# EPIC Online Technology Meeting on Datacenter Interconnects (in cooperation with COBO and EA)

https://www.youtube.com/watch?v=tHh-z724D2g

[00:00] afternoon everyone from the beautiful
[00:01] city of north like the netherlands
[00:04] we still do not know who the next u.s
[00:06] president is going to be
[00:08] but we are here to know what is going to
[00:10] happen in the data centers of the future
[00:12] welcome
[00:13] to another one and a special one edition
[00:16] of the epic online
[00:18] technology meetings
[00:23] oh there you are you caught me gaming in
[00:26] the cloud
[00:26] that would have been unthinkable a few
[00:28] years back but now optical communication
[00:31] networks
[00:31] made the impossible possible and not
[00:34] only for cloud gaming
[00:36] but think about remote medical
[00:38] operations autonomous driving hands-on
[00:40] remote training
[00:41] virtual reality in an open world and the
[00:44] list
[00:44] goes on all of these only work when
[00:47] there is near zero latency
[00:50] a network faster than the human reaction
[00:52] time or the human interconnect between
[00:54] your eyes or brain
[00:56] other hands and the second challenge is
[00:58] that legacy electronics cannot keep up
[01:00] with the global demand
[01:01] for bandwidth parallel channel
[01:03] processing is reaching its limits
[01:05] while we watch youtube netflix amazon
[01:08] prime hbo disney plus and even the next
[01:10] olympics in 8k
[01:12] bank with demands are going through the
[01:14] roof but researchers
[01:16] already know that optical switching is
[01:18] agnostic to the speed at which you are
[01:20] sending information
[01:22] what you see in data centers every three
[01:24] to four years
[01:25] the entire infrastructure has to be
[01:27] ripped out because
[01:28] there's a new uh bitrate and we have to
[01:31] change everything
[01:32] the chip doesn't care what comes in what
[01:34] comes out and that that's unique to
[01:36] optical switching
[01:37] and if the next jump is for 800g you
[01:39] don't care
[01:40] whatever comes in comes out so on
[01:42] wednesday november 4th
[01:43] we are organizing a top level discussion
[01:46] about data center interconnects
[01:48] what will the optical nego architecture
[01:50] look like and what demands will it put
[01:53] on the physical layer
[01:54] this is a key opportunity for the
[01:56] companies in photonic and electronic
[01:58] integration
[01:59] and optical network solution providers
[02:02] to find ways to collaborate
[02:04] we'll have the data center operators
[02:06] like telefonica or microsoft
[02:08] in your zoom room together with
[02:10] suppliers who are ready to solve
[02:12] tomorrow's challenges today and i am
[02:15] thinking of a vip list including sienna
[02:18] huawei
[02:19] juniper infinera adva and cisco
[02:22] so sign up to participate in your online
[02:25] technology meeting or watch it real time
[02:27] on our youtube channel because our
[02:29] urgent need for speed
[02:31] is up for discussion
[02:35] so everybody full of energies we are
[02:37] about to
[02:38] create business opportunities that's
[02:40] what this meeting is all about
[02:42] and what we have to say we start with
[02:44] this we have to always
[02:46] thank the support that we keep getting
[02:48] from the industry thank you very much
[02:50] many of you are aware that a couple of
[02:51] weeks ago epic became the largest
[02:53] photonic association on this planet
[02:57] so we are extremely extremely happy to
[02:59] have the support of the epic members for
[03:01] this 600
[03:02] 620 members that we have right now
[03:06] thank you very much all of you our job
[03:08] is to keep knowing each of you
[03:10] individually i'm trying to find ways for
[03:12] you to contact your suppliers customers
[03:14] and partners
[03:15] at our beautiful meetings thank you very
[03:17] much for this and also
[03:19] thank you for all the support that our
[03:20] online technology meetings are having
[03:22] this is the agenda
[03:23] from now until christmas remember today
[03:26] we have data center interconnects
[03:28] and we are going to have other topics
[03:30] like water quality monitoring quantum
[03:32] communication
[03:33] peak testing all the way to the last
[03:35] meeting of the year
[03:36] which is going to be on food and
[03:38] beverage and it's just gonna be just
[03:40] before
[03:41] the christmas holidays i just want to be
[03:44] extremely extremely thankful for all the
[03:46] community and all the support
[03:47] that we keep getting from all of you i
[03:49] just want to make sure
[03:51] that you get what you deserve from the
[03:53] epic network and that you
[03:55] little by little keep telling us how to
[03:57] do our job because
[03:59] ultimately we are here to support you i
[04:01] just would also
[04:03] like to acknowledge that this meeting
[04:05] wouldn't have been possible
[04:07] with the support of for of our sponsors
[04:09] today
[04:10] we have a long list of sponsors first of
[04:12] all modulite all the way from beautiful
[04:14] finland
[04:14] they make semiconductor lasers this
[04:16] company actually has
[04:18] mbe growth themselves and they go all
[04:20] the way to making semiconductor lasers a
[04:22] wafer level
[04:23] at the eye level and even make
[04:24] instruments that go even to opticians
[04:26] they tejema from the netherlands these
[04:28] people are ex-specialists on the active
[04:30] alignment of components
[04:32] they make equipment for the active
[04:34] alignment
[04:35] vi systems in germany these pixels
[04:37] called the world record the fastest
[04:39] pixels in the world
[04:40] and you want to have any of the laser
[04:42] components then slice semiconductor
[04:44] all the way from singapore this company
[04:46] is the company of choice
[04:47] if you're looking for optics
[04:49] semiconductor laser fiber components
[04:51] from iterative photonics all the way to
[04:53] laser components and in switzerland we
[04:55] do have access
[04:56] this accent is a company specialized in
[04:58] the world of micro optics and when it
[05:01] comes to micro optics these people can
[05:02] actually design
[05:03] optics for the compact each optics
[05:06] optics for the integration and fiber to
[05:09] the chip and optics for the silicon
[05:11] photonics heavy integration components
[05:13] and if you're looking for a software
[05:14] provider you have to talk to synopsis
[05:16] you know this is accelerating the
[05:17] adoption of photonic and photonic
[05:19] integrated circuit technologies
[05:20] from software design to manufacture
[05:22] energy efficient high performance
[05:24] photonic devices systems
[05:26] integrated circuits anything synopsis is
[05:28] of course your panel of choice
[05:30] and finally last but not least we have
[05:33] seuss micro optics which is a company
[05:35] also specialized in the micro optics
[05:37] europe is
[05:38] leaving the next revolution of micro
[05:39] optics technologies and these people
[05:41] have actually also
[05:43] integrated micro optics in micro places
[05:45] bike
[05:51] optics reviews also a member of pixap
[05:53] and you're looking for members of pixab
[05:55] phycontek ficontec has positioned
[05:57] themselves in the center of the supply
[05:59] chain
[05:59] of packaging integrated photonics if
[06:02] you're looking for
[06:02] a specialized company for making the
[06:04] equipment of packaging integrity
[06:06] photonics
[06:07] to germany you go to bremen you go to
[06:08] hacking and you go to phi contact
[06:10] company in the segment and also i would
[06:12] like to acknowledge your media partner
[06:15] fiverr system thank you very much
[06:16] supporting today and
[06:18] our key players or cooperation players
[06:20] or panels in crying kobo
[06:22] and internet alliance but with all this
[06:24] i also would like to thank
[06:26] my very good friend better are the
[06:29] manager
[06:30] anna gonzalez dr gonzalez thank you very
[06:32] much for being with us today what's in
[06:33] the menu
[06:35] thank you jose for this introduction so
[06:37] hello everybody
[06:38] they are from barcelona live and we have
[06:41] in front of us a really interesting
[06:43] meeting
[06:43] with talks from services providers such
[06:46] as telefonica companies that provide
[06:49] a provides vertical integrated
[06:51] networking solution
[06:52] such as unipair and footerway and
[06:54] companies providing transceiver modules
[06:56] such as
[06:56] h g t tech so as you know
[07:00] at epic meetings presentations are short
[07:03] and they are used
[07:04] to trigger the discussions between the
[07:06] entire supply chain in data center
[07:08] interconnects
[07:09] that is here today so let's explore new
[07:13] collaborations and synergy
[07:15] synergies driving by the needs of end
[07:17] users so these times
[07:19] these end users are service providers
[07:21] that will tell you about the needs
[07:23] and and then we have the companies doing
[07:26] transceivers and optical cables
[07:28] companies in the data center
[07:29] infrastructure management
[07:31] they want some very offering networking
[07:33] solutions and
[07:35] surrounding all these we have all the
[07:36] manufacturing supply chain
[07:38] companies offering packaging and testing
[07:40] companies active
[07:42] in photonic integrated circuits
[07:44] components and modules
[07:46] companies offering a materials fiber
[07:48] optic
[07:49] equipment software and much more also
[07:52] participating in this meeting we are
[07:54] very proud to support
[07:56] the european commission of photonics 21
[07:58] initiatives
[07:59] such as a passion that develops a very
[08:02] promising technology for metro network
[08:04] solutions
[08:05] and the pilot lines in photonics such as
[08:08] jpg spine online that offers services
[08:10] for manufacturing large
[08:12] volumes of indivosified chips as
[08:14] including design and testing
[08:16] also the pine online fabulous that offer
[08:18] services for free for
[08:20] micro optics for large volume and
[08:23] manufacturing
[08:25] that can become key to ramp up
[08:26] production of devices in terms of
[08:28] coupling light in photonic integrated
[08:30] circuits and
[08:32] a the pixel pile online that offers
[08:34] services for packaging and assembly
[08:36] photonic integrated circuits
[08:37] including optical thermal mechanical and
[08:40] electrical building blocks
[08:42] and talking about pixab i would like to
[08:45] announce
[08:46] a new open call for the pixab
[08:48] pre-commercial rounds
[08:49] so a pixab wants to help companies to
[08:52] work with
[08:52] standards and a process that are
[08:55] compatible
[08:56] um with large bowling production so
[08:58] please go to the pixab website a
[09:01] pixab.eco
[09:02] and check pixab packaging solutions and
[09:04] if you have some chips to package
[09:06] apply through the website or send me an
[09:09] email okay
[09:10] so all the ingredients for another
[09:12] successful meeting
[09:14] are in place so let's cook it back to
[09:17] you jose
[09:18] let's cook it thank you very much and at
[09:20] the end of the day this meeting in our
[09:21] cooperation i would like to also
[09:22] acknowledge that this meeting is being
[09:24] live stream
[09:25] in youtube so let me say hello now to
[09:26] the youtubers hello youtubers in the
[09:28] world thank you very much for being
[09:30] there
[09:30] welcome to the epic meeting if you want
[09:32] to get in touch with any of the
[09:33] participants today
[09:34] all you got to do is send me an email
[09:36] jose.pothotepicdashastroc.com and i will
[09:38] be more than happy of making that
[09:39] introduction you can also use the chat
[09:42] in the youtube stream
[09:43] at the same time this is also valid for
[09:45] the people in the zoom room
[09:46] we are going to have one-on-one
[09:48] discussions after the meeting so you
[09:50] want to have a private discussion with
[09:51] any participant let us know during the
[09:53] meeting
[09:54] and also if you want to get in touch
[09:56] with the meeting use the private chat
[09:58] and if you miss a chance to talk to any
[10:01] of your potential partner suppliers and
[10:02] customers
[10:03] also send me an email and i will be more
[10:04] than happy on making this
[10:06] introduction let's get the ball rolling
[10:09] let's
[10:09] start and we can start better than with
[10:12] our paneling crime
[10:13] cover tiger thank you very much for
[10:15] being with us today
[10:17] why is the onboard optics alliance
[10:21] cover helping supporting dci
[10:25] oh thank you very much so let me briefly
[10:27] introduce what kobo is
[10:29] for those who do not know about kobo but
[10:32] the cobo has uh
[10:34] started in 2015 to
[10:37] overcome some challenges we are foreseen
[10:39] for uh plugable optics where
[10:42] we need to increase bandwidth and lower
[10:45] the power consumption
[10:47] so we are to stand we standardize
[10:50] onboard optics uh modules that can be
[10:53] mounted mid
[10:54] board on the just just like described on
[10:56] the
[10:57] uh the slide and the that's been uh
[11:01] the standard is out there for a while
[11:03] now and now
[11:04] we are taking another uh task within
[11:07] kobo
[11:08] to um talk about discuss about
[11:11] uh callbacks optics as the optics needs
[11:14] to be
[11:15] closer to the h asic uh to support lower
[11:18] power consumption and increased
[11:19] bandwidth as well so cobo
[11:21] has a new working group for compact
[11:24] topics which i'm sharing
[11:26] so if you have any questions about it uh
[11:28] please reach out
[11:30] to uh the website onboard optics.com
[11:33] and the kobo has many materials
[11:36] uh of public available also we are
[11:39] having some uh
[11:40] discussions about kobach optics
[11:41] obviously so we
[11:43] encourage people who has an interest in
[11:46] kobach optics or anything about umbra
[11:48] optics
[11:48] reach out to ombudoptics.com
[11:52] has been instrumental for epic i would
[11:54] like to thank tiger and the whole kobo
[11:55] community for this i
[11:57] an initiative is supported by the top
[12:00] companies in the sector
[12:01] and has been open always with the arms
[12:04] open looking forward to receiving
[12:06] input from any company so anybody who
[12:08] has input for the
[12:09] package optics activity of kobo you can
[12:12] contact them
[12:13] and i think that was great we really
[12:15] need to know now about what are the
[12:17] demands for bandwidth
[12:18] and believe me believe me there is no
[12:20] better person that i know of
[12:22] thank my very good friend john ambrosia
[12:25] to take that chance
[12:27] john good morning in new york how are
[12:29] you doing
[12:30] i'm doing awesome how are you doing
[12:32] today jose you look fast
[12:34] fantastic you look lovely in action this
[12:37] morning
[12:38] everything is okay we don't know who the
[12:40] president is but we don't care because
[12:42] we care is about the bank with demands
[12:46] well thank you jose and uh yeah it's
[12:48] actually a really good time for us to be
[12:50] having this call
[12:52] i want people to realize that
[12:55] we've heard we've heard a lot of
[12:56] references already today to needing more
[12:58] bandwidth
[12:59] and i wanted to make the uh
[13:02] the epic community aware that next week
[13:05] i will be initiating requests within the
[13:07] ieee 802.3
[13:09] ethernet working group to start the next
[13:11] project to start development beyond 400
[13:14] gigabit ethernet
[13:15] let's see if you can flip to the next
[13:16] slide please
[13:19] just uh as this is a standards activity
[13:21] um i do need to use these official
[13:23] slides essentially basically any views i
[13:25] say today about the standard
[13:26] development is my view so i'm not here
[13:28] speaking on behalf of the ieee i'm the
[13:31] champion who's trying to make it happen
[13:32] inside of the ieee so jose next slide
[13:35] please
[13:37] okay so the state of beyond 400 gigabit
[13:40] per second ethernet
[13:41] a call will be held next week to
[13:44] initiate
[13:45] that effort and it will be i don't want
[13:48] to go through all the sausage making as
[13:50] as people like to talk about it um but
[13:53] come december we anticipate that there
[13:55] will be a new effort
[13:56] inside of the ieee 802 community to
[13:59] start defining
[14:00] um beyond 400 gigabit per second
[14:02] ethernet and that will be
[14:04] um i anticipate that the first meeting
[14:07] will be held in january of next year
[14:10] if you want to find information the the
[14:12] presentation explaining
[14:14] why we need to go to a higher speed and
[14:16] the issues with it
[14:17] are at the link there as well as
[14:20] believe the url i have here is the url
[14:23] that will be used
[14:24] for the future study group page um so i
[14:27] want everybody to be aware of that
[14:30] uh just you know as i know i got to keep
[14:31] this real quick let's go to the next
[14:33] slide jose
[14:35] um when once when somebody says ethernet
[14:39] uh you know this is my third time at
[14:41] starting a new speed
[14:42] and it seems like it grows every time in
[14:45] terms of what we're doing
[14:46] and we have this little diagram that i'm
[14:48] showing here today that kind of tries to
[14:50] put a picture in terms of
[14:52] what are we talking about for beyond 400
[14:54] gigabit per second ethernet now many of
[14:56] you will
[14:57] recognize that that data center style
[14:59] architecture over on the left
[15:02] and if you look at the different speeds
[15:04] and the different reaches and the
[15:05] different mediums
[15:06] you can see that we have a very big
[15:08] project in front of us to address a lot
[15:10] of different things
[15:12] in addition at the far right you can see
[15:15] that's assuming you don't have the video
[15:16] window over it but there's going to be a
[15:19] discussion about coherent optics
[15:21] versus direct detect optics so when i
[15:24] look at this diagram jose anna
[15:26] i see a huge don't don't go away go back
[15:29] to that slide jose
[15:31] i see a huge opportunity
[15:34] for the kobo membership to be getting
[15:36] involved and to be defining
[15:38] and helping to start the next generation
[15:41] of ethernet beyond 400 gigabit i mean
[15:43] just
[15:44] look at the different things that we
[15:45] have to worry about multi mode single
[15:47] mode
[15:48] single fi duplex fiber parallel fibers
[15:50] all we have to do in dwm systems
[15:53] you know there's a lot of stuff here
[15:55] that's going to have to happen and
[15:56] honestly it would be great
[15:58] if the kobo community could come
[15:59] together and really start looking at
[16:02] providing the industry with an optical
[16:04] roadmap
[16:05] um you know there's a lot of signaling
[16:06] going on there and it's not just about
[16:08] signaling it's about
[16:09] all of the different industries that
[16:11] will support making these higher speeds
[16:13] happen
[16:14] it really is an ecosystem style problem
[16:17] from the components from the testing
[16:19] i'm sure there are people on here who
[16:21] could contribute and say yeah you got to
[16:22] worry about this with higher speeds
[16:24] we need to get start thinking about this
[16:26] because honestly as we start to look
[16:28] ahead
[16:29] this is technologies that's going to be
[16:30] down the road in the second half of the
[16:32] of this decade so now is the time to
[16:35] start planning that road map
[16:37] and mapping out our journey
[16:41] and with that thanks everybody if you
[16:43] have any questions regarding the effort
[16:45] please feel free to reach out to me i'll
[16:46] be happy to talk to anybody
[16:48] jose knows how to get you in touch with
[16:49] me thank you very much john
[16:52] thank you very much and for what you're
[16:53] doing for for epic for the industry for
[16:56] for your call for interest for extra
[16:58] bandwidth
[16:59] uh we need people like you in our
[17:01] community i'm extremely happy extremely
[17:03] happy to have you as an advisor as a
[17:04] friend thank you very much
[17:06] anna let's continue
[17:09] yes let's start with the first
[17:11] presentation of today well
[17:13] the third presentation of today so it's
[17:14] my pleasure to introduce
[17:22] integrated transport center at
[17:24] telefonica so
[17:25] thank you again anna i think a lot lots
[17:28] of things i think to
[17:30] to tell you today short time i'll try to
[17:33] be fast
[17:34] uh i'll present you what is the
[17:36] telephonica vision about the sdn
[17:38] architecture for open optical networks
[17:40] and what is the relation of this topic
[17:42] with
[17:44] the with the data center interconnection
[17:47] currently intelco networks the
[17:50] intel optical networks as the ones from
[17:53] from telefonica
[17:54] are typically based on model vendor
[17:56] solutions where
[17:57] all the transponders the dams amplifier
[18:00] everything is provided by a single
[18:01] vendor in its domain
[18:03] also the network management system and
[18:05] if we want to
[18:06] introduce new data center
[18:08] interconnection solutions
[18:10] as the ones that are being developed in
[18:12] the last years there are parallel
[18:14] products
[18:15] for data center interconnection that are
[18:17] very cost
[18:18] effective and quite simple then we
[18:21] always need to go for the same vendor
[18:24] that
[18:25] we have in the network for example we
[18:26] have a sienna infinera huawei
[18:28] whatever network in the domain and we
[18:31] want to deploy a data center
[18:32] interconnection network
[18:34] or solution we have to integrate it with
[18:37] the same vendor
[18:38] otherwise everything all the
[18:41] configuration performance restroom
[18:44] protection and so on
[18:45] should be done manually because we can
[18:48] provide this alien wavelengths from
[18:50] data using data center interconnection
[18:53] solutions but this is very complex to
[18:56] maintain
[18:56] and operate the solution in order to
[18:59] integrate
[19:00] these dci solutions within the existing
[19:04] telco network while keeping performance
[19:07] monitoring
[19:08] optical restoration that is key for
[19:10] example telefonica we are using
[19:11] optical restoration of the optical layer
[19:13] for every channel
[19:15] we don't want to do it at the ip layer
[19:17] we want to do it in the optical because
[19:19] it's much
[19:20] cost effective okay we can do all these
[19:23] functionalities restoration performance
[19:25] fault management
[19:26] with an alien wavelength or for example
[19:28] a data center interconnection solution
[19:30] from one vendor
[19:31] over the open line system over the
[19:33] optical infrastructure
[19:35] the radam and so on from other vendor
[19:37] using sdn
[19:39] and a standard sdn architecture and this
[19:42] this standard sdn architecture is based
[19:45] typically in
[19:46] two main interfaces the transport api
[19:51] and that is offering uh information
[19:53] about their topology
[19:55] the network situation uh for example the
[19:57] performance default management and so on
[19:59] towards the applications on top
[20:01] the hierarchical controller on oss and
[20:04] very important also the support of
[20:06] netconf young and openconfig models
[20:09] in the sdn controller and also in the
[20:12] open terminals for example these
[20:14] solutions these
[20:14] dci solutions should support this
[20:18] netcom yank and the open config if we
[20:20] have these
[20:21] the support of these interfaces these
[20:23] standard interfaces in the terminals
[20:25] and in the controller then we can manage
[20:28] and operate
[20:29] an alien wavelength as with the same
[20:32] capabilities or with the same
[20:34] characteristics as a regular
[20:36] channel in a in an aggregated network
[20:40] okay in this case even if we have a
[20:43] an open line system where the sdn
[20:45] controller from the vendor from siena
[20:47] infinera huawei or whatever
[20:49] is not supporting the open config in the
[20:51] controller
[20:52] of its network we could also configure
[20:56] the open terminals with a hierarchical
[20:58] controller on top
[20:59] so we we have a backup in that case so
[21:03] we can always the key from the data
[21:05] center
[21:06] uh perspective or the the for the dci
[21:09] provider is to support
[21:11] netcom junk and the open conflict models
[21:13] because the configuration could be done
[21:15] by a third-party controller
[21:18] important okay we are intensely working
[21:21] in telefonica on this but we are not
[21:23] alone
[21:24] this vision about partial optical
[21:25] disaggregation
[21:27] and the standard sdn is not only a
[21:30] telephonica issue currently we have
[21:33] recently created a working group in tip
[21:36] the name is must and i think the name is
[21:38] clearly explained in the purpose
[21:40] mandatory use case specifications for
[21:42] sdn transport where telefonica vodafone
[21:45] mtn orange stelia
[21:47] and deutsche telekom are working
[21:49] together to
[21:50] sp specify to provide the technical
[21:52] specifications that
[21:54] for the for this interfaces the
[21:57] technical
[21:58] specifications that should be supported
[21:59] by both the
[22:01] network elements for example they open
[22:02] the the devices
[22:04] uh the the transponders the open
[22:05] terminals and the
[22:07] and there's the end controllers okay
[22:10] this this this network or at least this
[22:13] specification is something that
[22:14] we think and it should be led by by the
[22:17] telecom operators and we are doing that
[22:21] and okay in this picture just to show
[22:23] you that currently in telefonica we are
[22:25] hosting the oaf
[22:27] 2020 sdn interoperability demo
[22:30] the combining we are doing all the
[22:32] potential combinations we may can
[22:34] imagine in the open terminal in the
[22:35] controller domain in the hierarchical
[22:37] controller to test and check that
[22:40] the vendors are implementing the
[22:42] technical specifications
[22:44] as are requested in the in the standards
[22:47] currently we are working on on this on
[22:50] this testing
[22:51] and we are planning to also do to
[22:53] disseminate
[22:54] the results of this of this test plan in
[22:57] the
[22:58] in the coming month just to conclude
[23:02] three important messages for current
[23:04] optical dcr solutions
[23:06] are only applicable in monoventer
[23:08] networks if we want to deploy
[23:10] a dci solution from for example from
[23:13] juniper
[23:14] over asiana network this should be
[23:17] manually configured and is only
[23:19] applicable in basic
[23:20] point-to-point links for example we
[23:22] cannot have a
[23:23] optical restoration however if we apply
[23:27] an open sdn architecture
[23:29] using dappy and open config with
[23:32] a netcombine then we could have
[23:35] multi-vendor
[23:36] interoperability and enabling automated
[23:40] operation
[23:40] in these environments okay that's all my
[23:43] side
[23:43] thank you thank you very much
[23:47] from pedro super nice presentation so
[23:50] okay so i have a question um what do you
[23:54] think
[23:54] are going to be the needs for data
[23:56] center bandwidth in the next years
[23:58] i mean from the point of view of em of
[24:00] telefonica now how do you see also this
[24:03] transparents to to to this online
[24:07] online world knowing which everybody is
[24:09] working at home
[24:10] and not because of this a virus
[24:13] situation
[24:14] and yes how do you
[24:18] yeah i think most of the we are
[24:19] expecting most of the traffic pressure
[24:21] from the data centers at the edge of the
[24:24] network
[24:25] that is in the metro networks
[24:28] to the access and this is the network
[24:31] segment the metro regional i think like
[24:33] a distances
[24:34] of uh no more than uh 300 or 400
[24:38] kilometers
[24:39] this is metro regional no more than 300
[24:42] or 400 and most of the channels i think
[24:45] 100 gig 200 gig and 400 gig
[24:48] probably no more than this in the for
[24:51] small
[24:53] capacities above this may be needed for
[24:55] the data center interconnection for
[24:57] backups
[24:58] but the demand for these channels may
[25:01] not be
[25:01] so intensive as in the uh
[25:04] in the edge because most of the traffic
[25:06] is moving or the applications are moving
[25:08] to the edge
[25:09] and the consumer and most of the traffic
[25:12] will be
[25:12] in this in this segment and in this
[25:15] segment since we have or the operators
[25:17] as telefonica
[25:18] we have we have photonic meshes we have
[25:21] lots of different paths to reroute the
[25:23] traffic so we can absorb
[25:25] lots of of traffic using multiple paths
[25:29] according to that we don't need very
[25:31] high capacity links
[25:33] we can have it's better to have
[25:34] granularity
[25:36] lots of 100 200 gig or 400 gig and the
[25:39] main motivation to deploy a new
[25:42] technology or
[25:43] upgrade to for example to do for 200
[25:45] giga 400 gig
[25:46] is cost it's just the cost per bit if
[25:50] we have a lower cost per bit for 400 gig
[25:53] we will go for that in the in the matter
[25:56] otherwise we can absorb
[25:58] the 100 gig using multiple paths over
[26:00] the photonic mesh
[26:02] okay so you are asking here new
[26:05] technologies to reduce costs number bit
[26:07] i think we have um we have a comment
[26:10] from jim
[26:11] jim would you like to mute and make the
[26:13] comment by yourself
[26:15] uh yeah i can but just uh you mentioned
[26:18] that coffeeing a lot
[26:19] maybe um one sentence or briefly explain
[26:23] how those models are related to the to
[26:25] the network and why they're important
[26:27] in moving to the sofa model
[26:32] i'm i'm i'm sorry to be honest i didn't
[26:35] really catch
[26:36] the question what uh so what's a yet
[26:39] net conf yang you mentioned about every
[26:41] other word was netconfiance so maybe a
[26:43] brief description
[26:44] i'm sorry yes now very briefly
[26:48] discussing in the
[26:49] description these are the modeling in
[26:51] particular the open conflict modeling
[26:53] are the the models in order to configure
[26:55] the transponder so that
[26:56] from the controller we could decide what
[26:58] is the power
[27:00] the uh the wavelength
[27:03] so the in order to configure the basic
[27:06] parameters for
[27:07] the transponder in that case for example
[27:09] we can select the wavelength for our
[27:11] restoration path
[27:13] this is uh these are are like the
[27:15] configuration
[27:16] these are the protocols that enable uh
[27:19] the device configuration
[27:20] and if we use a standard one that could
[27:22] be implemented
[27:24] currently we typically use proprietary
[27:26] interfaces for that
[27:27] it is going for a standard one that
[27:29] could be
[27:30] integrated in an agnostic controller so
[27:33] that
[27:34] we can combine channels from one vendor
[27:37] over the optical infrastructure
[27:38] of a different vendor but this is device
[27:41] configuration
[27:42] so if i hear the answer correctly you're
[27:44] saying we have the capability today to
[27:46] do all that but to make it
[27:47] vendor agnostic um we need the net
[27:50] coffeeing models so that we can set the
[27:51] same things depending on no matter
[27:53] what's
[27:54] in the restoration path exactly we can
[27:56] do it
[27:57] now but this if we need to do an ad doc
[28:00] or a specific development or
[28:02] implementation per vendor this is too
[28:04] expensive this is too costly we should
[28:06] go for a
[28:07] standard in order to be vendor agnostic
[28:10] all right thank you
[28:12] welcome okay thank you very much for
[28:14] this
[28:15] very interesting conversation so now
[28:18] maybe this is a good moment to to see
[28:21] about the capacities for these companies
[28:23] that want to design
[28:24] a want to prototype or want to test them
[28:28] devices working at 400g or 200g
[28:31] so we have here giovanni from blc
[28:34] hello giovanni how are you doing hi hi
[28:37] anna
[28:38] i'm doing fine thanks thank you for
[28:39] joining will it be maybe it's a good
[28:41] moment for you to introduce what are the
[28:43] capabilities
[28:44] in design we have i mean as design house
[28:47] of a blc
[28:48] and maybe could you have an slide to
[28:50] show
[28:51] yeah yeah i have a i have a slide um
[28:54] [Music]
[28:56] let me just share my screen
[29:03] can you see it yes i mean it's not
[29:07] presentation no okay
[29:10] now he's in presentation okay uh
[29:14] thanks anna for the introduction uh so
[29:16] i'm i'm giovanni i'm
[29:18] rnd engineer in vlc photonics
[29:21] um so within vlc we
[29:24] consider ourselves a fabulous photonic
[29:27] design house okay we have about 10 years
[29:30] of
[29:31] existence we are located in
[29:34] in valencia spain and you have a network
[29:36] of
[29:37] international partners many of them in
[29:39] europe but also in the usa in
[29:42] and asia but we are a team of
[29:45] 17 uh people with extensive
[29:48] academic industrial experience uh
[29:52] what basically we do is that we provide
[29:54] engineering services
[29:55] for the develo development of of
[29:59] uh products based on photonic integrated
[30:01] circuits
[30:02] so we work with many platforms uh
[30:05] silicon photonics
[30:06] silicon nitride silica indium phosphide
[30:09] and
[30:09] and many others and we provide all kinds
[30:12] of services
[30:13] uh to to develop products in
[30:16] in picks from system concept to device
[30:20] design
[30:20] layout testing and packaging so the
[30:23] the fabrication is done outside
[30:27] uh we have clients on in all
[30:30] types of applications uh including
[30:32] telecom datacom
[30:34] and and other types of applications
[30:36] sensing signal processing
[30:38] and we are we are also involved in many
[30:40] uh
[30:41] uh projects uh rnd and pilot line
[30:44] european projects uh for example we have
[30:48] impulse which is a pilot line for uh
[30:51] indie phosphide
[30:52] photonic integrated circuits in in the
[30:54] japex
[30:55] association and you are developing um
[30:59] we are actually filling the gaps between
[31:04] japex today which is provides mpw
[31:06] services
[31:07] to to provide pilot line services
[31:10] and also for example pix4life which is
[31:13] in
[31:14] in a silicon nitride platform in imac
[31:17] so if you if you want to contact me to
[31:19] know more about vlc you have my email
[31:22] address here
[31:22] thanks anna thank you very much giovanni
[31:26] regarding indian phosphate chips
[31:29] regarding the inputs of jpg spine online
[31:31] and do you think it would be useful to
[31:33] to make
[31:34] transceivers that able to reduce this
[31:37] cost per chip that was mentioning
[31:39] juan pedro uh yes yes of course uh
[31:43] interphosphide has many advantages um
[31:46] which is of course uh the benefits of
[31:49] integrating the laser sources
[31:51] and and for example uh same conductor
[31:53] outgoing fires in the same chip
[31:56] uh i think the last years indus fight
[31:58] has
[31:59] um has improved a lot uh the technology
[32:03] um we are uh getting
[32:06] uh high levels of yield so
[32:09] and also increasing wafer size so this
[32:12] can can drive down the cost
[32:14] uh and of course bring the benefits of
[32:17] indus force into phosphite which is the
[32:19] the integration of all active active uh
[32:22] functionalities
[32:24] into the chip so yes i think is a is a
[32:26] great candidate for the next
[32:28] generation of of pigs for day to come
[32:31] thank you very much evan okay so then if
[32:34] there are no more
[32:35] well yes we have a question er for
[32:38] giovanni
[32:39] so tobias from genoptic would you like
[32:41] to make the question by yourself
[32:44] um yeah of course so giovanni you said
[32:48] that you also provide testing
[32:50] capabilities and
[32:51] services and once it comes to
[32:55] high volume production how do you go
[32:57] with this testing services
[33:00] do you perform them by yourself or do
[33:02] you transfer
[33:03] to somebody else test house for instance
[33:07] uh yeah that's a great question uh
[33:10] indeed
[33:11] uh high volume test is is uh uh is in
[33:14] our
[33:14] roadmap um indeed uh
[33:17] we i want to share that we we have uh
[33:21] we have received a wafer level test
[33:24] machine
[33:25] so we can now test at the wafer level
[33:28] and we are also developing
[33:32] automated scripts for testing our chips
[33:35] in thai level
[33:37] so we we strongly believe that we'll be
[33:40] capable
[33:40] in a few months to perform
[33:44] um high volume testing on several chips
[33:47] and also wafer level testing so this is
[33:49] something that we are looking into um
[33:53] actively because you know that uh uh for
[33:56] scaling up production uh we need a lot
[33:58] of tests so
[34:00] uh for the wafer level wafer level
[34:02] machine we for sure you'll be able to
[34:04] make electrical electrical tests
[34:06] and we are working on uh developing the
[34:09] also the optical test functionality in
[34:12] the ufr level machine
[34:13] so this is something that we are
[34:14] actively looking
[34:16] and uh developing our our capabilities
[34:19] for that
[34:21] let me stress this because this is one
[34:22] actually one unique thing that happened
[34:25] at the epic the magic of epic
[34:27] last week i saw this beautiful linkedin
[34:29] post from vlc photonics
[34:31] they actually got a machine from icon
[34:34] tech
[34:34] to do wafer level testing services and
[34:37] that has been a gap
[34:38] in the value chain for many years so
[34:40] congratulations vlc from valencia
[34:42] ficon tech working together we are
[34:44] stronger thank you very very much for
[34:46] this
[34:46] you should talk to gen optic because
[34:48] tobias is putting
[34:49] a probe for wafer level testing into
[34:52] different foundry
[34:53] systems and in my opinion it has to be
[34:56] incorporated
[34:56] to the vlc wafer level testing process
[34:59] thank you very much for that
[35:01] and i cannot wait for the next speaker
[35:03] because she's also a very good dear
[35:05] friend of mine
[35:06] michael levy thank you very much for
[35:08] being with us today
[35:09] you are the ceo of lightweight logic you
[35:12] are also in every role mapping activity
[35:14] i ever interacted with internet
[35:15] photonics
[35:16] and you having also helping epic thank
[35:18] you for this
[35:19] and the floor and the attention of
[35:21] everyone today
[35:22] all around the world is to you is to mr
[35:26] levy
[35:27] thank you let me see if we can get this
[35:30] screen to share
[35:33] i just want to double check everybody
[35:35] can see that
[35:36] we can see it is that clear perfect i've
[35:39] got
[35:39] two small factors in this presentation
[35:42] the first one is to the exciting
[35:45] um proposition of putting polymers into
[35:48] the data center
[35:49] and to upgrade your pick and then i've
[35:51] got a few slides on the work i've been
[35:53] doing with epic and
[35:55] photon delta aim photonics over the last
[35:58] year or so on the
[35:59] pick industry roadmaps and i'd like to
[36:02] share some of those roadmaps with
[36:04] everybody today so let's uh
[36:08] go to the next slide and i've got to get
[36:10] through this one because we're a public
[36:12] company
[36:13] and these slides will be on our website
[36:15] so just go to our website if you want to
[36:17] see the slides in more detail especially
[36:19] the roadmaps
[36:20] so the first slide is really market
[36:22] position in a dynamic environment
[36:25] an engineering advantage from our
[36:26] standpoint is polymers
[36:28] electro-optic polymers and clearly you
[36:31] want to address
[36:32] some of the major pain points facing
[36:34] network operators
[36:35] especially in the data center so if you
[36:37] have a good platform
[36:39] technology platform flexibility that
[36:42] will actually lead you into the middle
[36:43] box on this graph which is obviously
[36:45] faster networks
[36:46] you've got to be able to enable faster
[36:48] networks and we
[36:49] certainly can generate our own
[36:51] chromophores and polymers
[36:53] here in denver colorado and we design
[36:56] modulated devices from those
[36:58] and if you can actually enable some
[37:00] faster networks then hopefully you can
[37:02] do some energy savings within the
[37:04] network
[37:05] using these types of very low power very
[37:07] high speed devices
[37:09] and so um what we are really talking
[37:12] about is hybrid picks
[37:14] and hybrid from the standpoint of
[37:16] polymer plus so that's that purple arrow
[37:19] in the center
[37:20] and this is an interesting graph on the
[37:22] left hand side you've got indium
[37:23] phosphide which is a pick incumbent
[37:25] as everybody knows you know with the
[37:27] indium phosphide lasers and passives and
[37:29] we'll hear about some of those today
[37:31] those are the
[37:32] attributes and on the right hand side
[37:34] you've got silicon photonics
[37:36] which has also now become a big
[37:38] incumbent with a large
[37:39] couple of large companies really pushing
[37:42] this technology successfully
[37:44] but what we're seeing here is is that we
[37:46] believe
[37:47] you can upgrade your photonics platform
[37:49] by using polymer modulators which are
[37:52] very fast and very low power but how
[37:55] would you do that i mean you can do that
[37:56] certainly on
[37:57] indium phosphide you could do that on
[37:59] silicon photonics
[38:01] and in this slide here we're showing
[38:02] three ways to do this on the left-hand
[38:04] side
[38:05] is your traditional electro-optic
[38:09] polymer modulator which is your polymer
[38:11] stack where you typically have a core
[38:13] and two claddings and that's on a
[38:14] silicon substrate
[38:16] that is what a lot of people have done
[38:18] in the past especially over the last 20
[38:20] 25 years we've seen a lot of three-layer
[38:23] stacked solutions
[38:25] um and the one in the middle is the one
[38:27] that we are working on
[38:28] uh pretty hard right now we put a press
[38:30] release out a couple of weeks ago once
[38:32] we called polymer plus
[38:33] so instead of doing three layers onto a
[38:35] silicon substrate
[38:37] we've actually simplified the
[38:38] fabrication and put a single layer
[38:40] right down onto an integrated silicon
[38:43] photonics layer
[38:44] it could be any other integrated
[38:46] photonics layer 2
[38:48] so that's a polymer plus add in the
[38:50] polymer
[38:51] in the fab onto an existing platform to
[38:54] upgrade your modulators
[38:55] and the one on the right is a lot of
[38:57] companies a lot of
[38:59] universities and academics are working
[39:00] on polymer slots we also have done some
[39:02] ourselves
[39:03] it's the smallest form factor these slum
[39:06] modulators are certainly not
[39:08] commercially available yet but they're
[39:10] certainly showing a lot of interest in
[39:12] performance traits and they're very very
[39:14] small
[39:16] and so polymers we believe can deliver
[39:18] radical innovation
[39:20] certainly faster devices we've already
[39:22] demonstrated greater than 70 gigahertz
[39:25] eos 21 bandwidths very low voltage down
[39:28] at the vault level which opens the
[39:30] opportunity for direct drive from
[39:32] cmos circuitry um we very much focused
[39:35] on lower cost easy fabrication because
[39:38] in the end that's the secret to
[39:41] implementing new technology
[39:42] can you make it easy can you implement
[39:44] it very
[39:46] efficiently in normal standard fab
[39:48] scenarios
[39:49] and the one on the right the questions
[39:50] that come up a lot can you make the
[39:52] technology robust and stable
[39:54] and we're doing a lot of stability and
[39:56] reliability measurements so
[39:57] labeling low power faster unique
[39:59] differentiating solutions
[40:01] and with that with a couple of minutes
[40:04] i'm going to brief everybody on the
[40:05] roadmaps because i think this is
[40:07] important as we
[40:08] have this meeting today on data centers
[40:11] on board optics
[40:13] so if we go back to 2016 because i've
[40:16] been
[40:16] working in road maps for a long time
[40:18] this is what we predicted in 2016. i
[40:21] know it's a little bit of an eye chart
[40:23] that's why the slides will be available
[40:25] on our website and you can see what we
[40:27] predicted for 2020 was 25 terabits
[40:31] um you know below five dollars per
[40:34] gigabit these sort of metrics
[40:36] and the same side i'm going to show next
[40:38] is where do we actually
[40:40] get in terms of our assessment and
[40:43] uh some of the the uh the comments on
[40:46] the right hand side may be covered
[40:48] by the screen but what they are is we
[40:51] predicted in 2020 that we would get to
[40:53] 25 terabits
[40:54] we will be less than five dollars per
[40:56] gigabit we will be using 50 gig devices
[40:59] and even 100 gig devices
[41:00] would be using pam4 so some of these
[41:03] things that we predicted four years ago
[41:05] actually came to fruition we actually
[41:07] were fairly accurate
[41:08] and so last year when we worked on the
[41:11] road maps again
[41:13] uh 2019 to 2000 2020
[41:17] where are we going what's the new
[41:19] metrics we're looking forward at
[41:21] in 2023 2024
[41:24] and you can see numbers on that top line
[41:27] you can see 800 gigs per second
[41:29] you can see 1600 coming in
[41:32] you can see less than two dollars per
[41:34] gigabit per second
[41:36] you'll see in a whole uh section on
[41:40] the variety of pick technologies for
[41:42] minimum phosphide
[41:43] monolithics to silicon photonics and
[41:45] hybrids
[41:46] dielectrics polymers gallium arsenide so
[41:49] you've seen a lot of different device
[41:51] technologies are included in this
[41:53] and this is the the last slide is you
[41:55] know where do we predict
[41:58] the industry is going for pick
[42:01] technologies especially they relate to
[42:03] data center type opportunities
[42:05] in the next three years well you know
[42:08] the top part of this slide
[42:10] low power is going to be key hybrid
[42:12] picks are going to be key
[42:14] faster devices are going to be key
[42:15] you're going to see devices move towards
[42:18] 70 gigs
[42:19] you can see folks focus on low power
[42:21] ethernet designs
[42:22] 100 gig solutions um you're going to see
[42:26] folks come up with 100 terabits
[42:28] per second one one are you um so
[42:31] 2023 i think you're going to see that
[42:33] you've already seen people design
[42:35] 25 and 50 right now and so um
[42:38] you know from what john said at the
[42:40] start of this session
[42:42] beyond 400 gig it's certainly happening
[42:44] and you can see the technologies below
[42:47] certainly are there to support that but
[42:49] i think low power and hybrid is
[42:51] certainly going to be some of the key
[42:53] metrics going forward
[42:55] and that's the end of my slide thank you
[43:02] very much michael for a great
[43:04] thank you very much michael for a great
[43:04] presentation and i love
[43:06] i love this slide with the roadmap we
[43:08] talk about it
[43:09] you can talk about this live for hours
[43:11] and we will have to because we
[43:13] really need to use this in many many of
[43:15] the epic activities
[43:17] two things the first one in the top top
[43:20] right part of the screen
[43:21] you mentioned 3.2 terabits per
[43:24] second is that realistic
[43:28] what did you base yourself let me share
[43:31] the
[43:31] that beautiful slide once again you put
[43:34] the no wait
[43:36] wait this is the magic of the life i
[43:38] don't have the slide but you put in the
[43:40] top left corner the top right corner
[43:42] that we are going in 20 27 to 3.2
[43:45] per second what did you base your
[43:48] your research or the discussion with the
[43:50] with supply chain on reaching that
[43:52] conclusion
[43:53] so i think everybody should be aware
[43:55] that you know this is not michael levy's
[43:57] roadmap this is a roadmap
[43:59] where there were a number of meetings
[44:01] both in the u.s and in europe
[44:03] driven in europe by photon delta driven
[44:06] in the us by aim photonics inemmie
[44:08] there was a group of technical experts
[44:10] that certainly looked at
[44:12] where things could go in the next decade
[44:15] and certainly if if we're looking at
[44:19] you know taking things
[44:22] beyond 25 and 50 terabits per second
[44:25] and as you say going up to 1600 and
[44:27] beyond
[44:28] what we're seeing right now in the
[44:30] community is people
[44:32] doing prototype demonstrations at 800.
[44:35] folks are trying to figure how to do
[44:37] 1600
[44:38] and certainly when you start looking at
[44:40] some of the challenges
[44:42] form factors may need to change maybe
[44:44] the plugable needs to change
[44:46] maybe we need to really encompass the
[44:49] co-packaging techniques
[44:51] just to to really allow us to utilize
[44:54] the technologies in ways
[44:56] that today was sort of constrained i
[44:59] mean if you look at a plugable form
[45:00] factor was constrained a little bit
[45:03] and if we take some of those constraints
[45:05] away in order to
[45:06] increase the performance and the data
[45:08] traffic
[45:09] i think there are opportunities here and
[45:11] certainly
[45:12] the excitement of the folks in the
[45:15] committee that were looking at the
[45:16] roadmap
[45:17] saw ways through this albeit we don't
[45:20] know the exact
[45:21] technological solutions but people did
[45:24] feel
[45:25] that there was a route forward together
[45:28] you and me we are going to go to
[45:29] indovent because you mentioned photon
[45:30] delta in eindhoven we have
[45:32] photonex networks and we have a comment
[45:35] from my friend other rats oh that
[45:37] what's on your mind yes it's it's a
[45:40] great presentation michael it's
[45:41] excellent to see you
[45:42] i wish we could meet face to face of
[45:44] course um
[45:46] it it's this the idea that we're going
[45:48] to go to 100 terabits and
[45:49] uh people have been speculating and
[45:52] doubting whether we can make a 12 and a
[45:54] half and 25 gig
[45:56] a terabit silicon devices
[45:59] out of uh in cmos nodes is the idea that
[46:02] we have 100 terabit
[46:04] chip coming from broadcom soon or are we
[46:06] hoping to do optical switching instead
[46:08] of
[46:08] layer 2 layer 3 switching oh that's a
[46:11] really tough question i mean i i first
[46:14] when i first came to this country in
[46:15] 1985 about labs everybody was talking
[46:18] about optical switching and that was
[46:20] a long time ago um we will see optical
[46:24] switching but
[46:24] i think you know the the role of optics
[46:27] is to help the electronics
[46:29] and if you think about electronics what
[46:31] some of the big impediments are today
[46:33] five nanometers getting the
[46:35] power consumption down we all know some
[46:38] of these 25
[46:39] and 50 turbo chips actually consume tons
[46:42] and tons of power
[46:43] so we have to think very carefully about
[46:45] a thermal solutions
[46:46] low power solutions and trying to keep
[46:50] the speed up
[46:51] and so i i do believe we will see
[46:54] electronic switching but i think you'll
[46:56] see optics do what it does
[46:58] best which is help get those signals in
[47:01] and out of that chip
[47:02] now whether the optics will be on the
[47:04] silicon chip or next to the silicon chip
[47:06] or a little bit away from the trip is to
[47:08] be sort of discovered over the next
[47:11] few years but i do see optics becoming
[47:14] closer to that chip and we are seeing a
[47:17] lot more
[47:17] sort of proper what i call proper
[47:20] electronic and photonic integration it's
[47:22] actually tough to do that
[47:24] but i think we will see more of that as
[47:25] we move forward
[47:28] michael you thanks a lot you keep
[47:31] showing this 70 gigahertz modulator
[47:33] and i understand this is a paradigm
[47:35] shift to go to the 100g channel
[47:37] um do you think this is the way to go to
[47:41] incorporating 70 gigahertz modulators in
[47:43] iterative photonics or there are other
[47:45] potential solutions that you think maybe
[47:47] could also be incorporated well it's
[47:50] always been
[47:51] a challenge for the industry and if you
[47:53] think about what the industry's done
[47:54] over the last decade is is that they've
[47:56] increased the symbols per bit
[47:58] so they've gone from nrz you know one
[48:00] zero one zero to pound
[48:01] four and then we've got the different
[48:04] types of encoding scene
[48:05] schemes that we've seen in telecom but
[48:08] in
[48:09] but they're getting more complex all the
[48:11] time so the question is
[48:12] is it's a trade-off if you start going
[48:15] that route
[48:16] and some of the lot of the what john
[48:17] indicated there before at the start of
[48:20] the show was coherent typically requires
[48:23] dsps which are expensive
[48:25] so there's a balance between cost and
[48:28] trying to get the speed of the devices
[48:30] up
[48:30] and i i'm not sure that we really fully
[48:33] understand that yet except
[48:35] we've done a great job in symbols per
[48:37] bit
[48:38] in terms of pam-4 type solutions but
[48:41] certainly if we can increase
[48:43] the speed of the devices keep the power
[48:45] consumption low
[48:47] there's less pressure on the symbols per
[48:49] bit so
[48:50] i do believe faster devices whether
[48:52] they're in silicon in your phosphite
[48:54] polymers
[48:55] or any other material are going to be
[48:57] important
[48:58] what they have seen a lot is from many
[49:00] companies trying to
[49:02] reach desperately 70-year hertz i'm not
[49:04] going to mention the company's suggested
[49:05] technologies plasmonics incorporating on
[49:07] silicon photonics for example
[49:09] with low trls i do
[49:12] believe that if we need a 70 gigahertz
[49:15] modulator it's going to be possible with
[49:16] indium phosphide definition of proving
[49:18] with polymer
[49:18] it is going to be a challenge for
[49:20] silicon photonics which as of today is
[49:22] the platform of choice and
[49:24] at least that's what i see from the
[49:25] different companies in the higher parts
[49:26] of the supply chain
[49:28] do you see this as an opportunity to
[49:30] bring new
[49:31] new technologies to this market
[49:33] absolutely if we look at the roadmap you
[49:35] know
[49:36] the first two phrases were low power and
[49:38] hybrid solutions
[49:40] and so pure play silicon photonics pure
[49:44] plane and phosphate i think we're going
[49:45] to see hybrid solutions
[49:47] folks are going to look at ways to
[49:49] upgrade their platform
[49:51] now i've just talked about a modulator
[49:52] but i think there's other ways to do
[49:54] that too
[49:55] and so you know folks are going to look
[49:57] for creative ways to do that
[49:59] we have to get faster devices we have to
[50:01] get lower power we have to get better
[50:03] at thermal performance and these are
[50:06] criterias that are not going to change
[50:08] and so the challenge is on so i think
[50:10] you're absolutely right jose we've seen
[50:12] a lot of different innovative solutions
[50:14] there
[50:14] and this is exciting times i wanted to
[50:16] start with you because now we know where
[50:18] we are going i know we know the
[50:19] challenges and now we know what
[50:21] what the industry is going to look like
[50:22] in 2028 so let's now bring all the
[50:25] different parts of the supply chain
[50:27] michael come with me we are going to go
[50:28] to juniper networks
[50:30] we are going to get my friend shinjuan
[50:32] meng to talk to us about how
[50:34] to put all these different pieces of the
[50:35] puzzle together and how to help
[50:38] juniper be even greater the floor and
[50:41] the attention of
[50:42] everyone goes to see you and ben goes to
[50:44] you
[50:49] okay okay thank you
[50:52] uh thank you uh it's nice to meet
[50:55] everyone here uh
[50:59] i'm from juniper uh the
[51:02] writer of uh
[51:06] uh foreign
[51:09] uh specification uh let me see
[51:16] uh can you see my screen yes christina
[51:20] okay uh okay uh disclaimer
[51:23] it's a routine one uh
[51:27] here uh i want to first
[51:30] i want to introduce the progress on the
[51:33] 400 g
[51:35] transfer modules here
[51:39] these are available from features
[51:42] on the market currently the main stream
[51:46] from feature is osip and the ksvdd
[51:51] kvdd has three type uh
[51:54] tawa the size of taiwan
[51:57] is the same with the legacy ksp plus
[52:01] kv-28
[52:03] tribal two sizes bigger
[52:08] it makes the nose outside of the cage
[52:12] longer so currently
[52:16] most modules use the type 2 so
[52:20] module integrator could put
[52:23] all the components into
[52:26] this type of kvdd form factor
[52:30] so from the beginning it seems
[52:33] um the task
[52:36] is daunting because we have a new
[52:40] uh optical space specs uh
[52:43] such as pam4 and also 50
[52:47] perlin electro interface second
[52:50] we have a new form factors crazy
[52:53] kpd and osap uh
[52:56] the third we have this will
[53:00] uh need a new managed module management
[53:03] interface
[53:04] so uh surprisingly the progress
[53:08] is quite smooth smooth uh so
[53:11] let me go to next page
[53:15] so uh we begin from the
[53:19] uh double density legacy energy optical
[53:23] pulse
[53:26] because we think uh this would be easier
[53:29] uh to start uh uh
[53:33] these uh modules uh have
[53:36] um a new uh form factor kftd
[53:41] and a new main
[53:44] interface but uh legacy optics so
[53:48] uh currently uh juniper has qualified
[53:52] uh 2x1 gsr4 twin digital dm4 to any lfo
[53:58] so double the capacity of
[54:01] 1g optics
[54:05] next we qualified a 100g single lambda
[54:10] kb 28 optics uh this co
[54:13] optics uh could be plugged into the
[54:17] legacy kp28 uh
[54:20] cage and also uh there could be also be
[54:23] inserted into
[54:24] this qrp dd's cages
[54:27] so uh with the legacy electric interface
[54:32] and with the new 100g painful of its
[54:35] uh their car interoperates
[54:38] with uh for bioenergy
[54:42] objectives of 5g platforms so currently
[54:46] we qualify uh one gtr one jfr
[54:49] one dlr1
[54:52] next page ah this is a foreign g optics
[54:56] uh with 50g electrical lens
[55:00] uh this is uh we create
[55:05] the whole portfolio of hiring
[55:08] optics uh with different
[55:11] link distance uh currently we qualify
[55:15] 400 dr4 from the fr4
[55:19] lr8 and the phobia gfr
[55:22] so actually 400 uh gdr4
[55:25] uh for the same module uh it can
[55:29] also can support 4x1 gtr
[55:35] okay next page so uh
[55:39] i want to introduce uh uh
[55:43] the new manager interface of this kfd or
[55:47] sap
[55:49] modules with the
[55:53] effort of the whole industry
[55:56] we made a common many many interface
[55:59] specifications
[56:01] uh it
[56:04] is now uh being used by uh
[56:07] kfcd or sap cover kv56
[56:11] and sfg dd format with
[56:15] we use a simpler simplified version of
[56:18] cms
[56:19] this new main menu interface introduce
[56:24] more customer facing features
[56:28] i want to uh
[56:31] introduce uh main two main new features
[56:34] uh
[56:35] uh these features are is quite different
[56:39] friendly to module users
[56:44] why is uh this
[56:47] uh similar
[56:51] make it possible that for the same
[56:54] module
[56:54] it can support multiple applications
[56:58] so each application has a
[57:02] host interface id and a media
[57:05] interface id hosts hardware
[57:09] just to select different operating code
[57:12] so uh to swap
[57:16] among different applications
[57:19] uh
[57:22] for example foreign gsr8
[57:25] it can support a foreign g protocol
[57:28] right
[57:29] it can also support two 200g sr4
[57:33] or 400 gsr2
[57:36] or ua 850gsr
[57:41] foreign gdr4 uh could support
[57:44] uh pharyngeal objects and also could
[57:47] support 400 gtr
[57:49] so this gives the mode user flexibility
[57:52] to do the configuration
[57:56] the next new feature is from from
[58:00] update we all have the experience
[58:04] on the software update of the cell phone
[58:07] right
[58:09] like the cell phone the devices
[58:13] are more and more relying on the
[58:18] former object to for example fix
[58:22] the problem or add a new application
[58:26] so uh this feature uh
[58:29] use a cb function
[58:32] defined in cma's format zero and foreign
[58:35] wine
[58:38] of this double items uh include the
[58:41] microcontroller code the sap code marine
[58:44] map
[58:45] and uh so not
[58:48] before we passive here give you 28
[58:51] modules
[58:52] when we found some problem we have to
[58:55] create physically create
[58:57] the modules from around the world
[59:01] to understand them back to the module
[59:03] integrator
[59:07] so it's it's not good so
[59:10] currently with this format update
[59:14] feature so the modules could be updated
[59:18] remotely by host software so
[59:21] this the same software code
[59:24] is a very applicable to different
[59:27] objects
[59:27] from different windows so far uh
[59:32] uh junior house software engineer had
[59:35] implemented this feature
[59:37] successfully so it's quite
[59:39] straightforward
[59:42] okay uh this is my presentation
[59:45] thank you thank you very much
[59:49] yeah okay so thank you very much for
[59:53] your
[59:53] presentation i guess that when telephony
[59:56] guy is asking for and reducing the cost
[59:58] per bit
[59:59] you feel the pressure right yeah yes
[01:00:02] yeah yeah
[01:00:03] fabricating your transceivers okay so
[01:00:05] let me know in order to achieve this no
[01:00:08] in order to reduce the
[01:00:09] cost per bit what would be the the
[01:00:12] technologies well the technologies what
[01:00:14] is the aspect
[01:00:15] of the packaging of the transceiver or
[01:00:17] maybe of the design of the transceivers
[01:00:20] that you want to take into account
[01:00:22] i am asking because we have here plenty
[01:00:24] of companies one thing they want that
[01:00:26] want to collaborate with you
[01:00:28] so which will be your your
[01:00:31] preferred collaboration
[01:00:34] yeah uh currently
[01:00:37] all 400 objects uh
[01:00:40] most of us are still using separate
[01:00:43] components
[01:00:46] so to further
[01:00:50] decrease uh i
[01:00:53] still need the a4 effort
[01:00:57] from outside to for example
[01:01:00] uh the dsp
[01:01:05] has a uh cost much right
[01:01:09] also the power consumption of the sap uh
[01:01:12] is large now so currently
[01:01:15] uh the industry is uh using
[01:01:19] uh 14 nanometer dsp and uh
[01:01:23] current as the currently the seven
[01:01:25] nanometer also available
[01:01:27] uh so uh in order to uh
[01:01:31] to decrease the power construction i
[01:01:34] need to continue to decrease
[01:01:37] the power consumption of dsp also
[01:01:41] uh i think
[01:01:44] many colleagues mentioned the uh
[01:01:48] integrity objects right uh
[01:01:52] in reality uh i think uh uh
[01:01:58] cause the reductions to uh issue
[01:02:01] the mass production right so how to
[01:02:04] make it real uh to be a mass producing
[01:02:08] uh technology uh also
[01:02:11] uh to decrease the cost below the
[01:02:14] several components
[01:02:16] yeah yeah well you cannot imagine
[01:02:20] a how active was the chat during your
[01:02:22] presentation so i don't know how
[01:02:24] to start the course do you want to make
[01:02:26] it i want to start with valerie i want
[01:02:28] to start with valerie he made a very
[01:02:29] interesting comment about
[01:02:31] package optics valerito ticking from
[01:02:34] intangent
[01:02:34] one of the indian phosphide suppliers
[01:02:36] that we have in our network
[01:02:38] valerie when it turns when it comes to
[01:02:40] call package optics you had a very
[01:02:42] interesting comment could you could you
[01:02:44] share your mind with the rest
[01:02:50] i don't know we're talking online after
[01:02:52] mute yes
[01:02:53] do you hear me i hear you and i see you
[01:02:56] and i'm so happy to see you
[01:02:58] hi jose and hello everybody yeah i don't
[01:03:01] think it's a very original comments
[01:03:03] which we made i believe it's pretty
[01:03:05] common these days
[01:03:06] people who spend you know as many time
[01:03:09] as much time as i did with all these
[01:03:10] plugables
[01:03:12] ask themselves where does it go so i
[01:03:14] don't believe personally that you can
[01:03:16] squeeze more than 800 g per lane
[01:03:20] into pluggables whatever it is for
[01:03:23] c8 or 16. so what's next so with all
[01:03:27] these thermograph
[01:03:28] switches and likes going into
[01:03:31] many tenths of terabit what do you do
[01:03:35] so i i do not see how pluggables can do
[01:03:37] this so
[01:03:38] co-packaging or anything like that i
[01:03:41] believe
[01:03:41] is logically comes up as a solution and
[01:03:46] besides i believe
[01:03:47] that if silicon photonics per se
[01:03:50] has a real advantage this is where you
[01:03:54] have
[01:03:54] this kind of very massive parallel
[01:03:56] integration like in this co-package
[01:03:58] devices so that's the view i believe
[01:04:01] shared by many by this time there is a
[01:04:04] view shared by by many
[01:04:06] not by everyone but by everyone there
[01:04:07] are many who actually consider cpo
[01:04:10] as an intermediary step but i want to
[01:04:13] steer the pot
[01:04:14] a little bit here i'm going to go to
[01:04:16] john john future way what's on your mind
[01:04:20] well i think this all that you know the
[01:04:22] form factors
[01:04:23] discussions and with discussions all
[01:04:26] played back into the original discussion
[01:04:28] that michael started when you start
[01:04:30] talking about the next rate
[01:04:32] okay and if you look at market adoption
[01:04:35] what we've seen is that the serial and
[01:04:37] the by four solutions have really been
[01:04:39] those
[01:04:40] that have been adopted okay and forget
[01:04:42] the form factors just saying in general
[01:04:44] if you go and look at that for example
[01:04:47] um a lot of the
[01:04:48] uh by eight solutions for 400 gig really
[01:04:51] aren't ramping up as quickly as some
[01:04:52] would like all right so
[01:04:54] i think that you can always achieve
[01:04:58] a higher speed through an aggregation of
[01:05:00] lanes but then it comes back to the
[01:05:01] other point that michael was making
[01:05:03] what's the cost and we're seeing this
[01:05:06] whole by four
[01:05:07] sort of thing so i i do think that um
[01:05:10] as we start to talk about the next
[01:05:11] speeds and we start to talk about the
[01:05:13] widths
[01:05:14] you know i don't know if the market is
[01:05:15] going to go to to buy eight solutions
[01:05:17] um i think on michael's road map he had
[01:05:20] showed 800
[01:05:21] then 1.6 that's one way of looking at it
[01:05:25] in the conversations that um i have had
[01:05:28] people look at the per lane rate of 200
[01:05:31] gigabit per second as really the key
[01:05:33] ingredient
[01:05:34] um and then you know and that then
[01:05:36] that's really starting from the
[01:05:38] the switching chip out you know going
[01:05:40] from electrical to the optical world
[01:05:42] um and then you start to think about
[01:05:45] okay well what am i going to put 200 in
[01:05:47] well if i put 200 in a by 4 it is a
[01:05:50] um 800 gig solution now if i put
[01:05:54] 200 into a by 8 solution the maximum
[01:05:58] capacity is 1.6
[01:06:00] so you don't look at it as the speed of
[01:06:03] the
[01:06:04] of the ethernet rate essentially in in
[01:06:06] my world i look at it as okay
[01:06:08] what is the lane rate and then what can
[01:06:10] i get into it because what we're seeing
[01:06:12] right now
[01:06:13] in discussions is that people are
[01:06:15] starting to gravitate
[01:06:16] and we're hearing more play about a buy
[01:06:18] eight solution
[01:06:20] and by eight i mean by a module for buy
[01:06:22] eight because they can then put in
[01:06:24] two 400 gigs so now it's a density play
[01:06:27] all right so it's not
[01:06:29] you can't just look at it as a single
[01:06:30] rate you've got to look through the
[01:06:32] through the applications so now as we
[01:06:34] start to talk about 801.6 coming
[01:06:36] together
[01:06:37] i see that next step not being 3.2 but i
[01:06:41] see that next step being
[01:06:42] 400 gig a bit per second per lane to get
[01:06:46] that 1.6
[01:06:47] down into a by four module okay
[01:06:50] so i i i those are my viewpoints that's
[01:06:53] why i
[01:06:54] i think this is a really interesting
[01:06:55] conversation and we've got to bring all
[01:06:57] of this stuff together so people have a
[01:07:01] a good idea of what their you know
[01:07:02] what's going on you can't just look at
[01:07:04] it from one perspective there are
[01:07:05] multiples that are happening
[01:07:06] we are of course asking the million
[01:07:08] dollar question we don't expect to have
[01:07:10] the million dollar answer
[01:07:11] because that will give us million
[01:07:12] dollars but we all have a very
[01:07:15] interesting approach here i saw a very
[01:07:17] interesting comment from karen liu from
[01:07:19] lightweight logic
[01:07:21] so karen hello how are you doing today
[01:07:23] hey jose good to see you guys
[01:07:26] i just wanted to to follow up on the
[01:07:29] discussion i
[01:07:30] these are not my thoughts i i saw two
[01:07:32] very interesting talks
[01:07:33] in the last week and when i put them
[01:07:35] together in context i think
[01:07:37] it very much relates to what john was
[01:07:39] saying um one was a chart from juniper
[01:07:41] who i'm sure that you know perhaps uh
[01:07:43] genma we'll talk to that instead i think
[01:07:46] i saw jeff mackey present
[01:07:48] a chart in which he showed the switch
[01:07:51] basic capacity over time and then showed
[01:07:55] how um
[01:07:56] the you know by one by four and then by
[01:08:00] eight was actually required to keep
[01:08:02] up because the slopes were different so
[01:08:03] you know maybe juniper could speak to
[01:08:05] that one
[01:08:06] and of course you know i've looked at
[01:08:07] arrays for a long time
[01:08:10] and one of the frustrating things has
[01:08:13] always been that
[01:08:14] um photonics as you know has has
[01:08:17] struggled with
[01:08:19] certain fundamental issues and and it's
[01:08:23] impacted the ability to build
[01:08:25] arrays um in terms of both yield
[01:08:28] and then also in terms of failure and so
[01:08:31] you needed to keep the granularity low
[01:08:33] for
[01:08:33] field replaceable units and so forth for
[01:08:36] years i thought you know we were just
[01:08:37] stuck and there was no hope but then i
[01:08:39] saw this talk from
[01:08:40] facebook in which there was sort of a
[01:08:42] glimmer of hope right
[01:08:44] that the observation was yeah it's all
[01:08:46] this data historical data we've had for
[01:08:48] decades
[01:08:48] are coming from direct mods and now the
[01:08:50] world is different right so not that
[01:08:52] we're not
[01:08:53] no longer abusing these laser junctions
[01:08:55] right in the same way
[01:08:57] maybe things can be more reliable he's
[01:09:00] not seeing the same failures
[01:09:01] once we go into you know silicon
[01:09:03] photonics and external mods and
[01:09:05] possibly even amls you know in a
[01:09:08] previous
[01:09:09] online meeting we have been sensing
[01:09:10] presenting about the reliability role
[01:09:12] in beyond 400g so it was a great
[01:09:15] presentation we have everything
[01:09:17] in youtube there is a lot to discuss on
[01:09:19] that very point and i do think
[01:09:21] there are more points to be addressed
[01:09:23] but everything is going to come together
[01:09:25] much more clear when we hear from
[01:09:29] jim theodoras and gene has been one of
[01:09:32] these
[01:09:32] souls that we connected during the
[01:09:34] online technology meetings
[01:09:36] he got connected to so many epic members
[01:09:38] and he followed up with all of us
[01:09:40] you are one rock star jim thank you very
[01:09:42] much into the rest for an edgy genuine
[01:09:44] for being with us today
[01:09:46] what's on your mind
[01:09:49] well i first wanted to uh comment on the
[01:09:52] previous statement
[01:09:53] and that was um yes as you get wider
[01:09:58] it's not just reliability but also the
[01:10:01] i call it the hidden cost of complexity
[01:10:03] and it's not an exponential
[01:10:05] it's an it's a factorial so
[01:10:08] i currently make a lot of transceivers
[01:10:10] that have four you know four channels
[01:10:13] and the the odds that one of the four
[01:10:16] channels
[01:10:17] are marginal and remember this would be
[01:10:19] like mass production at hot
[01:10:22] wit at the ninth corner of you know low
[01:10:24] voltage and
[01:10:25] hot case and everything is at its
[01:10:27] extreme
[01:10:28] to get everything all the stars to align
[01:10:31] across four channels
[01:10:32] really as dramatically to
[01:10:35] um the end cost of the module and these
[01:10:37] are hidden costs because it may not show
[01:10:39] in the bomb but those yield hits
[01:10:41] are eventually uh do make their way to
[01:10:44] the end customer so
[01:10:46] i i cringe when i hear eight channels
[01:10:49] not because it's not doable
[01:10:51] but um and even not because of the
[01:10:53] component count but just because
[01:10:55] my history of always seeing trying to
[01:10:58] get
[01:10:59] four of something for instance into cwm4
[01:11:03] to to yield out and meet customers
[01:11:06] expectations on pricing
[01:11:08] to try to do that with eight um oh my
[01:11:11] goodness especially when they're eight
[01:11:12] different colors
[01:11:13] wow um let alone 16. so
[01:11:16] that scares me a bit
[01:11:20] well let's see you you ever need to to
[01:11:22] do the testing of these devices i can
[01:11:24] connect you with all the different
[01:11:25] companies
[01:11:26] from from road microtech to vlc
[01:11:28] photonics will make it happen
[01:11:29] but indeed if you are scared a company
[01:11:31] genuinely is scared to go to the 16
[01:11:33] channel then we all should be a bit
[01:11:35] concerned
[01:11:36] yes yes so it wasn't clear do you want
[01:11:39] me to uh pop-up slides now or do you
[01:11:40] want yes i want to see your presentation
[01:11:42] because after this i expect a really
[01:11:44] really strong discussion you always set
[01:11:46] the
[01:11:46] bar very high for a good communication
[01:11:48] okay
[01:11:50] let me uh pull this up
[01:11:56] nice cycling i live in the netherlands
[01:11:57] when i see this picture i feel
[01:11:59] represented
[01:12:01] yeah i'm showing that because i uh my
[01:12:03] zoom is updated and it's not showing the
[01:12:05] typical
[01:12:06] zoom so is that full screen for all of
[01:12:07] you yes chris are clear
[01:12:10] okay so just real quickly uh so hd
[01:12:13] genuine is not a common name so just to
[01:12:15] briefly say who we are
[01:12:16] so hd genuine is the optical transceiver
[01:12:20] part of a larger company called hd tech
[01:12:23] it's based in
[01:12:24] wuhan china and uh the parent company is
[01:12:28] listed on the shenzhen stock exchange
[01:12:31] a few years ago they had they pretty
[01:12:35] much dominate the pond market in china
[01:12:37] and they wanted to grow and there's two
[01:12:39] ways of doing that you can
[01:12:40] grow geography and or you can grow
[01:12:44] market segments and they quite
[01:12:48] showed some courage and decided to try
[01:12:50] to do both simultaneously so we
[01:12:51] moved into datacom as well and moving
[01:12:53] outside of china simultaneously
[01:12:56] which um was a bit risky but i think
[01:12:58] it's it's paid off and we're seeing a
[01:13:00] lot of success
[01:13:01] um uh so we just opened a new 100
[01:13:04] million dollar factory
[01:13:05] um just dedicated to 400 gig
[01:13:08] um in higher speed ethernets is just
[01:13:10] datacom so
[01:13:11] um our uh our current two factories are
[01:13:13] full upon and other things
[01:13:15] but um for datacom we have a brand new
[01:13:18] factory
[01:13:18] um so this is the information we took
[01:13:21] over it was an old building that was a
[01:13:23] printing press that made newspapers and
[01:13:26] you think well
[01:13:27] how could you do optics there well the
[01:13:29] floors could hold the weight of our
[01:13:30] machinery
[01:13:31] and there was a lot of power available
[01:13:33] because printing presses took a lot of
[01:13:34] power and
[01:13:35] it went from this dusty old moldy
[01:13:38] newspaper factory to now being a
[01:13:40] state-of-the-art data com
[01:13:42] factory we have uh 14 100 meter long
[01:13:46] production lines of which remember i
[01:13:49] said that the cost of complexity
[01:13:51] so our more complicated parts actually
[01:13:53] have two full 100 meter lines where
[01:13:56] the optical engine consumes an entire
[01:13:58] line and then the transceiver consumes
[01:14:00] an entire line so that's
[01:14:01] a lot of equipment um just to to make a
[01:14:05] transceiver
[01:14:06] so these are some pictures from the
[01:14:08] factory these are screen grabs from the
[01:14:10] video because it's grown a lot since
[01:14:12] the lockdowns and so i've not seen it in
[01:14:15] a bit and these uh were some screen
[01:14:17] grabs showing that
[01:14:18] um they're uh they're going full steam
[01:14:21] there
[01:14:21] so we've actually pre-sold um uh
[01:14:25] 60 of the capacity has already been
[01:14:27] pre-sold for the next three years with
[01:14:29] another 15 percent
[01:14:30] um on allocation so we're sitting about
[01:14:34] 85 percent um capacity uh and we're
[01:14:37] still
[01:14:38] we're still finishing up production
[01:14:39] lines so the point i want to make this
[01:14:41] is actually one of jose's slides that
[01:14:43] i've edited from our previous webinar is
[01:14:46] that
[01:14:46] people think oh you build a transceiver
[01:14:48] and you don't just buy parts
[01:14:51] and then build a transceiver that would
[01:14:53] make you a contract manufacturer um
[01:14:55] trans
[01:14:56] to get these latest greatest uh
[01:15:00] data com transceivers it's an entire
[01:15:02] ecosystem has to support that
[01:15:04] um uh so i talked a lot last time about
[01:15:08] the optical engines so there's a lot of
[01:15:10] various technologies that go into making
[01:15:12] an optical engine work but there's also
[01:15:14] the cabling like you can't do a dr4
[01:15:16] without a senko connector
[01:15:18] the vixel technology needed to support
[01:15:20] not only active optical cables but our
[01:15:22] highest volume
[01:15:23] part is an sr4 100 gig sr4
[01:15:26] just the sheer vixel technology that
[01:15:29] goes into that
[01:15:31] and people often forget about the other
[01:15:33] supporting
[01:15:34] things such as market intelligence
[01:15:36] because we have to build and buy
[01:15:38] our parts way in advance of actually
[01:15:40] needing them and we can't do that with
[01:15:42] the market intelligence reports
[01:15:44] um uh that did testing so i know jose
[01:15:47] you bring up a lot about the taste and
[01:15:48] capabilities of the eptic members um
[01:15:51] not just at the transceiver level but
[01:15:53] i'm glad the topic of wafer scale
[01:15:55] testing came up
[01:15:56] because i'm not sure how many of you
[01:15:58] know but our
[01:15:59] end customers require us to provide
[01:16:02] wafer level
[01:16:03] test data and we we'd only make the
[01:16:05] wafers so that means our end customer
[01:16:08] before they our hyper scales that buyer
[01:16:10] transceiver
[01:16:11] they want to see in the cloud uploaded
[01:16:14] all the way back to if we we buy a tosa
[01:16:17] and then we then part of that's a laser
[01:16:19] and that comes from another vendor that
[01:16:20] came from a wafer
[01:16:22] the source wafer that
[01:16:25] ends up being our product that data must
[01:16:27] be uploaded
[01:16:28] in the cloud and there's these ai
[01:16:30] routines constantly sifting and sorting
[01:16:32] and
[01:16:32] and giving that data and showing is the
[01:16:34] process
[01:16:36] tight is the process still in control
[01:16:38] and i can't sell a transceiver without
[01:16:40] that wafer level test date even though i
[01:16:42] don't make wafers
[01:16:43] so i'm glad the topical way for level
[01:16:45] testing came up
[01:16:46] it's an ecosystem and i think um
[01:16:49] all the all the members of epic need to
[01:16:51] come together and work together to
[01:16:53] to create this ecosystem that's going to
[01:16:56] make these
[01:16:56] these 800 gig and 1.6 terabyte and god
[01:16:59] forbid 3.2
[01:17:01] terabit uh transceivers um so i just
[01:17:04] wanted to quickly talk about
[01:17:05] to know where you're going you have to
[01:17:06] kind of know where you've been and so
[01:17:09] this is a slide i like to show because
[01:17:11] it shows what we used to do
[01:17:12] what we do today and maybe what we're
[01:17:14] going to do tomorrow
[01:17:15] because data center isn't just one set
[01:17:18] of data com optics there's different
[01:17:20] parts of their networks that have very
[01:17:22] different needs which
[01:17:23] can get confusing sometimes so i i split
[01:17:26] them up into what's duplex
[01:17:29] and protected by like two to one three
[01:17:31] to one end by protection
[01:17:32] and then what has to fan out because in
[01:17:35] one case you care about
[01:17:37] uh how much can i get on a duplex fiber
[01:17:39] and i still have to go pretty far
[01:17:41] in the other case you can about i've got
[01:17:42] to touch as many ports as i can
[01:17:44] so this is what we used to do what we do
[01:17:47] today and where we're headed tomorrow
[01:17:48] and i don't want to spend too much time
[01:17:50] on that what i want to do is instead
[01:17:51] jump to this
[01:17:53] because if i really put on and this you
[01:17:55] know this is epic so i've got to put on
[01:17:56] my really longer term hat and say
[01:17:58] where's this headed so that we make the
[01:18:00] investments and technologies today
[01:18:02] so we get to that 3.2 terabit and that
[01:18:05] is
[01:18:06] there's there's this divergence in
[01:18:08] what's needed for the data center and we
[01:18:10] saw that
[01:18:11] 10 years ago john walked all of us
[01:18:13] through that one
[01:18:14] when the optics need diverged from
[01:18:16] telecom also meeting data compton you
[01:18:18] know what datacom means something a
[01:18:20] little different than telecom
[01:18:21] and today we have two different product
[01:18:23] sets i think that's happening in data
[01:18:25] center optics as well to where
[01:18:27] everything in green here is you know
[01:18:30] what i need duplex connectivity i need
[01:18:32] to go distances
[01:18:33] um but i still need um you know i still
[01:18:36] need a lot of bandwidth and speed
[01:18:37] um and that's very different than below
[01:18:41] in yellow here is
[01:18:43] i need to touch as many ports as
[01:18:45] possible because i need what's called
[01:18:47] fan out uh it's known as breakout in the
[01:18:49] industry but to me it's fan out you're
[01:18:51] you want to touch everything and so as
[01:18:54] we
[01:18:54] as i put in my long-term hat here's my
[01:18:56] dunce cap um
[01:18:57] i see z you know coherent eventually is
[01:19:00] going to meet
[01:19:01] most of these needs which is good for
[01:19:03] epic members because we've seen all
[01:19:05] these phenomenal technologies that's
[01:19:07] going to play into
[01:19:08] 480 800 gig zr coherent uh and then
[01:19:12] also anything that must fan out wide for
[01:19:14] these ai and these uh switch
[01:19:16] matrixes and i mean google now is making
[01:19:18] their these fan out stories again just
[01:19:20] like the old cray supercomputers used to
[01:19:22] be which i find really
[01:19:24] ironic but it's it's cool too um that
[01:19:27] is going to be you know maybe not cpo
[01:19:30] maybe not kobo but it's going to be
[01:19:31] something
[01:19:32] very much like it it's going to be
[01:19:34] something co-packaged something
[01:19:35] integrated
[01:19:36] and again this is great news for the
[01:19:38] epic companies because
[01:19:39] all those technologies are going to be
[01:19:41] needed to get to that level of
[01:19:43] integration
[01:19:44] so that's why i'm glad when we're
[01:19:45] talking about the polymer
[01:19:47] modulators it was brought up dry voltage
[01:19:50] because
[01:19:51] it's not just about the switch power
[01:19:53] it's not just about
[01:19:54] how much power the modulator needs it's
[01:19:57] also
[01:19:58] the overall channel power and so
[01:20:01] co-package gets rid of driving a trace
[01:20:03] on a board okay i've saved some power i
[01:20:05] removed the driver
[01:20:06] um but there's still drivers for the
[01:20:08] modulators okay going to load drive
[01:20:10] modulators
[01:20:11] i got rid of those drivers um on these
[01:20:13] little chiplets on my coat package
[01:20:15] now i can drive straight from the
[01:20:16] tomahawk but there's still certies on
[01:20:18] the tomahawk driving probably
[01:20:20] 75 percent of the power of the tomahawk
[01:20:22] so oh well my dry voltage got even lower
[01:20:24] down to
[01:20:25] like uh you know um you know
[01:20:28] to to uh logic levels now i can get rid
[01:20:30] of the surreys
[01:20:32] now being fully integrated makes a lot
[01:20:34] more sense because i'm
[01:20:35] switched straight driving into what's
[01:20:39] going to make my optical signal and then
[01:20:41] maybe
[01:20:41] that makes sense so you know long story
[01:20:44] short this is good news for all the epic
[01:20:45] members
[01:20:46] i think we all need to come together and
[01:20:48] and you know make this a reality
[01:20:50] and i think we have all the right talent
[01:20:52] on these calls and thank you to epic for
[01:20:54] for uh driving this and making this a
[01:20:56] reality no thank you to you jim for
[01:20:59] giving these extraordinary presentations
[01:21:01] um okay so then i i like it very much
[01:21:04] the
[01:21:04] the slide in which you put our our parts
[01:21:08] of the supply chain at least the parts
[01:21:10] of the supply chain in which
[01:21:11] you are interested not for developing
[01:21:13] your transceivers so
[01:21:14] let me focus on this because probably
[01:21:16] you you are working with you need
[01:21:19] software right a software for designing
[01:21:22] okay so we have here aura from synopsis
[01:21:27] how are you here yes hayanna
[01:21:30] and maybe do you think this is a good
[01:21:32] opportunity for you
[01:21:33] to to talk about which are the
[01:21:34] capabilities of a synopsis software
[01:21:37] maybe do you have one slide yes of
[01:21:40] course
[01:21:40] uh let me capture my
[01:21:43] screen
[01:21:49] if you can see this one
[01:21:54] yes but this is the note mode if you
[01:21:56] come now
[01:21:57] very good yeah now you see the normal
[01:22:00] one right
[01:22:01] okay perfect yes as uh as it was
[01:22:04] mentioned uh
[01:22:06] getting into the ecosystem all the parts
[01:22:08] is really important one of them is
[01:22:10] software so hi everybody i'm aura i'm
[01:22:13] working at synopsis photonics solutions
[01:22:16] and i'm a technical account manager
[01:22:18] but without further ado let me introduce
[01:22:21] you the
[01:22:22] synopsis industry's first unified peak
[01:22:26] design platform
[01:22:27] so actually at synopsis we are
[01:22:29] accelerating the adoption of photonic
[01:22:31] and
[01:22:32] peak technologies with software to
[01:22:34] design and manufacture
[01:22:35] energy efficient high performance
[01:22:38] photonic devices
[01:22:39] systems and integrated circuits so the
[01:22:41] photonic solutions portfolio that we
[01:22:43] have
[01:22:44] offer a seamless design flow from
[01:22:47] concept to manufacturable design
[01:22:49] and it is supported by photonic experts
[01:22:52] as you can see on the slide
[01:22:54] so with synopsis software uh all the
[01:22:57] design teams have access to a widely
[01:22:59] used
[01:23:00] high quality electronic and photonic
[01:23:02] design solutions from a single provider
[01:23:05] and with a single support channel and
[01:23:07] you can see here
[01:23:08] our newly unveiled product which is
[01:23:11] named optocompiler
[01:23:12] this is the electronic photonic cockpit
[01:23:14] of synopsis
[01:23:16] and it offers an interactive environment
[01:23:19] enabling electronic photonic co-design
[01:23:22] from system
[01:23:23] from schematic to the layout photonic
[01:23:26] aware could be linear layout synthesis
[01:23:28] and a complete solution from concept to
[01:23:30] validated and verified mask
[01:23:32] so this is uh quite an important step on
[01:23:35] the peak value chain
[01:23:36] and of course if if you are interested
[01:23:39] please contact us so we will be happy to
[01:23:41] to support you on developing your
[01:23:44] photonic design flow thank you very much
[01:23:48] okay so now um we have the software
[01:23:51] and i would like now to go for the
[01:23:55] packaging and we have
[01:23:56] here today our company barry optics and
[01:23:59] we have here nicolas
[01:24:02] hello yes okay so can you tell us a
[01:24:06] little bit about your capabilities for
[01:24:08] packaging and maybe
[01:24:09] you also have a one slide to show yes i
[01:24:11] hope you see my slide here
[01:24:13] so thanks for this opportunity um we are
[01:24:17] at the various we are developing and
[01:24:18] manufacturing electro-optic circuit
[01:24:20] boards
[01:24:21] um so this is a platform that can be
[01:24:23] used to couple
[01:24:25] picks um to the outside world on the one
[01:24:28] hand it you can for example
[01:24:30] think of a picture you want to interface
[01:24:32] with a fiber array but
[01:24:34] as you see here on the left you could
[01:24:35] also think of a more complex board
[01:24:38] with more than one pick and what you
[01:24:41] need then is a
[01:24:43] way to optically communicate between
[01:24:45] these peaks
[01:24:46] so our solution here consists of a
[01:24:48] electro optic
[01:24:49] circuit board platform that means you
[01:24:51] have a pcb you can also have a rather
[01:24:54] complex electrical circuit
[01:24:57] and we add on this our planar waveguide
[01:24:59] technology
[01:25:00] based on on polymers which can be used
[01:25:03] as fan out structures they can be used
[01:25:06] to efficiently couple to the peak
[01:25:08] by changing the pitch or the remote view
[01:25:10] diameter
[01:25:13] and all of this you see it here on the
[01:25:14] right really enables this complete
[01:25:17] platform of having a simultaneous
[01:25:19] electrical
[01:25:20] and optical interface so this is an
[01:25:23] example that we realized
[01:25:25] in a consortium you see assuming here
[01:25:28] of the area where the peak can be
[01:25:32] mounted and you see here finally you see
[01:25:35] a
[01:25:36] fully assembled device here with a flip
[01:25:38] chip
[01:25:39] bonded pick that is used here and that
[01:25:41] results
[01:25:42] so our partners have measured here nice
[01:25:45] data rates
[01:25:45] also with very low power consumptions so
[01:25:48] what we offer
[01:25:49] to answer this these epic questions to
[01:25:52] the epic members is really this complete
[01:25:54] platform of having an
[01:25:56] electro optical circuit board and what
[01:25:59] we are looking for are partners
[01:26:01] that have chips that have an application
[01:26:04] in mind
[01:26:05] and that needs such an innovative and
[01:26:07] versatile
[01:26:08] packaging platform so if you if you feel
[01:26:12] like you are in this
[01:26:13] region you feel free to catch up and get
[01:26:16] in contact with me
[01:26:19] thank you very much nicolas for this for
[01:26:22] this
[01:26:22] slide okay yes of course if any of the
[01:26:24] companies and making transceivers
[01:26:26] want to contact with barry optics so
[01:26:28] also with synopsis
[01:26:30] uh we will be very happy to make the
[01:26:32] introduction for you
[01:26:33] and now let's go for one of the
[01:26:35] questions that is in the chat
[01:26:37] is so we have a chris here from
[01:26:40] cta chris would you like
[01:26:44] to to ask your question about the volume
[01:26:48] capabilities
[01:26:51] okay i i don't see increase here but the
[01:26:54] game what about your production
[01:26:58] capability
[01:26:59] for 100 gm on 100 and 400 g
[01:27:03] in these new facilities sorry i was on
[01:27:06] sorry i was on newt ah hello greece
[01:27:09] uh hi jim how are you fine how are you
[01:27:13] doing
[01:27:14] not too bad so i was just yeah
[01:27:15] congratulating you on your on your new
[01:27:18] facility and i don't know because one of
[01:27:19] the big topics that
[01:27:21] we always have i've lived in this space
[01:27:23] for quite a while cti right now is my
[01:27:26] consulting company i'm advising a few
[01:27:28] companies in the uh
[01:27:29] transceiver space but one of the big
[01:27:31] challenges we have on the transceiver
[01:27:33] side is always the substantial upfront
[01:27:35] investment on capex and i just saw you
[01:27:37] mentioned
[01:27:38] on the order of 100 million dollars to
[01:27:40] get ready for the
[01:27:42] the wave of 100 g and 400 g modules
[01:27:45] can you share roughly what capacity that
[01:27:49] that translates to i don't think it's
[01:27:53] public information yet
[01:27:54] and we are publicly listed so i will
[01:27:57] have to be cautious
[01:27:58] i will say we uh we have the capability
[01:28:02] of doing
[01:28:03] um a half million cwdm4 modules
[01:28:07] per year and that is not even half of
[01:28:10] that factory so
[01:28:12] there's a lot of capability there and
[01:28:14] when i say 100 million obviously that's
[01:28:16] not
[01:28:16] wasn't uh something where you walked you
[01:28:19] know into a bank and just got that money
[01:28:21] it's uh it's done the same way that most
[01:28:24] governments do it where
[01:28:25] uh if you can show there's investment
[01:28:28] it's released in trenches
[01:28:29] if you can show you're hiring people and
[01:28:31] you're you're building something that
[01:28:33] then
[01:28:33] benefits the economy um and you're
[01:28:35] buying uh
[01:28:37] equipment locally things like that so
[01:28:40] i think we've burned through about 85
[01:28:42] million of the 100 million
[01:28:44] uh credit line and we uh have most of
[01:28:47] the 16 lines in
[01:28:49] so we're just kind of uh getting going
[01:28:52] there but we do have uh we have a 100
[01:28:55] gig
[01:28:56] uh sr4 line running full speed
[01:28:59] and uh we have the cwm4 line running but
[01:29:03] when you run that many parts that fast
[01:29:05] you got to do a re-qualification
[01:29:07] of the parts to qualify the process that
[01:29:10] makes them not the parts or design
[01:29:11] themselves so
[01:29:13] that's currently on ongoing it takes at
[01:29:15] least two thousand hours
[01:29:16] on a large statistical sampling to do
[01:29:19] that
[01:29:20] um and we've we've had already had
[01:29:23] hyperscale auditors come through
[01:29:25] um because they use a lot of cd4 parts
[01:29:28] as well and active optical cables
[01:29:30] and our sr4 is the basis for our 100 gig
[01:29:33] active optic cable
[01:29:34] so i guess it's a long way of answering
[01:29:36] can't tell you but i mean
[01:29:37] we're easily a million parts a year and
[01:29:40] that's probably not the full capacity
[01:29:42] sure no thank you i think it's also good
[01:29:44] to illustrate for people who are maybe
[01:29:46] not as close to the action on the
[01:29:48] production side on
[01:29:49] what the substantial investments are and
[01:29:52] the
[01:29:53] first they all have lead times and then
[01:29:54] you have all the calls you do your own
[01:29:56] calls and then the customers want to do
[01:29:58] the calls so this is always a
[01:30:00] a massive uh a massive undertaking and
[01:30:02] sometimes when we
[01:30:04] when we talk about all the different
[01:30:06] generations we lose sight of
[01:30:08] um what that does to the uh the
[01:30:11] manufacturing side at the
[01:30:12] at the tail end before you can actually
[01:30:14] roll the product out the door
[01:30:16] oh yeah yeah thank you chris for
[01:30:18] bringing that up and anna cut me off if
[01:30:19] there's not
[01:30:20] time for this but it's important that
[01:30:21] everyone does understand that process
[01:30:23] you just brought up because
[01:30:25] we have to buy the equipment long in
[01:30:26] advance and then when the equipment does
[01:30:28] come in
[01:30:28] it just doesn't do what you need you
[01:30:30] have to build a fixture and write
[01:30:32] software
[01:30:33] to make it do what you want and then
[01:30:35] when that's even running
[01:30:37] product you're not allowed to sell that
[01:30:38] to customers you then have to what's
[01:30:40] called close the process
[01:30:41] where you run enough product that
[01:30:43] statistically you have a meaningful data
[01:30:45] set that shows everything's repeatable
[01:30:47] um and
[01:30:48] then you've got your reliability and
[01:30:50] even when you're at that point you have
[01:30:51] to start bringing in auditors because
[01:30:53] even if it's the same part you made
[01:30:54] somewhere else they want to audit the
[01:30:56] process to see that it's tight
[01:30:57] and as i mentioned earlier they even now
[01:30:59] want to see your wafer scale
[01:31:01] data even when we don't make the wafers
[01:31:03] so it's um
[01:31:04] there's a lot involved it takes a lot of
[01:31:06] planning and that's why i brought up
[01:31:07] also early it's important these industry
[01:31:10] consultants and you know data
[01:31:13] that are plotting and charting the
[01:31:15] future help us pre-plan so
[01:31:17] obviously the lead time on these
[01:31:19] machines can be a year or more so we
[01:31:21] have to pre-plan advance and get the
[01:31:22] credit lines and things like that it's
[01:31:24] very complicated process
[01:31:25] and i really thank you so much for
[01:31:26] bringing that up so that the epic
[01:31:28] members can understand all
[01:31:30] that's involved but we are not done
[01:31:33] because i still have another company to
[01:31:36] introduce to you
[01:31:37] a wilfred from sus micro optic
[01:31:41] how are you doing good very well thank
[01:31:44] you i had to laugh when i saw jim slice
[01:31:46] about the newspaper factory
[01:31:48] because that's what we just did so we
[01:31:50] got an old building from an old
[01:31:51] newspaper factory and got our
[01:31:53] equipment in there i just like to share
[01:31:55] a slider too is that okay or
[01:31:58] yes yes go ahead so uh what one is light
[01:32:01] please
[01:32:01] one slide okay when i show one slide
[01:32:03] then i'll show this one
[01:32:06] and uh i don't know if you see it
[01:32:12] can you see it yes yes oh sorry
[01:32:15] this is so this is what we offer so that
[01:32:17] means
[01:32:18] so we we we are we realized that in
[01:32:21] micro optics
[01:32:22] we're doing micro optics for
[01:32:23] transceivers so for telecom data com
[01:32:25] since
[01:32:26] 15 20 years now this is nothing new but
[01:32:28] what we see
[01:32:29] is all these packaging issues so that's
[01:32:31] why we try to integrate more and more
[01:32:34] features for packaging so this is all
[01:32:36] known
[01:32:36] for cwm4 and for prison but we
[01:32:40] also offer all of these these things in
[01:32:42] the front and back side
[01:32:43] but we we want to also show actually
[01:32:45] what what what is the future we believe
[01:32:48] so this is
[01:32:48] integrating mirrors directly into the
[01:32:50] silicon and we're also looking into
[01:32:52] other materials now where we can
[01:32:54] directly integrate a 90 degrees
[01:32:57] mirror in into the micro optics and the
[01:33:00] question is
[01:33:02] if this is something that co-packaging
[01:33:04] optics
[01:33:05] needs or not so this is this is still
[01:33:07] something i would like to do
[01:33:09] with uh different companies who offer
[01:33:12] co-packaging or looking for co-packaging
[01:33:14] optics
[01:33:15] this is basically my question how does
[01:33:19] the community see this so um
[01:33:21] the technology is out there it all has
[01:33:23] to be re-qualified
[01:33:24] like like jim just pointed out so you
[01:33:27] can't just come up with a new product
[01:33:28] and then it might work or might not work
[01:33:30] so it's
[01:33:30] there's a long way of course but i think
[01:33:33] there's there's
[01:33:34] we are still not at the end of what can
[01:33:36] be can be integrated in a single device
[01:33:39] so you can so for now micro optics was
[01:33:42] just lenses
[01:33:43] on on a slab but but now we can
[01:33:45] integrate more and more features just
[01:33:47] like you do on the pic
[01:33:49] and and there is where are we looking
[01:33:52] for more
[01:33:52] cooperations with other companies in the
[01:33:55] transceiver business
[01:33:57] thank you anna okay thank you very much
[01:34:00] this is a very very important question
[01:34:02] this is a very important question
[01:34:04] we have great technologies so micro
[01:34:06] optics a wafer level and also with two
[01:34:08] photon polymerization
[01:34:10] i think multiphoton optics my friend
[01:34:11] ruth is also in the room we have great
[01:34:13] technologies to do
[01:34:14] micro optics and bringing the optics
[01:34:16] very close to the to the pcb board
[01:34:19] is that something needed is there more
[01:34:22] high-end or more know-how on the micro
[01:34:24] optics jim what do you think is there a
[01:34:26] paradigm is there a need for
[01:34:28] new approaches for the micro optics
[01:34:30] bringing them into the
[01:34:32] packaging yes because the
[01:34:35] amount of things we're squeezing into
[01:34:38] the
[01:34:38] traditional transceivers is is growing
[01:34:41] dramatically
[01:34:42] and yet the size of the transceiver
[01:34:44] hasn't and so you need something
[01:34:46] other than the full kobo pick because
[01:34:48] transceivers despite my graph showing
[01:34:50] everything's going zr
[01:34:52] and uh and uh and cpo that that's the
[01:34:55] long-term vision there's
[01:34:56] there's a long long future for
[01:34:59] transceivers
[01:35:00] and we just can't put any more stuff
[01:35:04] in our toasters roses anymore we
[01:35:06] currently use z prisms
[01:35:08] discrete z prisms and 45 degree prisms
[01:35:12] um and there's just not room even for
[01:35:14] four
[01:35:15] because you know you have four in the
[01:35:16] toes of four and the rest of this eight
[01:35:18] channels
[01:35:18] squeezed into the width of an sfv that
[01:35:21] you know it's just uh
[01:35:25] yeah yeah definitely in the views of
[01:35:29] and it's very interesting the views of
[01:35:30] the ceo of photonex networks because a
[01:35:32] couple of years ago he thought
[01:35:34] there is a big business on the complex
[01:35:36] packaging of pixels
[01:35:38] and now that all these micro optics
[01:35:40] technologies are really booming right
[01:35:41] now
[01:35:42] is there a process or is there an idea
[01:35:45] that you have to
[01:35:47] add this to the packaging technologies
[01:35:49] of photons networks for pixels
[01:35:51] yeah i jose it's it's it's a fantastic
[01:35:54] opportunity but also it's a very well
[01:35:56] organized workshop so congrats on this
[01:35:58] um
[01:35:59] i i i think um i think photonics is
[01:36:03] looking at this uh
[01:36:04] slightly from a different i i i'm very
[01:36:06] impressed with the uh slide just shown
[01:36:08] by sus micros micro optics but
[01:36:10] we're looking at as a silicon bench kind
[01:36:12] of solution we are we are seeing that
[01:36:14] there's a need for a platform and you
[01:36:17] have different ways of packaging optics
[01:36:18] of course
[01:36:19] but we have we have started by looking
[01:36:21] at pixels as a
[01:36:22] as a main platform for integration maybe
[01:36:25] i can share a slide maybe i can
[01:36:28] always always so that always go ahead
[01:36:31] yeah i'm just trying to figure out which
[01:36:32] one is the best to
[01:36:34] do i think this this should work i'm not
[01:36:37] sure it's the highest resolution but
[01:36:40] i'm afraid to press any button in in
[01:36:42] fear of losing what i now have on the
[01:36:44] screen so uh
[01:36:45] go ahead so you just see what i'm
[01:36:48] showing here so basically
[01:36:50] we have looked at this from the
[01:36:52] originally from the research perspective
[01:36:54] and we found that silicon is is a lot of
[01:36:56] very very attractive proposition
[01:36:59] uh as as the people in sus can
[01:37:01] definitely uh of course
[01:37:02] vouch for it's it's a very uh robust
[01:37:06] technology
[01:37:06] it offers a lot of opportunities for
[01:37:08] processing i mean there is a
[01:37:10] there's a whole field of fab of founders
[01:37:14] doing mems
[01:37:14] processing and what we think is that
[01:37:16] silicon is really an exciting platform
[01:37:19] compared to ceramics or all the very
[01:37:22] complicated
[01:37:23] pcb materials which organic base or
[01:37:26] others
[01:37:27] to create these platforms we can we can
[01:37:29] do all this packaging we can trace all
[01:37:31] these
[01:37:32] all these high speed traces we have we
[01:37:35] have really
[01:37:36] nice data showing the speed is really up
[01:37:38] there at 56 gigabyte 112 gigabits per
[01:37:41] second
[01:37:42] with pum4 and we can really do a very
[01:37:45] massive parallelization so we
[01:37:47] demonstrated 48 channels of course
[01:37:50] 14 channels nobody is even dreaming
[01:37:52] about it and i really like the comments
[01:37:53] about reliability and the complexity of
[01:37:56] of increasing the channel count this is
[01:37:57] definitely not a linear
[01:37:59] complexity problem it's definitely as we
[01:38:02] know yield anyway is exponentially
[01:38:04] degrading thing if you have 99 on one
[01:38:07] device
[01:38:08] try calculate it for 16. so but we are
[01:38:11] really optimistic about this and we
[01:38:12] really think that the way to go is
[01:38:14] is some sort of silicon interposer maybe
[01:38:16] with v grooves
[01:38:18] like the the guys in the sous are
[01:38:19] looking at so these kind of solutions
[01:38:21] are really interesting
[01:38:23] congratulations first following how you
[01:38:25] are
[01:38:26] really helping all the pixel community
[01:38:28] interacting with the companies who
[01:38:30] really need
[01:38:31] the complex packaging of pixels but i'm
[01:38:34] also going to push a bit more if you
[01:38:35] allow me on that on the micro optics
[01:38:37] part because i'm super excited
[01:38:38] what is happening and i really want to
[01:38:39] understanding a bit better we have a
[01:38:41] company in epic
[01:38:42] a company that is a petrochemical giant
[01:38:45] a company is called savik and when they
[01:38:47] went to visit them they told me no we
[01:38:48] also do optics for
[01:38:50] packaging of transceivers and i want to
[01:38:52] say okay what is the link
[01:38:54] gabriel from savic gallery i want to
[01:38:56] know your vision and also i want you to
[01:38:58] dream a way of your preferred future
[01:39:00] for the optics needed on the packaging
[01:39:03] of this
[01:39:04] co-package optics yeah thanks jose
[01:39:07] yeah that's that's correct i mean we
[01:39:09] sell plastics for the plugables but we
[01:39:12] also can sell plastics for the onboard
[01:39:14] optics
[01:39:15] especially the plastics that have reflow
[01:39:17] capable materials
[01:39:20] so my question is is this interesting
[01:39:21] for the industry
[01:39:24] and that's also my question and it's
[01:39:26] difficult always to to have an answer
[01:39:28] there is interest that's really
[01:39:29] interesting why don't we go to our
[01:39:31] friend from juniper network uh why don't
[01:39:33] we go to universe network
[01:39:35] we had a really interesting presentation
[01:39:36] a couple of uh
[01:39:38] half an hour ago and you talked to us
[01:39:40] about different concepts on the 100g and
[01:39:42] 400g
[01:39:43] we have great technologies for adding
[01:39:46] optics
[01:39:47] on the top of silicon for bringing the
[01:39:49] optics very close to the electronics
[01:39:51] we're manufacturing a wafer level very
[01:39:54] reliable with
[01:39:55] suitable to high temperatures is there a
[01:39:58] need on the industry for these
[01:40:00] optics
[01:40:03] oh i'm not sure how i understand
[01:40:07] question or not
[01:40:10] we have great optics we have great
[01:40:12] optics and my my
[01:40:13] members are telling me i hear a lot
[01:40:15] about co-package optics
[01:40:17] i hear a lot about this i want to know
[01:40:19] about the optics part of the cpu i want
[01:40:21] to know about the
[01:40:22] what kind of optics still require for
[01:40:24] the whole package optics
[01:40:27] for cool package
[01:40:30] i think i still need some time
[01:40:34] currently we are talking about the 800
[01:40:37] year right
[01:40:43] sorry about yeah well still uh it is
[01:40:46] still
[01:40:46] talking about to buy uh it's still the
[01:40:49] propagal
[01:40:50] so for cool package i'm not sure
[01:40:53] sorry about it yeah no but i just don't
[01:40:56] worry because nobody is sure so i'm
[01:40:58] going to ask some of the people in the
[01:40:59] room to see what is their vision and
[01:41:01] their need for
[01:41:02] high-end optics wafer level optics and
[01:41:05] even printed optics for the packaging
[01:41:06] with three photonics
[01:41:08] let me go first to lightweight logic let
[01:41:10] me start with a polymer foundry let me
[01:41:11] go to michael levy
[01:41:13] michael you have seen especially in
[01:41:14] europe a boom on micro optics
[01:41:16] manufacturing
[01:41:18] is there a need on the packaging of
[01:41:20] iterative photonics especially in the
[01:41:22] co-packaging which seems to be
[01:41:24] the future well i think the general
[01:41:26] answer is
[01:41:27] is yes um and the reason i'm saying yes
[01:41:31] is is that we haven't
[01:41:33] seen all the solutions yet over the next
[01:41:35] decade
[01:41:36] what we do know is is that we've got to
[01:41:38] figure out things in a more miniaturized
[01:41:41] format
[01:41:42] you know whether it's plugable or
[01:41:43] otherwise i don't think that really
[01:41:46] matters what what does matter is is that
[01:41:48] we've got to get light in and out
[01:41:50] of the situation using different
[01:41:52] techniques
[01:41:53] and whether it's going to be on board
[01:41:55] with different layers and waveguides in
[01:41:57] a silicon photonics medium phosphide
[01:41:59] that's certainly going to be one
[01:42:00] solution but we'll also
[01:42:02] try we also need micro optics because
[01:42:06] we'll want to simplify how the light
[01:42:08] gets to the fiber
[01:42:10] and everybody has their own solutions to
[01:42:12] do that whether it's a grating solution
[01:42:14] or a bug couple or you've got some
[01:42:16] transition from one waveguide to another
[01:42:18] waveguide
[01:42:19] um i think you're going to see a whole
[01:42:22] portfolio of optical solutions
[01:42:24] which is why it's sort of exciting
[01:42:26] there's no one
[01:42:27] single solution that is the right
[01:42:29] solution everybody's going to design for
[01:42:32] you know look we've been in all of us
[01:42:34] have been in the optics business for a
[01:42:35] long time and it's really easy to lose
[01:42:38] light
[01:42:39] one of the most difficult things is to
[01:42:40] not lose light i mean for being in a lab
[01:42:43] i mean light just goes everywhere and so
[01:42:45] micro optics is is a
[01:42:47] key component that's going to help us
[01:42:50] preserve the light
[01:42:51] and so if we can preserve the light our
[01:42:53] optical budgets are going to be better
[01:42:55] and
[01:42:56] and everything becomes easier so it's
[01:42:58] going to be important
[01:42:59] so for me what i have seen in the last
[01:43:01] 10 years is two tendencies on one hand
[01:43:03] the tendency of bringing active
[01:43:05] alignment to a fast speed so it's as
[01:43:08] fast and as
[01:43:09] low cost as passive alignment and on the
[01:43:11] other
[01:43:12] uh the opposite on the other trying to
[01:43:14] do everything a wafer level
[01:43:16] try to do they wait for level packaging
[01:43:18] they wait for level optics
[01:43:19] on the packaging try to align them
[01:43:21] passively i want to understand this a
[01:43:23] little bit better so i'm going to ask a
[01:43:24] company that we have a as a
[01:43:26] as a player on the on the passive
[01:43:28] alignment in ep
[01:43:29] is called fine tech in berlin herman you
[01:43:32] have seen the same thing the tendency of
[01:43:33] doing everything a wafer level and at
[01:43:35] the same time
[01:43:36] companies going into active alignment
[01:43:38] components individually
[01:43:40] from the point of view of fine tech how
[01:43:41] is this co-package optics industry gonna
[01:43:44] look like
[01:43:45] uh so it's a good question jose
[01:43:49] um truth to be told actually we can see
[01:43:52] both directions being followed um it
[01:43:55] seems companies have totally different
[01:43:56] philosophies on that some are really
[01:43:58] saying
[01:43:58] okay we're going for the active
[01:43:59] alignment some others say okay we are
[01:44:01] buying an equipment
[01:44:03] that is super precise and we're trying
[01:44:05] to do that in a passive way
[01:44:07] um as you know the the active alignment
[01:44:11] is not really our cup of tea
[01:44:13] this goes to our colleagues in our theme
[01:44:15] whereas our focus is on the on the
[01:44:17] passive alignment so
[01:44:18] that typically we we don't touch each
[01:44:21] other much in the field because
[01:44:23] companies going to this field with a
[01:44:24] totally different approach
[01:44:26] everybody thinks that his approach is is
[01:44:28] a one the right way to go
[01:44:30] so what we are doing basically is to
[01:44:32] support our customers that decide to go
[01:44:35] as a passive way
[01:44:37] and yeah as i said trying to support
[01:44:40] them by
[01:44:40] by trying to build some samples for them
[01:44:43] and probably trying to develop this as a
[01:44:46] process with them
[01:44:48] thank you very much herman i i honestly
[01:44:50] believe there is a huge
[01:44:51] room for cooperation here and i do love
[01:44:54] when companies going to waffle level
[01:44:56] i'm going to go to a friend of mine a
[01:44:58] very dear friend of mine when i go to
[01:45:00] gloria
[01:45:00] infinera and the reason why i'm coming
[01:45:03] to you first because i like you very
[01:45:04] very much and second because as a
[01:45:06] foundry of indian phosphate chips who
[01:45:08] managed to go all the way up to the
[01:45:10] value
[01:45:10] chain to sell optical networking systems
[01:45:14] for data centers and telecom operators
[01:45:16] gloria how do you see these co-package
[01:45:19] optics is the future
[01:45:20] do they need extra different more
[01:45:22] difficult more complex optics to make
[01:45:24] the future possible
[01:45:27] thank you jose i think
[01:45:30] that the issue of co-packaging and the
[01:45:33] alignment and all
[01:45:34] all of those related packaging
[01:45:37] constraints really are
[01:45:38] very much application dependent i think
[01:45:40] as michael had mentioned
[01:45:42] you can't lose photons it's very very
[01:45:45] you know they're very expensive to make
[01:45:47] and so efficiency is really critical but
[01:45:49] it cannot come at the
[01:45:51] at the at the cost of um you know more
[01:45:54] adding more cost
[01:45:55] either by uh slowing down the alignment
[01:45:58] process
[01:45:59] which is one of the areas that you have
[01:46:00] to fight in active alignment
[01:46:02] or by losing photons which is you know
[01:46:05] essentially what you do when you do
[01:46:07] passive alignment uh especially in
[01:46:09] indian phosphide which is
[01:46:11] more um what we do so you know i guess
[01:46:14] the answer really is it depends depends
[01:46:17] on on
[01:46:17] on the application low cost applications
[01:46:20] will will continue i think to be looking
[01:46:22] more for
[01:46:23] passive alignment uh and in the issue of
[01:46:26] co-packaging uh with the electronics uh
[01:46:28] especially um you have to be looking at
[01:46:31] a thermal solutions
[01:46:32] so at infiner we have done a lot of
[01:46:35] micro optics uh in in our packages
[01:46:38] they're some of the most
[01:46:39] compact uh modules and dcos
[01:46:43] in the industry and um i think
[01:46:46] it can be done it it really just depends
[01:46:49] on
[01:46:49] on the cost and and the
[01:46:53] is another room for application gloria
[01:46:54] has been a great entry point for epic
[01:46:56] members into
[01:46:57] potential discussions with infinera
[01:46:58] thank you very much gloria
[01:47:00] i want to understand a bit better and
[01:47:02] the best way to do it is to talk
[01:47:04] to the semiconductor manufacturers
[01:47:06] electronic ics manufacturers
[01:47:08] the ones that are talking cmos business
[01:47:10] and for that we have
[01:47:11] aem in the room thank you very much
[01:47:14] kharshan and thank you also jim
[01:47:16] childress for making this happen
[01:47:17] harrison pandya is here from aem the
[01:47:21] general manager
[01:47:23] thank you very much for being with us
[01:47:24] the floor and the attention of everyone
[01:47:26] for closing this
[01:47:27] spectacular meeting goes to you
[01:47:30] thank you jose and thank you for giving
[01:47:33] me the opportunity to
[01:47:35] close this meeting let me try to share
[01:47:38] one of my slides if i can
[01:47:43] can you see my slide
[01:47:47] yes crystal clear okay
[01:47:50] so aem is a singapore-based publicly
[01:47:53] listed company
[01:47:55] which is focused on system level test
[01:47:57] handlers
[01:47:59] like very large scale device testing
[01:48:02] handlers
[01:48:04] and also we have test solutions
[01:48:07] test and measurement solutions
[01:48:11] okay just give me a minute so some of
[01:48:14] our products
[01:48:16] as i mentioned this is on the left you
[01:48:18] what you see is the system level test
[01:48:21] handler solution which is capable of
[01:48:25] doing massively parallel testing of
[01:48:28] large number of devices uh in most
[01:48:32] cases we have customers who have their
[01:48:35] own
[01:48:36] uh test systems that get integrated into
[01:48:39] our handlers
[01:48:40] but we also have a possibility to
[01:48:43] provide
[01:48:44] test solutions for specific devices and
[01:48:47] the devices range from
[01:48:49] semiconductor chips very complex chips
[01:48:53] to even modules right so we have a wide
[01:48:56] range of
[01:48:56] duties that get tested through our
[01:48:59] handlers
[01:49:02] in addition to the system level test
[01:49:04] which is the larger
[01:49:05] business unit we also have
[01:49:09] a facility in finland where we do
[01:49:13] mems testing at wafer level
[01:49:16] mainly for the sensors so
[01:49:20] pressure gyroscopic accelerometers
[01:49:22] sensors and so on
[01:49:24] and then the third business unit that we
[01:49:26] have is test and measurement solution
[01:49:28] where we make vector network analyzers
[01:49:31] for rf
[01:49:32] testing we focus particularly on passive
[01:49:35] devices like cables
[01:49:37] so both fiber optic and copper cable
[01:49:40] testing
[01:49:41] including single mode and multi-mode
[01:49:44] fiber optic certification
[01:49:45] and and category copper cable
[01:49:49] certification
[01:49:51] and we also you know have solutions for
[01:49:54] automotive and 5g infrastructure
[01:49:56] cable testing just to give an idea about
[01:49:59] one of our recent introduction of
[01:50:01] test tier this is a multi-mode fiber
[01:50:05] bandwidth test solution so when
[01:50:07] multimode fiber is used
[01:50:09] one of the challenges of course is the
[01:50:11] dmd
[01:50:12] which severely restricts the bandwidth
[01:50:15] capability
[01:50:16] and we have a solution to measure the
[01:50:19] bandwidth of
[01:50:20] multi-mode fiber cables but our
[01:50:23] test solution basically covers more on
[01:50:26] the
[01:50:28] more on the enterprise side uh for
[01:50:31] testing
[01:50:32] all applications that go with ethernet
[01:50:36] networks
[01:50:38] thank you
[01:50:42] very much thank you very much
[01:50:44] thank you very much
[01:50:44] thank you very much i also for joining
[01:50:45] this event and for presenting presents
[01:50:49] this presentation and what do you think
[01:50:52] in terms of
[01:50:53] in terms of collaborations the companies
[01:50:55] that you saw at this meeting
[01:50:57] the different talks what will you be
[01:51:00] yeah the perfect partners for you to
[01:51:01] collaborate from our industry
[01:51:04] yeah i think uh like like i mentioned
[01:51:06] our system level test handlers
[01:51:09] provide uh i think uh very good solution
[01:51:12] for testing
[01:51:14] devices in large quantities and
[01:51:17] we can collaborate in terms of making
[01:51:20] specific test
[01:51:21] test systems per cell or you know we can
[01:51:25] also you incorporate customers
[01:51:27] testing environment one thing i did not
[01:51:29] mention is that
[01:51:31] our test handlers can also provide
[01:51:35] thermal excitation so like you know heat
[01:51:38] and heat and cooling
[01:51:40] extreme temperatures so you can you can
[01:51:43] subject the duties to
[01:51:45] different environmental conditions while
[01:51:48] performing the test
[01:51:49] right so i think the opportunity is to
[01:51:51] collaborate with
[01:51:52] device manufacturers to provide uh
[01:51:56] testing for the i mean a post-production
[01:52:00] testing
[01:52:00] system level
[01:52:03] okay thank you very much so maybe i
[01:52:05] would like to introduce you one of our
[01:52:07] companies uh maybe you could
[01:52:11] you could think about the collaboration
[01:52:12] so we have here danslight
[01:52:17] so i don't know if um
[01:52:22] hello how are you and maybe
[01:52:25] it's a good moment for you to present
[01:52:27] the capabilities of your company and
[01:52:28] maybe do you have a slide
[01:52:30] yeah okay i can share slides just give
[01:52:33] me a moment
[01:52:34] and
[01:52:35] [Music]
[01:52:41] okay yes if you can go to presentation
[01:52:44] mode
[01:52:45] yeah okay so uh just go on let me get to
[01:52:48] the
[01:52:48] share the right screen so screen tool
[01:52:54] and i'm going to go into the
[01:52:55] presentation mode
[01:52:57] okay right you should yeah it's very
[01:52:59] good
[01:53:00] okay yeah yeah thank you um yeah
[01:53:04] uh hi hello my name is eloy from dance
[01:53:06] lights
[01:53:07] um benzler is the interim phosphites
[01:53:10] device supplier with vertically
[01:53:12] integrated intel phosphite fat
[01:53:15] from epidermal growth to device fab and
[01:53:17] testing actually located here in
[01:53:19] singapore together with aem um i think
[01:53:22] for today's session i'd like to take the
[01:53:24] opportunity to introduce our d5
[01:53:28] and our tensor photonics hybrid
[01:53:30] integrations
[01:53:31] which is a one-step integration service
[01:53:33] for interphosphite laser onto silicon
[01:53:36] photonics
[01:53:36] optical engine so on the left we show
[01:53:39] the
[01:53:39] manufacturing flow whereby the denser
[01:53:42] engine phospite lasers or gain trees or
[01:53:45] all the soas are passively annoying and
[01:53:48] free chip onto
[01:53:49] sacred photonics wafer and this wafer
[01:53:52] comes from our partner
[01:53:53] silicon country to create ready to use
[01:53:55] optical engine
[01:53:57] so in this slides uh we illustrate an
[01:54:00] example of a 400g optical engine for dci
[01:54:04] and as you can see it can be examined
[01:54:06] into practicable
[01:54:08] transceiver or it can be cheat mounted
[01:54:11] on board and uh as a as a photonics
[01:54:14] engine
[01:54:15] so lastly i like to mention that densely
[01:54:18] is platform agnostics
[01:54:20] in supporting our customers indian
[01:54:21] phosphite needs
[01:54:23] so for instance our streets of laser
[01:54:25] solutions
[01:54:26] support both the remote and integrated
[01:54:30] laser kobo architectures for
[01:54:32] data centers interconnects and switches
[01:54:35] thank you
[01:54:39] okay so thank you very much if any of
[01:54:41] the companies present
[01:54:42] here that are here in the meeting want
[01:54:44] to contact the then slide please drop me
[01:54:46] an email and i will be
[01:54:47] very happy to make the the introduction
[01:54:50] okay so then if there are no
[01:54:52] more comments so then back to you jose
[01:54:57] thank you very much anna for this i i do
[01:55:00] have a lot of comments i do have a lot
[01:55:01] of comments but what i want to say first
[01:55:03] of all
[01:55:04] to all of you is that this meeting only
[01:55:06] lasted two hours we could have lasted
[01:55:08] several days
[01:55:10] it could have lasted several days it
[01:55:11] will last for two days soon when we
[01:55:13] can't travel again and i cannot
[01:55:15] wait and we had a meeting at microsoft
[01:55:17] because microsoft has told us that we're
[01:55:19] going to have
[01:55:20] a two days meeting for the epic members
[01:55:22] in seattle at the headquarters
[01:55:24] all of us together trying to understand
[01:55:26] this and bringing
[01:55:28] the high-end data center and telecom
[01:55:30] operator it's gonna be epic i don't have
[01:55:32] a date yet i'm dying
[01:55:34] there's a virus going on out there we're
[01:55:36] gonna kill it we're gonna beat it and
[01:55:37] we're gonna have this amazing meeting at
[01:55:39] microsoft i want all of you there it's
[01:55:41] gonna be epic i cannot wait to have a
[01:55:43] beer with you
[01:55:43] it's been a great meeting it's been a
[01:55:45] great meeting and i know that all of you
[01:55:47] are going to say well because it was a
[01:55:48] big meeting a horse gonna say it
[01:55:50] it was a great meeting because we had
[01:55:52] high-end
[01:55:53] companies telling us about their needs
[01:55:55] and we didn't understand
[01:55:57] that we don't know where the industry is
[01:55:58] going and that was clear that was clear
[01:56:00] we had a beautiful roadmap slide from
[01:56:02] like a levy but we had a lot of
[01:56:04] different options and maps and how many
[01:56:06] channels are we gonna multiplex are we
[01:56:08] gonna go to onboard optics yet and now
[01:56:10] we're gonna stay with plug out for a
[01:56:11] while is this an intermediary step
[01:56:13] anyway
[01:56:14] regardless regardless there are a lot of
[01:56:16] business opportunities here and
[01:56:18] companies are going to get in touch with
[01:56:19] each other
[01:56:20] why first we talk about fast modulation
[01:56:22] like with logic
[01:56:23] michael levy told us i think the polymer
[01:56:25] is the way to go there are of course
[01:56:27] other approaches but we need
[01:56:28] fast modulation that's clear the second
[01:56:29] one is do we need
[01:56:31] micro optics a wafer level and all of
[01:56:34] you we had a really beautiful round that
[01:56:36] was for me the highlight of the meeting
[01:56:37] although you agreed that we believe so
[01:56:39] so let's explore that because we have
[01:56:41] the best micro
[01:56:42] optics right now in the epic network the
[01:56:43] best micro optics in the world and we
[01:56:44] need to push that
[01:56:46] the second one it was clear wafer level
[01:56:48] packaging and high level packaging
[01:56:50] and the wafer level packaging is of
[01:56:52] course the challenge
[01:56:54] but the testing a wafer level is
[01:56:57] one of the enablers of the challenge and
[01:56:58] for that we have seen in the last years
[01:57:00] huge developments on that and what i
[01:57:03] want to say also
[01:57:04] is that this is just a part of the of
[01:57:07] what's going to happen now
[01:57:08] because for me introducing the companies
[01:57:11] is just a way of life introducing the
[01:57:14] companies are making people
[01:57:15] happy for working together it's what
[01:57:18] they do for a living that's what the
[01:57:19] epic people do for a living so remember
[01:57:21] if you want to get in touch with any of
[01:57:23] the participants today all you have to
[01:57:24] do is send me an email visit
[01:57:28] [Music]
[01:57:30] epicdashasset.com
[01:57:31] and i will make that introduction on
[01:57:33] behalf of
[01:57:34] an amazing amazing team for epic members
[01:57:38] and epic staff i would like to wish all
[01:57:40] of you
[01:57:41] first be ready for the best and prepare
[01:57:44] for the worst we don't know what's going
[01:57:45] to happen
[01:57:46] but always be happy join the epic
[01:57:49] meetings we have a great selection of
[01:57:50] topics for the coming
[01:57:52] weeks until christmas and we're gonna
[01:57:54] continue i can already tell you that
[01:57:55] after christmas
[01:57:56] again two meetings per week always
[01:57:59] looking for the end-users and
[01:58:01] understanding what they need until the
[01:58:03] next time remember from the beautiful
[01:58:05] city of northride just next to the beach
[01:58:07] with a little bit of voice left all my
[01:58:10] energies
[01:58:10] go to the epic level have a really
[01:58:12] fantastic day
[01:58:14] good night everyone