WEBVTT 00:00:00.000 --> 00:00:03.899 Welcome to episode 242 of the 6G podcast. It's 00:00:03.899 --> 00:00:06.400 the latest inside scoop on everything 6G, and 00:00:06.400 --> 00:00:08.919 we cover six topics in about 30 minutes, and 00:00:08.919 --> 00:00:10.380 it's brought to you by More Insights and Strategy. 00:00:10.980 --> 00:00:13.519 I'm Anshul Saag, and joining me this week is 00:00:13.519 --> 00:00:16.199 style analyst Mike Dano from Ookla, who will 00:00:16.199 --> 00:00:18.839 co -host the podcast with me from here on out. 00:00:19.359 --> 00:00:21.780 Yeah, no, Anshul, thanks so much for inviting 00:00:21.780 --> 00:00:23.559 me. I really appreciate the opportunity. I know 00:00:23.559 --> 00:00:26.379 I have big shoes to fill with Will heading out. 00:00:26.800 --> 00:00:29.120 into the sunset and wishing him the best of luck. 00:00:29.199 --> 00:00:31.899 And I'm going to do my best to, to work inside 00:00:31.899 --> 00:00:34.159 of the long shadow that he cast. And I'm happy 00:00:34.159 --> 00:00:36.380 to have this opportunity to talk to you. I listened 00:00:36.380 --> 00:00:38.820 to the podcast, a long time listener, a first 00:00:38.820 --> 00:00:41.140 time caller in this regard. And so I've listened 00:00:41.140 --> 00:00:43.000 to your pod, your, the podcast that you guys 00:00:43.000 --> 00:00:45.420 have done for a long time, and it's really useful 00:00:45.420 --> 00:00:47.399 and a lot of good stuff in there. And I'm excited 00:00:47.399 --> 00:00:49.740 to keep it going. And I think the rebrand is 00:00:49.740 --> 00:00:52.060 what we were just talking ahead of the recording 00:00:52.060 --> 00:00:54.420 of this. And I think the rebrand to the six cheek 00:00:54.420 --> 00:00:57.539 podcast is. Exactly right. It's the right time 00:00:57.539 --> 00:00:59.979 to do that. I think that was a good call on your 00:00:59.979 --> 00:01:02.500 part. Now we're really hitting, I think, the 00:01:02.500 --> 00:01:05.700 5G lull. And I think now is a good time to start 00:01:05.700 --> 00:01:08.359 talking about all things 5G, but also getting 00:01:08.359 --> 00:01:11.459 ready for whatever's next. Yeah. And honestly, 00:01:11.620 --> 00:01:16.120 this is a great segue because I feel I've been 00:01:16.120 --> 00:01:18.659 wanting to do the transition to fixture even 00:01:18.659 --> 00:01:21.900 before Will left. This was, I think, just a lot 00:01:21.900 --> 00:01:24.480 of great timing. We've got MWC coming up. We'll 00:01:24.480 --> 00:01:26.439 talk about that in another episode leading up 00:01:26.439 --> 00:01:28.799 to the show. I didn't mention this, but I will 00:01:28.799 --> 00:01:31.819 be going to Samsung Galaxy Unpacked, which is 00:01:31.819 --> 00:01:35.560 next week. That'll be a bunch of new 5G phones, 00:01:35.780 --> 00:01:38.000 but I'm sure there'll be some 5G references and 00:01:38.000 --> 00:01:40.340 we'll talk about that next week. But I wanted 00:01:40.340 --> 00:01:43.239 to throw to you in your first topic, which is 00:01:43.239 --> 00:01:46.420 related to 5G Americans. That's right. And yeah, 00:01:46.579 --> 00:01:49.859 another, it's an unfortunate situation, but certainly. 00:01:50.459 --> 00:01:53.379 a sign of the times and a sign of where we are 00:01:53.379 --> 00:01:55.939 right now. We talked about 5G kind of hitting 00:01:55.939 --> 00:01:59.120 a lull, and this is really representative of 00:01:59.120 --> 00:02:01.359 that. So if you're not familiar with it, there's 00:02:01.359 --> 00:02:04.620 a trade association called the 5G Americas, and 00:02:04.620 --> 00:02:07.140 they have been around for a long time. I still 00:02:07.140 --> 00:02:11.000 have pens and some notepads that say 3G Americas, 00:02:11.139 --> 00:02:12.759 and I think there's a lot of us out there who 00:02:12.759 --> 00:02:14.819 have saved some of that schwag over the years. 00:02:14.960 --> 00:02:17.400 But they've been a great trade association over 00:02:17.400 --> 00:02:20.080 the years. They usually have an annual event 00:02:20.080 --> 00:02:22.539 for analysts and they put out a lot of white 00:02:22.539 --> 00:02:25.759 papers, do a lot of work on the spectrum front 00:02:25.759 --> 00:02:28.879 in terms of trying to set aside spectrum for 00:02:28.879 --> 00:02:31.080 3G or 4G. And they've been around for a long 00:02:31.080 --> 00:02:33.560 time, Chris Pearson over there and the new guy 00:02:33.560 --> 00:02:35.699 who showed up a couple of years ago. But the 00:02:35.699 --> 00:02:38.560 news is that so they were 3G Americas, then they 00:02:38.560 --> 00:02:40.740 turned into 4G Americas during the 4G phase. 00:02:40.860 --> 00:02:43.039 And then, of course, they turned into 5G Americas. 00:02:43.120 --> 00:02:45.780 But the news that hit this week was that they're 00:02:45.780 --> 00:02:48.860 shutting down operations. So they will no longer 00:02:48.860 --> 00:02:51.780 be around. The beat over there wrote a real nice 00:02:51.780 --> 00:02:54.199 thing about there. It's a bittersweet moment 00:02:54.199 --> 00:02:57.340 for him. And it's just, it is unfortunate. I'm 00:02:57.340 --> 00:02:59.599 sure that we wish everybody over at 5G America 00:02:59.599 --> 00:03:01.900 is just the very best. It was a great association. 00:03:02.039 --> 00:03:04.139 They did a lot of really good stuff, put out 00:03:04.139 --> 00:03:06.580 a lot of really useful information. But I think 00:03:06.580 --> 00:03:08.719 it is, it's definitely a sign of the times in 00:03:08.719 --> 00:03:12.360 that 5G sales are slowing down. A lot of the 00:03:12.360 --> 00:03:15.840 RAN sales are flat to negative. The interest 00:03:15.840 --> 00:03:19.740 in 5G has really cooled off. It's just the cycle 00:03:19.740 --> 00:03:23.259 is ending and that era has passed, really. And 00:03:23.259 --> 00:03:26.659 so the closing of the 5G Americas Trade Group 00:03:26.659 --> 00:03:29.960 is really representative of, I think, the fact 00:03:29.960 --> 00:03:31.379 that we're, I wouldn't say we're closing the 00:03:31.379 --> 00:03:34.199 book on 5G, but it's really entering a new era 00:03:34.199 --> 00:03:36.900 where I think everybody's starting to look forward 00:03:36.900 --> 00:03:41.439 to what's next. Yeah, I think I agree in a lot 00:03:41.439 --> 00:03:44.719 of ways. I enjoyed a lot of their events. I think 00:03:44.719 --> 00:03:48.000 I joined during the 4G America days, but I do 00:03:48.000 --> 00:03:50.319 think it was really interesting to see things 00:03:50.319 --> 00:03:53.740 start to peter off with time. Every year, there 00:03:53.740 --> 00:03:58.419 were less people and less vendors. It was never 00:03:58.419 --> 00:03:59.919 a bad event because there were always really 00:03:59.919 --> 00:04:01.900 interesting people and discussions to be had. 00:04:02.099 --> 00:04:03.919 And I feel like I always learned something from 00:04:03.919 --> 00:04:06.020 attending. I think, honestly, there would have 00:04:06.020 --> 00:04:08.000 been better sales not trying to do it in October 00:04:08.000 --> 00:04:11.159 when now everybody does everything in October 00:04:11.159 --> 00:04:13.759 and it's impossible to squeeze it in. And that 00:04:13.759 --> 00:04:16.180 was honestly the main reason why I didn't attend 00:04:16.180 --> 00:04:20.120 for the last year or two. But I think it's sunsetting 00:04:20.120 --> 00:04:22.459 5G in there because it's a fit for the phase 00:04:22.459 --> 00:04:25.500 where we are today with 5G and just the fact 00:04:25.500 --> 00:04:27.819 that it's not really a focal point for anybody. 00:04:28.019 --> 00:04:30.339 Even if there is a spectrum discussion, it's 00:04:30.339 --> 00:04:33.000 a 6G spectrum discussion, not a 5G one. I feel 00:04:33.000 --> 00:04:36.180 like everyone's focus has shifted, which is kind 00:04:36.180 --> 00:04:38.000 of the reason why I think this podcast has changed. 00:04:38.199 --> 00:04:40.100 Even if Will didn't leave, we were going to make 00:04:40.100 --> 00:04:43.079 the transition to 6G because... MWC is going 00:04:43.079 --> 00:04:46.040 to be all 6G. Yeah, I brought that one up. I 00:04:46.040 --> 00:04:48.019 totally forgot about that, even though I talked 00:04:48.019 --> 00:04:50.220 about it a couple of days ago. I think that will 00:04:50.220 --> 00:04:53.220 allow me to switch to my topic, which is that 00:04:53.220 --> 00:04:56.540 Samsung and KT, which is Korea Telecom, were 00:04:56.540 --> 00:05:01.740 able to hit record 6G speeds in the 7GHz band. 00:05:02.339 --> 00:05:06.579 Obviously, there isn't a 6G network today. So 00:05:06.579 --> 00:05:10.860 this is really about using 7GHz potentially for 00:05:10.860 --> 00:05:15.069 6G. using Keysight's FIXG test equipment. This 00:05:15.069 --> 00:05:18.009 is very much a test trial kind of thing. They're 00:05:18.009 --> 00:05:23.110 using the xMIMO configuration. And I think they 00:05:23.110 --> 00:05:25.810 hit something like three gigabits per second. 00:05:25.910 --> 00:05:28.189 Yeah, they achieved a downlink of three gigabits 00:05:28.189 --> 00:05:30.589 per second. What's interesting is none of the 00:05:30.589 --> 00:05:33.069 coverage of this I could find out would tell 00:05:33.069 --> 00:05:35.790 me how much bandwidth they were using to achieve 00:05:35.790 --> 00:05:38.209 that three gigabits per second. It could be like 00:05:38.209 --> 00:05:42.800 100 megahertz or 200, I'm unclear. I would assume 00:05:42.800 --> 00:05:45.519 that most would probably be 100, just considering 00:05:45.519 --> 00:05:49.360 how spectrum is chopped up today. But it's unclear. 00:05:50.180 --> 00:05:53.319 Also, I feel like represents the state of where 00:05:53.319 --> 00:05:58.779 we are in 6G today, which is a lot of 6G is test 00:05:58.779 --> 00:06:04.420 and experimental and like making sure the idea 00:06:04.420 --> 00:06:08.040 works, but not really like commercially, remotely 00:06:08.040 --> 00:06:10.149 commercially ready. And I feel like we're getting 00:06:10.149 --> 00:06:12.949 signals on when commercialization will hit because 00:06:12.949 --> 00:06:16.110 I hear 2030, I hear 2029, I hear pre -commercial 00:06:16.110 --> 00:06:18.170 2028. I think we're going to have a lot of these 00:06:18.170 --> 00:06:20.470 discussions this year and next, especially as 00:06:20.470 --> 00:06:22.829 more trials and things happen. I'd love to hear 00:06:22.829 --> 00:06:24.670 what you think about it. For sure. No, I totally 00:06:24.670 --> 00:06:27.550 agree. And I think that I am not surprised that 00:06:27.550 --> 00:06:29.389 they're talking about 7 gigahertz. That really 00:06:29.389 --> 00:06:34.209 seems like ground zero for 6G. And we were talking 00:06:34.209 --> 00:06:36.560 about the 5G America Street Association. They 00:06:36.560 --> 00:06:39.180 really did do a good job in ushering in the era 00:06:39.180 --> 00:06:42.759 of 5G, particularly by helping to release mid 00:06:42.759 --> 00:06:45.620 -band spectrum. In the US, the 2 .5 gigahertz 00:06:45.620 --> 00:06:47.519 for T -Mobile, and then of course the C -band 00:06:47.519 --> 00:06:51.240 for AT &T and Verizon, and then 3 .45 a little 00:06:51.240 --> 00:06:54.360 bit later. But anyway, all that mid -band spectrum 00:06:54.360 --> 00:06:59.560 was really ground zero for 5G. And as 6G comes, 00:06:59.899 --> 00:07:03.879 a big focus that I have been watching over the 00:07:03.879 --> 00:07:06.699 past few years is to see... Where is 6G going 00:07:06.699 --> 00:07:09.339 to land? Because I think that certainly 6G is 00:07:09.339 --> 00:07:11.779 going to be more spectrally efficient, right? 00:07:11.839 --> 00:07:13.480 It's going to have new technologies. It's going 00:07:13.480 --> 00:07:15.579 to do this, that, and the other. My hope is that 00:07:15.579 --> 00:07:17.259 it has, I can't remember what the name of that 00:07:17.259 --> 00:07:19.779 technology is, but it's a side link. It's where 00:07:19.779 --> 00:07:22.060 the phones can talk directly to each other. They 00:07:22.060 --> 00:07:24.620 do not need to go to a cell tower in order to 00:07:24.620 --> 00:07:26.079 have a communication. So you can have a walkie 00:07:26.079 --> 00:07:28.029 talkie, basically. I think that would be super 00:07:28.029 --> 00:07:30.129 helpful at places like ski areas or on a road 00:07:30.129 --> 00:07:31.649 trip where you're going in and out of coverage. 00:07:31.730 --> 00:07:33.889 I think that'd be awesome. But that's one of 00:07:33.889 --> 00:07:36.410 the technologies that is going to be most likely 00:07:36.410 --> 00:07:39.689 included in the next set of standards. You can 00:07:39.689 --> 00:07:42.370 only get so efficient. And like my understanding 00:07:42.370 --> 00:07:46.490 of 5G was that it was about like 10, 20, 30 % 00:07:46.490 --> 00:07:49.250 more efficient than 4G, but it was really the 00:07:49.250 --> 00:07:51.110 addition of additional spectrum. That was the 00:07:51.110 --> 00:07:53.709 big deal. And so I think the same is true with 00:07:53.709 --> 00:07:55.980 6G, right? The 6G standard is probably going 00:07:55.980 --> 00:07:57.279 to be a little bit more efficient. It's going 00:07:57.279 --> 00:07:59.720 to have more bells and whistles, new things that 00:07:59.720 --> 00:08:02.779 you can do. But I think part of the objective 00:08:02.779 --> 00:08:05.899 of rolling out another generation of technology 00:08:05.899 --> 00:08:08.980 is including more spectrum. And so the fact that 00:08:08.980 --> 00:08:11.800 they have been talking about 6G, I know the Trump 00:08:11.800 --> 00:08:15.420 administration signed an executive order naming 00:08:15.420 --> 00:08:20.939 7 gigahertz as one of the specific bands that 00:08:20.939 --> 00:08:23.399 they're going to look at for 6G. So I think that's 00:08:23.399 --> 00:08:25.360 interesting to hear from the perspective of KDDI 00:08:25.360 --> 00:08:29.439 and KDDI or KT, KT, that they're also looking 00:08:29.439 --> 00:08:33.659 at 6G in that particular spectrum band. You also 00:08:33.659 --> 00:08:36.539 made me think about Sidelink. This would be the 00:08:36.539 --> 00:08:39.000 third attempt to get Sidelink to Appian because 00:08:39.000 --> 00:08:42.639 we had it in release 12 and in release 15. And 00:08:42.639 --> 00:08:44.700 that means that this might be the time for Sidelink 00:08:44.700 --> 00:08:46.919 to actually be a thing. I mean, if a carrier 00:08:46.919 --> 00:08:49.679 offered it, that's a competitive advantage. I 00:08:49.679 --> 00:08:51.259 just, I feel like that's something that somebody 00:08:51.259 --> 00:08:53.580 ought to do. Get on that. Whoever's listening 00:08:53.580 --> 00:08:55.220 out there, get on that. It's just funny because 00:08:55.220 --> 00:08:58.220 there's a lot of opportunities there, especially 00:08:58.220 --> 00:09:00.240 when you consider other types of devices other 00:09:00.240 --> 00:09:02.899 than smartphones. Because it's not that different 00:09:02.899 --> 00:09:06.559 from V2V or V2X. But what I also was thinking 00:09:06.559 --> 00:09:11.659 about was China tried to block off 6GHz or 6G, 00:09:11.740 --> 00:09:15.419 not allowing it to be used for Wi -Fi. But now 00:09:15.419 --> 00:09:18.139 it's like everyone wants to use 7GHz. So they 00:09:18.139 --> 00:09:21.120 might have not even picked the right frequency. 00:09:21.629 --> 00:09:23.190 But the one thing that I'm always concerned about 00:09:23.190 --> 00:09:26.230 with 7GHz is propagation. Because, you know, 00:09:26.269 --> 00:09:30.009 I have Wi -Fi 6, which runs in the 6GHz band. 00:09:30.470 --> 00:09:34.230 And Wi -Fi 6E or Wi -Fi 7 goes to Wi -Fi 6. But 00:09:34.230 --> 00:09:37.129 it doesn't propagate very well. You need a lot 00:09:37.129 --> 00:09:39.070 of power. And it's going to be interesting to 00:09:39.070 --> 00:09:40.509 see, because I've had these discussions with 00:09:40.509 --> 00:09:43.409 Qualcomm already about what they're doing to 00:09:43.409 --> 00:09:46.549 mitigate some of those 7GHz challenges with coverage. 00:09:46.809 --> 00:09:49.029 But the one thing I was thinking about also is... 00:09:49.549 --> 00:09:51.289 Based on the conversation I've been having with 00:09:51.289 --> 00:09:55.549 carriers, it sounds like 6G is going to be fairly 00:09:55.549 --> 00:09:58.409 limited in its deployment initially. So it might 00:09:58.409 --> 00:10:01.970 not even be an issue to have limited coverage 00:10:01.970 --> 00:10:04.049 because they're going to have limited deployments 00:10:04.049 --> 00:10:06.529 to begin with. But they will have maybe a little 00:10:06.529 --> 00:10:09.909 bit more time to figure out how to deploy 6G 00:10:09.909 --> 00:10:14.090 at 7 gigahertz. I'm assuming they're not refarming 00:10:14.090 --> 00:10:20.460 spectrum right off the bat. Few people are talking 00:10:20.460 --> 00:10:24.240 about the terahertz bands anymore for 6G. I think 00:10:24.240 --> 00:10:27.919 that is a good sign that we're not really having 00:10:27.919 --> 00:10:29.720 any serious discussions about that. Because in 00:10:29.720 --> 00:10:32.100 the early days of 6G, a couple of years ago, 00:10:32.200 --> 00:10:33.940 when people started kicking around the idea, 00:10:34.080 --> 00:10:35.840 there was some discussion about those terahertz 00:10:35.840 --> 00:10:38.879 bands, which are super high. Propagation is just 00:10:38.879 --> 00:10:41.759 awful. And I'm sure that somebody can get it 00:10:41.759 --> 00:10:44.379 work, but get a commercial network up and running. 00:10:44.460 --> 00:10:46.600 But to what purpose when your transmissions only 00:10:46.600 --> 00:10:50.019 stretch a couple hundred feet? And your bandwidth 00:10:50.019 --> 00:10:52.779 is affected by humidity. Yeah, exactly. I just, 00:10:52.860 --> 00:10:55.740 I can't imagine. I can't see that working very 00:10:55.740 --> 00:10:58.740 well. At least seven gigahertz is presumably 00:10:58.740 --> 00:11:00.679 the propagation would be a little bit better 00:11:00.679 --> 00:11:02.600 in seven gigahertz than the terahertz bands. 00:11:03.179 --> 00:11:06.659 Definitely. I think next topic should be yours. 00:11:06.720 --> 00:11:09.519 Yes. Moving along. We've got a couple of good 00:11:09.519 --> 00:11:12.679 topics to cover. That's right. Now I have a little 00:11:12.679 --> 00:11:14.940 bit of shameless self -promotion to go over. 00:11:15.019 --> 00:11:17.539 As you may know, I work at Ookla. For the past 00:11:17.539 --> 00:11:20.860 25 years, I've been a journalist, worked at places 00:11:20.860 --> 00:11:23.659 like Light Reading, Fierce Wireless, covering 00:11:23.659 --> 00:11:26.580 the industry. And last year, I decided to take 00:11:26.580 --> 00:11:31.740 the exciting step into the analyst game and took 00:11:31.740 --> 00:11:34.500 a job with Ookla. And if you're familiar with 00:11:34.500 --> 00:11:36.039 Ookla, Ookla is the company behind the speed 00:11:36.039 --> 00:11:38.519 test app. And they also have a couple of other 00:11:38.519 --> 00:11:40.720 lines of business. They've got RootMetrics, which 00:11:40.720 --> 00:11:43.940 does drive testing, which is Wi -Fi installation 00:11:43.940 --> 00:11:46.779 and Wi -Fi monitoring. And then also DowDetector, 00:11:46.879 --> 00:11:49.940 that's a global outage tracking service. And 00:11:49.940 --> 00:11:53.559 so it's been a very interesting year in terms 00:11:53.559 --> 00:11:57.200 of learning all of the data swimming through 00:11:57.200 --> 00:11:59.539 the Ookla data lake. There's all kinds of sources 00:11:59.539 --> 00:12:01.500 of data, all sorts of interesting things to look 00:12:01.500 --> 00:12:04.679 at. And along with that, my colleague, Luke. 00:12:05.049 --> 00:12:07.330 who is probably one of the smartest guys I think 00:12:07.330 --> 00:12:10.029 I've ever met. But he just put out a report about, 00:12:10.169 --> 00:12:13.429 he's citing Ookla speed test data, and the report 00:12:13.429 --> 00:12:18.909 was about 5G SA, standalone. And it's pretty 00:12:18.909 --> 00:12:20.970 interesting. There's a lot to go through in the 00:12:20.970 --> 00:12:23.409 report. It's about all the operators that are 00:12:23.409 --> 00:12:26.289 launching standalone 5G, what kind of speeds 00:12:26.289 --> 00:12:28.429 they're seeing, what the implementations look 00:12:28.429 --> 00:12:30.610 like, stuff about the core network. This report 00:12:30.610 --> 00:12:32.690 just goes on and on. But the thing that I think 00:12:32.690 --> 00:12:36.289 is... pretty interesting and something that I 00:12:36.289 --> 00:12:39.929 didn't really know about. Like I had heard rumors 00:12:39.929 --> 00:12:42.090 about this, but what's nice about this report 00:12:42.090 --> 00:12:43.950 is that it really kind of looks at this specifically, 00:12:44.110 --> 00:12:49.070 which is how 5G standalone affects battery life. 00:12:49.529 --> 00:12:53.389 And apparently it affects it pretty well. Luke's 00:12:53.389 --> 00:12:55.509 findings was specific. So I'm going to read this 00:12:55.509 --> 00:12:58.250 from the report. Specifically, EE in the UK, 00:12:58.389 --> 00:13:02.370 they have a standalone network and the data shows 00:13:02.370 --> 00:13:07.840 that The device that's on 5G SA, its battery 00:13:07.840 --> 00:13:11.740 lasts 22 % longer than if it were on non -standalone 00:13:11.740 --> 00:13:17.179 NSA. And same with O2 in the UK, 11 % better 00:13:17.179 --> 00:13:21.399 battery life on 5G SA compared to the 5G NSA, 00:13:21.740 --> 00:13:24.440 the first version of 5G, 5G non -standalone NSA. 00:13:24.860 --> 00:13:27.559 So I just think. That's pretty interesting, right? 00:13:27.679 --> 00:13:29.919 And again, I feel like here's yet another missed 00:13:29.919 --> 00:13:32.100 opportunity for operators out there is, dude, 00:13:32.200 --> 00:13:34.500 launch Sidelink. Hey, we've got walkie -talkie 00:13:34.500 --> 00:13:36.820 phones for you. Dude, talk about 5G standalone. 00:13:37.279 --> 00:13:40.080 Your battery life is literally better. These 00:13:40.080 --> 00:13:42.919 are things, I feel like the things that normal 00:13:42.919 --> 00:13:45.080 people really do care about, better battery life. 00:13:45.179 --> 00:13:46.919 Like that is something that would get me to change 00:13:46.919 --> 00:13:48.980 carriers, depending on the phone that I had. 00:13:49.080 --> 00:13:50.799 But it would certainly be a selling point, right? 00:13:51.620 --> 00:13:54.769 Yeah. What's funny is I have never... Been able 00:13:54.769 --> 00:13:57.029 to discuss that as a selling point for standalone 00:13:57.029 --> 00:13:59.269 when talking to people about different carriers' 00:13:59.490 --> 00:14:02.330 experiences. But it's a no -brainer because you're 00:14:02.330 --> 00:14:05.889 only connected to one network at a time. You 00:14:05.889 --> 00:14:08.809 don't have to connect to 4G and 5G. Who does 00:14:08.809 --> 00:14:12.309 think it would save you power? You're going to 00:14:12.309 --> 00:14:15.389 also benefit from lower latency and higher speeds. 00:14:15.990 --> 00:14:19.070 So those are all better things. And I do think 00:14:19.070 --> 00:14:22.009 some of them, I would say, are also coming from 00:14:22.009 --> 00:14:24.090 higher speeds. which then allow you to get off 00:14:24.090 --> 00:14:25.850 the network faster, right? You pull you down 00:14:25.850 --> 00:14:28.950 your data and you turn off the modem. Like, I've 00:14:28.950 --> 00:14:32.210 been following 5G modems and 4G modems for a 00:14:32.210 --> 00:14:34.850 very long time. And the thing they do is, the 00:14:34.850 --> 00:14:36.429 thing they do better is they just download things 00:14:36.429 --> 00:14:39.370 faster so they're on for less time. That's literally 00:14:39.370 --> 00:14:41.750 the power saving for most of what they do. Not 00:14:41.750 --> 00:14:43.450 a shock to me that if you have a more efficient 00:14:43.450 --> 00:14:45.750 network, you also have a more efficient device. 00:14:46.149 --> 00:14:48.929 I mean, it's always great to have data to support 00:14:48.929 --> 00:14:51.610 that thesis. And yeah, to your point, it should. 00:14:51.960 --> 00:14:54.580 push carriers to move faster, especially when 00:14:54.580 --> 00:14:57.639 you consider that it benefits to have an NSA 00:14:57.639 --> 00:15:00.559 network, like being able to support stuff like 00:15:00.559 --> 00:15:04.279 REDCap, which is also a great power saving opportunity 00:15:04.279 --> 00:15:07.860 in 5G, which still gives you low latency. And 00:15:07.860 --> 00:15:10.840 hey, guess what? All the new Apple Watches all 00:15:10.840 --> 00:15:13.600 have 5G REDCap. So if you want to be able to 00:15:13.600 --> 00:15:16.779 use a new Apple Watch to the best of its ability 00:15:16.779 --> 00:15:19.159 with the best battery life, you want to have 00:15:19.159 --> 00:15:21.320 a carrier that supports standard. Yeah. It's 00:15:21.320 --> 00:15:24.220 really interesting. It's always fun to have supportive 00:15:24.220 --> 00:15:27.659 data for a theory. Yeah. Interesting stuff. But 00:15:27.659 --> 00:15:29.580 yeah, you've got, and now I'm going to toss it 00:15:29.580 --> 00:15:32.159 back to you because on our list of stuff that 00:15:32.159 --> 00:15:34.399 we're going to talk about, you have, I think 00:15:34.399 --> 00:15:36.639 you actually win the award for the most exciting 00:15:36.639 --> 00:15:40.299 story topic or topic to talk about kinetic tokens. 00:15:40.399 --> 00:15:42.100 I don't even know what that is. This is the one 00:15:42.100 --> 00:15:44.360 that I was like, I'm going to listen very closely 00:15:44.360 --> 00:15:46.519 to see what you have to say about this. Yes. 00:15:46.600 --> 00:15:50.220 So this was a terminology. I believe, I looked 00:15:50.220 --> 00:15:53.080 this up, has been coined by T -Mobile's John 00:15:53.080 --> 00:15:57.000 Saw, who is formerly of Sprint and formerly of 00:15:57.000 --> 00:15:59.539 Clearwire. So he has been in the industry for 00:15:59.539 --> 00:16:02.480 a very long time and has been a very forward 00:16:02.480 --> 00:16:05.059 thinker in a lot of ways. So T -Mobile is very 00:16:05.059 --> 00:16:07.559 lucky to have acquired him with Sprint. And I 00:16:07.559 --> 00:16:08.980 worked with him when he was at Sprint as well. 00:16:09.100 --> 00:16:12.000 And he talked about kinetic tokens and the idea 00:16:12.000 --> 00:16:17.100 of physical AI and 6G and the idea that when 00:16:17.100 --> 00:16:20.220 you're doing AI. on a device that's in motion. 00:16:20.460 --> 00:16:23.639 You're going to want to have on -device AI, but 00:16:23.639 --> 00:16:25.299 you're also going to want to be able to service 00:16:25.299 --> 00:16:28.679 that device with AI from the edge and from the 00:16:28.679 --> 00:16:32.159 cloud. And it's like a continuum of compute that 00:16:32.159 --> 00:16:35.100 occurs from device all the way up to the cloud. 00:16:35.320 --> 00:16:37.080 And you're going to have different types of models 00:16:37.080 --> 00:16:39.559 and you're going to have different types of tokens. 00:16:39.740 --> 00:16:42.799 And the important part here to understand is 00:16:42.799 --> 00:16:45.860 that they're going to need to be very low latency 00:16:45.860 --> 00:16:49.279 tokens. as opposed to currently, which is a lot 00:16:49.279 --> 00:16:54.000 of the KPIs for AI tokens is mostly the tokens 00:16:54.000 --> 00:16:55.679 per second, right? Because you're getting this 00:16:55.679 --> 00:16:59.460 huge document spit out or inch of words or images, 00:16:59.580 --> 00:17:02.620 and it's all tokens per second. But the truth 00:17:02.620 --> 00:17:04.940 is with something like physical AI, you need 00:17:04.940 --> 00:17:07.559 the tokens to be very quick because you don't 00:17:07.559 --> 00:17:10.759 have the luxury of waiting for the tokens to 00:17:10.759 --> 00:17:12.500 arrive as you're driving down the road, right? 00:17:12.559 --> 00:17:14.599 That's just not a thing. Or if a drone's flying 00:17:14.599 --> 00:17:18.569 in the sky. These really low latency tokens are 00:17:18.569 --> 00:17:22.750 just kinetic tokens that serve the device appropriately 00:17:22.750 --> 00:17:26.650 between the application and the device and enable 00:17:26.650 --> 00:17:30.250 better use cases and better experience for the 00:17:30.250 --> 00:17:32.630 user or the device, depending on how it's being 00:17:32.630 --> 00:17:35.710 operated. And that these tokens will be generated, 00:17:36.230 --> 00:17:38.930 whether they're on device or at the edge for 00:17:38.930 --> 00:17:42.730 6G applications. And also talking about like, 00:17:42.750 --> 00:17:45.829 are they already on the road to this? with what 00:17:45.829 --> 00:17:47.789 they're doing with live translation in the network. 00:17:47.930 --> 00:17:51.150 With their 5G advanced networks where they're 00:17:51.150 --> 00:17:53.869 doing real -time translation for people and it's 00:17:53.869 --> 00:17:57.470 device -agnostic, only one person has to be a 00:17:57.470 --> 00:18:00.289 T -Mobile customer and it works with 15 languages. 00:18:00.549 --> 00:18:02.769 I think real -time translation is a very low 00:18:02.769 --> 00:18:05.849 -hanging fruit, but it's an interesting application 00:18:05.849 --> 00:18:08.150 nonetheless. We also talked a little bit about 00:18:08.150 --> 00:18:10.549 some 6G applications that are already in the 00:18:10.549 --> 00:18:13.309 standard or are almost certain to be in the standard, 00:18:13.450 --> 00:18:17.240 like ISAC. which is integrated sensing and communications. 00:18:17.700 --> 00:18:20.319 And there's just so many opportunities with that 00:18:20.319 --> 00:18:23.000 alone paired with something like kinetic tokens, 00:18:23.180 --> 00:18:25.640 where let's say you have a drone in the sky. 00:18:25.880 --> 00:18:28.500 The network can actually track the location, 00:18:28.539 --> 00:18:30.960 position, and speed of that drone and identify 00:18:30.960 --> 00:18:34.599 it as a drone and also send it packets or tokens 00:18:34.599 --> 00:18:37.220 as it's flying to make sure that it's getting 00:18:37.220 --> 00:18:39.660 the right data to perform its mission more effectively. 00:18:40.250 --> 00:18:42.309 So that way it doesn't need to have too much 00:18:42.309 --> 00:18:44.549 compute on device because we're always trying 00:18:44.549 --> 00:18:47.589 to reduce the weight and cost of these kinds 00:18:47.589 --> 00:18:49.890 of devices. And one of the few ways to do that 00:18:49.890 --> 00:18:53.430 is by removing compute as much as possible. And 00:18:53.430 --> 00:18:55.430 especially when you consider the memory crunch 00:18:55.430 --> 00:18:57.049 that we're going to be going through for the 00:18:57.049 --> 00:18:59.569 next year or two, then it could be an opportunity 00:18:59.569 --> 00:19:02.069 to do too much on device AI because you don't 00:19:02.069 --> 00:19:03.809 have the memory to enable that. Lots of cool 00:19:03.809 --> 00:19:08.049 ideas. Also some talk. How is 6G and AI going 00:19:08.049 --> 00:19:10.210 to work? Well, a lot of what they're thinking 00:19:10.210 --> 00:19:13.369 about is they're going to run the RAM with the 00:19:13.369 --> 00:19:16.970 6G RAM with NVIDIA and their hardware, which 00:19:16.970 --> 00:19:19.910 will have some added capacity for compute, for 00:19:19.910 --> 00:19:22.130 AI compute on top of that. And they'll be able 00:19:22.130 --> 00:19:24.930 to coexist in the same space at the same power 00:19:24.930 --> 00:19:29.390 levels as they were using before. I, so I agree. 00:19:29.490 --> 00:19:32.769 John saw is a super smart guy. And so I listened 00:19:32.769 --> 00:19:34.829 to everything he says very carefully. So if he's 00:19:34.829 --> 00:19:36.210 talking about this kind of stuff, I'm still, 00:19:36.289 --> 00:19:41.450 I am not sure that I am sold on the T -Mobile 00:19:41.450 --> 00:19:46.289 AI ran NVIDIA chips in the base station. This 00:19:46.289 --> 00:19:49.049 is what 6G is going to be. I know that's what 00:19:49.049 --> 00:19:52.289 they're pushing toward. And I think it's smart, 00:19:52.509 --> 00:19:54.970 right? This is the time to talk about it. Everyone's 00:19:54.970 --> 00:19:57.529 talking about NVIDIA GPUs and what AI can do. 00:19:57.769 --> 00:20:03.170 I'm still struggling to understand the reason 00:20:03.170 --> 00:20:07.049 why T -Mobile thinks that this is going to work. 00:20:07.190 --> 00:20:09.990 I get that you want to have processing closer 00:20:09.990 --> 00:20:15.750 to the users, but the idea of having an NVIDIA... 00:20:16.200 --> 00:20:19.059 A bunch of GPUs inside of every single base station, 00:20:19.220 --> 00:20:21.720 like that is an extremely high cost and it'll 00:20:21.720 --> 00:20:23.400 still be an extremely high cost five years from 00:20:23.400 --> 00:20:27.319 now. And so how to recoup that cost when you 00:20:27.319 --> 00:20:29.420 sell AI processing inside the base stations, 00:20:29.440 --> 00:20:30.859 are you going to really need to do that? And 00:20:30.859 --> 00:20:34.779 is there not anywhere else where you could potentially 00:20:34.779 --> 00:20:37.720 buy that kind of computing capacity? I think 00:20:37.720 --> 00:20:40.579 there are instances where it would make sense, 00:20:40.579 --> 00:20:43.599 but. This sort of broad canvas that T -Mobile 00:20:43.599 --> 00:20:45.839 is painting about, all the base stations are 00:20:45.839 --> 00:20:47.420 going to have NVIDIA in them, and we're going 00:20:47.420 --> 00:20:49.819 to be in the token, the AI token business. I 00:20:49.819 --> 00:20:52.440 wouldn't say I'm sold on that just quite yet. 00:20:52.579 --> 00:20:54.920 I need to see a little bit more. I think you're 00:20:54.920 --> 00:20:59.480 right. I had Vue maybe a year or two ago, and 00:20:59.480 --> 00:21:02.779 I still am in a lot of ways. I think what has 00:21:02.779 --> 00:21:05.519 convinced me about the strategy that will work, 00:21:05.680 --> 00:21:12.309 I was a very big critic. Not AI, XR and 5G, because 00:21:12.309 --> 00:21:14.750 you needed to have compute close to the edge. 00:21:14.869 --> 00:21:19.170 And there was no model to enable those GPUs to 00:21:19.170 --> 00:21:21.569 get close enough to the edge to make thin and 00:21:21.569 --> 00:21:26.089 light 5G glasses possible. But if the compute 00:21:26.089 --> 00:21:29.230 is there because of AI, then you can double dip 00:21:29.230 --> 00:21:32.809 on AI compute for AI workloads, but also for 00:21:32.809 --> 00:21:34.349 streaming graphics workloads, because they're 00:21:34.349 --> 00:21:39.859 still GPUs. Maybe there's an opportunity to have 00:21:39.859 --> 00:21:44.119 multiple types of industries take advantage of 00:21:44.119 --> 00:21:47.900 that compute. And I see an opportunity to help 00:21:47.900 --> 00:21:50.180 also get that GPU compute for other applications 00:21:50.180 --> 00:21:53.480 other than AI. But it's like the justification 00:21:53.480 --> 00:21:57.299 for it. But to your point, we still have to see 00:21:57.299 --> 00:22:00.059 these things actually get deployed and used and 00:22:00.059 --> 00:22:03.980 be commercially viable. I think what has maybe 00:22:03.980 --> 00:22:07.000 given me more confidence. is that Ericsson and 00:22:07.000 --> 00:22:09.680 Nokia are taking this as well. It's not just 00:22:09.680 --> 00:22:12.279 NVIDIA trying to push it themselves with T -Mobile. 00:22:12.480 --> 00:22:16.519 But to your point, we're still a few years out 00:22:16.519 --> 00:22:19.940 from having commercial deployments and the performance 00:22:19.940 --> 00:22:22.279 of these platforms will have to improve as well. 00:22:22.519 --> 00:22:25.000 And they'll probably have new GPUs. That might 00:22:25.000 --> 00:22:27.299 need to be more power hungry. Who knows? But 00:22:27.299 --> 00:22:29.599 power was the point of discussion that I had 00:22:29.599 --> 00:22:33.900 with John because power is a problem and he didn't 00:22:33.900 --> 00:22:37.059 seem concerned. which I hope is a good thing. 00:22:37.359 --> 00:22:40.880 But yeah, there's a lot of, we've talked about 00:22:40.880 --> 00:22:45.380 all kinds of open RAN and AI RAN kind of opportunities. 00:22:45.519 --> 00:22:47.539 And the truth is like, they're not as big as 00:22:47.539 --> 00:22:49.740 they used to be, but I will also say that there 00:22:49.740 --> 00:22:54.259 have been some improvements and I feel like we're 00:22:54.259 --> 00:22:56.940 closer today than we were five years ago. Yeah. 00:22:57.059 --> 00:23:01.509 Yeah. And I will say I even. Before John saw 00:23:01.509 --> 00:23:04.130 Neville Ray talked about, he thinks that wearables 00:23:04.130 --> 00:23:08.109 are going to be a big, it'll be big at some point, 00:23:08.130 --> 00:23:10.329 wearables and augmented reality glasses where 00:23:10.329 --> 00:23:12.230 you can, you move your head and use stuff moves 00:23:12.230 --> 00:23:15.089 around as you're looking around. Like I do. And 00:23:15.089 --> 00:23:18.430 even we see the meta Ray band glasses, which 00:23:18.430 --> 00:23:21.230 I have not yet played with, but is, oh, there 00:23:21.230 --> 00:23:23.670 we go. You've got some, I'm not surprised at 00:23:23.670 --> 00:23:26.009 all, but I've it's we're getting down that road. 00:23:26.109 --> 00:23:28.569 Like that kind of stuff is it used to be on PowerPoint 00:23:28.569 --> 00:23:31.920 and Apple. And now you're putting it on. You're 00:23:31.920 --> 00:23:33.460 literally putting those glasses on right now. 00:23:33.480 --> 00:23:36.059 It's hysterical. I know. I feel my glasses are 00:23:36.059 --> 00:23:39.559 not nearly as cool. Not as thick. Yeah, not as 00:23:39.559 --> 00:23:43.099 thick. That's for sure. Yeah. Yeah. I would also 00:23:43.099 --> 00:23:46.079 say you make a really good point. I actually 00:23:46.079 --> 00:23:48.940 think AI wearables might be it. Think about all 00:23:48.940 --> 00:23:51.140 the companies that want to bring some kind of 00:23:51.140 --> 00:23:54.640 AI wearable to market. Whether it's OpenAI and 00:23:54.640 --> 00:23:58.299 throughout the internet or Apple. I think that 00:23:58.299 --> 00:23:59.839 might even be one of the opportunities where 00:23:59.839 --> 00:24:04.759 they run 5G or 6G REDCAT modem in the device. 00:24:04.819 --> 00:24:06.700 So it's completely independent of your smartphone 00:24:06.700 --> 00:24:09.380 and it can do all the AI stuff it needs to do 00:24:09.380 --> 00:24:11.880 with the help of the cloud. So I don't know. 00:24:11.940 --> 00:24:14.000 But I think that there might be some opportunities 00:24:14.000 --> 00:24:17.799 with 6G and AI. And the timing seemed to be really 00:24:17.799 --> 00:24:19.799 good for T -Mobile. John was saying how we're 00:24:19.799 --> 00:24:22.920 early to 5G. So their hardware refresh for 6G 00:24:22.920 --> 00:24:25.160 might actually be perfectly timed for when they 00:24:25.160 --> 00:24:29.299 need to make that change. Yeah, I agree. Yeah, 00:24:29.359 --> 00:24:32.019 for sure. Yeah, do we have time for this last? 00:24:32.279 --> 00:24:34.359 I've got one more topic, but it's a bit heavy. 00:24:35.400 --> 00:24:37.259 Let's do it. All right, let's do it, because 00:24:37.259 --> 00:24:39.680 it is related to what we've been talking about. 00:24:39.759 --> 00:24:42.460 So I've got one more thing on my list to talk 00:24:42.460 --> 00:24:44.079 about, but it's definitely related to all this, 00:24:44.180 --> 00:24:46.160 which is just the data center boom that's going 00:24:46.160 --> 00:24:48.359 on this year. And it's related to what we're 00:24:48.359 --> 00:24:50.720 talking about, because T -Mobile is talking about 00:24:50.720 --> 00:24:55.500 putting GPUs inside of its base stations. and 00:24:55.500 --> 00:24:59.180 running AI applications that way. And I can see 00:24:59.180 --> 00:25:02.000 why T -Mobile is interested in that because, 00:25:02.180 --> 00:25:04.339 and here's the newest numbers that are fresh 00:25:04.339 --> 00:25:07.359 as of about this week. The new number for the 00:25:07.359 --> 00:25:10.119 amount that the top hyperscalers in the world 00:25:10.119 --> 00:25:13.640 are going to invest into the creation of additional 00:25:13.640 --> 00:25:21.420 data centers for AI is $710 billion in 2026 alone. 00:25:21.539 --> 00:25:23.819 That is how much they're spending. in order to 00:25:23.819 --> 00:25:26.039 build additional data centers in the U .S. and 00:25:26.039 --> 00:25:29.700 globally just this year, which is, I realize 00:25:29.700 --> 00:25:33.200 we're at the peak of the bubble of AI. I don't 00:25:33.200 --> 00:25:35.059 know if it's going to continue or if it's going 00:25:35.059 --> 00:25:37.660 to pop or what, who knows. But certainly the 00:25:37.660 --> 00:25:39.640 idea that these companies are investing that 00:25:39.640 --> 00:25:41.640 much money in the creation of additional data 00:25:41.640 --> 00:25:44.799 centers is just mind -boggling. $710 billion 00:25:44.799 --> 00:25:49.079 is a staggering figure, I think. But here's the 00:25:49.079 --> 00:25:50.579 thing that I've been trying to look at is, what 00:25:50.579 --> 00:25:52.500 does that mean? Where's all that money going? 00:25:52.619 --> 00:25:54.460 And what I think is interesting is that it is 00:25:54.460 --> 00:25:56.259 going into additional data centers, but many 00:25:56.259 --> 00:25:58.279 of them are the sort of big, giant, sprawling 00:25:58.279 --> 00:26:00.880 data centers that sit out in the wilds of Montana 00:26:00.880 --> 00:26:04.579 and are not at all the idea of an edge computing 00:26:04.579 --> 00:26:06.799 location like at the base of a cell tower. These 00:26:06.799 --> 00:26:10.359 are gigantic, factory -sized warehouses that 00:26:10.359 --> 00:26:14.440 are just literally full of AMD chips or NVIDIA 00:26:14.440 --> 00:26:17.380 chips or whatever. And what I think, I'm trying 00:26:17.380 --> 00:26:20.160 to understand the evolution of the data center. 00:26:20.700 --> 00:26:22.940 investment cycle from the perspective of what 00:26:22.940 --> 00:26:25.160 does it mean for telecom. And there's a couple 00:26:25.160 --> 00:26:27.359 of interesting things. One thing is they keep 00:26:27.359 --> 00:26:29.519 struggling to get additional power to all these 00:26:29.519 --> 00:26:33.140 data centers, require a ton of electricity in 00:26:33.140 --> 00:26:36.460 order to run. And it's putting big strains on 00:26:36.460 --> 00:26:39.000 the power grid. And so according to this one 00:26:39.000 --> 00:26:41.539 report from JLL that came out this week, natural 00:26:41.539 --> 00:26:45.279 gas is still the dominant approach in order of 00:26:45.279 --> 00:26:47.039 powering these data centers. They're using a 00:26:47.039 --> 00:26:49.259 lot of natural gas. The other big power generation. 00:26:49.900 --> 00:26:51.839 method for these data centers is gas turbines. 00:26:52.019 --> 00:26:53.740 And I've seen they've, they're literally buying 00:26:53.740 --> 00:26:57.539 jet engines. Like companies are selling jet engines 00:26:57.539 --> 00:26:59.700 to data centers in order to power the data center, 00:26:59.799 --> 00:27:01.900 which just tells you how, like what an insane 00:27:01.900 --> 00:27:03.599 world we live in. Like the future is just crazy. 00:27:03.819 --> 00:27:05.519 But again, I'm trying to understand like how 00:27:05.519 --> 00:27:07.779 these data centers, what does it mean for telecom? 00:27:07.960 --> 00:27:09.940 What I thought was two things that I think are 00:27:09.940 --> 00:27:12.500 interesting. One is that a company called Black 00:27:12.500 --> 00:27:15.839 Blaze, Back Blaze, put out a new report this 00:27:15.839 --> 00:27:18.900 week talking about traffic flows over networks. 00:27:19.480 --> 00:27:22.579 And the sort of the summary of that report is 00:27:22.579 --> 00:27:26.299 that it's really the traffic flows are huge. 00:27:26.539 --> 00:27:29.039 There are large data sets consisting of images 00:27:29.039 --> 00:27:31.480 and videos and metadata, and that's all being 00:27:31.480 --> 00:27:35.400 used to train AI models and stuff. But those 00:27:35.400 --> 00:27:39.380 data sets are traveling over super high bandwidth 00:27:39.380 --> 00:27:41.279 private connections. They're not going over the 00:27:41.279 --> 00:27:43.200 regular Internet, basically. And so that's why 00:27:43.200 --> 00:27:45.579 you're seeing companies like Lumen sign all these 00:27:45.579 --> 00:27:48.710 agreements in order. to do AI transport. It's 00:27:48.710 --> 00:27:50.470 specific. It's basically data center to data 00:27:50.470 --> 00:27:52.670 center. Like those are the transport type things 00:27:52.670 --> 00:27:55.109 that we're seeing. And that's where, from what 00:27:55.109 --> 00:27:57.690 I understand, the bulk of AI traffic today is 00:27:57.690 --> 00:28:00.829 really that. It's huge data sets traveling from 00:28:00.829 --> 00:28:03.769 one data center to another. And that's the impact 00:28:03.769 --> 00:28:08.210 to the telecom industry today. But then also 00:28:08.210 --> 00:28:11.970 companies like Akamai and Backblaze and stuff, 00:28:12.170 --> 00:28:15.029 there are sort of these neo -clouds. And so they're 00:28:15.029 --> 00:28:20.279 setting up GPU rental sites in order to sell 00:28:20.279 --> 00:28:24.299 GPU by the hour, basically. So if you are a company 00:28:24.299 --> 00:28:26.740 that wants to experiment into AI, you can rent 00:28:26.740 --> 00:28:29.779 a GPU by the hour and you can do it from companies 00:28:29.779 --> 00:28:32.640 like Akamai. But again, here's, and it's relevant 00:28:32.640 --> 00:28:34.839 to T -Mobile is that, so Akamai announced this, 00:28:34.880 --> 00:28:36.259 they're doing really well. They're totally sold 00:28:36.259 --> 00:28:38.819 out of the 20 cities where they built these GPU 00:28:38.819 --> 00:28:41.930 farms, but Akamai stock fell. Because people 00:28:41.930 --> 00:28:44.750 are basically Wall Street is unsure as to whether 00:28:44.750 --> 00:28:46.769 or not that is a sustainable business model. 00:28:47.049 --> 00:28:49.109 And they're worried that the investment into 00:28:49.109 --> 00:28:51.869 some of that sort of AI bubble stuff is just 00:28:51.869 --> 00:28:53.910 too much. And so I'm thinking from the perspective 00:28:53.910 --> 00:28:55.829 of a T -Mobile or a telecom company that wants 00:28:55.829 --> 00:28:58.750 to put GPUs in the base station. I don't know, 00:28:58.809 --> 00:29:01.069 like even Akamai is not being rewarded for the 00:29:01.069 --> 00:29:04.230 work that is doing in building these AI focused 00:29:04.230 --> 00:29:08.140 neoclouds with full of GPUs and stuff. I think 00:29:08.140 --> 00:29:10.180 the whole data set, it's obvious there's a ton 00:29:10.180 --> 00:29:11.940 of investment going on at the data centers, but 00:29:11.940 --> 00:29:14.579 I'm just, I'm still unconvinced that it will 00:29:14.579 --> 00:29:17.160 trickle down into the telecom companies that 00:29:17.160 --> 00:29:19.660 like you and I have been talking to. I feel like 00:29:19.660 --> 00:29:22.700 this is another area where companies like Google 00:29:22.700 --> 00:29:26.480 and Meta are going to really capture the lion's 00:29:26.480 --> 00:29:28.559 share of things. At least right now. I don't 00:29:28.559 --> 00:29:30.599 know. Like I could be proven wrong. We'll have 00:29:30.599 --> 00:29:32.200 to see. And I also, here's the last thing I'll 00:29:32.200 --> 00:29:34.970 say is I think it will matter. where this is 00:29:34.970 --> 00:29:36.490 going to go down. So in the US, there's like 00:29:36.490 --> 00:29:39.130 already 4 ,000 data centers all over the country. 00:29:39.250 --> 00:29:42.009 Like, I feel like that's a lot. Other countries, 00:29:42.089 --> 00:29:45.710 far fewer. And so in those cases, if you're a 00:29:45.710 --> 00:29:47.829 much smaller company, or sorry, a much smaller 00:29:47.829 --> 00:29:50.269 country with very few data center operations, 00:29:50.589 --> 00:29:53.029 there, a telecom company could build data centers 00:29:53.029 --> 00:29:55.529 for AI. And that would make sense in terms of 00:29:55.529 --> 00:29:57.869 data sovereignty and all that stuff. It's keeping 00:29:57.869 --> 00:30:00.470 data inside of that country within the telecom 00:30:00.470 --> 00:30:03.160 operators network. And we're seeing some telecom 00:30:03.160 --> 00:30:05.200 operators doing that, building their own data 00:30:05.200 --> 00:30:08.180 centers for sovereign AI applications. That makes 00:30:08.180 --> 00:30:10.440 sense to me. But for a company like T -Mobile 00:30:10.440 --> 00:30:13.039 in the U .S. that already has so many data centers, 00:30:13.059 --> 00:30:18.140 I don't know. We'll see. I'm not convinced. I 00:30:18.140 --> 00:30:20.779 think you make a good point on the countries 00:30:20.779 --> 00:30:23.619 that don't have as many data centers as the U 00:30:23.619 --> 00:30:26.619 .S. I also think there's some challenges every 00:30:26.619 --> 00:30:30.680 time the carriers have tried to be more than 00:30:30.680 --> 00:30:33.170 a big dump hike. They haven't necessarily been 00:30:33.170 --> 00:30:36.829 very successful, but I also think if they establish 00:30:36.829 --> 00:30:39.410 the right partnerships, they could be partnered 00:30:39.410 --> 00:30:42.869 with the Googles and Amazons and crescents of 00:30:42.869 --> 00:30:46.109 the world to enable those clouds to be more close 00:30:46.109 --> 00:30:49.250 to the user. So we'll see. But yeah, the spend 00:30:49.250 --> 00:30:53.490 is absurd. The power demands building nukes as 00:30:53.490 --> 00:30:56.910 fast as possible. I think one of the few things 00:30:56.910 --> 00:30:59.670 that I see in terms of the benefits of this potential. 00:31:00.659 --> 00:31:03.079 infrastructure buildup. Maybe we end up with 00:31:03.079 --> 00:31:05.880 a bunch of new nuclear reactors that give us 00:31:05.880 --> 00:31:08.880 a lot of clean power. That would be a good, I'd 00:31:08.880 --> 00:31:11.039 appreciate that. It's cheaper power, like I'm 00:31:11.039 --> 00:31:13.720 down. Yeah, I'm just thinking about even if this 00:31:13.720 --> 00:31:16.960 doesn't end up being what we hoped, it maybe 00:31:16.960 --> 00:31:19.900 incentivizes more clean energy. I'm optimistic, 00:31:20.099 --> 00:31:22.640 but I also think, I think a lot of this CapEx 00:31:22.640 --> 00:31:26.160 is really coming down to ZOMO and not being outspent 00:31:26.160 --> 00:31:28.900 by your competition. And not having the capacity 00:31:28.900 --> 00:31:31.359 when the, when it's needed, because right now, 00:31:31.420 --> 00:31:33.700 a lot of them are pretty capacity constrained. 00:31:34.319 --> 00:31:37.259 And Google has their own stuff and they build 00:31:37.259 --> 00:31:38.839 their own chips, but they still have to get power 00:31:38.839 --> 00:31:41.160 from somebody. And yeah, it's just, it's really 00:31:41.160 --> 00:31:43.579 interesting to see the crazy things people are 00:31:43.579 --> 00:31:46.000 doing to get power. I know at the end, they got 00:31:46.000 --> 00:31:49.460 in trouble with Tennessee for doing impermitted 00:31:49.460 --> 00:31:52.859 power solutions. What we're emitting more fuel 00:31:52.859 --> 00:31:58.039 excess than they were permitted to have. So it 00:31:58.039 --> 00:32:00.700 is crazy, but as somebody who has to live it 00:32:00.700 --> 00:32:04.880 every day, it's hard to contextualize $710 billion 00:32:04.880 --> 00:32:09.140 in CapEx. Yeah. Yeah. That's just, it's a factor 00:32:09.140 --> 00:32:11.140 of, it's a huge factor above what we're used 00:32:11.140 --> 00:32:13.259 to, the kind of CapEx that we're used to. And 00:32:13.259 --> 00:32:14.839 it also doesn't help that there's all kinds of 00:32:14.839 --> 00:32:17.700 weird financing models that are being deployed. 00:32:17.819 --> 00:32:21.400 I saw a few recently that I'm like very confused 00:32:21.400 --> 00:32:25.299 by. I'm just hoping we're not in a bubble that 00:32:25.299 --> 00:32:28.150 cannot be. Slowly deflated. Yeah, it's true. 00:32:28.269 --> 00:32:32.369 Yeah. Okay. So I think I will wrap it up with 00:32:32.369 --> 00:32:35.890 my third and final topic. And that was an announcement 00:32:35.890 --> 00:32:40.609 from Microsoft to Ericsson. And they are integrating 00:32:40.609 --> 00:32:44.589 Ericsson Advanced Enterprise Mobility into Windows 00:32:44.589 --> 00:32:48.670 11. They are building in enterprise 5G connectivity 00:32:48.670 --> 00:32:52.890 with support for stuff like Intune and eSIM provisioning. 00:32:53.279 --> 00:32:55.640 into Windows itself. The way I look at it, this 00:32:55.640 --> 00:32:59.740 is a way to enhance Microsoft's security for 00:32:59.740 --> 00:33:02.900 Windows, but also to make it easier for IT departments 00:33:02.900 --> 00:33:06.240 to manage 5G laptops and to do that securely 00:33:06.240 --> 00:33:08.339 within the environments that they're already 00:33:08.339 --> 00:33:10.660 working within. So it's really just like about 00:33:10.660 --> 00:33:14.619 enabling AI, of course, with constant connectivity 00:33:14.619 --> 00:33:18.900 via 5G, but doing it in a easy to manage and 00:33:18.900 --> 00:33:21.519 a cost -effective way. I think this is just like 00:33:21.519 --> 00:33:25.089 a net positive. Because Windows needed something 00:33:25.089 --> 00:33:28.910 like this, like in order to just have more 5G 00:33:28.910 --> 00:33:32.609 first kind of thoughts and features. What's interesting 00:33:32.609 --> 00:33:35.769 is that this is being piloted by the Surface 00:33:35.769 --> 00:33:38.430 5G devices. So this is not all Windows devices 00:33:38.430 --> 00:33:42.430 yet, but they are building it native into Windows 00:33:42.430 --> 00:33:44.690 11. They're making Windows 11 natively cellular 00:33:44.690 --> 00:33:47.789 and enterprise managed. I think it's a great 00:33:47.789 --> 00:33:52.160 thing. I wish I had a Surface 5G, I think. My 00:33:52.160 --> 00:33:55.059 current Surfaces are not 5G or I don't have the 00:33:55.059 --> 00:33:58.539 Surface 5G laptop, which they did release last 00:33:58.539 --> 00:34:02.039 year. But I'm going to try and go through my, 00:34:02.160 --> 00:34:05.019 I think I actually do have one. I'd have to go 00:34:05.019 --> 00:34:07.180 back and look through all my laptops. I have 00:34:07.180 --> 00:34:09.860 too many of them. It's another problem. But yeah. 00:34:11.659 --> 00:34:13.980 Problems of device analysts, right? Yeah. Yes. 00:34:14.179 --> 00:34:17.159 And I do get a lot of crap from my colleagues 00:34:17.159 --> 00:34:19.739 who don't get as many devices because they would 00:34:19.739 --> 00:34:22.219 like to have new laptops too. The truth is, I 00:34:22.219 --> 00:34:25.079 give away it. Yeah. This is not a very deep topic, 00:34:25.119 --> 00:34:26.659 but it is an interesting one because I think 00:34:26.659 --> 00:34:29.820 that does set the stage for 6G. Yeah. To say, 00:34:29.880 --> 00:34:33.179 I'm very surprised. When 5G first rolled around, 00:34:33.300 --> 00:34:35.900 I thought for sure, this is the time when laptops 00:34:35.900 --> 00:34:39.159 would have cellular integrated. And in a couple 00:34:39.159 --> 00:34:41.639 of years, the only laptop you'd be able to buy 00:34:41.639 --> 00:34:44.280 was one that had cellular in it. And that hasn't 00:34:44.280 --> 00:34:45.739 really happened. And I'm honestly surprised. 00:34:45.840 --> 00:34:47.820 If I had placed bets, I would have bet that would 00:34:47.820 --> 00:34:49.340 have happened and it really hasn't. But when 00:34:49.340 --> 00:34:52.199 you think about it, from... the perspective of 00:34:52.199 --> 00:34:56.059 just a regular everyday person that would you 00:34:56.059 --> 00:35:00.719 buy the laptop that's $100 or $200 cheaper? You'd 00:35:00.719 --> 00:35:02.820 probably buy the one that's $200 cheaper. If 00:35:02.820 --> 00:35:05.400 you were given the option, you'd buy the one 00:35:05.400 --> 00:35:08.300 that was $200 cheaper. And so I think just that 00:35:08.300 --> 00:35:10.840 stumbling block of the cellular version is always 00:35:10.840 --> 00:35:13.480 going to be more expensive, I think is a real... 00:35:13.480 --> 00:35:15.820 And then on the other side of things, if you 00:35:15.820 --> 00:35:18.760 have a 5G capable one, now you have to figure 00:35:18.760 --> 00:35:21.139 out how to do the service plan and... I can't 00:35:21.139 --> 00:35:22.860 share the data on your smartphone. You've got 00:35:22.860 --> 00:35:26.239 to get a separate service plan. And so I think 00:35:26.239 --> 00:35:28.679 there's a lot of reasons why that hasn't taken 00:35:28.679 --> 00:35:31.599 off. And the last thing I'll say, too, is because 00:35:31.599 --> 00:35:34.059 my laptop doesn't have 5G, I'm glad about it 00:35:34.059 --> 00:35:35.739 because it has great battery life. And when I 00:35:35.739 --> 00:35:38.760 need to, when I'm out and about, I can just tether 00:35:38.760 --> 00:35:42.480 with my phone. And I like that. I think that 00:35:42.480 --> 00:35:44.260 works really well, especially I've got a new 00:35:44.260 --> 00:35:47.210 phone, so the battery life is really good. Yeah, 00:35:47.289 --> 00:35:50.269 I think, will that situation change in 6G? Is 00:35:50.269 --> 00:35:52.789 6G going to bring us integrated, more integrated 00:35:52.789 --> 00:35:54.730 laptops? Or are all the MacBooks going to have 00:35:54.730 --> 00:35:58.269 6G in them when that rolls around? What's funny 00:35:58.269 --> 00:36:02.469 is a lot of PC OEMs have been holding on to Apple 00:36:02.469 --> 00:36:06.250 tests. So I think we might actually get that 00:36:06.250 --> 00:36:09.590 moment when Apple does it with their own PCs. 00:36:09.750 --> 00:36:12.769 With their own silicon. Because they have their 00:36:12.769 --> 00:36:14.809 own modems now. They have their own modems now. 00:36:15.179 --> 00:36:17.599 The cost of doing that is much lower than if 00:36:17.599 --> 00:36:20.099 they use a Qualcomm modem, which is, I actually 00:36:20.099 --> 00:36:23.400 did a recent 5G survey of people who are actively 00:36:23.400 --> 00:36:26.940 deploying 5G in PCs and all interesting data, 00:36:27.039 --> 00:36:29.159 which I'll be publishing in a report fairly soon. 00:36:30.500 --> 00:36:33.059 The survey concluded right when I had my second 00:36:33.059 --> 00:36:35.119 kid, so I wasn't spending any other time to actually 00:36:35.119 --> 00:36:38.539 get that survey data. A little busy, yeah, for 00:36:38.539 --> 00:36:41.320 sure. Yeah, but the truth is, the number one 00:36:41.320 --> 00:36:45.400 complaint is actually coverage. Inconsistent 00:36:45.400 --> 00:36:48.519 coverage, not enough coverage. And a lot of companies 00:36:48.519 --> 00:36:51.639 are working on that right now by offering 5G 00:36:51.639 --> 00:36:54.719 PC services that have all three carriers at the 00:36:54.719 --> 00:36:57.440 same time and just switch to the best carrier 00:36:57.440 --> 00:36:59.760 where you are, which I think is going to solve 00:36:59.760 --> 00:37:01.800 a lot of those challenges. But to your point, 00:37:01.880 --> 00:37:04.539 the cost is still an issue. Enterprises are more 00:37:04.539 --> 00:37:07.360 willing to spend that money because they can 00:37:07.360 --> 00:37:10.059 secure their connection. No connecting to coffee, 00:37:10.300 --> 00:37:13.599 Wi -Fi, no hotel Wi -Fi. All those are wildly 00:37:13.599 --> 00:37:16.190 insecure. And then also just being always be 00:37:16.190 --> 00:37:18.050 connected to somebody who goes to a lot of client 00:37:18.050 --> 00:37:20.369 sites. The amount of times that I have to authenticate 00:37:20.369 --> 00:37:23.789 my laptop on their network and get all the permissions 00:37:23.789 --> 00:37:26.289 and then still be filtered to not be able to 00:37:26.289 --> 00:37:28.730 use AI or whatever things I want to do. It's 00:37:28.730 --> 00:37:31.090 actually quite nice to have 5G. There are definitely 00:37:31.090 --> 00:37:34.210 types of personas, which we also found in who 00:37:34.210 --> 00:37:37.929 uses 5G the most and why. But yes, it's, this 00:37:37.929 --> 00:37:39.730 is still a fundamental thing that should have 00:37:39.730 --> 00:37:41.869 been done all the time. And it's just another 00:37:41.869 --> 00:37:45.650 piece of the equation. making 5G easier to deploy 00:37:45.650 --> 00:37:48.829 and use regularly. And yes, the carriers are 00:37:48.829 --> 00:37:51.010 also a problem in terms of mixing. We'll get 00:37:51.010 --> 00:37:53.250 there, I think, eventually. I think Apple might 00:37:53.250 --> 00:37:56.789 help quite a bit. I can tell you, once you get 00:37:56.789 --> 00:38:00.130 a 5G laptop, it's very hard to go back. All right. 00:38:00.329 --> 00:38:02.909 So my objective now is not to get a 5G laptop. 00:38:03.369 --> 00:38:06.809 I'll keep my Wi -Fi for as long as I can. Yeah, 00:38:06.829 --> 00:38:09.449 it'll come to a cheap price. Yeah, exactly. Yeah, 00:38:09.489 --> 00:38:11.960 interesting. Yeah. I will say how it develops. 00:38:11.980 --> 00:38:14.000 And that's everybody out there. That's your job, 00:38:14.000 --> 00:38:17.239 right? Make the stuff easy. Definitely. Okay. 00:38:17.500 --> 00:38:20.639 And I wanted to say thank you very much for joining 00:38:20.639 --> 00:38:23.000 me today. I really, I know we ran a little long 00:38:23.000 --> 00:38:25.179 this week and I think we'll tighten it up a little 00:38:25.179 --> 00:38:27.260 bit for our listeners so they can get it under 00:38:27.260 --> 00:38:30.559 30 minutes. Otherwise, we hope our viewers and 00:38:30.559 --> 00:38:31.980 listeners found this week's topic interesting. 00:38:32.619 --> 00:38:35.659 And please don't forget to join us next week. 00:38:36.059 --> 00:38:38.360 And don't forget to rate and subscribe. And once 00:38:38.360 --> 00:38:40.019 again, I want to say thanks a lot from Mike for 00:38:40.019 --> 00:38:43.559 joining me. It's been great so far and excited 00:38:43.559 --> 00:38:46.940 to see where this goes. Let's do it. Sounds good.