The POET Technologies AGM 2018 took place on June 21th in San Jose, California, USA. This document comprises a transcript of the slide presentation given by CEO Dr. Suresh Venkatesan. The subsequent Q&A section is also included. The transcript is the result of the dedicated work of a couple of POET Technologies shareholders. It is based on Suresh Venkatesan’s slide presentation (PDF) and on the audio recording of the AGM webcast (MP3). The (formal) AGM as such has not been transcribed. The transcript statrts at slide 7.
Zusammenfassung auf Deutsch: Die POET-Technologies-Hauptversammlung 2018 fand am 21. Juni in San Jose, Kalifornien, USA statt. Dieses Dokument enthält eine Abschrift der Folienpräsentation von CEO Dr. Suresh Venkatesan. Die anschließenden Fragen und Antworten sind ebenfalls enthalten. Das Transkript ist das Ergebnis der engagierten Arbeit einiger POET-Aktionäre. Es basiert auf der Folienpräsentation von Suresh Venkatesan (PDF) und auf der Audioaufzeichnung des Webcasts (MP3). Die (formelle) Hauptversammlung als solche wurde nicht transkribiert. Das Transkript beginnt bei Folie 7.
Mit Google Translate erstellte automatische Übersetzung ins Deutsche
Transcript
The numbers at the start of each paragraph denote the position in the webcast, i.e., hour:minute:second.
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
00:22:31 – Suresh Venkatesan: We have transitioned our dielectric waveguide development from a university lab, which was the University of Swansea in England, to Silterra. It was done both to complete our development in a manufacturing environment on 8″ wafers, as opposed to smaller wafer sizes that we were working with, but also to provide a seamless ramp to high-volume manufacturing.
00:22:58 – We established a memorandum of understanding with Accelink, one of our lead customers and development partners for datacom and telecom applications. We also demonstrated a 100 Gbps-capable quad PIN receiver, so each pin is a 25 Gbps lane. Four of them provide a cumulative of 100 Gbps operation. And we have developed that for use with our optical interposer, and it represents a critical part of our first product offering, which would be a receive optical engine for 100 Gbps. Also, looking forward to the applications of our platform, we understand that we would require a hermetic seal capability, and we wanted to do that in a low-cost, wafer-scale manner, as opposed to the more traditional gold box, individual module, seam-sealing process that is used in the industry today. And we have invented a new wafer-scale, wafer-level, hermetic sealing technique that is compatible with our optical interposers. So that was, in one page, detailing a lot of work on the part of a lot of people in this company over the past year since we changed our strategy and put most of our investments behind the development of our optical interposers and on our indium phosphide devices.
Slide 8
00:24:32 – So a little bit on each of these agreements: With Accelink, they are our lead customer and also a development partner in the sense that we are able to provide prototypes and get feedback on their performance as it relates to market applications that we are going after. Our first products would be 100/400G receive optical engines. Subsequent products would be an extrapolation of that interposer capability from receive to also including the transmit chain as well as an expansion of our portfolio to telecom applications like GPON. It’s the next generation GPON, so it’s the 10G/2.5G combination solution. The early prototypes to validate the functionality of our interposers will be in Q3 of this year, with production-ready prototypes being delivered to the customer by the end of the year. We would expect production revenues associated with this product, subject to successful qualification and meeting all the standards for that application space in 2019.
Slide 9
00:25:52 – The other main agreement that we signed was with SilTerra, it’s our development and manufacturing partner, with sufficient foundry capacity to meet the requirements of the optical interposer production. We made a fairly significant capital investment in Q1 of this year, and we procured 8″ wafer tools for the optical interposer development. Those tools were delivered in Q2, early in Q2, and as of this week, they’ve actually been released for process development. You can see the pictures of the tools installed at SilTerra’s clean-room facility in Malaysia. We picked SilTerra for their unique combination of lithography capability and copper metallization, which is critical for high-frequency RF, and there aren’t too many foundries with that combination at 8″ scale, so we were very pleased to have a relationship with SilTerra to continue development, and then manufacturing, of our interposers. So this was another significant agreement that we announced in Q2 of this year.
Slide 10
00:27:17 – A third announcement that we just made, actually this morning, was to partner with another company, based in France, to help us develop the transmit-side of the equation. So we are partnering with Almae to jointly develop and manufacture optical engines, the transmit optical engines, that are compatible with our interposer. We are doing this to accelerate our time to market for the receive-and-transmit chain. And the first products that we would be working on together are, again, the high-speed lasers that are required for a full 100G transmit-and-receive engine. So we are internally working on the receive engine and the interposer, and we are partnering on the high-speed lasers. I think this was another strategic decision on our part, which is to partner to accelerate time to market, because otherwise we would have to do all these developments sequentially, internally. I think the interposer concept has been well-received in the industry, and we do have companies like Almae wanting to partner with us, because they see an opportunity to participate in what we think is a good disruption as it relates to photonics packaging. So, on the transmit side, the prototypes are expected in Q1 of 2019, along with joint sales in 2019.
Slide 11
00:28:55 – Technology development and progress: I think there’s been a lot of questions that we have had relative to charting out what are the steps needed to get to a real product? And so we kind of look at it, at least in the context of our first product, and spell it out and talk about the progress we have made in each of these areas. Obviously we need processes and tools. In that context, as I have talked about, we have the interposer process development and initiation of qualification with a target completion in Q3 of 2018. As of this week, our tools are now ready, and we have transferred the process from a lab environment to a full-fledged 8″ foundry environment. We are pleased with the results that we have seen so far out of the tools we have procured. So we’ll spend the rest of this quarter optimizing and fine-tuning the development processes for the interposer itself.
00:30:03 – The second key is the PIN detector itself. The active portions of the PIN detector we announced in February of this year. Since then, we have partnered with an electronics provider and have validated a full 25 GHz per second, per channel, performance. A critical piece of qualification is to ensure that the PINs that we are building internally don’t have any latent reliability issues. So ahead of qualification, we have gone through a little over 1800 hours of what we call life-test, which is a stress test of these detectors, to ensure that we don’t see any issues with the PINs themselves. We have gone through that with no failures. We are now on track to being able to populate the qualification of our PIN detectors, which would start in July. The qualification requirement is 2000 hours, which is a better part of a quarter. So we would expect to start in July, go through 2000 hours of stress, and complete the qualification of the PIN itself by the end of the third quarter.
00:31:19 – We continue to optimize the performance of the passive portion of the PIN. So I think our differentiating features on this technology are the use of waveguides to couple light in and out of active devices. The active devices are built in indium phosphide, and the waveguides are built in indium phosphide as well as the dielectric. So while we have got the actives completed, we are now focused on optimizing the performance of the passives, both in the dielectric as well as on the indium phosphide, to ensure we meet the specifications of the final product.
00:31:55 – The optical interposer itself. So all the required silicon and dielectric process flows have now been established at the foundry; we actually completed that this week at Malaysia. Actually, as of this morning, I think we have got a formal notice that the process is now ready for ongoing development. There are some process optimizations to be done since we are transitioning wafer size from a lab into a fab. There’s some optimization of key lithography, deposition, and etch processes that we continue to work on.
00:32:32 – Packaging and assembly is the next critical piece. What we are building is an interposer that allows us to flip-chip or place, and we’ll talk about this a little bit later in the presentation, our products with good alignment accuracy. We have validated that we are able to place our die on the interposer, and in doing so we are not negatively impacting the performance of the devices. We have to still ensure that the interposer still performs at the required speeds and has the required RF characteristics and performance. Remember that for 100 Gbps application, we are talking about a 20+ GHz RF frequency, and for a 400 Gbps, it’s close to 40 GHz RF frequency. So these are very high-frequency RF performance requirements, and we have validated that the interposer in and by itself is capable of supplying the transmission lines and the RF traces needed for these high-frequency applications.
00:33:47 – For the dielectric filters and the waveguides, which is the last key component of the interposer itself, we have transferred the process over to SilTerra, which is our foundry. The optimization of the design of the filter is now ongoing. We have just taped out a new version of our designs to help close the gap to meeting the requirements of the process, and we expect those optimizations to be completed through the course of this quarter.
Slide 12
Slide 13
00:34:22 – We want to spend a little bit of time just talking about what an interposer is. I think we have made a press release in January, but I haven’t had a chance to actually sit and talk about what is an interposer, and why is it versatile and flexible and in the way we think it is.
00:34:40 – The concept of an interposer has been around for a while. In fact, in the world of electronics, interposers have been in production since about 2015. It was initially invented for close proximity placements of electronic components. So as the data speeds and frequencies have gotten faster, it is very difficult to place, let’s say, a memory die and logic die on a printed circuit board (PCB), because the spacing in a printed circuit board is in order of millimeters to a centimeter and that is typically too large to support very high-frequency operations. So the concept of an electrical interposer was invented and practiced at most of the semiconductor foundries today, that allows a very close proximity placement of two electrical dies to promote high-speed communications.
00:35:35 – What we have done with the optical interposer is we have taken the concept of a electrical interposer and provided a level of optical interconnectivity to it, using our waveguide technology. And the key piece of our waveguides is they are compatible with CMOS. They are actually build a CMOS chip and then, post completion of this CMOS process, we are able to deposit waveguides onto the chip, and without affecting the underlaying CMOS in a way.
00:36:11 – So we are able to build a interposer with pre-built copper metallization required for interconnectivity and then deposit the waveguides on top of that. That allows for both an electrical interconnectivity to occur as well as an optical interconnectivity to occur. So we just now have the ability to communicate optically between the optical components on the interposer, and we communicate electrically through the electrical connections underneath it.
00:36:39 – And it is really the first practical application that we know of with the dielectric waveguides technology, because we have deposited these dielectrics in a way that is compatible with underlaying metallization and underlaying CMOS.
00:36:56 – So that is in essence what a interposer is. The interposer platform in itself can be targeted for multiple applications. If the filters on the interposer are, let’s say, a CWDM for wavelength-division multiplexing applications, then that becomes CWDM interposer. If we design those filters for a different protocol like LR4, then it becomes a LR4 interposer. So the interposer, once the mechanics of depositing, patterning, designing waveguides are figured out, then you can spin-off variants of these interposers for different applications.
00:37:40 – And the active devices are then done separately, and the key piece of these optical interposers are, because you have separated out the interposer from the active die, we can engineer them separately, and we assemble what we call „known good die“. It allows us to lower the cost of photonics packaging quite dramatically.
Slide 14
00:38:05 – So the next slide gives you a cartoon of what an interposer multi-chip module might look like. It has metallization at the bottom, it got the waveguides on top, and then we would place our active die flip-chip or rather upside-down onto this interposer, and we would have built into the interposer, the appropriate mechanical stops, such that when a die is placed on the interposer, the optical axis of the die that we are placing aligns itself with the optical axis of the waveguides that are on the interposer. And that is the key! It allows us to place in a passive way the components onto the interposer and communicate optically between this dies without requiring active alignment or individual handling of these components. So you can assemble them on wafer-scale using conventional pick-and-place tools, and we can hermetically seal them at wafer-scale as well. It dramatically lowers the cost of the end-solutions that utilize these interposers.
00:39:26 – A critical piece of this is the deposition process that is proprietary that we have worked to develop over the past year and half. It needs to be extremely low-loss, that is the optical loss as you go through the waveguide, has to be very low.
00:39:47 – As a benchmark, Silicon Photonics that uses silicon waveguides, typically have, let’s say, an optical loss – it is measured in decibels – that is 2 decibels per centimeter length, let’s say, of the waveguide. We have been able to improve that by order of magnitude. So we now have demonstrated that we are at 0.2 dB/cm, so very, very low-loss waveguides.
00:40:14 – It also needs to have very low stress. If you typically deposit material on a silicon wafer and if the stress of that material is very high, then you typically get bowing or warping or chipping of the wafers. A critical piece of any deposition process is to insure that it can be deposited with very low stress. And it has to be low-temperature.
00:40:37 – So that unique combination of low loss, low stress, and low temperature is kind of what makes the ability for us to make these interposers in a CMOS-compatible manner and make it versatile in its application space.
00:40:55 – A key thing with our dielectric waveguides also is that it is transparent from the visible to about two-micron wavelength. So we can apply it, like we are doing right now, in what is called the O band spectrum of light, which is around 1310 nm, which is what we using for our data communications products, but it is equally applicable to the C band and the L band, which is 1550 nm, 1650 nm, it is also applicable down at 850 nm and 900 nm, and then below that into the visible.
00:41:30 – The versatility and flexibility of the platform really comes into play when we talk about expanding the portfolio of our products from a specific wavelength band that we are focused on today to alternate wavelength bands that we will work on in the future. And to a certain extent, that is a critical piece of why customers are excited about working with us on the interposer, because when they look at their portfolio of products, it’s not just one type of product at one wavelength – telecom uses multiple wavelengths –, and the ability for this interposer to be used as a more pervasive packaging approach is quite appealing. In each application that we’ve looked at we can provide the kind of cost savings that we’ve talked about in the datacom products as well.
Slide 15
00:42:26 – The next slide gives you a visual picture of what this interposer looks like on an 8″ wafer. You can see the copper traces, the metal traces that are built into the wafer. The waveguides are built on top of them. There’s a 3D model of a PIN detector in this case, being placed on that interposer and then sealed with a hermetic cap.
00:42:53 – This is all done at wafer-scale. One of the pictures on the bottom-right actually shows you a populated PIN die on the wafer. You can imagine the entire wafer being populated with, in this case, quad PINs, or it could be lasers. And then it would eventually be singulated into individual multi-chip modules.
00:43:15 – The concept of a multi-chip module has been around for a while in the RF world and in the electronics world. But there has never really been a discussion of a solid-state multi-chip module for photonics. And that’s what we are talking about. It is solid-state – there’s no free-space optics, there’s no micro-optics elements – it is a solid-state device that has photonics functionality. And that’s really what we are trying to qualify: that base platform that allows us to then have multiple product applications.
00:43:50 – And our first proof-point and also the first product is the 100G receive product that is shown here. These are actual 3D solid works images to scale of our PIN die being placed onto 8″ wafers.
Slide 16
00:44:10 – This is an example of how the interposer would be used in the context of a laser assembly. This is actually a product that we’re working on now for another customer on the transmit side. So this is a CWDM, which stands for „coarse wavelength division multiplexing“. It’s four lasers, four wavelengths that are 20 nm apart, and a specific application could require either 4 or 12 or up to 16 of these lasers being placed onto the interposer. Why is this important? Because if you do it without an interposer, you have to place these individually and then align each one of them individually with the lenses that are required to insure the right amount of coupling. With our interposers, we promote a very high coupling efficiency between the lightsource and the waveguide. Our target is to be over 90 percent coupling efficiency versus – the norm in the industry is about 60 to 70 percent.
00:45:15 – That is one piece of it. The second piece is, we have advanced thermal management capability built into the interposer. The lasers are placed on the interposer, we have the ability to sink the heat out of these lasers in an efficient way. So that improves the performance of the lasers.
00:45:15 – And the third and probably most important is we passively place these lasers onto the interposer. So there’s no active alignment. When we sell a module with four to twelve lasers on it and provide it to the customer, they don’t need to – now actively – align twelve lasers or four lasers, it’s all pre-aligned for them, and it makes it easy for them to apply it in to their applications. So completely separate from the receive optical engines that we’ve been talking about over the past couple of quarters, this is another product application that we’re working with for another customer, also in the data center applications, utilizing the concepts of the interposer.
00:46:22 – The opportunities for us to make a dramatic difference in the way photonics chips are used and packaged is there, and our discussions with every customer that we have had over the past couple of quarters has borne out the fact that the strategy we put in place a year ago, and that we’ve been executing on, is in fact the right one. The market opportunities are very large: eight billion dollar plus growth opportunity over the next two to three years. A couple of these products with a couple of customers could drive a revenue opportunity for us north of seventy-five million dollars.
00:47:04 – It is incumbent on us of course to deliver the capabilities the platform to the promise that we know we can deliver, finish its qualification. But we are in a unique position right now with having promoted this platform to have the kinds of reception we are having with customers. And it has also been encouraging that different customers have come up with slightly different applications of the interposer, and so we know we are solving problems for them using this approach and using this capability. So it’s been exciting that the strategy that we set forward is gaining momentum and traction in the market. We have been able to have three, what we consider, critical agreements in place over the course of the past couple of quarters with companies and partners that have the heft and the capability to take this into high-volume production for us.
Slide 17
00:48:10 – Of course, if you look out in the future, the interposer does allow and enable the capability for electronic die, ASICs in this case, to coexist with optics. There are some slides that we took snapshots of at the OFC conference, that were presented by either Microsoft or Arista Networks, talking about the need for interposer connectivity, so that we can bring the optics from outside of the server chassis to eventually onto the ASIC itself. And, of course, with the concept of the interposer and its known-good-die assembly capabilities, we believe that there is a path to being able to provide this capability as we get up in frequency beyond 400 Gb/s. So it is a roadmap-play in terms of the platform. It’s not a one-and-done, you set a product.
00:49:16 – Once that platform is done, it has legs, because the platform, in and by itself, is speed-agnostic. The active die can go up in frequency, but the platform itself is speed-agnostic. Hence the term „versatile“ and hence the term „flexible“ when we talk about the potential capabilities that the platform can offer.
Slide 18
00:49:40 – To recap then, the key differentiators, we believe, are:
- Using dielectric waveguides to essentially create a solid-state, multi-chip module for photonics, eliminating the need for active alignment and doing the packaging at wafer-scale rather than at a module-scale or a die-scale.
- It’s a system architecture that allows for optimization of power and performance.
- It’s a fairly small form-factor that can be placed onto a PCB (printed circuit board).
- The cost doesn’t scale with the number of channels, so we make a four-channel device, or an eight-channel, or sixteen-channel, the cost of an interposer doesn’t scale linearly – where as if you did it using more conventional means, as you add more channels your cost goes up linearly – so that allows for flexibility.
00:50:41 – The full wafer-scale integration of components, wafer-scale test the potential, as well as burn-in, and what we talked about as hermetic-sealing, and the fact that we can do this in a silicon fab, compatible with coexisting CMOS processes, makes it quite advantageous. There are waveguide technologies out there today. Silica PLCs, planar light-wave circuits, have been around for a long time, 15+ years, and they’re very good, but they require very high temperature processing. If you look at a silica PLC-based waveguide grating, it’s a separate chip. And it’s not something you can co-integrate or use for co-packaging with interconnects and so on and so forth. So I think that is a key differentiator that we can actually take our waveguides and co-integrate it with electronics capability, metals, metal traces, high-frequency traces, that enables the photonics packaging applications that we are talking about.
Slide 19
00:51:57 – So the platform itself has a broad range of applications. Of course our key focus has been in data centers, data communication, and telecommunication. I think that will remain our focus over the next year. But when it comes to high-performance computing, multi-channel applications, automotive for LIDAR applications, industrial fencing, it’s got a broad range of applications. Wherever you use photonics chips and you want a low-cost, or high-performance, or solid-state configuration, this platform has applicabilities there. And that’s true for space, it’s true for automotive.
Slide 20
00:52:41 – But in our near, immediate focus, we talked about datacom and telecom, and these are very large market segments. For Ethernet data centers today, the prevalent speed form-factor and growth is a 100 Gbs, that’s been in production for about a year and a half now, so we would be entering the 100G market about halfway into its growth, but we do provide a cost benefit. When typically markets expand and volumes grow, there’s always pressure on ASPs, and so, although we enter the market a little bit later at 100G, the margin benefits that we provide to customers are still quite attractive, especially in the high-volume growth phase of the 100G market. The 400G market is the next step up. That’s expected to ramp into production in 2020 – 2021. And at 400G we believe we will be early in terms of providing interposer-based samples in 2019. That’s kind of how that roadmap builds up.
Slide 21
00:54:02 – Cumulatively, we’re talking about an 8 billion dollar market that is growing at greater than 25 percent. We do believe that we have a source of competitive differentiation in the space that we’re excited about. I think, in general, and we’ve done this, and this is unique for our interposers, because it’s a known-good-die, hybrid integration of active components, that the cost benefits of our platform actually extends from a single-channel 10G to a quad or an eight- channel 400G, and that is not something a lot of people can claim. We do have that benefit of providing wafer-scale integration and cost benefits across a variety of different applications. In general for most photonics components, in this particular case we’re looking at a transceiver, the optic – and that includes the die as well as the packaging – is about 75 percent of the total bill of materials. And we believe that as a consequence of eschewing free-space optics and active alignment and doing things at wafer-scale, we can lower the cost of these solutions by about a factor of two. First of all, it enables that ASP curve to keep coming down, but while it’s coming down, it enables module makers that use our solutions to maintain healthy margins for themselves. That’s why even though we are late on 100G, relative to adoption in the market of 100G solutions, we believe we have opportunities in 2019 with a few of the customers that we’ve been talking with.
Slide 22
00:56:05 – This is an example of market pricing in 2019 for a CWDM. I think the pricing now is probably in the $200 – $250 range for a transceiver, depending on who you talk to. But we believe our engines can be provided in the range of about $50 – $70. That means if you utilize the POET engine, then a transceiver could cost somewhere in the $150 range, as opposed to the $200 – $250 as it is today. It does allow for a reduction in ASP on the part of the module maker while maintaining margins.
00:56:52 – So we’re excited about this. And we talked to customers about it and they get excited about it. And whether it be one customer who wants a CWDM for a data centre, or another one who wants it more for LR4, or someone else who wants something for telecom, they are all looking to utilize the interposer concept that we have put out there to effectively lower their cost in the markets that they serve so they can be more competitive, and that’s the interactions that we’ve been having over the past six months. I was out in Asia a couple months ago again and the level of interest is humbling. We knew that it was the right strategy but it’s good to get that response back from the market. The onus it’s on us to be able to deliver. That’s what we have been really focused on with the team, and we believe we have made a lot of progress.
Slide 23
Slide 24
00:57:55 – In terms of milestones, there’s been a lot of discussion about “hey what’s next” and a lot of conjecture on timelines, and I thought it’s important to kind of level-set where we are and where we are going.
00:58:12 – I think, like I’ve said, the most important piece of what we are trying to do is to establish the platform capability. We do think that it’s time well-invested to have that be robust, so that products can be spun-out on a robust platform without significant re-engineering. So the process transfer to SilTerra on the interposer we expect to complete in the first half with the installation of our tools and the processes. We get into a multi-chip-module qualification in the second half of the year. We would use a single vehicle to do that, which would be a coarse WDM data centre solution. And then we would expand, at that point in time, the portfolio to other protocols like LR4, or to other market verticals like telecom, GPON, or sensing.
00:59:11 – We have intentionally, actually, not invested in growing in a dramatic way our sensing business, because we do believe that once the interposer is available as a platform it can be applied to our sensing products, and then we can start growing sensing with the basis of a differentiation and a competitive edge. Otherwise we are just marketing sensing solutions and competing on thinner margins as opposed to having a real competitive differentiation or competitive edge. And so that’s what we intend to do on the interposer-platform itself.
00:59:50 – As it relates to spin-off products, I think our key is the 100G, so we’d finish like I said our PIN qualification in the back-half of this year and put out our receive optical engine prototypes and complete that qualification as well. So we drive orders and revenue off of these products in 2019. The partnership that we have announced with Almae allows us to accelerate our time to market on the transmit/receive side. Obviously the ASPs, if you will, on the receive-only is less than the ASPs on a transmit plus receive for exactly the same unit-volume that we would have to sell. So moving from receive to receive-plus-transmit represents another growth in revenue opportunity for us.
01:00:48 – As we have mentioned before, our PIN receivers are very close to already meeting the requirements of a 400G market. Typically, a 100G PIN receiver needs to meet a bandwidth of about 17 GHz, and a 400G receiver needs to meet a bandwidth of about 36–38 GHz. Our PIN receivers today are at about 30 GHz, so we know they meet the requirements of a 100G application, and we are looking to optimize the last 20 percent to be capable of 400G. We believe that by the middle of next year we would be in a position to put 400G applications or prototypes out, with effectively the same interposer. The interposer does not change going from 100 to 400, once it’s done it can be applied to multiple protocols and multiple speeds. That’s kind of the beauty: We are not co-engineering these, we engineer them separately, and assemble known-good-die.
Slide 25
01:02:04 – So, in summary then, we are addressing large and rapidly growing sensing and datacom markets. We have highly differentiated technology. You know, I’ve been in technology for a long time in my career, and I remember a mentor of mine told me, “If you don’t have competitive differentiation then don’t compete.” And this is an area that I do believe we have got an edge. And I think the markets are starting to understand what we are doing and they understand the edge that we could provide them.
01:02:45 – We are participating in a high-volume space where cost is critical, and we do have an architecture that enables low cost. We have made a lot of progress in technology development. You know, I think technology development in something like this is hard, and it takes a lot of dedication and focus. It’s not always linear, and we have had to change from a lab to a fab. So given all of that, the progress that the team has made has been good. We expect to make faster progress now that we are actually in a foundry that can turn wafers on a faster timescale over the back-half of this year, and we would be tracking to an initial customer revenue ramp in 2019.
01:03:38 – We are not doing all of this ourselves. SilTerra is doing the manufacturing of the interposer. We are internally doing the manufacturing of the indium phosphide chips. We also have partners for the assembly, the pick-and-place test, and we are establishing those supply-chain partners along the way.
01:04:01 – And, like I said, we are getting increasing customer traction. Of course we announced one potential lead customer, subject to us meeting all of their requirements. But there are other customers that we are engaged with that we can’t announce, and I think that we need to understand that there are sometimes confidentiality requirements that prevent us from talking about everything that we are doing, but I did mention that we have got another customer looking at one of our laser products utilizing an interposer that we are actively working on in the company right now.
01:04:37 – So, I’d say, fast-forward a year from last year when we made a strategic decision to invest in the concept of the dielectric at that time – we hadn’t even invented the concept of the interposer – I think, that came by along the way as we really focused our resources towards commercializing the dielectrics that we knew we had something, but we converted that into something that is very tangible and something that we can actually talk to a lot of customers about, and they can start assessing the value that we bring. But the ability to have these discussions with tier-one players in the US and Asia has been staggering, and I think collectively, I believe we have made a lot of progress over this year.
01:05:30 – I know there is a lot of sentiment out there about, well, you know, I thought we would get revenues in on the interposer in 2018, but think of it in the context of, you know, we started, in earnest, the development in July of last year, and in the course of the 11 months, we have signed up a customer that is eager to work with us on a product, a few other product developments based on customer needs, signed a manufacturing partner, signed a development partner for our lasers, and product development does take some period of time.
01:06:08 – Qualification itself is a quarter. The qualification on the lasers is two quarters: 5000 hours of qualification. So, I mean, from that perspective, we believe we have an aggressive and achievable plan for growth in 2019. So that’s how I would end it, and then we’ll now take Q&A from both the web as well as the audience.
Slide 26
Q&A
01:06:42 – Leanne Sievers: Good morning, I’m Leanne Sievers with the Shelton Group, POET’s investor relations firm. As many of our shareholders know, we made available on POET’s website the opportunity to submit questions in advance of this meeting, for those who were not able to attend in person. There was a number of repetitions in those questions, so we have consolidated them, so that we can cover as many as possible. Management can address those, and then we open the floor up to your questions.
01:07:15 – To get it started, Suresh, this first question will be to you. The shareholder had asked that the transceiver, as he understood it, consisted of a silicon-based ASIC, ROSA, TOSA, and the dielectrics, assembled on a module. He was curious as to how the company could market a transceiver that does not have a TOSA in it.
01:07:47 – Suresh Venkatesan: Yes, good question! I’m hoping that my presentation added some clarity to what we are talking about. I think, we have been careful to talk about what we offer is a transceiver optical engine, not a transceiver. Our goal is not to make a full transceiver. We are an optics provider, a photonics provider. A transceiver has more than just optics in it, and we are not providing the electronics. What we would provides is a optical engine.
01:08:11 – So we would start with a receive optical engine, which would have photodetectors as well as the demux in a chip-onboard-compatible MCM [multi-chip module] and then move to a full transmit-and-receive optical engine, which would have the lasers, the modulators, the mux/demux and the photodetectors.
01:08:36 – Outside of that optical engine are laser drivers, trans-impedance amplifiers, limiting amplifiers, and a bunch of micro-controllers and electronics, which our intent is not to make ourselves. We would partner on the high-speed, because we do believe that once we get to 400G and beyond, these electronics components need to be on the interposer in close proximity to the optics, they cannot be on the PCB [printed circuit board]. So we are providing that flexibility and a potential to partner with electronic players and integrating them into the interposer, but we will not be making electronics chips. I think this is an important point to note.
01:09:11 – Transceiver module makers do buy individual components and assemble them in a module. What we are doing is partially assembling all of the optics for them and providing them an MCM, or photonics MCM in this case, that can be used in the transceiver, so they don’t have to now individually assemble and test and so on and so forth the optics piece of the chain.
01:09:49 – Leanne Sievers: I guess a follow-on question to that was, you kind of addressed it in some way, about the degree of difficulty of developing these products to ensure the success for POET?
01:10:01 – Suresh Venkatesan: I think the question in particularly is related to a statement I made earlier about, you know, we are focused on the receiver first, and then we do the transmitter later. The transmitter is more difficult, is I think the word I used. And I think, the reason is, just from a device physics perspective, that the laser is a high-current device that generates a lot of heat. The receiver is not. And it is also very difficult to modulate a laser at very, very high frequencies.
01:10:30 – So while we have internal capability on lasers that are non-modulated, so they are continuous wave, as well as high-frequency receivers, we decided that we would partner on the high-frequency laser side of things with our recent announcement with Almae to accelerate time to market.
01:10:53 – So I think with that in place, as we jointly develop these products, we now feel like we are better positioned to provide a complete solution versus where we were a quarter ago.
01:11:11 – Leanne Sievers: Then there were also a number of questions related to the India Space Research Organization, also referred to as ISRO. Suresh, can you confirm whether the company has submitted a bid or bids for transfer technology for GaAs-based devices?
01:11:29 – Suresh Venkatesan: We typically will only make announcements or discuss agreements or contracts that we have entered into, so at this point we will not comment or speculate on anything associated with ISRO or SCL in this case.
01:11:46 – Leanne Sievers: Okay. And then the next question, also to you, Suresh, in regards to sales from DenseLight: What are the prospects for more significant progress, given the investments and promises for driving near-term revenue?
01:12:03 – Suresh Venkatesan: Yeah, our strategy with DenseLight was to use that facility and grow that team to focus our investments on new products in data communications. I think we realized that while sensing was a big market, it is a very fragmented market, and for us to compete in that fragmented market, I didn’t want to do it with old technology, so we needed to develop the capabilities on the interposer before we went and started to grow that in any significant way. So our investment, I’d say, roughly 80 % of our investments, have been focused on data communication products, and we’re systematically starting to qualify lasers first, the receive optical engine we’ve talked about, and move in that direction, which is where the growth for this company is going to be.
01:12:55 – With that said, our product sales at DenseLight have grown 25 % year-on-year. We started at a low level, so we’ve been growing that, but we’ve not been growing that with significant investments. We’ve basically been growing it with more sales focus and support in specific regions of the world that we didn’t have much exposure into. But we expect the growth in terms of revenue for this company to be primarily coming from new data communications products where the volumes are larger, the market is not quite as fragmented, and where we believe we have a competitive edge with which we can market ourselves.
01:13:39 – Leanne Sievers: This next question is for Tom. Tom, do you expect more analysts to follow POET? Or could you comment on progress there or opportunity?
01:13:51 – Thomas Mika: So, as you know, we are already covered by H.C. Wainwright, and they put out regular coverage on us, so if you’re not an H.C. Wainwright client, maybe you would want to think that you should be. But, in addition to that, we have been either contacted by, or we’ve contacted a number of analysts in the industry at the OSC conference that was referenced earlier, that occurred back in March. Many of the analysts were there, and we met with several of them. I think that there is an appreciation for what the optical interposer can bring to the industry, and there’s interest from analysts. And I think that within the next six months, we hope to pick up some other coverage.
01:14:54 – But I think it’s important for people to understand that, generally speaking, analysts at investment banks cover large companies. And so we want to be covered, but it’s a hurdle that we have to overcome. We have to ensure that they understand the story and they understand not only the potential for POET, but also how this might affect the other companies that they cover. So that’s the basis on which we’ve had these discussions, and I’m pretty optimistic we’ll get some coverage, and if not coverage, we’ll certainly get mentions in industry reports.
01:15:38 – Leanne Sievers: Yeah, I would agree with that as well. And now I want to thank Kevin for being here today, Kevin Dede. So the next question is for Suresh: What should shareholders look for by the end of the third quarter, or perhaps by the end of the year?
01:15:53 – Suresh Venkatesan: Yeah, I think, I mean, we covered that. I believe it was the second to last slide in the presentation, so I think we’ve been transparent on what we’re working on and what we believe our critical milestones along the way to get us to where we need to be in the first part of 2019.
01:16:14 – Leanne Sievers: Yeah. Okay, a next question: Are all the resources spent on qualifying the interposer platform and producing the DenseLight SLEDs, or are there other projects?
01:16:16 – Suresh Venkatesan: No, I think we have our hands full. So, yeah, no, there are no other projects. I think, we’re 100 % focused in the company on delivering on our current strategy, which is the interposer, with its first application being a 100G receive module.
01:16:44 – Leanne Sievers: Okay. This next question, the next few questions actually, are for Tom. Tom, when does POET expect any new revenues, and will you give any guidance or estimates associated with the optical interposer.
01:16:58 – Thomas Mika: I think that Suresh made it clear that our expectation for revenue is in 2019 on the interposer-based products. I think that guidance is only … you can only give guidance when you have a high confidence that the orders are going to be there and the sales are going to be there and you have the capacity to deliver. So typically that comes with backlog, and I think at the point in time in which we have backlog and a reasonably predictable revenue stream, that’s the point in which we would give guidance.
01:17:41 – Leanne Sievers: Okay. And this next shareholder had noticed that, and was inquiring as to why, in the latest presentation the POET platform was removed, acknowledging that there was in previous presentations it depicted three versus the two?
01:18:00 – Thomas Mika: So, last year at this time, we told shareholders that we were essentially shelving the gallium arsenide project until such time as we found a strategic partner that had the resources to move that forward. And what I’m finding, because I give some of these presentations to investors and to generalists, that it’s difficult enough to communicate about what we’re doing, because it’s highly technical, and what I found was that the mention of the gallium arsenide platform kind of interfered with the message we were trying to deliver, because we are really very much focused on the indium phosphide platform and the optical interposer. So that it was just a matter of trying to communicate crisply to generalists and investors.
01:18:58 – Leanne Sievers: And the next question is in regards to the goals the company has set in the past like being cash-flow positive, and they were inquiring as to, kind of, are we currently at an inflection point, or do we have any better forecasts?
01:19:14 – Thomas Mika: Well, we don’t … I think we won’t be at an inflection point until we’ve introduced and qualified these new products, and we start to receive customer orders. We haven’t really increased our burn rate that much. So our break even is still pretty low. So I think it’s … I can’t say that we’re at an inflection point yet. I think we will be soon, and I think we’ll give guidance when it’s appropriate to do so. As far as estimates for cash flow break even previously given, I think, you know, the best we can do is try to figure out how much sales we would need to achieve in order to get to cash flow break even, and with the focus on the new products, which is where we believe the future is, you know we specifically decided to invest in those new products. Largely what we’re doing in Singapore is related to that, as Suresh has said, so I think we made a choice: to develop a company that has a great potential, based on new products, rather than trying to chase cash flow break even at the expense of the future of the company.
01:20:53 – Leanne Sievers: The next question – Tom, you can start on this, and then, Suresh, if you want to add anything – the question is in regards to there were other goals stipulated in presentations and MD&A about obtaining multiple partners or collaborations. Can you comment on what the progress is there and any other partners that the company is seeking?
01:21:18 – Thomas Mika: So, Suresh covered three major partnerships that we have, so we’ve actually delivered on that promise, and there are likely to be more in the future. But we’re not in a position to comment on anything specifically until we actually have some agreement with a partner.
01:21:41 – Leanne Sievers: Okay. And then, Suresh, this to you. Given the advent of artificial intelligence and quantum computing, and the more advanced chips designed for quantum computing and the length of time it takes to develop, will POET be outdated by the time this is finally developed?
01:22:02 – Suresh Venkatesan: Yes, so, I mean, that’s … so we don’t make quantum computing chips or AI processors. What we are building, is a communication fabric for these processors to communicate large amounts of data at high speed. So from that perspective our goal is to ensure that we establish an interposer capability that can house these chips and provide them with an optical backbone, or an optical fabric, for communication. So that’s really our strategy and our path, not to build the chips themselves.
01:22:48 – Leanne Sievers: Okay. And I’ll continue with you, Suresh. Do we still have ongoing projects with A*STAR, and if yes, at what stage is it, and what is the scope of those projects?
01:22:59 – Suresh Venkatesan: We had started a project with A*STAR, or IMRE specifically, and it was initially targeted at some advancements that we wanted to make on the GaAs technology. This was put in place back in 2016, I believe.
01:23:20 – We continue engaging with them, but we have changed their focus as well, in terms of what they do. They are working with us on DFB lasers, for example, at this point in time. So again, even with A*STAR, we have changed their focus, to be focused more on where we want to go, which is the indium phosphide and the interposer platform. So, yes, we do continue to work with them, but what they working on is different.
01:23:55 – Leanne Sievers: Very good. – I’m going to shift to you, Dave. There was a question in regards to if management can explain how, based on both the short and long-term progress and share price, they believe the options granted are justified?
01:24:09 – David Lazovsky: Sure. So, I think it’s important to talk about the stock options plan from ??? and the critical role that stock options plans play for companies, particularly for companies at our stage. POET has been focused on – you know, we’re pre-profitability – we’re focused right now on preserving cash. Equity plays a vital role in a company at our stage and ability to attract and retain talent. Suresh just walked you through the development activity. We went through last year this time the strategy, focused on data communications. Suresh walked through the progress we’ve made against that strategy. A key component of our ability to execute against our strategy is the ability to attract and retain the right people. An equity plan facilitates our ability to do that, and we appreciate the support of the shareholders that we saw today in the vote in supporting the equity plan to approve that.
01:25:15 – Now, I do know that there was another question that was submitted regarding the management team potentially seeing an incremental grant at the AGM, because the timing of a grant, from year to year, can vary. I will clarify, the management team received a grant in the first quarter, in March of this year, I believe the strike price, the exercise price for that grant, was well above the current stock price, it was about 51 ½ cents. I can tell you that the executive management team will not receive another grant at this AGM nor at any point in 2018. So, any further grant that takes place in the near term would just be the standard board grant that would exclude myself and Suresh as we have operating roles. But no incremental grant for the management team in 2018, just to clarify.
01:26:15 – Leanne Sievers: Thank you for addressing that. Then the last two questions are back to Tom, financially-related. Tom, can you expect profit margins to continue at the approximate 60 % range with continued growth and development?
01:26:32 – Thomas Mika: So we do. We’ve been able to maintain those margins in the sensing products, and all of our plans going forward on the datacom side are at about the 60 % gross margin level overall. I think we probably won’t see it in the financials, but obviously when you introduce a new product, you’re going to take lower margins at first, because the cost is higher at first, but I don’t see that really bleeding through to the overall financials in any significant way. So, I think, we’ll be in the 50 – 60 % gross margin range if we meet our plan.
01:27:17 – Leanne Sievers: Okay. And then the next question was in regards to a potential up-listing and when the company may be ready to either move to the TSX big board and/or a dual listing on NASDAQ?
01:27:30 – Thomas Mika: Okay. So, we’re really happy being on the TSX Venture Exchange. And there are specific rules about moving to the TSX, which has to do with the length of time that we’re profitable and we have revenue. Regarding NASDAQ, a couple of us here have the experience of actually being in a company that is on the NASDAQ, and, you know, it can be quite difficult for a company that is under 100 million dollars, that doesn’t have a very predictable revenue stream, to get any attention whatsoever on the NASDAQ. So it’s not that we don’t have ambitions to be on NASDAQ, it’s just that it’s not the right time really to talk about it. There will be a right time. And when that right time comes, we’ll see what we can do. But having run a company that was on the NASDAQ, it’s a very difficult place to be if you’re under 100 million dollars and you’re not wildly profitable, or you’re not some kind of social-media company.
1:29:00 – Leanne Sievers: Okay, thank you. So that covers the questions that were submitted, and now we’ll open it up to the floor if anyone has questions. Yes, Mike?
01:29:21 – Michael White: Hi! Mike White, IBK Capital. Suresh, this question is for you. On your milestones slide, you mentioned customer orders, and I am just wondering what are the expectations of POET and the expectations of your customers and partners as to initial and early-stage customer orders?
01:29:56 – Suresh Venkatesan: I think they come in sequence, right? We have to provide a prototype that meets all the specifications for the applications. It need to work within their module. It has to be interoperable, because it has to eventually work with all other people’s modules. I think once that it is validated, then you turn it into orders. So that is why we have a sequence of things we need to do and an expectation of when we believe orders could be placed, which then ultimately translates into sales and revenue.
01:30:34 – That said, I think there are some products that are currently in a sampling mode, so those sample quantities do generate revenue. So we do have orders from that perspective, but they are small orders, sample orders for test and validation purposes.
01:30:55 – Michael White: So for the ROE and the TROE, what would your initial size expectations be in terms of the initial and early stage orders?
01:31:05 – Suresh Venkatesan: I mean, I can’t, I don’t want to speculate on the fraction of the wallet, if you will, that we would get right from the get-go. I would expect that if we deliver on our cost targets, it could be significant, but it would be a ramp. I can say that the size of the market, the revenue opportunity, like I said, with a couple of these players is … is quite significant. It’s, like I said, north of seventy-five million dollars just for a couple of products for a couple of customers. But that does require – I mean, that’s annual – so it requires us to be qualified and then ramp to a significant portion of their wallet share. I am confident that if we are qualified and we deliver on our cost targets that we think we can deliver on, then ramping to a significant portion of their wallet share is going to be not hard to do. But it’s also subject to … you can’t speculate on their agreements with other suppliers and so on and so forth. So I think I would leave it at what is the size of the opportunity that we could get if we executed, and then focus on the execution.
01:32:31 – Michael White: And do you have the ability in terms of capacity in the supply-chain to ramp to those potential expectations?
01:32:39 – Suresh Venkatesan: Yeah, I believe we do. We’ve been investing in capex. I think the agreement that we have with SilTerra provides us with all the capacity we probably need on the interposers for the near future. There are some incremental investments we would continue to make internally in our factory to meet the capacity requirements, but I don’t believe at this point in time we’re going to be capacity-constrained, post some investments we do need to make on the capex, which we’re actively doing right now.
01:33:19 – Michael White: Thank you.
01:33:35 – Darius: So my name is Darius. I am one of the investors. First I have to compliment the team on what appears to have been a lot of hard work in the last twelve months. Having said that, I have two questions. One of them is: What could cause the POET Technologies’ train to go off-track? And the other one is: What is the likelihood of us needing to raise funds one more time?
01:34:03 – And, lastly, I want to say that even though I think that for the good work that’s being done people deserve to get options, but it would really inspire a lot of confidence in the investor’s mind if the options are priced at a slightly higher price. I understand you could give yourselves more options, but at a higher price so that it would appear that ok, well the goals are higher than otherwise. Ok, I’ll just sit down and listen for the answers. Thank you.
01:34:47 – Suresh Venkatesan: Do you want to talk to the options question?
01:34:52 – David Lazovsky: Yes. I’ll address first the options question. I discussed that previously so … but I’ll recap. I think there’s standard practice in terms of how stock options plans are structured and how they’re executed. What we don’t want is our stock option plan to be an outlier relative to market standards in the US or in Canada or anywhere else in the world.
01:35:17 – It is important that the employees and … the management thing is one thing. Every level of employee throughout the company we want to ensure is well-aligned and focused on execution of the company’s objectives, which are well-aligned and focused on delivering return to our shareholders, right? Delivering on a strategy and executing a plan that’s going to allow us to grow this company, that strategy laid out last year and recapped today. It’s crucial to build the right team, so the ability to retain, attract the right people, the right high-caliber talent, which we’re compensating with equity more heavily today than with cash for … by design. Because it’s very expensive, right, for a company to pay premiums and in a very competitive work environment, like we have today in the photonics sector [where they] pay high salaries with significant bonuses, which you have not seen out of POET throughout the employee days. So we have to compensate. We want to make sure that … The biggest risk that we can have, coming back to your earliest question, is not having the right people involved with this company to allow us to execute a plan that we know there is a market out there, and customers that are eager to help us achieve our business objectives. Right? So, folding the two elements of your question together, we need to attract the right people and we need to retain the right people.
01:36:56 – Thomas Mika: So, obviously, I agree 100% with what Dave and Suresh have said. You know, last year we identified as a source of financing for the company our outstanding warrants, and actually with the last capital raise we added to that pool of warrants at a higher price. I still regard that as being the best source of additional capital for us, because it’s effectively non-dilutive. Those shares are already included in the fully-diluted share count. So we’re definitely going to focus on that as our first alternative, and we’ve got an arrangement in order to try and encourage people once the price is where it needs to be to get those warrants converted as quickly as possible. You’ll notice that in the most recent raise, the life of those warrants is significantly shorter, so I think we’re doing better as far as raising capital is concerned.
01:38:12 – I think though … I see a lot of comments about dilution, and, you know, I’m just as concerned as every other shareholder, not just at POET but at other companies, about dilution. But the reality is two things: one is, if you’ve got an increasing share price, which comes as a result of your performance, dilution really doesn’t matter. I mean, as long as the shares that you own are increasing in price and you can sell them for a higher price, then, you know. And that’s essentially how large companies get created.
01:38:52 – The second point I would make, and it’s not specific to POET, but it’s just a general point, is that I’ve never come across, ever, a technology company that doesn’t require capital, because it’s always doing some new things, it’s always creating new products, it’s always thinking about new places to produce those products. And so over the lifetime of a company, for a CFO or anybody else to say “well, this is going to be our last capital raise”, I think is just wrong. It’s not something I would say whether it’s POET or any other company.
01:39:34 – But again, we think that the best source of capital for us are those warrants. They can bring in substantial capital without any dilution, and what we need to get those warrants converted is a lot of liquidity, a lot of trading volume, and a higher stock price … which we’re only going to get through the performance of the company.
01:40:05 – Darius: So you’re not speculating as to the likelihood of needing to raise more funds in the coming year, for example?
01:40:17 – Thomas Mika: I’m not! I’m not saying “no”, and I’m not saying “yes”.
01:40:23 – Darius: I get it. Okay, thanks.
01:40:29 – Dennis: Hi, my name’s Dennis. I had two questions. One was, Luxmux was, and hopefully still is, a customer of us with DenseLight. It looked like they had quite a good growth-pattern. I was wondering if we can expect increased sales to them as a customer?
01:40:53 – Suresh Venkatesan: Yeah, they’re still a customer and we still sell our SLEDs to them, you know, at whatever volume they need for their business. I can’t comment on their business. They are a customer of ours and we are increasing sales with them.
01:41:10 – Dennis: Good. And my other question was that data centers will want transceivers, and then we have our optical engines, but apparently there’s an electronics side that needs to be combined to complete and become a transceiver. So are we going to have to partner with an electronics-side company to make a complete package, a transceiver that we can sell to data centers?
01:41:40 – Suresh Venkatesan: We don’t intend to sell transceivers. I think I want to be very, very clear. We are not going to be a transceiver module company or a transceiver engine company, we are a transceiver component company. In the 100G space, module makers, who are customers, have their source of electronics. We are not in that supply chain. We are going to sell the optical engines. That includes the optical active components and the optical passive components, put together into a photonics package. That’s our business model.
01:42:20 – Now as frequencies go up and we get to the 400G range, it is possible that the electronics needs to be housed very close to the optics, and we have very elegant solution for that, and I think it will be a three-way partnership, if at all, between the module maker, anybody that is sourcing in electronics, and ourselves, to insure that we provide a solution that is elegant and can scale in frequency. But we will not sell a transceiver.
01:42:50 – Dennis: So we’ll sell our optical engines to a transceiver company that makes the complete transceiver?
01:42:58 – David Lazovsky: Yeah, that’s right. There are module manufacturers that currently buy the optical engines and currently source the electronics for those optics: drivers, transimpedance amplifiers, etc. So we are engaged with those electronics providers, ensuring that we understand the performance requirements of their electronics and are compatible with our optics including some of the data that I believe Suresh has shown around the PIN receiver technology. And the customers that we’re engaged with, including the large Chinese customers that Suresh reviewed today, have specified specific electronics providers that they would like us to work with and we are working with those electronic providers in many cases to ensure compatibility of our optical engines with the existing electronics that they’re currently using.
01:43:56 – Dennis: So, we’ll sell our optical engines to transceiver manufacturers? Is that correct?
01:44:02 – David Lazovsky: Yeah, so Accelink for example. We have a MOU with Accelink, we’ve disclosed that. They’re a module maker. So what Accelink does is, they would buy the optical engine from us, and they would buy the electronics to run that optical engine from an electronics provider.
01:44:17 – Dennis: Got it. Okay, thank you.
01:44:26 – Eric: My name is Eric. A little more on that note. So, is it true then that the sales of the interposer ultimately depend on successfully integrating somebody’s transceiver? Because right now you have the ROSA, the receive part integrated, but we’re still waiting.
01:44:46 – Suresh Venkatesan: Maybe I’m not clear. The transceiver has in it the receive engine, the transmit engine, and the electronics. So we will be selling the receive engine initially, that would be used to make a transceiver. So we sell a receive engine that meets the specifications of the protocol, and it would be incorporated into a transceiver for subsequent sales to the system providers.
01:45:26 – Eric: Okay, so currently with the … so you have the end of the year, or the beginning of 2019, before you get the TOSA integrated to the interposer, but do you anticipate sales even if … prior to that?
01:45:45 – Suresh Venkatesan: Yeah, I think … because they’re separate chains in the transceiver, there’s a receive chain and a transmit chain. So you can sell a receive chain, and the transmit chain will be someone else’s. And then when we incorporate the transmit chain in the interposer, then we would own all the optics in the transceiver. But initially we will not own all the optics in the transceiver, so our share of the transceiver’s bill of materials will be less for the receive-only and will increase dramatically for the receive-plus-transmit.
01:46:32 – Thomas Mika: One thing that I’ve found in doing other presentations is that sometimes there’s a misunderstanding about these products. These products when they’re bought and sold meet specific specifications, and they have to communicate at a certain frequency with a whole range of specs that they have to meet. Once they meet those specs, they are interchangeable, and so you can take one from one manufacturer or from another manufacturer, and what that essentially means is that the basis for competition in this industry is cost. If you can come in delivering the same performance with a substantially better cost, you’re going to win, and that’s the whole basis for competition in this industry, and that’s why these parts are continuously interchangeable with one another, particularly among the module manufacturers who are simply assemblers of products. There are some that are very much vertically-integrated, where they will manufacture every component in the supply chain. And they’re driving their costs down by perfecting every … perfecting their device manufacturing, perfecting their packaging, and even those use, some large ones use, conventional optics, they’re just doing it better and better and they’re committed to that. So they’re never going to …, well, they wouldn’t switch over to our optical engine unless they were really making an entire revamp of their whole product line. But on the assembly side, which is characteristic by the way of China, virtually all those companies are assemblers, so they’re buying from suppliers all the time as long as the component meets the spec.
01:48:39 – Michael White: Hi, it’s Mike White, IBK Capital, again. Suresh, will POET’s technology ultimately be sold through … POET’s optic-engine, will it ultimately be sold through to the Chinese titans Alibaba and Tencent? Are we dependent on those types of companies to better understand what we might be able to expect in terms of product sales and revenues going forward?
01:49:17 – Suresh Venkatesan: Yeah, I mean I think we’re … as long as we’re going to be selling into a data center market, we’re obviously dependent on the growth of the hyper scale guys. That’s not limited to just the three in China’s BAT: Baidu, Alibaba, Tencent, and there’s your other household names in hyper scale. So, yeah, absolutely, we’re dependent on hyper scale growth to continue to drive demand, and that is projected to grow at a 25 % CAGR. But there’s a system guys, they are serviced by module providers, so it’s unlikely that we’re going to get a direct sale, if you will, but we would be in the supply chain.
01:50:15 – Michael White: And they continue to expand in terms of their offerings, they all have digital wallets now, their own payment systems, they’re collecting information from many different sources. I can only imagine that their need for increased speed is rapidly growing. Will POET’s optical engines be able to keep up in terms of the speed demands of the near future and medium-term future?
01:50:48 – Suresh Venkatesan: Yeah, absolutely. The interposer, as I said, is speed-agnostic. It’s a packaging platform that can lower the cost of photonics assembly. So I think there’s a roadmap for increased speed on the actives that we’ve laid out going from 100 to 400, which is really the next standard in the industry, and beyond. I think it’s not necessarily all speed, it’s bandwidth. It’s how many bits you can transmit per second. You can do it two ways. You can do it by increasing the speed in a lane, or you can do it by keeping the speed in the lane constant and adding more lanes. Either way you transmit more data. It’s about … the speed is not important, it’s the amount of data that can be transmitted per second. So if you get to a 400 Gbps solution, then extending it to 800 means adding more lanes. I think we have the ability with the interposer to add more lanes while keeping a cost structure that’s very competitive, because we’re doing it in an integrated fashion with the waveguides and the interposer. Adding more lanes in a more conventional optics configuration is more expensive, because each lane is done individually with separate components. So we do have that flexibility to increase bandwidth, data bandwidth, which is really the more important metric.
01:52:32 – David Lazovsky: I’ll make one additional comment, echoing what Suresh has said. As you think about increasing bandwidth requirements as more machine learning applications, AI applications, cryptocurrency applications are being used, the more channels that are required, the higher the competitive differentiation of the POET optical interposer platform. Suresh sort of modestly covered this a few times, but it’s a very import thing for our shareholders to understand, the scalability of this platform with small, incremental, additional costs, going to much, much larger bandwidth applications. As the industry progresses over the course, especially the mid to 400G and beyond applications, our competitive differentiation expands.
01:53:26 – Michael White: I can see why you’re getting such great potential customer traction. I just want to say thank you personally to all the directors and management team of POET. I think you’ve done a tremendous job over the last year, and it looks like your future is very, very bright, so congratulations to you all.
01:53:54 – Q: You mentioned coming to the 100G market at a time, maybe at the middle of the market, and then getting to the 400G more towards the beginning of that market rise. Will companies go ahead and buy your 100 when the 400 is coming soon thereafter, or will they just wait?
01:54:22 – Suresh Venkatesan: I mean we’re … it’s your traditional S-curve, right? So as the volume goes up, ASPs come down, so there’s always a need for a solution that can provide better margins. So we do believe that there’s an opportunity over the next several years to be competitive in the 100G space as well, right? But obviously, I mean, there will be a transition from 100 to 400, and we want to be well-positioned for that transition to occur by having capability and prototypes out next year, but that ramp is not expected till 2020 – 2021. In the meantime, the growth of all 100G … there’s a nuance, the initial ramp of 100G was on a protocol that’s called PSM4, that’s parallel quad-lane, as opposed to the new growth is intended and expected to be in what is called CWDM, or coarse wavelength division multiplexing, which is where we add a lot of value, because we’re incorporating the de-multiplexer and multiplexer into the interposer. So that’s where the next wave of 100G growth is, it’s moving away from PSM and into WDM, and so that’s where our opportunity is for both long-reach as well as medium-reach in data communication. So, long answer, but in short, yes there is an opportunity at 100G.
01:56:05 – David Lazovsky: One other comment on the 100G market is that even when the 400G market begins to ramp in the 2020 – 2021 timeframe, the size of the 100G market is going to be much, much larger and continue to be stable.
01:56:20 – Q: So there won’t be an exodus out of the 100 altogether?
01:56:24 – David Lazovsky: The 100 will be much, much larger than the 400 at that point.
01:56:26 – Q: And, Suresh, if we went from this point up until the point where the interposer is done and ready to be put into the hands of customers, whatever point that might be, what would you say your biggest concern is getting from here to there?
01:56:45 – Suresh Venkatesan: I mean, I think … We had initially to deal with the logistics: transitioning out of a lab, buying tools, installing tools, bringing up processes, that consumed us over the past quarter. Going forward, it’s the design optimization. You know, I think between now and the end of the year, we expect to get increased cycles of learning to optimize the design to meet the required specifications. Thereafter it’s going to be yield and … more operational, manufacturing, ramp-related challenges that anybody would face. But technically, it’s the design-optimization, primarily on the passives, the passive waveguides in the dielectric, the passive waveguides in the indium phosphide. I think that’s where our focus is going to be over this next quarter to optimize performance. I think there’s a gap to performance requirements on the order of about 10 or 15 %, and it’s kind of the last mile. You just got to work on fixing those issues. That’s what we’re working on right now.
01:58:06 – Q: Thank you very much.
01:58:11 – Leanne Sievers: I think that concludes our Q&A session and I turn it over to Dave.
01:56:15 – David Lazovsky: Okay, well, with that, we truly appreciate participation of all the shareholders, both in person as well as those of you who are listening into the webcast, and with that, we’ll bring the meeting to conclusion. Thank you very much.
Bitte beachten Sie die Hinweise zu Risiken und zum Haftungsausschluß!
Rainers Blog 