig models
[46:06] to to do that so that we can configure
[46:09] the
[46:09] transceiver for example for the
[46:11] wavelength that we need for the
[46:13] restoration for the provisioning or
[46:15] or for network optimization and
[46:19] even having a transponder from one
[46:22] vendor over the optical network of
[46:24] other vendors so that the
[46:27] opera operability could be could be
[46:30] assured
[46:31] so considering this for example that the
[46:34] main drawbacks the scalability and the
[46:37] interoperability can be solved using
[46:39] both
[46:40] a standard sdn and new plugables then
[46:43] is really the time for the optical and
[46:46] packet integration according to that for
[46:48] some
[46:48] telefonica were intensively actively
[46:51] working in the telecom infra project
[46:54] there were some boxes cassini
[46:57] there are different boxes doing packet
[46:58] and optical integration with different
[47:00] capacities
[47:01] are being developed and we have
[47:04] starting the trials for deployment in
[47:06] the network with these technologies
[47:09] yes to conclude my presentation my
[47:11] message is
[47:12] uh traffic impact in the edge only for
[47:16] regional
[47:16] and metro networks distances lower than
[47:18] one thousand kilometer
[47:20] mainly 100 200 and 400 channels we do
[47:23] not fully see
[47:24] anything more but very important is the
[47:26] dynamicity
[47:27] dynamic provisioning and dynamic
[47:30] restoration
[47:31] we are deploying messages in order to
[47:32] have multiple paths so that
[47:34] we don't have problems for a spectrum if
[47:37] we have many paths
[47:38] many alternatives we don't we do not we
[47:41] don't need to optimize
[47:43] the optical spectrum but the key is the
[47:45] opti is the cost-effectiveness
[47:47] optical desegregation could help us to
[47:49] do it
[47:51] and find other applications are in the
[47:54] 5d
[47:55] the passive passive rings
[47:59] migrated to to 2 stars
[48:02] and the network sharing together with
[48:05] the integration
[48:06] packet and optical integration motivated
[48:08] by new
[48:09] optical pluggables okay that's all from
[48:11] my side thank you very much juan muchas
[48:13] gracias it was fantastic to hear from a
[48:15] service provider that the move to 400g
[48:18] cost you to cost per bit and not because
[48:20] of throughput or because of all the
[48:22] ideas that we are having in mind i love
[48:24] having telefonica tell us that and we
[48:27] have two questions for you in the chat
[48:28] the first one is coming from
[48:30] iponix francois what's on your mind
[48:36] francois from iponix i
[48:39] was muted no longer am i so uh
[48:43] question two juan so should 5g go to way
[48:46] of optical front haul
[48:48] with wdm and go
[48:51] from ring to start topologies
[48:54] is there an ongoing need for optical
[48:58] uplink redundancy so-called dual homing
[49:04] okay thank you yeah we are forcing the
[49:07] redundancy
[49:08] in the uh just in the in the physical
[49:11] path
[49:13] but not for example in the in the router
[49:17] uh let's imagine the fortune for the 5d
[49:20] the number of users per side would be
[49:23] very low
[49:25] the idea is okay and if we have
[49:27] redundancy dual human redundancy in
[49:29] terms of for example the head ends for
[49:31] its
[49:33] for its 5d side the cost
[49:36] could be extremely high but if we have
[49:39] for example
[49:39] a ring in a ring configuration
[49:43] we can have a star but with two
[49:46] separated paths
[49:47] over the same the physical structure and
[49:49] the cost for this would be
[49:51] minimum so most of the risk that is the
[49:53] fabricat could be solved
[49:55] using a star topology with redundancy
[49:58] but typically
[50:02] terminating the connection with a in a
[50:04] single
[50:05] hyper router in other not to not to
[50:09] increase the the cost
[50:14] you did a good job answering i think he
[50:16] looks satisfied we have another question
[50:18] andrew from signal ia
[50:21] thank you so it was interesting to see
[50:24] your presentation on coherent at the
[50:26] edge of the network
[50:27] i think people have a pretty good idea
[50:29] of what the cost is
[50:31] of a 400 zr module and
[50:34] when you look at 100 gig coherent
[50:36] pluggable module that the cost is about
[50:38] the same but
[50:40] obviously for the the adoption to be
[50:43] successful i think the price needs to be
[50:44] a little bit lower
[50:45] i was wondering if you could talk a
[50:47] little bit about you know what your
[50:48] price expectations are in this
[50:50] application
[50:51] relative to your existing 10 gig
[50:53] solutions
[50:54] or you know relative to a hyperscale
[50:57] based 400 zr solutions
[51:00] okay yeah i think for example as a
[51:04] reference i think the 200 gig
[51:06] currently the cost for a 200 gig is
[51:09] pretty similar to 100 gig
[51:10] is 10 percent higher 200 gig than
[51:14] uh 100 gig and so
[51:18] according to that i think the the cost
[51:20] for 400 gig
[51:21] with respect to a 100 gig
[51:24] shouldn't be more than let's say uh 30
[51:27] percent
[51:28] in order to be competitive or
[51:32] in order to to have a business case for
[51:35] for that
[51:36] and that's uh i think that's important
[51:40] also if we consider that most of the the
[51:43] the applications
[51:44] or most of the demands would be for
[51:46] channels that are
[51:47] with the distances below 1000 kilometers
[51:52] literal regional networks and and so on
[51:55] so let's say as a reference we can say
[51:58] that the
[51:59] the 400 gig is cost the target cost
[52:02] is around 30 percent is
[52:05] higher than the existing for for 100g
[52:11] we have one more question coming from a
[52:13] beautiful country in europe coming from
[52:15] finland coming from modulite
[52:17] pekko what's on your mind
[52:20] yeah i was wondering i wanted to ask
[52:23] ask one about this
[52:26] wdm wavelength scheme that the telephone
[52:29] consists for
[52:30] four six for 5g front hole i mean
[52:32] obviously i
[52:33] assume that the reuse of the cwdm
[52:37] wavelengths is
[52:39] first thing but when when increasing the
[52:42] channel numbers they are i think they
[52:44] are different possible options
[52:48] so how do you see that yeah when
[52:51] expanding from four to four to eight
[52:53] twelve
[52:53] channels now yeah i think
[52:57] uh we are going for for for dwdm only i
[53:01] think the number of channels is
[53:02] is is very important even if
[53:06] i think the even the the front hole
[53:08] using for example cpri
[53:10] or ecpri is very seen in very limited
[53:14] uh places it's no
[53:17] we are not con assuming or at least
[53:19] considering a way to spread deployment
[53:22] of optical front hall using
[53:25] [Music]
[53:26] let's say or cloud run or using ecbri
[53:30] the main target is to provide a
[53:32] connectivity
[53:33] a connectivity of 10 gig
[53:36] or 100 gig to the bbu
[53:40] in the in the 5g site
[53:44] and trying to the main goal is try to
[53:48] optimize the cost
[53:49] of the routers in the access and
[53:53] if we want to optimize the cost of the
[53:54] routers that is try to
[53:57] increase the and the the number the
[54:00] scalability the number of
[54:01] ports uh or in this route and the number
[54:04] of sites that are connected to this
[54:06] to this router the idea is keeping the
[54:08] routers in the same place we
[54:10] have nowadays and absorb also the
[54:14] increased number of of 5g sides so the
[54:17] key is
[54:18] the number of wavelengths we need if we
[54:20] need dedicated wavelengths per side
[54:23] we need a wdm dwdm
[54:27] and okay the front hall
[54:30] or eca ecbri is not seen as a
[54:33] top priority is it it's not it's not
[54:36] discarded but is not the
[54:38] top priority okay juan thank you very
[54:41] much being with us today
[54:42] the r and d of department of telefonica
[54:45] is a really important one they are
[54:47] looking for partnerships cooperations
[54:49] collaborations juan is here to talk to
[54:51] all of you
[54:52] i would like to remind you right now
[54:54] that after the meeting
[54:55] for those of you who sign up for one to
[54:57] one meetings we have arranged
[54:58] separate zoom rooms for all of you all
[55:01] you have to do is to stay and we'll
[55:02] arrange it for you
[55:03] all those of you who are interested in
[55:05] one to one
[55:07] meetings just check your inbox you got a
[55:08] message from carlos lee
[55:10] why don't we continue with the program
[55:12] we go to a company who i love become a
[55:14] silic i'm a silicon photonics fan who
[55:16] doesn't love
[55:17] infi thank you ralph for being with us
[55:19] this beautiful afternoon here in the
[55:21] netherlands
[55:22] the floor and the attention of everyone
[55:24] goes to you
[55:30] i think we have you muted we can see the
[55:32] san francisco beautiful red bridge but
[55:34] we cannot hear you
[55:35] can you hear me now yes okay all right
[55:38] let me
[55:40] let me ah all right
[55:52] we can hear we can see your screens
[55:54] quite clearly you go
[55:56] fantastic talk to us okay
[56:00] thank you very much as always jose you
[56:02] are just as animated as i have always
[56:04] known you
[56:06] uh having this uh meeting open with mark
[56:10] is always
[56:10] interesting i'm going to be talking
[56:12] about uh pluggable modules for
[56:14] coherent optical interconnects
[56:18] all right so
[56:22] technology choices in data center
[56:24] interconnects we've been talking about a
[56:26] number of different technology choices
[56:28] here
[56:29] the view at 100 gig which is inside of
[56:31] full deployment
[56:32] you have the ieee 802.802.3
[56:36] psn4 cwdm4 under the two kilometers
[56:40] lr4 under 10. between 10 and 100
[56:44] there are a mixture of coherence and
[56:47] pam4
[56:48] direct effect deployments and anything
[56:50] beyond about 100 kilometers
[56:52] you have a variety of coherent formats
[56:54] from gps gate acorn to 16
[56:57] as the distances get larger beyond um i
[57:00] think telefonica called it uh
[57:02] edge up to a thousand kilometers uh
[57:04] metro
[57:05] and and beyond uh you have qpsk
[57:10] i'd like to propose a view at 400 could
[57:13] be a little different
[57:15] uh with the advance of 400 gzr
[57:20] the zr has been adopted at oif ieee as a
[57:23] working group
[57:24] itu-t has the early adoptions of the
[57:29] cr for data center interconnects under
[57:32] 120
[57:34] kilometer links so there is an evolution
[57:39] from what the data center looked like
[57:41] say about three four five years ago
[57:43] to what it could look like uh this year
[57:46] and the next um
[57:47] volumes aside uh there is a real move to
[57:50] deploy zr zr being a plugable dco
[57:58] so what does a a
[58:02] switch plugable actually looks like so
[58:05] this is a silicon photonics based
[58:07] 400 gzr coherent module
[58:10] looks like any qsfp 28 module
[58:15] except that this is a qsfpdd
[58:19] and the dsp in there is an advanced
[58:21] seven nanometer cmos process
[58:23] internally it's called canopus and we
[58:26] have an advanced silicon photonics
[58:28] uh from an advanced silicon photonics
[58:30] node from a by cmos foundry
[58:32] it's a eight by fifty gigan gigabyte
[58:36] four
[58:36] and output is a single wavelength 16
[58:38] qual
[58:40] nominal reach about 120 kilometers
[58:44] and if i were to um get to my
[58:48] point or options here um so you have the
[58:51] dsp
[58:51] chip the silicon photonics the tin
[58:53] drivers are external and have
[58:55] you have a tunable laser
[58:59] uh silicon photonics is a highly
[59:01] integrated silicon photonics chip with
[59:03] both the transmit and receiver on a
[59:05] single chip
[59:07] so a single tunable laser splits between
[59:10] the receiver and the transmitter and the
[59:13] split ratio depends on the trade-off
[59:15] between output power and receiver
[59:17] sensitivity
[59:18] the input light comes in you have a
[59:19] polarization beam splitter and a rotator
[59:22] here they're all t by the way and the
[59:24] output is polarization
[59:25] beam rotator combiner and
[59:28] uh this is a compact single chip
[59:32] architecture and the performance at
[59:36] c-band for coherent links
[59:37] uh bandwidths in excess of 30-some
[59:39] gigahertz 40 gigahertz
[59:41] it's not as difficult to make photo
[59:43] detectors with a high enough bandwidth
[59:46] one of the important things about
[59:50] building the silicon photonics chips for
[59:54] c-band coherent communications
[59:57] is that you need a polarization beam
[01:00:00] splitter combiner technology
[01:00:03] that has got low loss performance over
[01:00:05] entire c band
[01:00:06] and low loss and high crosstalk
[01:00:10] so this has got -25 db crosstalk and
[01:00:13] probably about a half a db or so uh loss
[01:00:17] so you need a high speed max under
[01:00:20] a photo detector and uh low loss high
[01:00:23] speed polarization
[01:00:25] maintaining components so that's what's
[01:00:27] inside
[01:00:28] that module so if you were to go down
[01:00:32] um this is the marginal performance at
[01:00:36] 80 kilometers so it seems to be the
[01:00:38] deployment scenario
[01:00:41] single span optical links and
[01:00:44] uh this is a multi-channel this has got
[01:00:47] about 32 channels
[01:00:48] maybe a little more a typical deployment
[01:00:51] scenario these days people are talking
[01:00:53] about 64 channels at 400 gig giving it
[01:00:56] a fiber capacity of 25.6 terabits
[01:01:00] per second per fiber pair
[01:01:03] and this performance at 80 kilometers is
[01:01:06] with polarization scrambled
[01:01:07] real-world conditions and it's got less
[01:01:10] than
[01:01:12] half a db polarization penalty and this
[01:01:15] is actually on a real 400g
[01:01:17] switch in which we are running traffic
[01:01:21] uh right now so before i end um there
[01:01:24] was a lot of questions about
[01:01:26] um how far can this go i was limited to
[01:01:30] about three or four slides as an entire
[01:01:32] zr plus mode uh that this chip also has
[01:01:36] with a lot of vendors out there have i'm
[01:01:39] not professing that this is going to go
[01:01:40] up to 10
[01:01:41] 000 kilometers but in a plugable form
[01:01:44] factor
[01:01:46] you can probably cover up to about a
[01:01:48] thousand or two thousand kilometers
[01:01:51] imagine a a single pluggable form factor
[01:01:54] that can
[01:01:55] cover the whole gamut from two to about
[01:01:58] say two or three thousand
[01:01:59] kilometers and that is sort of the
[01:02:02] promise of four gzr
[01:02:04] and we are well on our way to uh
[01:02:07] to shipping shipping we are shipping
[01:02:09] these modules in
[01:02:11] in uh sampling volumes to
[01:02:14] several customers and uh we'll go to
[01:02:17] general availability end of this year
[01:02:19] early next year
[01:02:21] thank you thank you very much
[01:02:24] it was a super interesting presentation
[01:02:26] and it seems that you have
[01:02:28] plenty of challenges for our supply
[01:02:30] chain in photonics so
[01:02:31] well what would you say so
[01:02:34] yes uh this is always an interesting
[01:02:37] question
[01:02:39] uh the the challenges are first in in
[01:02:42] in silicon photonics part the
[01:02:45] the the common complaint which people
[01:02:48] have about silicon photonics is you can
[01:02:50] demonstrate one or you can demonstrate
[01:02:52] two
[01:02:53] but can you do this in volume all right
[01:02:55] that's the whole goal
[01:02:56] um can we produce these things uh we
[01:02:58] make these in um
[01:03:00] in eight inch wafers um can we produce
[01:03:03] this in volume
[01:03:04] and can you have the performance
[01:03:06] reliably
[01:03:08] way for two wafer uh that's
[01:03:10] manufacturable for coherent applications
[01:03:13] we we have a large number of 100g
[01:03:16] pam 4 modules in deployment and that's
[01:03:20] where we had learned to make these
[01:03:21] things manufacturable
[01:03:23] one is the manufacturability of silicon
[01:03:26] photonics but the same time
[01:03:27] manufacturability of
[01:03:29] high-end drivers and tias as well as
[01:03:31] seven nanometer dsp
[01:03:33] so right across if you want to make
[01:03:35] something that small
[01:03:37] consuming anywhere between 15 to 17
[01:03:39] watts
[01:03:40] thermal and layout and
[01:03:44] supply chain becomes critical
[01:03:48] okay got it so well we have a few
[01:03:50] companies well several companies doing a
[01:03:52] silicon photonics here in this meeting
[01:03:54] so
[01:03:54] i guess they would like to talk to you
[01:03:56] afterwards
[01:03:57] but now we have a question from from
[01:04:00] whose microoptics so wilfred yes go
[01:04:03] ahead please
[01:04:04] hi thank you um enough for giving you
[01:04:06] that word
[01:04:07] uh super interesting especially the the
[01:04:10] talk from infi
[01:04:11] and from brad because i'm more on the
[01:04:13] component side so that's why i like this
[01:04:15] stuff
[01:04:16] and so we were talking about thermal
[01:04:18] management
[01:04:19] and and and thermal gradients as well
[01:04:22] interesting point that was raised by
[01:04:23] microsoft from my point of view
[01:04:25] so how can all these components handle
[01:04:28] this
[01:04:29] on on a chip level maybe even so not
[01:04:31] maybe on a system level this usually
[01:04:32] works but on a on a
[01:04:34] component level what does that mean for
[01:04:36] me as a component provider
[01:04:38] thermal gradients or thermal stability
[01:04:41] because cisco
[01:04:41] cisco i think they wanted to do
[01:04:43] co-package uptake at elevated
[01:04:45] temperature like 100 120 degrees
[01:04:48] and we we make optics now for the
[01:04:49] automotive industry we're getting used
[01:04:51] to these high temperatures and this is a
[01:04:53] different business now so
[01:04:55] um we know now 120 degrees is is pushing
[01:04:58] it
[01:04:58] but but it's all feasible of course and
[01:05:01] i just wondered
[01:05:02] how that what does it mean for for us
[01:05:05] for
[01:05:05] for our micro optics or integrated
[01:05:08] optics
[01:05:09] how does these gradients so
[01:05:12] you are asking a very relevant question
[01:05:15] uh before brad jumps in here um
[01:05:18] so the cooling at module level
[01:05:22] is is one of one challenge but cooling
[01:05:25] at the chip level because
[01:05:28] uh all these chips do not have a uniform
[01:05:31] thermal distribution
[01:05:33] is a bigger challenge it is something
[01:05:37] it is starting to become a challenge at
[01:05:40] 400g
[01:05:41] plugable and we are starting to look at
[01:05:43] it
[01:05:44] and i would love to see materials that's
[01:05:47] out of
[01:05:48] source that would allow us to actually
[01:05:51] uh manage the thermal gradient within a
[01:05:54] chip
[01:05:55] as opposed to trying to manage it at the
[01:05:57] board level because
[01:05:58] everybody assumes it's uniform it's
[01:06:00] never gonna fall
[01:06:01] so uh yeah that that's that's an
[01:06:04] exciting development
[01:06:08] okay great thank you very much and we
[01:06:10] have a mark from five reality that the
[01:06:13] right side his hand so mark yes thank
[01:06:16] you uh
[01:06:17] i had a question uh actually a couple
[01:06:19] questions uh you know
[01:06:20] when it came to silicon photonics uh i'm
[01:06:22] not sure that uh
[01:06:24] anyone who got into the datacom game uh
[01:06:27] is making money uh whether they're part
[01:06:30] of another company or not
[01:06:32] uh certainly it's been more successful
[01:06:34] on the telecom side but
[01:06:36] it's been you know kind of hidden uh as
[01:06:39] to what the actual costs
[01:06:40] are in terms of dealing with loss and
[01:06:42] that kind of thing
[01:06:44] but my bank my main question is on the
[01:06:47] forge of zr the price of admission now
[01:06:49] is fifteen hundred dollars uh or less
[01:06:52] uh which is substantially lower than
[01:06:54] what was initially anticipated
[01:06:57] uh does that mean that infi has to
[01:07:00] [Music]
[01:07:01] necessarily consider going back to just
[01:07:04] uh you know maybe just offering the dsps
[01:07:07] for that
[01:07:08] or uh do you have some wiggle room there
[01:07:11] uh given the you know the price crash
[01:07:14] which has kind of happened uh
[01:07:17] even before the market has taken off
[01:07:20] the pricing is always sort of dynamic
[01:07:24] here mark uh by the way to answer your
[01:07:26] uh the other question we do offer both
[01:07:29] actually we have a business model in
[01:07:31] which we offer the dsp chip
[01:07:34] as well as make the module
[01:07:37] so they're both business models and we
[01:07:40] also offer the tiaa
[01:07:42] and driver
[01:07:45] but the driver that use internally is
[01:07:46] optimized for our silicon photonics
[01:07:49] um so the the
[01:07:52] the business model for n5 is
[01:07:56] uh it's flexible so we are not we
[01:07:59] like to treat the zr module as
[01:08:02] a packaging vehicle and if somebody were
[01:08:06] to take the pieces and and come up with
[01:08:08] a more creative package
[01:08:10] that'll be just as fine as well
[01:08:16] okay thank you very much any question
[01:08:18] additional comment about this
[01:08:20] oh well we have a i saw johanna for
[01:08:24] from a interest right hey please
[01:08:27] go ahead with your question uh hi
[01:08:30] hi guys uh i'm working from interest
[01:08:32] where we're dealing with a lot with
[01:08:34] thermal management of the chips
[01:08:36] and also at the hot spots of the chips
[01:08:38] so then immediately i i wonder did you
[01:08:40] guys try to use
[01:08:42] a diamond placed as a heat
[01:08:45] heat spreader directly at the hot spots
[01:08:47] of of such a turbo chips
[01:08:50] uh we haven't gone that far yet uh we
[01:08:52] don't usually disclose all the things we
[01:08:54] do
[01:08:55] but uh that is in in in some part of my
[01:08:58] life uh we had used industrial diamond
[01:09:02] before people get any ideas
[01:09:04] yes yes that's a very interesting idea
[01:09:08] uh
[01:09:09] no we we are not using it um
[01:09:12] not not for any other reason um except
[01:09:15] uh
[01:09:16] except cost uh from the last time i
[01:09:19] looked at it
[01:09:20] if they had come down uh and cost
[01:09:22] favorably uh
[01:09:23] we might we might end up considering it
[01:09:25] as well but so i'm
[01:09:27] sure it is not the first time that
[01:09:29] people answer that to you
[01:09:31] right johanna how do you how did you
[01:09:33] answer that people say oh diamonds are
[01:09:35] too expensive
[01:09:36] how did you answer that well that's
[01:09:38] exactly wrong answer because uh
[01:09:40] we're talking about very small piece of
[01:09:42] synthetic diamond
[01:09:44] so we're not using like big chunk of 1
[01:09:47] by one centimeter we
[01:09:48] using for instance one by one millimeter
[01:09:50] and that's already quite big piece
[01:09:52] so if you can put that kind of piece or
[01:09:55] even smaller
[01:09:56] chip into the hotspot then you're
[01:09:58] actually helping to remove the
[01:10:01] the heat generated over there and of
[01:10:02] course then you have some additional
[01:10:04] thermal uh
[01:10:05] package that helps to uh extract that
[01:10:08] and diamond is
[01:10:09] actually the fastest extractor of the
[01:10:11] hit and the most efficient here
[01:10:13] so then the question the price could be
[01:10:15] even uh ten dollars per piece
[01:10:18] so the depends actually really on the
[01:10:20] size of such a synthetic time
[01:10:22] so so where is placed also yeah your
[01:10:26] points exactly right
[01:10:28] uh it's not a a large piece um so
[01:10:31] the way commercially we move is we go
[01:10:33] from
[01:10:35] corporate vias to gold uh
[01:10:38] graphite tapes uh and then to diamond uh
[01:10:42] we might be well on our way uh
[01:10:44] to using it and your point is absolutely
[01:10:46] right it's a very small piece of uh
[01:10:49] uh what you call industrial diamond or
[01:10:51] synthetic diamond right
[01:10:52] we are very excited what minters is
[01:10:54] trying to to do to bring they are very
[01:10:56] already
[01:10:56] very very active bringing interactive
[01:10:58] photonics we die among other
[01:10:59] applications we need to push the telecom
[01:11:01] market
[01:11:01] i fully agree with joanna it is not a
[01:11:03] cost definitely not a
[01:11:05] reason i'm going to read a question from
[01:11:07] the youtube channel all the way to my
[01:11:09] friend rajya i'm going to read a
[01:11:10] question from a fantastic company in
[01:11:12] florida called
[01:11:13] light path douglas aguirre is wondering
[01:11:16] in the b e r versus osnr plot
[01:11:20] that was shown you had a prefect for the
[01:11:23] bitter rate about one
[01:11:24] exponential minor 3 even with 30 db of
[01:11:27] osnr
[01:11:28] and he wonders what are the impacts of
[01:11:31] noise floor
[01:11:32] have on the operation of the links is it
[01:11:35] more difficult to compute linked
[01:11:36] margins for example uh
[01:11:41] no no no what we what we do is um
[01:11:45] the the um error floor
[01:11:49] is an indication of how some things are
[01:11:52] put together inside the system
[01:11:54] but what error floor the yeah i
[01:11:56] understand your point
[01:11:57] um the the uh the question is uh
[01:12:01] okay the well this is a much longer
[01:12:05] answer than i thought
[01:12:06] uh the bit error rate uh correction
[01:12:09] limit is one e minus two so if we were
[01:12:11] to look at the
[01:12:12] post effect system you'll find that the
[01:12:16] uh it's it's completely false uh it
[01:12:19] completely corrects itself at one e
[01:12:21] minus two
[01:12:22] uh thus uh floor uh about one e minus
[01:12:26] three floor uh these are the
[01:12:28] uh initial modules they've built the new
[01:12:30] one ones that we have
[01:12:32] uh do quite a bit more quite a bit
[01:12:34] better uh in our air floors
[01:12:37] but they don't go to like 10 e minus
[01:12:40] four or ten you minus five but they
[01:12:42] straddle between minus three and minus
[01:12:43] four
[01:12:44] i think that's a good that's a good
[01:12:47] answer then
[01:12:54] we're going to move to the next
[01:12:55] presentation we're going to go out to
[01:12:57] the beautiful wall of indian fossils
[01:12:58] we phosphide we have now two booming
[01:13:01] companies one on the
[01:13:03] united states one of the beautiful
[01:13:04] netherlands we go first to
[01:13:06] infinera gloria muchas gracias forest
[01:13:09] thank you very much for being with us
[01:13:10] this beautiful afternoon morning for you
[01:13:12] in pacific time
[01:13:13] the floor and the attention of everyone
[01:13:16] especially
[01:13:18] thank you jose let me uh navigate
[01:13:21] through here can you
[01:13:22] see my screen not yet almost here is now
[01:13:25] through the atlantic
[01:13:26] and it's here yes okay well you must
[01:13:29] mean
[01:13:30] we must need to use uh more infinite
[01:13:32] networks so
[01:13:35] good morning and good afternoon my name
[01:13:36] is gloria hoffler and i will be
[01:13:38] presenting on the integration
[01:13:40] competencies
[01:13:41] that we've developed at infinera to
[01:13:44] commercialize uh
[01:13:45] our our products um for optical network
[01:13:48] solutions so
[01:13:50] we believe that these technologies are
[01:13:53] relevant
[01:13:53] uh for uh to reach you know kobo
[01:13:56] objectives and plugable transceivers
[01:13:59] and as many of you already know we're an
[01:14:02] optical network
[01:14:03] solution company and our approach to
[01:14:06] developing optical networks
[01:14:08] has been a high degree of vertical
[01:14:10] integration
[01:14:13] so um
[01:14:18] um so we are organized um to to leverage
[01:14:20] um
[01:14:21] the synergies of integration across
[01:14:23] across the group so
[01:14:24] the group that i'm part of on the
[01:14:26] optical innovation center we develop
[01:14:29] asics dsap analog electronic designs
[01:14:32] highly customized drivers
[01:14:34] for the pics and the modules we also
[01:14:37] developed the modules but we also have a
[01:14:39] manufacturing lines
[01:14:41] for the packaging and the optical chip
[01:14:43] fabrication
[01:14:45] and over the years we've developed
[01:14:48] competencies in rf interconnects which
[01:14:50] are necessary to meet the demands for
[01:14:52] higher and higher data rates we believe
[01:14:55] that
[01:14:56] you know these um these technologies
[01:14:58] need to come together to understand the
[01:15:00] trade-offs and optimize the system level
[01:15:02] designs
[01:15:03] uh we we've heard about crosstalks uh
[01:15:05] thermal and electrical and so this is
[01:15:07] all really important to look at in the
[01:15:09] context of
[01:15:10] co-designing so ultimately uh
[01:15:13] the result of a high performance and
[01:15:16] more reliable
[01:15:17] optical engine really comes from from
[01:15:19] merging and looking at all these
[01:15:22] technologies uh together
[01:15:25] so um and and so these competencies
[01:15:29] um an integrated approach is what we
[01:15:31] believe is relevant for
[01:15:32] um for for the objectives of of
[01:15:34] performance and pluggable transceivers
[01:15:36] as well
[01:15:37] so today um i will share an example of
[01:15:40] our i6 engine
[01:15:41] and show how vertical integration
[01:15:42] enables uh leading system performance
[01:15:45] um we are um we're we drive uh
[01:15:49] innovative solutions so i will also
[01:15:51] highlight coherent point-to-point and
[01:15:53] optical networking in our foundry
[01:15:55] services
[01:15:57] um so infinera has pioneered the use of
[01:16:00] large-scale
[01:16:01] photonic integration for multi-channel
[01:16:03] transceivers since 2004
[01:16:05] and over multiple generations of
[01:16:07] coherent optics our picks have become
[01:16:09] much more complex
[01:16:10] uh requiring uh intimate integration
[01:16:12] with electronics
[01:16:14] to meet um the speed requirements and
[01:16:16] signal integrity
[01:16:18] so let me first start out by describing
[01:16:20] the architecture of our latest
[01:16:23] uh module which is our 1.6 terabit
[01:16:26] module it's a
[01:16:27] a two-channel uh pick uh cons
[01:16:30] each uh operating at 800
[01:16:34] gigabits per lambda so let me walk you
[01:16:36] through
[01:16:37] some of the details of the indian
[01:16:39] phosphi pick first i'd like to note that
[01:16:42] this is rx and tx functions
[01:16:45] integrated on one same chip each of the
[01:16:48] transmitter and receiver channels
[01:16:50] includes its own independent widely
[01:16:51] tunable laser with line widths less than
[01:16:54] 50 kilohertz
[01:16:56] and this is um so basically a total of
[01:16:59] four lasers um
[01:17:00] this allows for independent tuning of
[01:17:02] the rx and the tx channel
[01:17:04] and uh network configuration and
[01:17:07] uncompromising power
[01:17:09] we also have two sets of dual
[01:17:11] polarization iq modulators
[01:17:13] and two sets of dual polarization
[01:17:15] coherent receivers
[01:17:17] so for system advantages and because we
[01:17:19] can't do that in indian phosphi also
[01:17:22] we amplify and control by using power on
[01:17:25] chip
[01:17:25] soas and these are monolithically
[01:17:27] integrated at the
[01:17:29] output of each polarization stream
[01:17:32] um to support the high electro-optic
[01:17:36] bandwidth the pix
[01:17:37] had a matching high density um
[01:17:40] two-channel sigi asic 180
[01:17:44] nanometer with four modulator drivers
[01:17:47] and four
[01:17:48] transmission impedance amplifiers
[01:17:51] we deployed um fiber v-groove um
[01:17:55] arrays and arrays of polarization being
[01:17:56] connectors to couple the light out of
[01:17:58] the fiber
[01:18:00] so due to the high level of integration
[01:18:02] this transceiver requires significantly
[01:18:04] less assembly
[01:18:05] and uh less alignments
[01:18:09] than components needed for even a single
[01:18:12] channel transceiver using discrete
[01:18:13] functional components
[01:18:15] so i'm playing around with my controls
[01:18:17] here i apologize
[01:18:18] um both the uh
[01:18:22] the the pick and the a6 are co-packaged
[01:18:24] with free space optics and polarization
[01:18:26] control elements
[01:18:28] and have demonstrated bandwidth in
[01:18:29] excess of 55 gigahertz
[01:18:31] and a record of a hundred gigabytes
[01:18:34] using the 64 qualm
[01:18:35] operation and this was demonstrated a
[01:18:38] few years ago at ecoc
[01:18:40] we have uh productized this now into an
[01:18:43] 800 gigabits
[01:18:44] per lambda transceiver which is
[01:18:47] assembled into a
[01:18:49] dca optical module a dco
[01:18:53] with our latest 800 gigabit um or 800g
[01:18:57] dsp and uh the support electronics
[01:19:00] so the dsp asic incorporates also
[01:19:03] digital
[01:19:04] um nyquist of carrier generation
[01:19:06] probabilistic constellation shaping
[01:19:09] cross-channel coding via multiple
[01:19:12] dimensional modulations
[01:19:13] and channel forward error correction so
[01:19:16] we have already completed
[01:19:18] many live trials and various fiber
[01:19:22] types closing up to 970 kilometer
[01:19:24] lengths at 800
[01:19:26] g what i'm showing here is an example of
[01:19:29] the white stream live demonstration you
[01:19:31] may have seen some of our
[01:19:35] press release on that and you can see
[01:19:37] that the
[01:19:38] um receiver constellation is nice and
[01:19:42] clear
[01:19:42] we divide the spectrum into subcarriers
[01:19:45] and we have also
[01:19:46] demonstrated 800 uh gigabits that
[01:19:49] transition across
[01:19:50] a thousand kilometer link um using this
[01:19:53] uh dco
[01:19:54] so this product really brings out the
[01:19:57] benefits of vertical integration from
[01:19:59] the monolithic
[01:20:00] pic that is hybridly integrated with
[01:20:02] high density two-channel a6
[01:20:04] uh and co package with free space optics
[01:20:08] uh up to the uh assembly of um of the
[01:20:11] dsp and the
[01:20:12] uh into a dco now
[01:20:16] the uh commercial line that um that
[01:20:19] offers
[01:20:20] or that builds this uh this pick um has
[01:20:23] also been now
[01:20:25] um open uh to foundry uh
[01:20:28] service uh and founder capability and we
[01:20:30] did that by partnering with aim
[01:20:32] photonics
[01:20:33] to develop a pdk capability that was uh
[01:20:36] completed very much like our commercial
[01:20:38] products uh using a very system-centric
[01:20:40] approach um the platform is designed
[01:20:42] with electronics and packaging in mind
[01:20:44] so providing uses a turnkey access uh
[01:20:48] in reference designs uh both uh
[01:20:51] for the functions as well as for the uh
[01:20:54] electronics
[01:20:55] so it allows users to really focus on
[01:20:57] the prototype
[01:20:58] and a path to commercialization
[01:21:02] we did this you know also by developing
[01:21:06] design automation
[01:21:07] tools for simulation and layout
[01:21:11] and this minimizes transition to
[01:21:13] manufacturing
[01:21:14] we've worked very closely with
[01:21:16] simulation vendors um to generate
[01:21:18] compact models that are
[01:21:20] derived from commercial baselines if
[01:21:22] you're interested to hear more about the
[01:21:24] details of
[01:21:25] of what we're doing in this area we're
[01:21:27] going to be
[01:21:28] having a talk tomorrow um with at
[01:21:31] 8 am pdt pacific standard time it's an
[01:21:34] osa webinar
[01:21:36] jointly hosted with vpi photonics so
[01:21:39] welcome to
[01:21:40] to to hear some of what the things that
[01:21:42] we're doing there uh for
[01:21:43] for the foundry capability um
[01:21:47] i just want to say that uh along with
[01:21:49] with what we're doing there we have some
[01:21:51] ideas on how to transform network from
[01:21:53] point to point to
[01:21:54] point to multiple uh point uh improving
[01:21:57] operational efficiency cutting the
[01:21:59] number of transceivers in half and
[01:22:00] eliminating other in
[01:22:02] inefficiencies um and as you see on this
[01:22:04] graph we have a hub
[01:22:06] where all the data goes onto the
[01:22:07] aggregation point and then connects
[01:22:09] through a two end transceiver coherently
[01:22:11] to the edge
[01:22:12] of the network and then on to off key
[01:22:15] to the users however if the
[01:22:18] um aggregation point um
[01:22:22] could be you if we could use a plugable
[01:22:24] form factor gadget like an xr
[01:22:26] optics is uh what we're calling it with
[01:22:29] digital subcarrier technology that
[01:22:30] transmits
[01:22:31] uh from one point to many the network
[01:22:33] can be transitioned from 2n to 2n plus
[01:22:36] and uh offering multiple advantages
[01:22:40] so to sum it up a multi-point
[01:22:42] transceiver and pluggable
[01:22:44] format that enables 2x reduction in the
[01:22:46] number of transceivers and simplifies
[01:22:48] um the network and we've announced
[01:22:50] several partnerships in this area
[01:22:52] and along with all the work that i
[01:22:55] presented were all
[01:22:56] open to discussions and collaborations i
[01:22:59] hope that um
[01:22:59] i have shown technical competencies and
[01:23:01] need to come together
[01:23:03] to develop these leading edge
[01:23:04] performance through holistic co-design
[01:23:06] and monolithic integration
[01:23:08] so i just want to point out that the
[01:23:09] level of integration that is shown here
[01:23:11] is noble in that
[01:23:13] all the optical functions needed for
[01:23:15] coherent transmission and detection
[01:23:17] are optimized in and monolithically
[01:23:21] entered or embedded in a single indian
[01:23:23] phosphide chip
[01:23:24] that reduces the size and the complexity
[01:23:26] of the packaging and assembly
[01:23:28] and holistic integration in our view is
[01:23:31] really
[01:23:31] uh critical it's a fundamental for more
[01:23:34] innovation
[01:23:34] um but it's really critical because it
[01:23:37] enables um
[01:23:38] us to uh understand the trade-offs and
[01:23:41] um
[01:23:42] reduce the you know optimize um you know
[01:23:45] the the areas that
[01:23:46] that are critical and
[01:23:50] maintain a high level of signal
[01:23:52] integrity
[01:23:53] thank you thank you very much gloria it
[01:23:56] was a great presentation because
[01:23:58] you presented different ways of
[01:24:00] collaborating with
[01:24:01] our members we're with uh with the the
[01:24:03] people in this room
[01:24:04] so well if maybe you can summarize what
[01:24:07] kind of challenges
[01:24:08] would you like to what would you like
[01:24:11] our members to help you
[01:24:12] with sure so our challenges are always
[01:24:16] in the area of
[01:24:17] uh you know uh thermal uh crosstalk uh
[01:24:20] rf and you know rf integrity so thermal
[01:24:23] and electrical crosstalk are
[01:24:25] are usually challenges that we're faced
[01:24:26] with uh in the
[01:24:28] um on on the foundry side some of the
[01:24:31] challenges
[01:24:31] are you know uh in in the area of
[01:24:35] um interoperability interoperability of
[01:24:38] the
[01:24:38] um of the tools of the eda tools
[01:24:41] um and that's those are just some of the
[01:24:44] challenges that we face
[01:24:45] when we develop this products okay so
[01:24:48] maybe it's a good
[01:24:49] one yes because um greg from ficontech
[01:24:53] had a comment related to the packaging
[01:24:56] where
[01:24:56] i just right to the packaging and
[01:24:58] assembly so break a
[01:24:59] if you want and also i think you have a
[01:25:03] thank you yes
[01:25:06] uh no i don't actually have a slide
[01:25:08] ready on this and it was more like
[01:25:10] spontaneous expansion
[01:25:12] okay so well just a comment about the
[01:25:14] what is phi context doing and
[01:25:16] how can you have the packaging um we're
[01:25:19] listed in the in the packaging um
[01:25:21] technology area this is if you look at
[01:25:24] packaging as a as a set of the new
[01:25:27] individual processes it's probably
[01:25:29] nothing
[01:25:30] um that we have not seen already in the
[01:25:32] sum of our
[01:25:33] activities um we are of course
[01:25:36] um very interested in learning about
[01:25:39] where
[01:25:40] manufacturability of these units is on
[01:25:43] the timeline
[01:25:44] for the developers and we've heard about
[01:25:47] cost
[01:25:47] concerns and we've heard about issues
[01:25:51] these are all of course tied deeply
[01:25:53] together with the
[01:25:56] information that we'd be keen to hear
[01:25:58] about
[01:26:02] okay thank you very much greg and also
[01:26:04] we have a
[01:26:05] question from a from a wilfred from
[01:26:08] seuss micro optics
[01:26:11] do you want to go ahead with the
[01:26:12] question with friends yeah i just
[01:26:14] wondered if um
[01:26:15] what the reason why you opened up your
[01:26:18] you kind of opened up your foundry with
[01:26:20] a pdk do you
[01:26:22] are you looking for other applications
[01:26:24] besides data and telecom besides this
[01:26:27] classical application do you see other
[01:26:29] markets that are growing or emerging
[01:26:31] or i'm curious
[01:26:35] no it's it's a good question uh yes we
[01:26:37] are certainly
[01:26:38] uh opened uh open to other applications
[01:26:41] uh
[01:26:41] in fact um where we offer you know uh
[01:26:44] our expertise is not
[01:26:46] it's it's obviously in the in in the um
[01:26:49] telecom
[01:26:49] area but uh but this expertise is
[01:26:52] transferable
[01:26:53] uh at least on the um monolithic
[01:26:55] integration to other areas like sensors
[01:26:57] uh we see applications in defense as
[01:27:00] well as
[01:27:01] you know lighter for high volume
[01:27:04] here we go i i figured lighter or
[01:27:07] medical but lighter is probably
[01:27:09] i'm very curious about lidar because
[01:27:11] this is uh
[01:27:13] coming and going and i don't know
[01:27:17] how do you see the future of lidar in
[01:27:18] this context in this optical
[01:27:21] that would be interesting to know i
[01:27:24] think the last time i saw there was
[01:27:25] probably over 100 companies that were
[01:27:27] trying to do lidar
[01:27:29] so i think uh there's a lot of there's a
[01:27:32] lot of
[01:27:34] different approaches and but i think at
[01:27:37] the end of the day
[01:27:38] uh a lot of the lighter approaches that
[01:27:40] i have seen require high power lasers
[01:27:42] and um and that typically requires uh
[01:27:45] some level of integration within your
[01:27:47] phosphite
[01:27:48] so we see that as an up and coming okay
[01:27:51] thank you okay and we have another
[01:27:53] question for you gloria
[01:27:55] from chris from vpi photonics please
[01:27:57] would you like to
[01:27:58] make the question yeah hi gloria um
[01:28:01] this is chris from bpi um i just have
[01:28:04] one question i'm
[01:28:05] i'm curious um in terms of your open
[01:28:07] foundry
[01:28:08] um what type of support does infinera
[01:28:11] provide to your customers to get their
[01:28:13] um
[01:28:14] their ideas uh fabricated that's that's
[01:28:17] a good question uh chris
[01:28:19] thanks actually infinera works very
[01:28:21] closely with the users um what we've
[01:28:23] found in the past uh from the runs uh
[01:28:26] the founder runs that we have done is
[01:28:28] that a lot of the users that come to us
[01:28:30] uh really are system level users and so
[01:28:33] they
[01:28:34] um they welcome our support in the in
[01:28:37] in in the way of um integrating the pic
[01:28:40] into their system and we work closely
[01:28:42] with them in terms of providing the
[01:28:44] tools
[01:28:45] for the simulations as well as the
[01:28:48] controls that are necessary
[01:28:50] not just at the system level but also
[01:28:53] for the assembly
[01:28:54] so a lot of times um control loops uh
[01:28:58] that are required for uh for for
[01:29:01] managing the pick in
[01:29:02] in in the in the system um
[01:29:05] require integration of of of components
[01:29:08] that they
[01:29:09] have not thought through and so that's
[01:29:11] that's i think one of the areas where we
[01:29:13] uh we add value to our to our users
[01:29:17] okay and last thing i would like to
[01:29:18] introduce you a company maybe why not
[01:29:21] for a potential collaboration
[01:29:23] so i can see here a gravi from savik
[01:29:28] hello gravi would you like to introduce
[01:29:30] your company and maybe if i am not wrong
[01:29:31] again you have an slide to show to us
[01:29:34] i think you and i know i can do without
[01:29:36] slides no now
[01:29:38] indeed we make plastics that are used
[01:29:41] for the
[01:29:41] optical elements in optical transceivers
[01:29:44] so both from the transmitting as the
[01:29:46] receiving side and
[01:29:48] they're used because they are optically
[01:29:50] transparent but also have high
[01:29:51] dimensional stability
[01:29:53] at different temperatures and we're
[01:29:54] working on actually newer plastics for
[01:29:57] onboard optics
[01:29:58] that have even higher heat resistance
[01:30:00] that can withstand refill soldering or
[01:30:02] smt as we heard earlier
[01:30:05] and also even the polymers with the
[01:30:08] lower thermal expensive coefficients
[01:30:10] close to the substrate that these lenses
[01:30:12] are glued on
[01:30:13] so uh close to 20 ppm and actually
[01:30:17] we're looking for partners in the value
[01:30:18] chain to work with us to evaluate those
[01:30:21] materials and we can offer is obviously
[01:30:24] a global research center
[01:30:27] where we can do micro molding for
[01:30:28] optical elements we have
[01:30:30] multi-law analysis to help you moving
[01:30:32] away maybe from glass of german set to
[01:30:34] plastics
[01:30:35] and we have a global network for uh for
[01:30:38] micromotors
[01:30:39] yeah indeed after the presentation from
[01:30:41] effect photonics i would like to come
[01:30:42] back to gloria i would like to come back
[01:30:44] to
[01:30:45] to to understand a bit what are the
[01:30:47] needs for
[01:30:48] optics and micro optics for the
[01:30:49] packaging of transceivers because we
[01:30:51] have
[01:30:52] huge amazing technologies right now in
[01:30:54] epic and this is a huge potential market
[01:30:56] for them
[01:30:57] back to ana
[01:31:01] well no that's it if there are no more
[01:31:03] comments or questions for gloria i think
[01:31:05] that we move
[01:31:06] we can move ahead with the with the
[01:31:08] program and while talking about indian
[01:31:10] phosphite it's my pleasure to introduce
[01:31:12] you
[01:31:12] a team from effect photonics so tim
[01:31:15] if you want to share your screen yes
[01:31:18] thank you thank you very much
[01:31:19] um let me see if i can share the screen
[01:31:22] hold on there we go
[01:31:26] you should be seeing the screen now and
[01:31:28] also let me get the pointer options
[01:31:30] going so you know separately from some
[01:31:32] of my
[01:31:33] esteemed colleagues in this in this
[01:31:35] group i feel it's necessary to explain a
[01:31:37] little bit about what effect is and
[01:31:38] uh and what we do and where we come from
[01:31:41] um so i've taken
[01:31:43] carlos and jose's remit to to heart so
[01:31:47] the slides are in the format
[01:31:48] of who are we what are we doing and what
[01:31:51] are we looking for so
[01:31:53] let's uh let's take it away right so you
[01:31:56] should be able now to see
[01:31:58] the next slide
[01:32:01] yeah so effect photonics basically we
[01:32:04] are a transceiver vendor for dwdm
[01:32:08] transceivers and we're basing all of our
[01:32:10] module designs on our core technology
[01:32:13] which is monolithic integration of
[01:32:15] indium phosphide
[01:32:17] so what are we doing we're trying
[01:32:18] integrating lasers modulators photo
[01:32:21] detectors splitters
[01:32:22] but also network filter elements like
[01:32:24] awgs in our in our designs
[01:32:27] so we integrate all of the optical
[01:32:29] functionalities into one chip into one
[01:32:31] indium phosphide
[01:32:33] wafer we do that monolithically and that
[01:32:35] really allows us to scale into
[01:32:37] applications where the power and quality
[01:32:40] of high quality indian phosphide
[01:32:43] was not affordable yet so really taking
[01:32:46] the high performance that indian
[01:32:47] phosphate brings putting all of that on
[01:32:49] one wafer and then scaling that so
[01:32:52] we can use it at metro access and at
[01:32:54] some point also even for data center
[01:32:56] applications
[01:32:58] we combine that technology with the
[01:33:01] capability of
[01:33:03] non-hermetic optical packaging which is
[01:33:05] a key cost driver
[01:33:06] for pluggable modules and
[01:33:10] that together really brings a couple of
[01:33:12] very interesting benefits for
[01:33:14] for our modules onto the market so on
[01:33:16] the right we see
[01:33:18] one of our early chips this is an
[01:33:20] example of how much we integrated
[01:33:22] i think this must have been in 2016. so
[01:33:25] this is
[01:33:26] 10 tunable lasers that you see
[01:33:30] here together with a awg
[01:33:33] combiner optical amplification
[01:33:35] wavelength locking and many other
[01:33:37] functions on one
[01:33:38] on one die so we
[01:33:41] started in 2010 as a spin-out out of
[01:33:45] eindhoven university but
[01:33:46] here in eindhoven in the netherlands has
[01:33:48] been a strong background and heritage in
[01:33:50] diode lasers and integrated optics
[01:33:53] tracing all the way back to
[01:33:54] to philips optoelectronics we now have
[01:33:57] about 170 employees over
[01:33:59] three sites in the uk in the netherlands
[01:34:03] but also in taiwan
[01:34:04] i think one of the interesting points is
[01:34:06] that we are doing this indian phosphate
[01:34:08] technology
[01:34:09] fully fabulous so we own and we run
[01:34:12] our own pdk process development tool
[01:34:15] development epitaxial design
[01:34:18] but we run that through commercial
[01:34:19] volume values
[01:34:22] so okay that's who we are now what
[01:34:25] do we see as a next-gen vision
[01:34:29] so market segments for uh plugable
[01:34:31] modules now
[01:34:32] i apologize this is a gross
[01:34:34] oversimplification for how
[01:34:35] complicated the the internet really
[01:34:38] really is
[01:34:38] but in a nutshell we're seeing a
[01:34:42] four different parts of the market
[01:34:45] long haul metro and dci access and data
[01:34:47] center internet there's obviously
[01:34:49] ultra long haul and on the top left
[01:34:51] corner you would see
[01:34:52] um also things like onboard optics or
[01:34:55] maybe even
[01:34:56] even chiplets but in a nutshell you know
[01:34:59] in long haul
[01:35:00] your requested trend transmit data over
[01:35:03] thousands of kilometers so it may be a
[01:35:06] thousand or two thousand or maybe up to
[01:35:07] eight thousand
[01:35:09] and uh as a company for a particular
[01:35:11] product you
[01:35:12] may be able to sell ten thousands of
[01:35:15] those modules a year
[01:35:16] yeah maybe it's ten thousand maybe it's
[01:35:18] twenty five thousand maybe it's fifty
[01:35:19] thousand
[01:35:20] when you go into a successful metro dci
[01:35:22] application
[01:35:24] we're really now talking about hundreds
[01:35:26] of kilometers
[01:35:27] 300 500 we've heard discussions today
[01:35:30] about where the edge really stops
[01:35:32] but most of the applications and most of
[01:35:34] the module designs really top out at
[01:35:36] maybe several hundred k modules a year
[01:35:40] now there's obviously aggregate volume
[01:35:42] and there may be modules which are
[01:35:44] second source but
[01:35:44] roughly speaking that's correct now if
[01:35:46] you bridge up to
[01:35:48] the top left when we're talking about
[01:35:50] access that may be
[01:35:51] 10 kilometer 20 kilometers 30 kilometers
[01:35:54] but most of the applications are in the
[01:35:56] multiple millions a year so every time
[01:35:59] we go up we drop a factor of 10 in the
[01:36:01] transmission distance
[01:36:03] but we increase the volume that is
[01:36:04] needed to adequately fill that
[01:36:06] part of the market by the same amount so
[01:36:09] that's quite an interesting trend now if
[01:36:11] you want to go if you really want to be
[01:36:13] a
[01:36:14] high volume data center player yeah
[01:36:16] we're really talking about 10 20 30
[01:36:18] million pieces a year so
[01:36:20] the types of manufacturing the types of
[01:36:22] technology choices that you use
[01:36:24] really scale for these for these
[01:36:27] business cases
[01:36:28] now what we're seeing is that coherent
[01:36:30] is obviously a very powerful
[01:36:32] transmission technology
[01:36:33] brings many benefits in signal-to-noise
[01:36:36] ratio link distance and
[01:36:37] for some type of very long distance
[01:36:39] applications it really is the only way
[01:36:41] to get your data to the other side of
[01:36:42] the fiber
[01:36:43] now that means that up until now
[01:36:47] in 2020 and probably also for the next
[01:36:49] three to five years to come coherent is
[01:36:52] going to dominate
[01:36:53] you know long haul also long haul metro
[01:36:55] and and dci i mean that's in line with
[01:36:57] what we've seen today during the
[01:36:59] during the presentations and we believe
[01:37:01] that almost all of the links above
[01:37:03] arbitrarily 80 maybe even 40 kilometers
[01:37:06] will go to coherent in
[01:37:07] some way so what would it take
[01:37:11] to really move coherent into almost
[01:37:14] every part of fiber optic communication
[01:37:15] because now
[01:37:16] direct detect really is dominating the
[01:37:19] data center the ethernet and the access
[01:37:21] markets but what would it take to go to
[01:37:23] the next step
[01:37:25] so what we believe is it will take
[01:37:27] really a
[01:37:28] quantum leap a really next generation
[01:37:30] coherent
[01:37:31] to fully encapsulate almost all of the
[01:37:34] fibre optic links and bring coherent
[01:37:35] technology to that
[01:37:37] so you would need low power consumption
[01:37:40] low cost because you have to make it in
[01:37:42] massive volumes has to be high
[01:37:44] performance
[01:37:45] because you know the data rates of every
[01:37:47] single link length are going up they're
[01:37:48] going from
[01:37:49] you know 1g 10g 100g 400g we've been
[01:37:52] talking about 800 and beyond
[01:37:54] and that means that for an arbitrary
[01:37:57] link length
[01:37:58] if you want to go from 800 to 1.2
[01:38:00] terabyte you're going to hit the
[01:38:02] fundamental issue that coherent will be
[01:38:04] the only way to get your data to the
[01:38:05] other side
[01:38:07] so coherent will have to move up this
[01:38:10] diagonal and we'll need to move
[01:38:12] one or two magnitudes in production
[01:38:14] volume so what does that take well
[01:38:17] obviously it needs scalable optics
[01:38:19] and needs scalable electronics so in the
[01:38:21] format that
[01:38:22] jose has asked let's first look at which
[01:38:25] part of that puzzle
[01:38:27] we believe we have so um this is
[01:38:30] actually sort of a
[01:38:31] world first and uh happy to give the
[01:38:34] team at epic here the
[01:38:35] the scoop so in the long tradition of
[01:38:38] indian phosphide picks on a finger you
[01:38:40] may have seen the first part
[01:38:41] at the first part of this presentation
[01:38:44] that traces all the way back i believe
[01:38:46] to agility and or
[01:38:47] jdsu for the first tunable laser on a
[01:38:50] very well manicured finger we are
[01:38:53] continuing that noble
[01:38:54] tradition here with our latest addition
[01:38:58] to that
[01:38:58] uh to that technology so this is our
[01:39:02] single die monolithic multi-rate indian
[01:39:05] phosphate pick
[01:39:06] and based on this signal you can see it
[01:39:09] here it's on the
[01:39:10] ears on the finger and what's in it it
[01:39:13] has
[01:39:14] the coherent receiver it has the
[01:39:16] monolithic
[01:39:17] 700 kilohertz full band itla it has
[01:39:20] modulators but also it has optical
[01:39:22] amplification
[01:39:23] throughout also integrated wavelength
[01:39:25] locking and rf termination
[01:39:28] the interesting thing is this chip
[01:39:31] brings the full
[01:39:32] benefits that also gloria has discussed
[01:39:35] of
[01:39:36] indian phosphide integration high speed
[01:39:38] modulators we have
[01:39:39] also 50 or 53 gigahertz modulated
[01:39:43] bandwidth and
[01:39:44] similar pds we have high power i delay
[01:39:47] and it's all monolithic
[01:39:49] low loss on the chip that means that you
[01:39:52] can really cover um
[01:39:54] a wide range of applications we can
[01:39:56] configure this single chip
[01:39:58] into sort of a low power short range
[01:39:59] mode but also a high ocenar
[01:40:03] long distance mode and almost anything
[01:40:06] in between we can hook it up to 100g and
[01:40:08] 600g electronics
[01:40:10] and really use this chip design which we
[01:40:12] have invested
[01:40:13] the last you know decade of process
[01:40:16] technology
[01:40:17] development into and really make a
[01:40:19] massive
[01:40:20] range of different modules happen now
[01:40:24] the optics package that we've designed
[01:40:26] around it because we put everything in
[01:40:27] the chip the optics package actually
[01:40:29] becomes
[01:40:30] i would almost say deceptively simple so
[01:40:33] we can fit it into
[01:40:34] sort of the smallest but also the
[01:40:35] largest module so this optical package
[01:40:38] we can fit into a qcp 28 the qsf pdd
[01:40:41] but we can also easily put it into the
[01:40:43] incumbent standards like cfp2 and cfp
[01:40:46] right now
[01:40:47] so we really believe that to get to a
[01:40:49] million coherent
[01:40:51] anything you really need to have a
[01:40:53] monolithic semiconductor technology and
[01:40:55] just
[01:40:55] run lots and lots and lots of wafers
[01:40:57] that's the only way we believe we can
[01:40:59] get
[01:41:00] to a million or 10 million coherent
[01:41:03] modules being sold
[01:41:05] so we're working on this right so this
[01:41:07] is the first prototype it's sort of a
[01:41:09] global scope we've put a press release
[01:41:11] out
[01:41:11] early this year that we were working on
[01:41:13] this chip and now it's it's finally here
[01:41:15] and
[01:41:15] the real work starts but what do we need
[01:41:19] so if we go to the next slide and i
[01:41:23] have happily taken some of the people's
[01:41:27] products in this in this group and put
[01:41:29] them on the slide so
[01:41:32] really to get coherent to work through
[01:41:35] all ranges and all volume
[01:41:37] amounts you know we believe that we have
[01:41:38] the optics technology that can scale
[01:41:40] very well
[01:41:41] on terms of power and link distance but
[01:41:44] also with volume
[01:41:45] but it's quite obvious that pairing that
[01:41:47] with what's in a sense
[01:41:48] uh long-haul dsps even if they're very
[01:41:51] capable and
[01:41:52] of the seven nanometer type uh that
[01:41:54] we're currently using
[01:41:56] really doesn't get us to the point where
[01:41:58] we can really say okay it's going to be
[01:42:00] a million or 5 million or 10 million
[01:42:02] that's being used
[01:42:03] in all of the access networks and all of
[01:42:05] the data center and obviously this will
[01:42:06] be a couple of years out right this will
[01:42:08] be 2025 2026 and onwards
[01:42:11] so really what we are looking for as
[01:42:13] effect photonics
[01:42:15] where will the next you know super high
[01:42:18] volume
[01:42:19] next generation electronics both on the
[01:42:21] signal processing
[01:42:22] but also on the electronics where will
[01:42:24] that come from
[01:42:26] and which partners are really going to
[01:42:28] jump
[01:42:29] into this part of the market uh to
[01:42:31] support the optics that are now scaling
[01:42:34] towards that part of the market so i
[01:42:36] tried to encapsulate the
[01:42:38] the narrative a little bit uh jose but
[01:42:40] that that really is in a nutshell
[01:42:42] our presentation you know we know each
[01:42:44] other for a long time i think it's like
[01:42:46] 15 years
[01:42:47] i i would like first of all to say
[01:42:49] congratulations you are one of the
[01:42:51] founders of effect photonics
[01:42:52] last summer you became the cto i love
[01:42:55] when great things happen to great people
[01:42:57] and you are making a big difference on
[01:42:59] the european ecosystem
[01:43:01] playing with the big guys and i'm amazed
[01:43:03] i'm amazed by what you have achieved
[01:43:05] in the last 10 years so i would like
[01:43:08] effect photonics is the jewel
[01:43:10] and one thing that i love with the with
[01:43:13] the effect photonics in epic yogurts
[01:43:14] here looking for partnerships looking
[01:43:16] for
[01:43:17] potential people that can help you so i
[01:43:19] would like to as i promised in the
[01:43:21] previous presentation i would like to go
[01:43:22] back to
[01:43:23] the three companies in five infinite
[01:43:26] effect photonics and ask you
[01:43:27] two questions the first one need for
[01:43:30] micro optics in packaging
[01:43:31] the same kong one assembly and packaging
[01:43:34] needs in the supply chain and i'm gonna
[01:43:37] start with the cto of the photonics mr
[01:43:39] tim yeah thank you so
[01:43:42] really what we are looking for is you
[01:43:45] know every time you go into making
[01:43:47] something
[01:43:47] ten times as much as you're used to
[01:43:49] making it before
[01:43:51] almost all of the supporting
[01:43:52] technologies have to scale up to the
[01:43:54] next level right so
[01:43:55] it's all well and good to handcraft
[01:43:57] under a microscope your first gold box
[01:43:59] packages with
[01:44:00] manual assembly when you're making a
[01:44:02] thousand a year
[01:44:03] that doesn't really scale when you're
[01:44:04] making ten thousand a month or maybe a
[01:44:06] hundred thousand a month so
[01:44:08] from packaging technologies that really
[01:44:10] scale to for example a global network of
[01:44:13] contract manufacturers
[01:44:14] or from um uh
[01:44:18] you know as a sidebar one of the reasons
[01:44:20] why the dsps are now so expensive is
[01:44:22] that
[01:44:23] um you know taping out the seven
[01:44:24] nanometer dsp for a fifty thousand or a
[01:44:27] hundred thousand
[01:44:28] uh a year uh runway really is not a
[01:44:31] whole lot of silicon
[01:44:32] to be selling to offset the you know
[01:44:35] development cost so
[01:44:37] in optics and telecommunications we're
[01:44:39] really still using a high amount of
[01:44:41] very dedicated and in terms of silicon
[01:44:45] still relatively low volume components
[01:44:48] to support
[01:44:49] the tias the drivers signal processors
[01:44:52] but also things like custom ceramics and
[01:44:55] you know to really make a massive impact
[01:44:57] with the
[01:44:58] telecommunications and datacoms industry
[01:45:01] on sort of the global supply chain of
[01:45:03] pcps metalworks housing and silicon
[01:45:07] yeah we really have to graduate to you
[01:45:09] know the next
[01:45:10] the next step up so when the epic
[01:45:12] members see a slide like this they get
[01:45:13] really happy because they see
[01:45:15] a chip there and you talk about the
[01:45:17] optical input outputs you talked about
[01:45:19] the rf input outputs and you see a lot
[01:45:21] of
[01:45:22] active components packaged in a reliable
[01:45:24] way
[01:45:25] could you share with us maybe one two
[01:45:27] challenges on optics fibers packaging
[01:45:30] that
[01:45:30] you know maybe the people from epic can
[01:45:32] help you know for this product because
[01:45:34] it's already
[01:45:35] fully mature manufactured and user can
[01:45:37] be packaged in any form factor
[01:45:38] maybe for the 3.0 generation
[01:45:42] so you know what you see on the chip uh
[01:45:45] here
[01:45:46] um is you know you integrate more and
[01:45:48] more
[01:45:49] functionality into a single chip which
[01:45:51] also means that you have to have
[01:45:52] more and more electrical inputs in and
[01:45:54] out so you know the arrows don't really
[01:45:56] do it justice there's rfin and rf out
[01:45:58] but actually it's
[01:45:59] you know a couple of hundred contacts
[01:46:01] that need to go in and out and
[01:46:03] you know up until now wire bonding has
[01:46:05] been the
[01:46:06] idea of choice i know that uh infinera
[01:46:09] has been doing very well in flip
[01:46:10] shipping
[01:46:11] but i think getting to the maturity of
[01:46:13] flip shipping
[01:46:14] that electronics has for for optics
[01:46:17] packaging and really making that
[01:46:19] standard rather than the exception is
[01:46:21] something that we would
[01:46:23] really be interested in collaborating
[01:46:24] with excellent you're gonna get a lot of
[01:46:26] emails after this remember any company
[01:46:28] that wants to introduce i will make an
[01:46:30] introduction an email introduction
[01:46:31] gonna go now to gloria infinera same
[01:46:33] question goes to you
[01:46:34] a couple of challenges room for a
[01:46:36] collaboration in companies on packaging
[01:46:38] assembly and micro optics
[01:46:43] we have you muted and we really want to
[01:46:45] hear you yes
[01:46:47] okay um oh okay thank you for unmuting
[01:46:49] me
[01:46:50] i i think uh tim has really highlighted
[01:46:53] already uh
[01:46:54] the the key key areas for
[01:46:57] collaboration um i think it's important
[01:47:00] to note that there are emerging
[01:47:03] applications uh in indian phosphide
[01:47:05] beyond data common telecom so in the
[01:47:07] sensor area like
[01:47:09] lighter and there the challenge in the
[01:47:11] packaging would be
[01:47:12] quite different uh you have um
[01:47:16] environments that are uh very harsh and
[01:47:19] the cost has to be also competitive with
[01:47:22] uh
[01:47:22] with the application so finding uh you
[01:47:25] know either
[01:47:26] optics or free space optics whatever the
[01:47:28] solution is that is
[01:47:29] uh reliable uh under the environment
[01:47:33] of operation probably rf is not as
[01:47:36] important in that in that sense
[01:47:38] but the optics are gloria in the epic
[01:47:40] family we have companies that can do
[01:47:41] wafer level micro optics and it's a big
[01:47:43] thing with companies like access
[01:47:46] nanoscribe i apologize because i could
[01:47:48] talk for so so
[01:47:49] and we could mention so so many do you
[01:47:51] see a a
[01:47:53] challenge in the need of a low-cost
[01:47:56] volume production wafer level
[01:47:58] micro-optics for the co-package optics
[01:48:00] so just let's just take lidar for
[01:48:03] example because
[01:48:04] you know that's one of the things that
[01:48:05] we that i had mentioned earlier and that
[01:48:07] in that area not only do you have
[01:48:09] um you know uh very uh stringent
[01:48:12] requirements
[01:48:13] on on the environment uh operation
[01:48:15] environment very high temperatures you
[01:48:17] also have very high powers
[01:48:19] so uh so the packaging needs to be low
[01:48:22] cost but it also needs to be
[01:48:23] competitive with uh high power
[01:48:26] applications so
[01:48:27] you don't want to spend a lot of money
[01:48:29] on your chip making you know the photons
[01:48:31] and then
[01:48:32] losing them in the coupling process so
[01:48:34] um
[01:48:35] so that i think that is the challenge
[01:48:36] there the thermal and optical
[01:48:38] uh coupling if you this into you could
[01:48:41] be prepared because i'm gonna introduce
[01:48:42] a lot of potential
[01:48:43] collaborators from the u.s from europe
[01:48:46] from china from all over the world that
[01:48:47] they want to work with you
[01:48:49] and i really also want to work with you
[01:48:51] gloria and i go
[01:48:52] to rad i go to infi also the same
[01:48:55] question goes to you a couple of
[01:48:56] challenges on the micro optics
[01:48:58] on the wafer level optics on the fiber
[01:49:00] on the packaging for info
[01:49:02] i actually sent a uh just sent a private
[01:49:05] uh
[01:49:06] chat to uh the gentleman from savage
[01:49:09] uh yes we we are interested in
[01:49:11] [Music]
[01:49:13] in high temperature lens elements so
[01:49:16] um yeah we'd like to take a look at that
[01:49:19] fantastic this is going when the magic
[01:49:21] happens at epic i hope
[01:49:23] i i i didn't i there wasn't a slide but
[01:49:26] uh you had uh briefly mentioned about uh
[01:49:30] packaging uh for really high temperature
[01:49:33] uh applications um you know ours is not
[01:49:37] probably not
[01:49:38] i don't know how how high we're gonna go
[01:49:40] but anything in the range of
[01:49:42] um 80 to 120 would be of interest
[01:49:47] gallery i think that's something that i
[01:49:48] definitely can do yeah that's not a
[01:49:50] problem we can go all the way up to 260
[01:49:53] for reflo soldering yes oh
[01:49:56] yes operational temperature and this
[01:49:58] assembly assembly 260 would be good
[01:50:00] yeah 260 is about the number it's um
[01:50:04] you know it's one of those uh leadless
[01:50:07] tin based solder reflows
[01:50:09] 260 would be about i continue looking
[01:50:11] for partnerships i'm going to go to
[01:50:12] belgium i'm going to go to a skylane
[01:50:14] optics philippe how are you doing today
[01:50:17] fine on you
[01:50:18] i'm looking for friendships and
[01:50:20] collaborations you make fantastic
[01:50:22] transceivers say
[01:50:23] what kind of cooperations you see in the
[01:50:25] field of packaging assembly micro optics
[01:50:27] etc
[01:50:28] you know we work on interoptics you know
[01:50:30] we try to interoperable optics and i
[01:50:33] like to listen to all the presentations
[01:50:34] as you know
[01:50:35] a few years ago there was only a few
[01:50:37] manufacturers of dsp in the market
[01:50:40] and today there are more than 7 dsp
[01:50:42] manufacturer in the world and as you
[01:50:43] know
[01:50:44] everybody is somewhere proprietary that
[01:50:47] means
[01:50:47] my question that everybody wants to work
[01:50:49] together everybody wants to reduce the
[01:50:51] cost
[01:50:52] of the high speed optics but everybody
[01:50:54] is making his own dsp
[01:50:57] then i don't understand very well who
[01:51:00] wants to address that
[01:51:02] why is everybody making their own
[01:51:04] electronics their own dsp
[01:51:06] team you're smiling i think you have a
[01:51:09] vision for that because we discussed
[01:51:10] this over a decade ago when you were at
[01:51:12] the
[01:51:13] when we're together at t you are in
[01:51:14] dublin what is what is the
[01:51:16] the reason behind everybody going all
[01:51:19] the way or making the wrong electronics
[01:51:22] i'm not super sure if i'm perfectly
[01:51:24] qualified to answer that i think there
[01:51:26] are
[01:51:26] better colleagues in this group as well
[01:51:29] but what we are seeing is that
[01:51:32] especially when you are trying to make
[01:51:34] uh transceivers
[01:51:35] and you're working together with system
[01:51:37] integrators how to differentiate what
[01:51:39] you're offering into the market is very
[01:51:41] important and
[01:51:42] you know the dsp is a very strong
[01:51:45] ability to
[01:51:46] differentiate your offering but that
[01:51:48] almost goes directly against what the
[01:51:50] market needs which is the same modules
[01:51:52] which work together
[01:51:53] um and uh you know which are easy to buy
[01:51:57] and uh and cheap so it
[01:52:00] a large part of it has to do with sort
[01:52:02] of supplier and demand
[01:52:05] dynamics that doesn't explain anything
[01:52:07] but yeah i mean
[01:52:08] even from a module uh developer like
[01:52:11] ourselves
[01:52:12] the more uh the more we can find a dsp
[01:52:16] that works for most of our cases
[01:52:18] the more we can drive down the
[01:52:20] semiconductor volume
[01:52:21] curve with our optics and likewise with
[01:52:24] all of the electronics to
[01:52:25] follow yeah i do believe that there is
[01:52:28] just not a generic
[01:52:29] solution for everyone it's not available
[01:52:31] in the market if it was you would be the
[01:52:33] first one
[01:52:34] all of you would be the first one to to
[01:52:35] get it philip i'm looking forward to
[01:52:38] help you more
[01:52:39] with more potential partnerships skyline
[01:52:41] is a fantastic
[01:52:42] company but we have to move on because i
[01:52:44] want to feel something
[01:52:45] and i want to go to one key company in
[01:52:47] the ecosystem people is a great friend
[01:52:49] of epic and cisco
[01:52:50] is very important for us thank you very
[01:52:52] much people for
[01:52:54] being with us this beautiful afternoon
[01:52:56] hey uh
[01:52:57] good morning and good afternoon
[01:52:58] everybody i'm vipul patel from cisco
[01:53:01] uh and i prepared this presentation in a
[01:53:04] very short notice
[01:53:05] so i'm mainly going to touch upon some
[01:53:08] of the challenges that the next
[01:53:09] generation optics
[01:53:11] will face from product development point
[01:53:13] of view uh and with that
[01:53:15] if you can go to the next slide list
[01:53:18] sorry sorry i'm
[01:53:19] almost done okay no problem no problem
[01:53:23] at all don't worry
[01:53:24] now
[01:53:28] i think while you're changing the slide
[01:53:30] let me give you little history of
[01:53:31] silicon photonics also
[01:53:32] um in 87 around
[01:53:36] you know mid 80s richard sorry from air
[01:53:38] force
[01:53:39] developed some of the initial concepts
[01:53:41] uh unfortunately no soi was available at
[01:53:44] that time so it was based on adobe
[01:53:45] silicon waveguide
[01:53:46] very high loss real work and silicon
[01:53:49] photography started in early 2000
[01:53:51] and fortunately or unfortunately i got
[01:53:53] stuck with the fields since very early
[01:53:55] days of silicon photonics
[01:53:58] so what is possible with silicon
[01:54:00] photonics are what has been demonstrated
[01:54:02] so far um boardrooms up to 50 gigabit
[01:54:06] uh gigabot are already there in the
[01:54:08] products now
[01:54:10] um the screen is perfectly working
[01:54:15] people you come from yes it is working
[01:54:18] thank you
[01:54:20] so going to 100 gigabyte in silicon
[01:54:22] photonics uh
[01:54:24] from uh at least uh technical point of
[01:54:27] view
[01:54:27] uh it's achievable uh one of the promise
[01:54:30] of silicon photonics is that it allows
[01:54:33] a massive parallelism that you can do
[01:54:35] optics now on a silicon so
[01:54:37] you can add as many uh fiber links as
[01:54:40] you want
[01:54:42] and if you look at it one of our you
[01:54:44] know
[01:54:45] company that we acquired luxetera has
[01:54:47] done psm4
[01:54:48] uh light wire when we acquired we did
[01:54:51] also 10 by 10
[01:54:52] cpac module at that time
[01:54:56] uh wdm dwdm uh intel
[01:54:59] has done cwdm four best products uh we
[01:55:03] from lightweight team also did a dwdm
[01:55:06] product that was a
[01:55:07] cpec 100 gig lr4 uh
[01:55:10] modulation pamphlet modulation is very
[01:55:12] easily achievable
[01:55:14] and we have at the ofc 2019 demonstrated
[01:55:18] a single lambda m4 for 100 gig 50
[01:55:21] gigabyte
[01:55:23] um and and going to bam 8 also is
[01:55:26] possible
[01:55:27] on a px side it's rx becomes most
[01:55:28] challenging coherent you heard already
[01:55:31] from infi
[01:55:32] uh acacia that we are in process of
[01:55:34] acquiring also in that field and nokia
[01:55:36] also recently acquired uh elneon that
[01:55:39] also is in that space and now
[01:55:42] one of the discussion that happens
[01:55:45] or where silicon photonics comes up very
[01:55:47] often
[01:55:48] is in package optics or a copy package
[01:55:51] optics or a kobo
[01:55:53] so with that if you go to the next slide
[01:55:58] please
[01:56:01] so if you look at the history of silicon
[01:56:04] photonics again
[01:56:04] as i mentioned uh i have been working in
[01:56:07] it for two decades now
[01:56:08] lots of people started at that point
[01:56:12] and you heard from other players in this
[01:56:14] field also that
[01:56:15] major is a company it's a you know
[01:56:18] multi-multi-year of uh
[01:56:20] r d uh before a product can be realized
[01:56:23] so why is that one of the major major
[01:56:26] challenges
[01:56:27] that it's all you know co-designed
[01:56:30] you cannot invent things in a silo in
[01:56:33] silicon photo oxide
[01:56:35] just to give you an example uh the
[01:56:37] photonic ic that is in the blue here
[01:56:39] the two main components there are
[01:56:40] modulator and a photo detector one off
[01:56:42] for transmit one of them is for receive
[01:56:44] and then you have all the waveguides
[01:56:45] throughout signals and everything else
[01:56:48] and every company that is in silicon
[01:56:50] photonics right now
[01:56:51] has a individual pdk of their own and
[01:56:54] their own fab engagement so no
[01:56:56] two companies use identical fab process
[01:56:58] or have identical pdk elements
[01:57:03] so that doesn't mean that you cannot
[01:57:06] achieve almost every company can achieve
[01:57:08] the functionality that is needed once
[01:57:09] the light is
[01:57:10] in the silicon the biggest challenge is
[01:57:13] how to
[01:57:14] couple fibers to that silicon phonomic
[01:57:17] ic and how to get the laser light in
[01:57:19] while keeping the thermal electrical and
[01:57:22] all the mechanical aspects right
[01:57:24] intact and and that seems to be a major
[01:57:28] major challenge
[01:57:29] um now
[01:57:32] that is still for a trend you know uh uh
[01:57:36] for plugable optics that's still
[01:57:37] relatively simple
[01:57:39] uh because you are you know not
[01:57:41] constrained with the packaging that
[01:57:43] needs to be done for a you know board
[01:57:45] level packaging or
[01:57:46] smt level the more standard electronic
[01:57:48] packaging you can
[01:57:50] customize your packaging but when you
[01:57:52] take this situation and go closer to the
[01:57:54] ip or in package optics
[01:57:57] or also co-package optics what is known
[01:57:59] as now you are
[01:58:00] putting optics close to a a
[01:58:04] very large-scale heater uh in in that
[01:58:07] well they don't talk about
[01:58:08] power dissipation in watts but in
[01:58:10] hundreds of watts and sometimes
[01:58:12] in kilowatts when you especially talk
[01:58:14] about a 50 terabit or 100 terabit
[01:58:16] systems
[01:58:19] so now optics for example laser will
[01:58:21] fall off
[01:58:22] uh the cliff if you have to you know go
[01:58:25] beyond 90 degree
[01:58:27] centigrade temperature there same thing
[01:58:29] we talked about epoxies and everything
[01:58:31] else
[01:58:31] so a lot of those things uh the optical
[01:58:35] component uh reflowable compatible
[01:58:37] optical components uh epoxies a lot of
[01:58:40] those things
[01:58:41] plays a major major role and you need to
[01:58:43] know those constraints
[01:58:45] when you invest in your photonic ic
[01:58:47] platform so
[01:58:48] that's what makes it very very
[01:58:49] challenging so if you look at it the
[01:58:52] industry-wise challenge and one of the
[01:58:55] good things that epic is doing is
[01:58:56] bringing everybody together in a
[01:58:58] consortium way
[01:58:59] and and so common roadmaps and common
[01:59:01] challenges can be identified and people
[01:59:03] can work collaboratively
[01:59:04] so if you go to the next slide please
[01:59:09] so what are the main challenges i
[01:59:10] mentioned there is no standardization
[01:59:12] here each company
[01:59:13] has its own pdk own process flow
[01:59:17] same thing applies to light sources
[01:59:19] that's a very long and expensive
[01:59:21] development cycles for lasers
[01:59:23] and these lasers are also custom in many
[01:59:25] cases
[01:59:26] designed for a very specific silicon
[01:59:29] photonics
[01:59:30] another uh item that was discussed here
[01:59:33] was on a
[01:59:34] dsp side so why there are no you know
[01:59:37] why there are seven companies and and
[01:59:39] why there is no economy of scale
[01:59:41] because everybody asic has two
[01:59:43] components one of them is the service
[01:59:45] and
[01:59:46] and cdr and the other side and that can
[01:59:49] be
[01:59:49] probably standardized but the interface
[01:59:52] for silicon photonic ic is always custom
[01:59:55] and at least unless the pdks become
[01:59:57] common right that
[01:59:58] that part the tia interface the
[02:00:01] modulated driver interface and sense and
[02:00:03] control interfaces has to be custom
[02:00:06] now doing any customization on a 7
[02:00:09] nanometer node
[02:00:10] we all know how expensive it is so
[02:00:12] that's where the economy of scale
[02:00:14] uh or the cost margin reduction right
[02:00:17] is challenging in this field uh same
[02:00:20] thing
[02:00:21] happens to oset and uh especially for
[02:00:24] you know either wire boundary or cheap
[02:00:26] on wafer assemblies
[02:00:27] but again there are everybody is a low
[02:00:29] volume custom flow in a high volume
[02:00:31] facility
[02:00:32] and on assembly side also there are
[02:00:34] multiple models that uh at least we have
[02:00:36] still in the industry
[02:00:37] lots of people have their own cms
[02:00:41] contract manufacturers there are some
[02:00:43] joint development type of things that we
[02:00:45] have seen also in the industry
[02:00:46] but again there is no standardization
[02:00:48] even from equipment making point of view
[02:00:50] and that is a must so in summary what i
[02:00:53] can say is that supply chain steel for
[02:00:54] silicon photonics in infancy
[02:00:57] and we have to work collectively in many
[02:00:59] many areas
[02:01:00] to overcome some of these things and and
[02:01:03] standardize has to take place if
[02:01:05] we truly want to realize the potential
[02:01:07] of silicon photonics
[02:01:08] which you know when you go to very high
[02:01:11] density
[02:01:12] optics or when you have to make an
[02:01:15] optics that is a higher temperature
[02:01:17] comparable
[02:01:18] that will become the need of the hour so
[02:01:21] with that
[02:01:22] i would like to summarize this article
[02:01:24] uh the summarize this presentation if
[02:01:26] you can go to the next slide please
[02:01:32] so it is identified silicon phononics is
[02:01:34] one of the key enabler for scaling
[02:01:36] you know both the massive skill data
[02:01:38] centers as well as the coherent side
[02:01:40] uh it's a fast growth from you know
[02:01:43] aggregated
[02:01:44] growth point of view uh the
[02:01:46] heterogeneous
[02:01:47] integration meaning that wirebond
[02:01:49] solutions are disappearing or going away
[02:01:51] and people are going more and more
[02:01:53] towards
[02:01:53] doing a flip chip type of packaging
[02:01:55] between the electrical ics and the
[02:01:56] optical ics
[02:01:58] there is very strong momentum every
[02:02:00] company has a roadmap
[02:02:02] as a cisco we presented our roadmap to
[02:02:04] our
[02:02:06] investors in wall street and in fact if
[02:02:09] you do a search with
[02:02:10] one silicon strategy of cisco i think it
[02:02:12] was present last november or december
[02:02:15] and even in our quarterly
[02:02:18] reports optics has been identified as
[02:02:20] one of the the key growth area for
[02:02:22] and a focus area for cisco
[02:02:25] but again in order for us to realize
[02:02:28] that
[02:02:30] new silicon phononic ecosystem has to be
[02:02:32] developed and
[02:02:33] any company here who has a newer
[02:02:36] technology that can help
[02:02:37] leapfrog that and want to make a
[02:02:39] presentation to cisco
[02:02:40] you guys can reach me out and
[02:02:44] you know the focus also for long-term
[02:02:47] development point of view so one of the
[02:02:48] thing
[02:02:49] that
[02:02:53] was mentioned in previously right
[02:02:54] that was mentioned in previously right
[02:02:54] that you have to intercept the market at
[02:02:56] the right point also
[02:02:57] as i mentioned these are very very long
[02:02:59] development cycles the investment you
[02:03:00] make today
[02:03:02] has to meet a market uh intercept point
[02:03:05] has to be well defined and that can be
[02:03:07] five to seven years down the road
[02:03:09] so uh co-packaging and optics probably
[02:03:12] fall into that category that
[02:03:14] it's an imminent it is the need of the
[02:03:17] hour
[02:03:18] and people have to start thinking
[02:03:19] seriously about that because
[02:03:21] you know it this is very very long
[02:03:23] development cycle so if you if you don't
[02:03:24] make
[02:03:25] right bets then it's very difficult to
[02:03:28] catch up
[02:03:30] uh with that if you have any questions
[02:03:33] i'm
[02:03:33] more than happy to answer thank you very
[02:03:36] much this guy
[02:03:37] the people it's always a pleasure to
[02:03:39] have a you have to have you and to have
[02:03:41] cisco in our meetings
[02:03:43] and well you already answered the
[02:03:45] questions because you presented
[02:03:48] yes in this slide we can see
[02:03:51] several challenges and several a lot of
[02:03:54] room for collaboration with our members
[02:03:56] so maybe we can focus here on your
[02:03:59] yes on your code design and what your
[02:04:02] needs in
[02:04:03] in design because we have here in the
[02:04:05] room
[02:04:06] a company doing design tools a optiwave
[02:04:10] and would you like maybe rather would
[02:04:12] you like to comment
[02:04:13] i think you have a slide right to show
[02:04:16] you about the
[02:04:17] what are the tools that you offer
[02:04:21] yes hi uh everyone and let me share the
[02:04:24] screen here
[02:04:38] okay so if you can go to presentation
[02:04:40] mode well if not it's okay
[02:04:42] okay here we go and
[02:04:46] one more slide here
[02:04:53] okay so yeah so just quickly to
[02:04:57] to go through what optiwave offers
[02:05:00] as a system tool which can help like
[02:05:03] system companies or designers
[02:05:07] manufacturers of the
[02:05:08] of these transceivers is the
[02:05:11] the wide range of components and models
[02:05:15] which can be implemented in in the
[02:05:18] software which uh this software would be
[02:05:20] called optisystem simulation tool and it
[02:05:23] we offer in this simulation tool many uh
[02:05:26] like all the components needed for
[02:05:28] designing the transceiver from the light
[02:05:30] sources
[02:05:31] uh to the modulators to
[02:05:34] the uh we offer like for example
[02:05:37] different types of
[02:05:38] uh format which can be used uh to to
[02:05:41] modulate the lasers
[02:05:42] uh the also the for example the dsp
[02:05:46] component which is one of the things
[02:05:47] which uh what was discussed here
[02:05:49] we offer different uh algorithms in the
[02:05:52] dsp which can be used to
[02:05:54] to process the the data uh
[02:05:57] also we offer the uh um like some
[02:05:59] algorithms for forward earth effect
[02:06:02] corrections different types of detectors
[02:06:04] the amplifiers
[02:06:06] from semiconductor amplifiers can put
[02:06:08] them on the cheb to
[02:06:09] like external amplifiers web guide
[02:06:12] amplifiers
[02:06:13] plus the passive components from max's
[02:06:16] demoxes a red we have guides attenuators
[02:06:18] so all of these components can be put
[02:06:20] together
[02:06:21] in a module let's say and
[02:06:24] simulated for for your design in
[02:06:26] principle
[02:06:28] and at the end of the day we offer all
[02:06:31] these kind of visualizers
[02:06:32] from optical electrical uh from light
[02:06:36] wave analyzers
[02:06:37] uh vrs constellation diagrams where you
[02:06:40] can monitor your signal anywhere in the
[02:06:42] design and then you can manipulate the
[02:06:43] parameters of these components and
[02:06:45] be able to see for example synthesize
[02:06:48] your
[02:06:49] uh your chip when you design it or if
[02:06:51] you wanted to design
[02:06:53] certain chip you can look into the
[02:06:56] what what parameters require to achieve
[02:06:58] certain functionality so all of these
[02:07:00] components is available to to be
[02:07:03] addressed in
[02:07:04] in a unidirectional or a bi-directional
[02:07:07] uh
[02:07:07] capability of of the simulation tool so
[02:07:10] with this like uh again is is very short
[02:07:13] presentation here one minute to
[02:07:15] to finish and uh um we are open to
[02:07:18] uh collaborate with system companies
[02:07:21] with component companies as well
[02:07:23] uh to to to do their uh simulation store
[02:07:27] so can you also add the statistical
[02:07:30] parameters and do monte carlo
[02:07:31] simulations
[02:07:32] because one of the things you always
[02:07:34] look for is the yield right
[02:07:35] yes definitely we have this kind of
[02:07:37] optimization yeah yeah
[02:07:39] that's yeah that it's it's part of this
[02:07:41] uh process as well
[02:07:43] there is other software which is also
[02:07:44] can work in cohesively with this
[02:07:46] software it's called optispies where
[02:07:48] you can put up the electronics
[02:07:50] components together uh
[02:07:52] in also the uh you can do this
[02:07:55] the chip layout as well from because we
[02:07:57] collaborate with
[02:07:59] uh mentor graphic with cadence where and
[02:08:01] some fabs we are
[02:08:02] adding up uh pdks into our into our
[02:08:05] software tools to be
[02:08:06] enable the the designers as well to to
[02:08:09] use
[02:08:10] um the models into this simulation tool
[02:08:13] okay yeah so i think what i can do i can
[02:08:15] set up a meeting with
[02:08:16] our design team right and see if they
[02:08:18] have any interest
[02:08:20] that would be perfect yeah and also i
[02:08:22] sent an
[02:08:23] email to gloria like privately i hope
[02:08:25] also we can collaborate with infinera
[02:08:27] and their
[02:08:28] uh their uh kind of designs as well
[02:08:31] perfect this is exactly the goal of
[02:08:33] these meetings
[02:08:34] and also i would like to give the word
[02:08:36] to to
[02:08:38] the group of photonics from alto
[02:08:40] university
[02:08:41] i can see we have a representative here
[02:08:47] you can yes thank you
[02:08:51] hello hi yeah fine thank you
[02:08:54] um have something to explain i know yes
[02:09:02] uh if you look at the screen
[02:09:06] can you say everything
[02:09:10] i'm sorry are you trying to share your
[02:09:11] screen with the green boot on
[02:09:14] yeah yes
[02:09:23] okay we don't see this light oh
[02:09:27] sorry okay uh when that opens up
[02:09:29] probably i may
[02:09:38] [Music]
[02:09:45] screen
[02:09:48] don't worry too much yes tell us tell us
[02:09:50] what's on your mind and what room for
[02:09:52] cooperation are you looking for in this
[02:09:54] don't worry
[02:09:55] okay well um i just thought of sharing
[02:09:59] uh some
[02:10:00] research uh innovation which has
[02:10:02] happened last week uh thought of sharing
[02:10:04] from one of our research group
[02:10:06] uh they have made uh the some
[02:10:09] achievement in the
[02:10:10] silicon photonics especially the
[02:10:12] component uh
[02:10:14] waveguide laser atomically depos
[02:10:17] deposited airbn laser to be precise for
[02:10:20] especially for isp
[02:10:21] data center so airbnb is one of the base
[02:10:24] materials
[02:10:25] in a single and straightforward process
[02:10:28] compared to other competing technologies
[02:10:30] like
[02:10:30] indium phosphate or gallium arsenide are
[02:10:32] like
[02:10:34] the key advantages it is economical
[02:10:37] scalable compact and
[02:10:40] also the airbn is not a semiconductor
[02:10:42] does not conduct any heat so no cooling
[02:10:44] is required
[02:10:45] and further the processing is uh
[02:10:47] relatively done at a low temperature
[02:10:50] just 300 degrees celsius why this
[02:10:52] existing
[02:10:53] optical transistor is costlier is
[02:10:56] because a complex process
[02:10:58] that is not trying to show in the uh
[02:11:00] presentation any nevertheless
[02:11:02] okay um for example it's hundred euro
[02:11:05] per square centimeter for indian
[02:11:06] phosphate something like that and we
[02:11:08] talk about here
[02:11:09] approximately five five euro per square
[02:11:12] centimeter this is what we are
[02:11:14] uh achieved as a proof of concept uh
[02:11:17] last week
[02:11:18] and we are playing for this uh the
[02:11:20] funding with the
[02:11:21] finnish government here probably if we
[02:11:23] get through we may just proceed further
[02:11:25] with the
[02:11:26] uh this technology why we participated
[02:11:28] here now
[02:11:29] just to make sure that if there is any
[02:11:32] industry
[02:11:32] focus is to make a joint
[02:11:36] development of research ideas so
[02:11:39] we are open so this is what i wanted to
[02:11:41] share thank you very much
[02:11:42] thank you very much what i have seen
[02:11:44] lately is that there is a huge growth on
[02:11:45] atomic layer deposition
[02:11:47] in companies in epic if you're looking
[02:11:49] for a partnership alto university is a
[02:11:51] great place to start
[02:11:52] also the company alan 3d and we also
[02:11:54] have a lot of technologies for
[02:11:56] manufacturing of integrated photonics so
[02:11:58] thank you very very much
[02:12:00] and i think we have one quick question
[02:12:02] from photonics42 and the final market
[02:12:05] scope of joel
[02:12:06] so antonello what's on your beautiful
[02:12:08] mind
[02:12:12] antonello photon is 42
[02:12:17] i think what's in his beautiful mind
[02:12:18] photon 42 is a new company in the
[02:12:20] assembly and packaging of photonic
[02:12:21] devices
[02:12:22] they are looking for punishes on wafer
[02:12:24] level testing
[02:12:25] of indium phosphate and silicon
[02:12:27] photonics great partnership to be made
[02:12:29] there
[02:12:29] why don't we go to joel why don't we
[02:12:31] hear a bit about this
[02:12:33] beautiful market thank you very much for
[02:12:35] being with us on this beautiful
[02:12:36] afternoon
[02:12:37] send us away this is the last
[02:12:39] presentation of the show the floor and
[02:12:40] the attention of everyone is yours
[02:12:42] okay hello jose hello thank you
[02:12:46] for inviting me for today's meeting and
[02:12:49] giving me the opportunity to
[02:12:51] present yours activity in terms of uh
[02:12:54] optical communications uh
[02:12:58] okay so first i should share my screen
[02:13:05] share okay can you see my screen
[02:13:08] well
[02:13:13] okay so uh first i'd like to introduce
[02:13:17] uh
[02:13:18] the ul soyo provides market research and
[02:13:21] uh
[02:13:21] technology analysis and strategic
[02:13:23] consulting
[02:13:25] uh we have developed a unique
[02:13:27] understanding of
[02:13:29] technologies that enable enables us to
[02:13:34] accurately evaluate the markets uh the
[02:13:37] solutions
[02:13:38] the emerging applications and strategies
[02:13:41] so the production team of analysts
[02:13:44] is based in lyon in france and
[02:13:49] has a three division photonics and
[02:13:51] sensing semiconductor and software
[02:13:54] and power and wireless and you can see
[02:13:57] the the fields
[02:13:58] we cover in our in our reports so
[02:14:02] now i i'd like to uh share
[02:14:05] with you our uh latest uh product that
[02:14:08] uh
[02:14:08] uh where uh you should
[02:14:12] just uh recently optical transfers for
[02:14:14] data command telecom
[02:14:15] silicon photonics and and the vexil
[02:14:18] market uh
[02:14:21] for all type of applications
[02:14:24] so now i would like to start
[02:14:28] with the technology roadmap so the the
[02:14:30] roadmap
[02:14:31] uh shows the uh the evolution of
[02:14:36] uh of the transceivers using data
[02:14:39] command telecom
[02:14:40] along the the bandwidth bandwidth
[02:14:42] increase
[02:14:44] so the the new four effect terms
[02:14:48] are increasingly universal and they are
[02:14:51] designed
[02:14:52] to reduce the size and thus power
[02:14:55] consumption
[02:14:58] inside the modules the optics and
[02:15:02] integrated circuits are getting closer
[02:15:07] you know together to enable uh higher
[02:15:09] data
[02:15:10] data rates and and higher bandwidth
[02:15:14] uh different uh working groups
[02:15:17] are working uh on the standardization
[02:15:22] of the uh new uh
[02:15:25] form factors for for data common and and
[02:15:28] telecom and the leading optical
[02:15:30] uh manufacturers has have already
[02:15:33] unveiled their
[02:15:34] latest 400g optical transfers for
[02:15:38] osfp cfp8 and
[02:15:43] qsfpdd
[02:15:49] technology is quite new as
[02:15:52] and is in in in our roadmap
[02:15:56] but uh still require the standardization
[02:15:59] as
[02:15:59] um as we we
[02:16:03] all know and
[02:16:06] the industry struggles with
[02:16:10] with the the throughput and the data
[02:16:13] rates of all optical communication
[02:16:17] equipment as well as
[02:16:18] modules uh that don't grow as fast as
[02:16:22] the
[02:16:23] traffic of the the current and as well
[02:16:26] as
[02:16:26] new applications so the silicon
[02:16:29] photonics
[02:16:30] is is the promising
[02:16:34] solution for for enabling
[02:16:37] uh new new data rates
[02:16:43] also although the silicon photonics is
[02:16:47] quite well developed today
[02:16:50] uh the industry standards
[02:16:53] uh supply chain and the compatibility
[02:16:58] with with the whole ecosystem is need to
[02:17:01] be
[02:17:02] established yet
[02:17:06] yeah now i would like to share some
[02:17:09] [Music]
[02:17:11] market data of data com
[02:17:14] optical transceivers so the revenue
[02:17:17] generated
[02:17:18] of data complicated transceivers
[02:17:22] reached out around 4 billion dollars
[02:17:25] in 2019 uh and is expected to
[02:17:30] travel uh to around
[02:17:33] 12 billion dollars by 2025 at the
[02:17:38] kegger of 20 24 percent
[02:17:42] and then this this very high growth uh
[02:17:46] will be driven by adoption of uh much
[02:17:49] more expensive
[02:17:50] uh optical modules uh uh next uh
[02:17:54] next year that they will be based on on
[02:17:57] coherent technology and will migrate
[02:18:00] from
[02:18:00] long-haul application applications to
[02:18:03] short-haul applications
[02:18:05] however the
[02:18:08] cobor technology is still very emerging
[02:18:10] and as i said
[02:18:12] it needs standardization so
[02:18:15] we can see today some proprietary
[02:18:17] solution on the market but
[02:18:19] uh the cobo could
[02:18:22] in i mean the in a high volume
[02:18:25] application
[02:18:26] could uh enter the market
[02:18:29] uh maybe in uh in two three years
[02:18:36] now i i would like to share with you
[02:18:38] some uh
[02:18:41] recent uh mna activities
[02:18:44] for both platforms imp and silicon
[02:18:47] photonics so
[02:18:48] a high level of integration of optics
[02:18:52] with silicon integrated circuits
[02:18:56] lead to investment
[02:18:59] by by system vendors
[02:19:04] like cisco or nokias
[02:19:08] and um however
[02:19:11] today we still see that uh the strength
[02:19:14] and
[02:19:15] positions of uh traditional uh
[02:19:18] vertically integrated companies which
[02:19:20] are involved in
[02:19:22] in uh imp technology platforms
[02:19:25] uh if in case of silicon photonics
[02:19:29] i would like to give you an example of
[02:19:31] cisco
[02:19:34] cisco has been preparing um for
[02:19:37] for the future as is
[02:19:41] its customers transition from uh
[02:19:44] standard chassis based
[02:19:48] systems to to plug up the module
[02:19:53] compatible systems or
[02:19:56] ecosystems with switches and and routers
[02:20:00] and now when the the typical vendors
[02:20:03] of optical systems uh uh
[02:20:06] uh lose their um their exclusivity on
[02:20:09] the
[02:20:11] on on the on the market today and
[02:20:14] in a uh coming years
[02:20:17] um the the
[02:20:20] when the optical interconnect from chip
[02:20:22] to chip will
[02:20:23] will be required the
[02:20:26] the cisco and the similar uh
[02:20:30] companies need to bet on on the right uh
[02:20:33] technologies such as uh silicon
[02:20:35] photonics
[02:20:37] and then they they cannot rely on
[02:20:40] competitors
[02:20:41] uh from from broadcom or or intel
[02:20:44] okay so that's it for me thank you for
[02:20:48] your attention
[02:20:52] thank you very much for a great
[02:20:53] presentation and thank you for closing
[02:20:55] the event the way it should
[02:20:57] be we have a lot of room for cooperation
[02:21:00] for me it was great because first of all
[02:21:02] many many of you are very close friends
[02:21:04] of mine and the others are soon to
[02:21:05] become friends of mine
[02:21:06] very friendly and i'm hoping i can start
[02:21:08] traveling again as soon as the whole
[02:21:10] virus is over
[02:21:11] i i am sick about taking notes i take a
[02:21:14] lot of notes my god my wife is crazy i
[02:21:16] was taking notes on my wedding
[02:21:17] i'm going to show you what the kind of
[02:21:18] notes i took today i actually for me
[02:21:21] this meeting was very interesting for me
[02:21:22] to find out
[02:21:24] opportunities in the supply chain and i
[02:21:26] think at some point we got in there when
[02:21:28] we started talking with gloria with team
[02:21:30] we ran about their challenges and we
[02:21:32] understood that there is
[02:21:34] some room for materials in the heat
[02:21:36] dispensing there is definitely some room
[02:21:38] for optics for high temperatures and
[02:21:40] intellectual data com with cold high
[02:21:41] temperatures 130 degrees centigrade
[02:21:44] we see some tendencies we see some good
[02:21:47] ideas of intermediary steps within cpu
[02:21:49] intermediary step or fin
[02:21:51] the future it has been a lot of
[02:21:52] discussing that but that was not the
[02:21:54] purpose of the meeting the purpose of
[02:21:55] the meeting was
[02:21:56] to work together and i have seen in the
[02:21:59] zoom channel and also your private
[02:22:01] whatsapps
[02:22:02] that there is a lot of room for
[02:22:03] cooperation many of you have set up
[02:22:05] one-to-one meetings
[02:22:07] all of all of you please stay for the
[02:22:09] one-to-one meetings we have a range
[02:22:10] separate soon rooms for you
[02:22:12] the ones who are interested but they
[02:22:13] want to commit is you can also stay but
[02:22:15] for me
[02:22:16] from anna and from the epic thing i
[02:22:18] would like to thank you
[02:22:20] for a great discussion thank you for
[02:22:21] great two hours
[02:22:23] i hope you continue the follow up that
[02:22:24] was important and for me from anna and
[02:22:27] the whole team
[02:22:30] goodbye
[02:22:31] thank you