Scaling networks for AI infrastructure: a conversation with MOX Networks

1 00:00:00,583 --> 00:00:02,619 I am Bruce Hembree of Ciena. 2 00:00:02,619 --> 00:00:06,856 And I'm here with Allen Meeks, the CEO of MOX networks. 3 00:00:06,856 --> 00:00:09,843 So, Allen, tell us a little bit about MOX. 4 00:00:10,160 --> 00:00:10,910 Yeah. 5 00:00:10,910 --> 00:00:14,514 So MOX was founded, 13 years ago by Doctor Patrick Soon-Shiong. 6 00:00:15,031 --> 00:00:16,766 Initially it was to provide 7 00:00:16,766 --> 00:00:19,919 capacity for his health care mission of curing disease with big data. 8 00:00:20,537 --> 00:00:23,673 And we've evolved, you know, since then into a real carrier. 9 00:00:23,673 --> 00:00:24,374 A real carrier. 10 00:00:24,374 --> 00:00:29,412 And as a matter of fact, right now we're at Honolulu, Hawaii, at PTC. 11 00:00:29,679 --> 00:00:32,465 Where a lot of the carriers get together to talk about 12 00:00:32,465 --> 00:00:34,718 the next iteration needs and that of the network. 13 00:00:34,718 --> 00:00:37,804 So, I understand you've been to PTC a few times. 14 00:00:38,121 --> 00:00:40,607 Yeah, I think I've been coming since around 2009. Yeah. 15 00:00:40,607 --> 00:00:41,591 So definitely. 16 00:00:41,591 --> 00:00:42,742 Over 15 years. 17 00:00:42,742 --> 00:00:43,626 Yep, yep. 18 00:00:43,910 --> 00:00:46,679 And it's changed quite a bit in the years that I've been coming. 19 00:00:46,679 --> 00:00:48,181 Especially now. 20 00:00:48,181 --> 00:00:51,751 I mean, it's amazing with the subsea, landscape growing 21 00:00:51,751 --> 00:00:54,771 so much that, you know, this year is just so focused on subsea. 22 00:00:54,771 --> 00:00:56,956 And what a better place to do it than Hawaii. 23 00:00:56,956 --> 00:00:59,509 Yeah. Of course. It is where it all comes together. 24 00:00:59,509 --> 00:01:00,844 The crossroads as it were. 25 00:01:00,844 --> 00:01:02,662 So when you think of subsea, 26 00:01:02,662 --> 00:01:05,298 you think of MOX, you think of the changing landscape. 27 00:01:05,298 --> 00:01:07,033 You know, in some of our conversation 28 00:01:07,033 --> 00:01:10,253 before you talked about the dual mandate, of MOX. 29 00:01:10,270 --> 00:01:12,939 So remind us of what that means. 30 00:01:12,939 --> 00:01:13,406 Yeah. 31 00:01:13,656 --> 00:01:16,659 Again thinking of MOX as a, you know, the parent company, 32 00:01:16,676 --> 00:01:19,913 networks being like a healthcare cloud company, 33 00:01:19,913 --> 00:01:22,916 a healthcare AI company that had to found its own carrier 34 00:01:23,032 --> 00:01:25,852 to be able to supply its needs going out into the market, 35 00:01:25,852 --> 00:01:27,570 wasn't going to be able to give Patrick 36 00:01:27,570 --> 00:01:30,557 the type of capacity that he needed between certain locations. 37 00:01:30,623 --> 00:01:32,659 Our demand is different than a typical cloud 38 00:01:32,659 --> 00:01:33,676 provider in that, 39 00:01:33,676 --> 00:01:35,462 you know, cloud providers demand can come 40 00:01:35,462 --> 00:01:38,281 from kind of everywhere and anywhere all at once. 41 00:01:38,281 --> 00:01:40,049 We really know where our demand is coming from 42 00:01:40,049 --> 00:01:42,569 is we're doing patient trials in specific locations 43 00:01:42,569 --> 00:01:44,420 and doing our compute in different locations. 44 00:01:44,420 --> 00:01:47,724 But what's not, what is very similar and not different is the scale 45 00:01:47,724 --> 00:01:48,625 that we're dealing with. 46 00:01:48,625 --> 00:01:50,777 We're dealing with very similar scale challenges. 47 00:01:50,777 --> 00:01:54,397 So from the scale side you’ve dealt with, or you deal with, 48 00:01:55,415 --> 00:01:57,133 the unpredictability that the 49 00:01:57,133 --> 00:02:00,887 whole AI networking, infrastructure business is facing now. 50 00:02:00,887 --> 00:02:02,705 But you've got a bit more predictability 51 00:02:02,705 --> 00:02:06,276 with respect to the locations where you provide connectivity between. 52 00:02:06,676 --> 00:02:07,443 Yeah. 53 00:02:07,443 --> 00:02:08,361 And that's evolved, 54 00:02:08,361 --> 00:02:08,578 you know, 55 00:02:08,578 --> 00:02:10,280 through drug development and, you know, 56 00:02:10,280 --> 00:02:12,832 the methodologies that Patrick's doing to cure disease. 57 00:02:12,832 --> 00:02:13,066 You know, 58 00:02:13,066 --> 00:02:13,466 if we do 59 00:02:13,466 --> 00:02:15,718 a certain number of patient trials in one location, 60 00:02:15,718 --> 00:02:17,303 we're going to compute in one location, 61 00:02:17,303 --> 00:02:18,988 and we're sharing data in another location. 62 00:02:18,988 --> 00:02:20,807 These are national labs. These are universities. 63 00:02:20,807 --> 00:02:22,075 These are data centers. 64 00:02:22,075 --> 00:02:25,261 Interconnection points, and hospitals that we're dealing with. 65 00:02:25,261 --> 00:02:28,047 Right. So we know what the design needs to look like. 66 00:02:28,047 --> 00:02:31,050 But we, again, we don't know quite how much to invest for the scale. 67 00:02:31,201 --> 00:02:33,536 And that's the similar challenge that that we're seeing, 68 00:02:33,536 --> 00:02:34,470 you know, that we run into there 69 00:02:34,470 --> 00:02:36,472 a lot of other AI companies are running into now is 70 00:02:36,472 --> 00:02:39,359 how do you futureproof against the scale when you're not quite sure 71 00:02:39,359 --> 00:02:40,376 how much is going to? 72 00:02:40,376 --> 00:02:41,761 It's not just where we know where, 73 00:02:41,761 --> 00:02:43,213 but we don't know how much is coming. 74 00:02:43,213 --> 00:02:43,663 But you're, 75 00:02:43,663 --> 00:02:44,614 you're talking about things 76 00:02:44,614 --> 00:02:47,617 like genomic research and stuff like that, right? 77 00:02:47,750 --> 00:02:52,338 So, you know, when those universities are preparing 78 00:02:52,338 --> 00:02:55,325 for that scale, you know, what do they think about 79 00:02:55,325 --> 00:02:56,259 that's in common 80 00:02:56,259 --> 00:02:59,562 with the rest of the AI universe that we're working with now? 81 00:02:59,562 --> 00:03:00,914 So what can we learn from that? 82 00:03:00,914 --> 00:03:03,233 Yeah, I mean, from our perspective, it's it's, 83 00:03:03,233 --> 00:03:06,586 you know, we sequence a full genome and a full protein pathway as well. 84 00:03:06,586 --> 00:03:09,022 They call the proteome, we do that with healthy tissue 85 00:03:09,022 --> 00:03:11,057 and cancerous tissue for each patient. 86 00:03:11,057 --> 00:03:12,675 And that happens usually 3 or 4 times 87 00:03:12,675 --> 00:03:14,427 in a patient treatment life cycle. 88 00:03:14,427 --> 00:03:16,162 So you know, we're talking each sequence 89 00:03:16,162 --> 00:03:18,932 somewhere in the neighborhood of 12 to 15TB of data. 90 00:03:18,932 --> 00:03:20,550 And then you double that and then you times it 91 00:03:20,550 --> 00:03:22,168 by four over the life cycle. 92 00:03:22,168 --> 00:03:24,354 And that's not data we just have to store. Right? 93 00:03:24,354 --> 00:03:27,240 We have to elastically be able to compute on that data. 94 00:03:27,240 --> 00:03:30,593 And so really understanding again the scale where it's coming from 95 00:03:30,593 --> 00:03:32,061 and how we're going to deal with it. 96 00:03:32,061 --> 00:03:35,331 And preparing for that without. I like to call it future proofing. 97 00:03:35,431 --> 00:03:35,732 Right. 98 00:03:35,732 --> 00:03:37,400 Because you could look at a system, 99 00:03:37,400 --> 00:03:39,085 you might have enough capacity to give you a runway 100 00:03:39,085 --> 00:03:40,920 that gets you to a certain percentage of, 101 00:03:40,920 --> 00:03:42,538 you know, your demand at the time. 102 00:03:42,538 --> 00:03:44,424 But in our world, if something gets, 103 00:03:44,424 --> 00:03:46,759 you know, like an FDA approval on a methodology, 104 00:03:46,759 --> 00:03:48,745 we could go from a 10,000 patient trial 105 00:03:48,745 --> 00:03:50,930 to serving a hundred thousand patients overnight. 106 00:03:50,930 --> 00:03:52,248 And that could be multiple 107 00:03:52,248 --> 00:03:55,318 parallel systems between locations as opposed to just one system. 108 00:03:56,085 --> 00:03:56,920 That's really crazy. 109 00:03:56,920 --> 00:04:00,106 So a lot of the network providers, 110 00:04:00,106 --> 00:04:03,526 the service providers, as it were, out here at PTC, 111 00:04:05,295 --> 00:04:05,595 you know, 112 00:04:05,595 --> 00:04:08,598 they've made a living building networks, 113 00:04:08,615 --> 00:04:11,801 but you've got experience on both sides, both, you know, 114 00:04:11,834 --> 00:04:15,255 with the compute aspect of things like you're just describing, 115 00:04:15,455 --> 00:04:17,257 as well as the networking piece of it. 116 00:04:17,257 --> 00:04:21,194 So how does that expertise translate over into both 117 00:04:21,194 --> 00:04:24,781 how you build and operate the networks you have for 118 00:04:24,864 --> 00:04:27,050 non-medical applications out there? I mean it's a great question. 119 00:04:27,050 --> 00:04:30,086 We've had you know the traditional carrier methodology is really 120 00:04:30,086 --> 00:04:33,206 just driven around, you know success based opportunities. 121 00:04:33,623 --> 00:04:36,476 And that's really what's moved the industry forward for years. 122 00:04:36,476 --> 00:04:37,677 And it still is today. 123 00:04:37,677 --> 00:04:39,896 The uniqueness about us is that dual mandate 124 00:04:39,896 --> 00:04:42,181 and that we have our own capacity needs internally. 125 00:04:42,181 --> 00:04:43,549 And then if we need to overbuild 126 00:04:43,549 --> 00:04:45,501 to try to futureproof against that scale. 127 00:04:45,501 --> 00:04:48,388 But maybe that future scale doesn't come on the route that we build it. 128 00:04:48,388 --> 00:04:50,757 Well, now we've got excess capacity to sell to the market. 129 00:04:50,757 --> 00:04:52,458 And I think our reputation of being able to do 130 00:04:52,458 --> 00:04:55,461 these things, do fiber builds, you know, from scratch, permit, 131 00:04:55,728 --> 00:04:58,581 design permit, engineer, construct, you know, laying ducts 132 00:04:58,581 --> 00:05:00,717 in the ground on gas pipeline rideaways, 133 00:05:00,717 --> 00:05:02,902 rail rideaways, pulling fiber through that. 134 00:05:02,902 --> 00:05:05,788 You know, developing ILA sites all the way to, you know, 135 00:05:05,788 --> 00:05:09,959 developing, with you guys, high capacity C+ cell networks, 136 00:05:10,310 --> 00:05:12,979 as we go to light those. All of that, 137 00:05:12,979 --> 00:05:15,531 you know, from a traditional carrier standpoint, really 138 00:05:15,531 --> 00:05:17,216 was more based around the success space. 139 00:05:17,216 --> 00:05:19,168 But with that dual mandate, we kind of have a different 140 00:05:19,168 --> 00:05:20,103 economic model. 141 00:05:20,103 --> 00:05:21,821 Where it can be more intentional. 142 00:05:21,821 --> 00:05:22,438 That's right. 143 00:05:22,438 --> 00:05:23,106 Yeah, yeah. 144 00:05:23,106 --> 00:05:25,825 And that seems to be making a huge difference from the conversations 145 00:05:25,825 --> 00:05:29,562 we've had with whether it's the hyperscalers, the neo scalers, 146 00:05:29,562 --> 00:05:31,698 large enterprise, and whatever category. 147 00:05:31,698 --> 00:05:35,385 I mean, healthcare, you know, gas, oil, pipelines. 148 00:05:35,385 --> 00:05:35,985 You know, etcetera. 149 00:05:35,985 --> 00:05:38,988 It seems like y'all are really in the sweet spot, 150 00:05:39,138 --> 00:05:39,372 you know, 151 00:05:39,372 --> 00:05:43,309 for providing that unique expertise of knowing how to operate a network, 152 00:05:43,309 --> 00:05:45,528 but also the applications that are right on it 153 00:05:45,528 --> 00:05:47,630 to pull that together as an operator. 154 00:05:47,630 --> 00:05:49,015 That's right. Yep. 155 00:05:49,015 --> 00:05:51,167 I think a lot of the carrier experience has not 156 00:05:51,167 --> 00:05:52,552 they've not had that experience. Right. 157 00:05:52,552 --> 00:05:54,654 The carrier experience again has been success based. 158 00:05:54,654 --> 00:05:55,972 It's here's how much capacity I need. 159 00:05:55,972 --> 00:05:57,957 We respond to the request which is fine. 160 00:05:57,957 --> 00:06:00,960 And that's really what again pushes the industry forward in many ways. 161 00:06:00,960 --> 00:06:02,311 But yeah, our experience of being able 162 00:06:02,311 --> 00:06:04,680 to build and understand those scale challenges 163 00:06:04,680 --> 00:06:06,416 personally has really affected 164 00:06:06,416 --> 00:06:08,000 the way that we think about and build networks 165 00:06:08,000 --> 00:06:10,470 and how we not just approach it from an engineering perspective, 166 00:06:10,470 --> 00:06:11,971 but also just from a financial perspective. 167 00:06:11,971 --> 00:06:13,189 Oh yeah. Yeah, 100%. 168 00:06:13,189 --> 00:06:14,140 That makes tons of sense. 169 00:06:14,140 --> 00:06:17,677 And, you know, one of the foundations of this iteration, 170 00:06:17,994 --> 00:06:18,878 some folks have said 171 00:06:18,878 --> 00:06:22,031 this, PTC has gone through four different life cycles. 172 00:06:22,031 --> 00:06:22,482 I've heard. 173 00:06:22,482 --> 00:06:24,884 But, you know, this is much about that 174 00:06:24,884 --> 00:06:28,871 AI connectivity, you know, and also, you know, the subsea aspects of it. 175 00:06:28,905 --> 00:06:32,175 So one thing that's super cool, I believe with what you're doing, 176 00:06:32,175 --> 00:06:32,708 at MOX. 177 00:06:32,708 --> 00:06:34,927 You've got your terrestrial business 178 00:06:34,927 --> 00:06:36,496 that you've been doing for some time. 179 00:06:36,496 --> 00:06:41,134 But also really adding deep, into those subsea applications. Right? 180 00:06:41,134 --> 00:06:41,717 Yep. 181 00:06:41,717 --> 00:06:43,219 We hit a similar scale challenge 182 00:06:43,219 --> 00:06:45,588 on a handful of routes, you know, Trans Pacific, 183 00:06:45,588 --> 00:06:48,591 and even to Latin America, where leasing capacity 184 00:06:48,591 --> 00:06:50,460 wasn't going to make sense in the long term. 185 00:06:50,460 --> 00:06:52,728 So we started looking at options, you know, what was out there, 186 00:06:52,728 --> 00:06:55,681 what cables were being built, you know, could they be monetized? 187 00:06:55,681 --> 00:06:58,167 How would they be monetized? Could we be a player? 188 00:06:58,167 --> 00:07:01,170 And as we explored that, we found that we really kind of fit, 189 00:07:01,170 --> 00:07:02,271 you could call it a niche. 190 00:07:02,271 --> 00:07:04,006 But, you know, to your point earlier, it's 191 00:07:04,006 --> 00:07:05,758 maybe a little bigger than a niche. 192 00:07:05,758 --> 00:07:06,375 Super niche! 193 00:07:06,375 --> 00:07:08,177 Yeah, that we're one of the, 194 00:07:08,177 --> 00:07:11,447 if not the only one of the only North American focused 195 00:07:11,447 --> 00:07:14,784 carriers that's actually doing international subsea connectivity. 196 00:07:15,084 --> 00:07:18,121 But again, we we parlayed our, our legacy demand 197 00:07:18,387 --> 00:07:22,225 into new investment, into IRU, using full pairs on on multiple cables. 198 00:07:22,442 --> 00:07:24,143 And it's happened very quickly for us 199 00:07:24,143 --> 00:07:25,128 and it's satisfied 200 00:07:25,128 --> 00:07:28,131 our demand helped us achieve that future proofing of scale 201 00:07:28,131 --> 00:07:30,333 and really been a great business case for MOX at the same time. 202 00:07:30,333 --> 00:07:31,033 Right. 203 00:07:31,033 --> 00:07:33,536 And if you look at the customers that you've already signed up, 204 00:07:33,536 --> 00:07:34,654 the ones that you're talking 205 00:07:34,654 --> 00:07:38,624 with already, it seems to be exactly what they're asking for. 206 00:07:38,841 --> 00:07:40,726 They seem refreshed to deal with us 207 00:07:40,726 --> 00:07:42,528 is that we're not a traditional carrier, 208 00:07:42,528 --> 00:07:44,080 that we do understand these challenges. 209 00:07:44,080 --> 00:07:44,881 Bit more nimble? 210 00:07:45,031 --> 00:07:45,832 Yep. That's the 211 00:07:45,832 --> 00:07:47,483 other side of it, the flexibility 212 00:07:47,483 --> 00:07:48,868 and the understanding of their challenges 213 00:07:48,868 --> 00:07:50,386 and how to build these networks 214 00:07:50,386 --> 00:07:53,239 as opposed to the rigidity of some of the legacy. 215 00:07:53,239 --> 00:07:54,891 You know, I like operators, let's say 216 00:07:54,891 --> 00:07:56,309 that don't really see the world this way 217 00:07:56,309 --> 00:07:58,878 or have not productized things like spectrum and dark 218 00:07:58,878 --> 00:08:01,497 fiber and high capacity wavelengths the same way we have. 219 00:08:01,497 --> 00:08:04,901 from some of the, in conversation, you know, in end 220 00:08:04,901 --> 00:08:08,738 customer conversations that we've, have been part of. 221 00:08:08,738 --> 00:08:13,843 It sounds like, yes, you productize your solutions, but, 222 00:08:13,843 --> 00:08:15,995 you don't even have the rigidity of having 223 00:08:15,995 --> 00:08:19,482 to have a productized solution. That you can really customize 224 00:08:19,482 --> 00:08:23,202 a true solution based upon the needs of that particular customer data 225 00:08:23,286 --> 00:08:23,970 at that point in time. 226 00:08:23,970 --> 00:08:24,470 That's right. 227 00:08:24,470 --> 00:08:26,422 And for them, just like it is for us, 228 00:08:26,422 --> 00:08:28,224 it's also an economic consideration for them. 229 00:08:28,224 --> 00:08:30,226 And how are they investing in these networks. 230 00:08:30,226 --> 00:08:31,377 A traditional carrier 231 00:08:31,377 --> 00:08:33,563 really wants to put a lot of the equipment, 232 00:08:33,563 --> 00:08:36,082 the network build costs onto the customer to recoup that. 233 00:08:36,082 --> 00:08:38,251 And it makes sense. That's their economic model. 234 00:08:38,251 --> 00:08:40,253 But these these guys are building such large networks. 235 00:08:40,253 --> 00:08:41,971 They've outgrown the carrier, just like we outgrew 236 00:08:41,971 --> 00:08:44,273 the carrier ten years ago and did this ourselves. 237 00:08:44,273 --> 00:08:45,992 All these guys have outgrown carriers as well 238 00:08:45,992 --> 00:08:48,528 and are becoming their own providers to some extent. 239 00:08:48,528 --> 00:08:51,080 So when you watch how they're doing networks, 240 00:08:51,080 --> 00:08:51,497 we will look at 241 00:08:51,497 --> 00:08:53,966 how we're doing networks, we find those similarities. 242 00:08:53,966 --> 00:08:55,551 And then, you know, we communicate to them. 243 00:08:55,551 --> 00:08:56,819 That's that's how we do it, 244 00:08:56,819 --> 00:08:58,638 because we see that that's how you guys are doing it. 245 00:08:58,638 --> 00:09:01,507 And this is really what it requires to get to that scale. 246 00:09:01,507 --> 00:09:03,226 That's yeah, that's great expertise. 247 00:09:03,226 --> 00:09:07,513 And when we think about expertise, you know, 248 00:09:07,513 --> 00:09:11,500 a lot of the folks here, you know, have scientific backgrounds. 249 00:09:11,500 --> 00:09:13,386 They love the equations around stuff. 250 00:09:13,386 --> 00:09:14,587 Shannon's limit, whatever. 251 00:09:14,587 --> 00:09:19,125 But, you know, it seems like you all have really, 252 00:09:19,508 --> 00:09:23,262 you know, put your expertise together to understand better 253 00:09:23,529 --> 00:09:26,532 the equation around the balance of the compute. 254 00:09:27,700 --> 00:09:28,568 Yeah, yeah. 255 00:09:28,568 --> 00:09:32,321 How you need to design, GPU usage, you know, 256 00:09:32,321 --> 00:09:36,943 the pricing for that together with, you know, the capacity network 257 00:09:36,943 --> 00:09:37,843 and then really putting that 258 00:09:37,843 --> 00:09:41,080 all together to help your customers really design the optimal network 259 00:09:41,347 --> 00:09:43,065 for what they're looking to get accomplished. 260 00:09:43,065 --> 00:09:44,850 Yeah. Is that fair? That is fair. Yeah. 261 00:09:44,850 --> 00:09:46,736 And I think a lot of, you know, what we've 262 00:09:46,736 --> 00:09:49,105 what we've experienced in growing our own network 263 00:09:49,105 --> 00:09:51,107 again, raises the same types of challenges that 264 00:09:51,107 --> 00:09:53,626 that other AI firms are dealing with and what we can, 265 00:09:53,626 --> 00:09:55,278 you know, not just talk about that, 266 00:09:55,278 --> 00:09:58,281 we know how to do it, but actually show a proven record over time. 267 00:09:58,314 --> 00:09:59,815 Completing fiber builds, lighting 268 00:09:59,815 --> 00:10:02,868 systems, tying into subsea and commercializing capacity 269 00:10:02,868 --> 00:10:06,355 that frankly, not any other carrier was able to do in those situations. 270 00:10:07,940 --> 00:10:11,060 Yeah, we've you know, our capacity planning internally, I imagine 271 00:10:11,060 --> 00:10:14,063 is probably similar in some ways to a lot of the AI firms. 272 00:10:14,280 --> 00:10:16,349 And that's, that's, you know, it's not just network. 273 00:10:16,349 --> 00:10:17,867 I mean, network is a huge piece of it, 274 00:10:17,867 --> 00:10:19,785 but we're we're constantly looking at how much compute 275 00:10:19,785 --> 00:10:22,188 we're going to need, storage, where that's going to have to be. 276 00:10:22,188 --> 00:10:24,440 The network plays a really important role in that. 277 00:10:24,440 --> 00:10:28,728 But, you know, as much as Patrick is involved in, technology 278 00:10:28,728 --> 00:10:31,247 as it relates to health care and the science of health care, 279 00:10:31,247 --> 00:10:33,649 how to do personalized, evidence based medicine. 280 00:10:33,649 --> 00:10:36,636 We're researching how to, you know, 281 00:10:36,869 --> 00:10:40,756 own our own fabs, to generate our own electricity behind the meter 282 00:10:40,756 --> 00:10:43,759 with energy utility districts that we control. All of these things. 283 00:10:43,793 --> 00:10:44,960 So it's not just the network. 284 00:10:44,960 --> 00:10:47,880 It's all the way up the stack and to what it takes to deliver that. 285 00:10:47,880 --> 00:10:50,066 And yeah, the network is a vulnerability. 286 00:10:50,066 --> 00:10:52,918 There's a lot of challenges with capital and labor and equipment. 287 00:10:52,918 --> 00:10:53,152 Right. 288 00:10:53,152 --> 00:10:54,737 And getting things on time. 289 00:10:54,737 --> 00:10:57,973 But, you know, the compute side is almost more concerning 290 00:10:58,074 --> 00:10:59,091 in a lot of ways, right? 291 00:10:59,091 --> 00:11:01,827 Is that how not just compute but also equipment. 292 00:11:01,827 --> 00:11:03,946 How are we going to make all the semiconductors we need? 293 00:11:03,946 --> 00:11:05,548 That's something we think about a lot. 294 00:11:05,548 --> 00:11:08,751 And it drives our, methodology and how we're going to build networks. 295 00:11:09,118 --> 00:11:10,152 Yeah, 100%. 296 00:11:10,152 --> 00:11:14,357 No, I believe it's really that ability to put together 297 00:11:14,357 --> 00:11:17,810 the science, you know, put together the, expertise, 298 00:11:18,144 --> 00:11:21,497 you know, and then operationalize, you know, that all together 299 00:11:21,497 --> 00:11:24,500 that, you know, sets MOX, and MOX and Ciena as a partnership, 300 00:11:24,834 --> 00:11:28,003 a part, and that we're excited about the future together. 301 00:11:28,003 --> 00:11:29,388 When we think about the future. 302 00:11:29,388 --> 00:11:30,706 It is January. 303 00:11:30,706 --> 00:11:32,208 Now, we're hearing a beautiful Hawaii. 304 00:11:32,208 --> 00:11:36,545 If you had to think of one prediction, for 2026. 305 00:11:36,612 --> 00:11:37,997 What do you think? 306 00:11:37,997 --> 00:11:41,000 I, you know, I hesitate to predict because it's always 307 00:11:41,000 --> 00:11:42,017 a dangerous business. 308 00:11:42,017 --> 00:11:46,405 But if I'm going to, I'm going to try to look at, you know, tangible 309 00:11:46,405 --> 00:11:50,810 items that are driving the growth or lack thereof in certain segments. 310 00:11:51,143 --> 00:11:52,545 I think for me that really comes down 311 00:11:52,545 --> 00:11:56,215 to looking at, again, availability of, of actual equipment. 312 00:11:56,215 --> 00:11:56,482 Right. 313 00:11:56,482 --> 00:11:58,918 But all the way down to that semiconductor level. 314 00:11:58,918 --> 00:12:00,536 Like, where is this stuff going to come from? 315 00:12:00,536 --> 00:12:01,604 Who's going to make it? 316 00:12:01,604 --> 00:12:03,439 How are we going to make it? How are we going to get it to market? 317 00:12:03,439 --> 00:12:05,841 All of those things come into play in our thought process. 318 00:12:05,841 --> 00:12:08,127 So if I'm thinking about the future, I'm thinking about, 319 00:12:08,127 --> 00:12:10,463 okay, cost of capital going up, 320 00:12:10,463 --> 00:12:12,748 cost of labor going up, the shortage of labor. Right? 321 00:12:12,748 --> 00:12:14,400 These are all very concerning things. 322 00:12:14,400 --> 00:12:15,434 But at the same time, 323 00:12:15,434 --> 00:12:18,537 you see what human ingenuity has been able to do under 324 00:12:18,537 --> 00:12:19,822 all these stressful circumstances. 325 00:12:19,822 --> 00:12:21,474 And we're still able to grow this industry 326 00:12:21,474 --> 00:12:23,125 at the pace that we're growing it at. 327 00:12:23,125 --> 00:12:24,260 I think we'll figure it out. 328 00:12:24,260 --> 00:12:24,860 I think we will. 329 00:12:24,860 --> 00:12:26,662 That's really gives me confidence in the future. 330 00:12:26,662 --> 00:12:28,948 And, again, you know, thank you, to Allen. 331 00:12:28,948 --> 00:12:31,967 Thanks to all the folks, on the team at Ciena 332 00:12:31,967 --> 00:12:34,970 and at MOX, and we look forward to that future together. 333 00:12:35,087 --> 00:12:35,438 Thanks.