1 00:00:01,301 --> 00:00:02,942 Welcome to Let's Get Digital. 2 00:00:02,942 --> 00:00:05,264 I'm Carrie Charles, your host. 3 00:00:05,264 --> 00:00:09,328 And I'm so happy that you are here with me today. 4 00:00:09,328 --> 00:00:12,070 So it's gonna be a great show as always. 5 00:00:12,070 --> 00:00:16,994 I am thrilled to introduce to you Ivo Ivanov. 6 00:00:16,994 --> 00:00:21,317 He is the CEO and Chairman of the Board for DKICS. 7 00:00:21,317 --> 00:00:24,390 And Ivo, I've known you for a while now. 8 00:00:24,390 --> 00:00:28,183 I've seen you at many different events here in the US. 9 00:00:28,183 --> 00:00:31,125 So I just wanna say I'm so excited you're here. 10 00:00:31,139 --> 00:00:33,050 And thank you. 11 00:00:33,782 --> 00:00:35,236 Thank you so much for having me. 12 00:00:35,236 --> 00:00:37,322 I'm excited being part of your great show. 13 00:00:37,322 --> 00:00:42,516 uh yeah, let's have a chat about digital. 14 00:00:43,135 --> 00:00:44,927 Let's go, let's go. 15 00:00:44,948 --> 00:00:48,051 First, I wanna hear about you. 16 00:00:48,105 --> 00:00:57,775 A little bit about uh just the twists and decisions and maybe the risks that brought you to where you are today in your current seat with Deacon. 17 00:00:58,948 --> 00:01:02,630 my gosh, that's a very complex question. 18 00:01:02,630 --> 00:01:04,670 Let me see how to answer. 19 00:01:04,670 --> 00:01:08,712 uh Probably I will start about my origin, right? 20 00:01:08,712 --> 00:01:13,534 So I have been born back in time in Sofia, Bulgaria. 21 00:01:13,534 --> 00:01:15,955 And this is where my life journey started. 22 00:01:15,955 --> 00:01:25,528 uh During the high school, this is one essential element, probably driving oh my emotional side until today. 23 00:01:25,528 --> 00:01:26,960 I decided to... 24 00:01:26,960 --> 00:01:28,300 uh 25 00:01:28,300 --> 00:01:43,835 establish uh a radio show as moderator and producer so this was very close to media afterwards I did also one year with a TV broadcasting company uh on stage and in front of 26 00:01:43,835 --> 00:01:54,250 camera and microphones this gave me a lot of experience with media and the way how to communicate with a bigger audience then my father told me 27 00:01:54,250 --> 00:01:58,643 Son, listen, this is probably not the way you should take in your life. 28 00:01:58,904 --> 00:02:08,871 I want to make sure you will be able to uh grow and create a nice career and also take care of all the family one day. 29 00:02:08,871 --> 00:02:12,354 I don't think this radio stuff and TV stuff is good enough. 30 00:02:12,354 --> 00:02:17,718 uh He never watched a production or listened to production. 31 00:02:17,718 --> 00:02:24,099 I'm not sure how he came to that result because it was more a kind of entertainment Saturday afternoon thing. 32 00:02:24,099 --> 00:02:33,999 But he advised and I was really happy to take this advice to start something completely different to start my law studies. 33 00:02:33,999 --> 00:02:38,979 this is in fact one of the next important steps of my life. 34 00:02:38,979 --> 00:02:47,799 I started the first year in the University of Law, Science and Soviet and then I moved to Bonn, Germany for actually initial first year. 35 00:02:47,799 --> 00:02:51,479 But you'll see what happened out of this first year. 36 00:02:51,479 --> 00:02:53,999 28 years in Germany now. 37 00:02:54,055 --> 00:03:01,878 with uh my state exams, with my bar exams and creating my family here with two kids today. 38 00:03:01,878 --> 00:03:04,299 uh new home. 39 00:03:05,460 --> 00:03:13,864 And in fact my career started uh as a lawyer specialized on media and telecom law because of the media background. 40 00:03:13,864 --> 00:03:22,269 You see, their elements and puzzle pieces in life, uh they uh influence the career path later. 41 00:03:22,269 --> 00:03:22,919 and uh... 42 00:03:22,919 --> 00:03:30,204 so yes i did my career as a lawyer and one beautiful day and that's how i'll sum up with the founder of of the case uh... 43 00:03:30,204 --> 00:03:31,424 and and uh... 44 00:03:31,424 --> 00:03:33,616 working on the couple of projects uh... 45 00:03:33,616 --> 00:03:34,506 and uh... 46 00:03:34,506 --> 00:03:36,968 because of this i felt in love uh... 47 00:03:36,968 --> 00:03:38,488 with the internet uh... 48 00:03:38,488 --> 00:03:42,381 felt in love with digital infrastructure decided to to uh... 49 00:03:42,381 --> 00:03:47,274 to change my role being a lawyer in a in a big law firm to uh... 50 00:03:47,274 --> 00:03:50,116 becoming a general corporate council with the kicks 51 00:03:50,116 --> 00:03:58,756 and then i have been asked by colleagues after because my language skills are also related to eastern european languages. 52 00:03:59,356 --> 00:04:06,316 you help us out to start a strategy for GKX based on eastern european markets? 53 00:04:06,316 --> 00:04:07,596 I yeah why not. 54 00:04:07,596 --> 00:04:15,107 Let's start traveling to telco shows and this is how we got in touch with business development and 55 00:04:15,107 --> 00:04:20,587 more the business part and less the legal side of the business model. 56 00:04:20,667 --> 00:04:25,247 And I have been asked lately to manage the international expansion of DICEX. 57 00:04:25,247 --> 00:04:30,547 And this is how my journey has been for the past almost 17 years. 58 00:04:30,547 --> 00:04:44,355 I was for 12 years CEO international of DICEX and then three years back I have been asked after Harold stepped back from active operational duties to 59 00:04:44,355 --> 00:04:48,736 ah get the owner and become the new group CEO of DickX. 60 00:04:50,528 --> 00:04:57,490 So every time I talk to you, I just get so energized and inspired and I just love how you said why not, right? 61 00:04:57,490 --> 00:05:06,233 And you're just all about just going for it in life, in business, and I'm so excited for the rest of this interview. 62 00:05:06,233 --> 00:05:06,803 I can't wait. 63 00:05:06,803 --> 00:05:08,994 So thank you for sharing your story. 64 00:05:08,994 --> 00:05:10,995 Let's talk about DKicks now. 65 00:05:10,995 --> 00:05:12,635 What is the founding story? 66 00:05:12,635 --> 00:05:14,706 What problems do you solve? 67 00:05:16,323 --> 00:05:18,763 Yeah, the founding story is an amazing one. 68 00:05:18,763 --> 00:05:26,603 There are not a lot of companies we see in today's telco and digital infrastructure world that are that old. 69 00:05:26,703 --> 00:05:32,803 This year, a couple of weeks back, we celebrated the 30th anniversary of the company. 70 00:05:32,803 --> 00:05:39,303 In fact, Dickex has been founded back in time 1995 on the beautiful 26th of June. 71 00:05:39,303 --> 00:05:45,781 Exactly on this day, 30 years later, we celebrated the 30th anniversary. 72 00:05:45,781 --> 00:05:54,618 And the company has been found in a very special time, actually the time where the commercial internet uh has been born. 73 00:05:54,618 --> 00:05:57,980 So we have been born with the commercial internet. 74 00:05:57,980 --> 00:06:05,046 uh The company has been founded obviously in Frankfurt in 1995 by three different ISPs. 75 00:06:05,046 --> 00:06:13,532 One based in the north of Germany in Hamburg, one in the middle of Germany in Dortmund and one in the south of Germany in Karlsruhe. 76 00:06:15,892 --> 00:06:22,525 And now you can ask me why they decided to the start of the platform, the business of Digis in Frankfurt. 77 00:06:22,525 --> 00:06:29,688 Yeah, because Frankfurt has been always a kind of a crossroad in the middle of Germany and crossroad in the middle of Europe. 78 00:06:29,708 --> 00:06:41,534 And this is very essential for digital infrastructure because digital infrastructure is built on digital roads, our fiber installations, the telco cables, et cetera. 79 00:06:41,534 --> 00:06:41,894 Now... 80 00:06:41,894 --> 00:06:42,688 uh 81 00:06:42,688 --> 00:06:53,584 What happened later, so 30 years later we do stay for one of the market leaders on our planet for interconnection services. 82 00:06:53,584 --> 00:07:03,429 uh We operate on 60 plus markets and interconnect globally more than 4200 networks present on five continents. 83 00:07:03,429 --> 00:07:12,584 Obviously the US, Mexico, South America with Brazil, entire Europe, Middle East with Turkey, with Jordan, with Iraq, Pakistan. 84 00:07:12,738 --> 00:07:19,382 We have Dubai there, we have India, Southeast Asia with Singapore, Malaysia, Philippines, Indonesia and Japan. 85 00:07:19,382 --> 00:07:25,305 And you see the loop is now closed and in fact DECIX operates a global network these days. 86 00:07:25,305 --> 00:07:36,520 So this is the journey from 3 ISP uh early installation in Frankfurt back in time now to a global interconnection service company. 87 00:07:37,897 --> 00:07:41,729 talk about the evolution of the Internet Exchange. 88 00:07:41,729 --> 00:07:55,495 From the early pairing clubs to today's global software-defined interconnection platforms, what were the big inflection points and which ones or one are we living through right now? 89 00:07:56,929 --> 00:08:05,471 I think the essential service and this is how everything started, this is the peering service and peering, we will have this term a couple of times probably during our great 90 00:08:05,471 --> 00:08:06,702 chat today. 91 00:08:06,702 --> 00:08:13,393 Peering stays for direct interconnection of different types of networks, disregarding the type of networks in the business model. 92 00:08:13,393 --> 00:08:15,544 And it's a beautiful type of interconnection. 93 00:08:15,544 --> 00:08:24,460 It's probably one of the best ways to exchange traffic between different types of networks, those who deliver content and those who want 94 00:08:24,460 --> 00:08:28,393 to deliver this content to their end users, the high bone networks. 95 00:08:28,393 --> 00:08:31,565 So uh I think why is that so great? 96 00:08:31,565 --> 00:08:34,758 It's great because peering stays for the direct interconnection. 97 00:08:34,758 --> 00:08:40,232 So they are not third party networks between the content network and the internet access network. 98 00:08:40,232 --> 00:08:46,687 So self-explaining, this creates a very robust way to exchange the data packages without outages. 99 00:08:46,687 --> 00:08:51,134 So if you click at home on your TV and want to watch Amazon Prime or 100 00:08:51,134 --> 00:09:03,733 Netflix or Disney Plus video just to mention a couple of different operators uh So your request for this video for the show you want to see uh Needs somehow to approach the 101 00:09:03,733 --> 00:09:19,532 Netflix Amazon or Disney server where this show this movie is uh placed on and How this happens actually your request sits on the network of your internet access provider uh 102 00:09:19,532 --> 00:09:25,885 can be can be comcast can be can be the rise in whatever you name it all the different uh... 103 00:09:25,885 --> 00:09:36,640 network access operators they deliver access to homes and offices by the way the same is not for the office if those two networks are directly interconnected the comcast network 104 00:09:36,640 --> 00:09:36,900 uh... 105 00:09:36,900 --> 00:09:39,741 rise network at and at work with uh... 106 00:09:39,741 --> 00:09:46,943 netflix apple this network this office planning the packets they they flow immediately they are 107 00:09:46,943 --> 00:09:52,107 extremely fast available, no outages, no pixel failures. 108 00:09:52,107 --> 00:09:58,371 If we watch a real-time show, a live show, this became even more important. 109 00:09:58,371 --> 00:10:03,574 Watch uh sports or entertainment, different types of sports. 110 00:10:03,574 --> 00:10:11,600 We know uh how important live TV sports shows are nowadays on IP and everybody uses almost IP, right? 111 00:10:11,600 --> 00:10:15,446 So this type of quality is extremely important, but also security. 112 00:10:15,446 --> 00:10:24,189 because it's changing the packets in a direct way between the network that sends the request for the content and the network that has the content. 113 00:10:24,390 --> 00:10:33,614 Make sure that not a third party that might not have the best intent, can be a criminal third party, can approach this communication. 114 00:10:33,614 --> 00:10:44,790 So it's about uh high performance, how we call it in our business, very low latency uh and extremely resilient and secure way 115 00:10:44,790 --> 00:10:46,291 to communicate. 116 00:10:46,811 --> 00:10:48,522 So this is the peering service. 117 00:10:48,522 --> 00:10:49,593 This is the foundation. 118 00:10:49,593 --> 00:10:57,108 And now what happened over the decades or even the past couple of years, uh more services they have been added. 119 00:10:57,108 --> 00:11:04,332 Services like cloud peering, also uh peering related to specific industry segments. 120 00:11:04,332 --> 00:11:09,445 We call it close user groups uh for finance, for automotive, logistic industry. 121 00:11:09,445 --> 00:11:11,042 And now, nowadays, 122 00:11:11,042 --> 00:11:24,433 uh the latest news is the DICX uh internet uh exchange for the artificial intelligence communication and connectivity needs. 123 00:11:24,433 --> 00:11:26,535 call it AIIX. 124 00:11:26,535 --> 00:11:29,107 A lot of different developments. 125 00:11:29,107 --> 00:11:31,039 I think they have one thing in common. 126 00:11:31,039 --> 00:11:38,825 uh The logic of direct and very secure and resilient highly performable traffic uh exchange. 127 00:11:38,825 --> 00:11:39,926 uh 128 00:11:39,926 --> 00:11:41,646 This is extremely relevant. 129 00:11:41,646 --> 00:11:51,730 The more we talk about specific businesses, industries, are uh crucial and they are uh critical also for our economies. 130 00:11:51,730 --> 00:11:55,132 uh This is even more relevant nowadays. 131 00:11:55,132 --> 00:12:00,384 And Kerry, one uh more element I want to add to uh the picture here. 132 00:12:00,384 --> 00:12:01,844 It's not about the services. 133 00:12:01,844 --> 00:12:08,077 It's also how the way the governance on Internet exchanges has been developed over time. 134 00:12:08,077 --> 00:12:09,171 uh 135 00:12:09,171 --> 00:12:21,721 early days there were two types of exchanges the so-called European model this is the model we stay for that is uh very neutral as we are owned entirely by the one of the 136 00:12:21,721 --> 00:12:32,850 largest internet associations of the world so we don't have one single share holder as a carrier or a data center operator so very neutral and the other model is the data center 137 00:12:32,850 --> 00:12:39,073 owned internet exchange there are examples in the US companies like Equinix they have their own exchange 138 00:12:39,073 --> 00:12:41,794 CoreSight as well, another example for. 139 00:12:41,794 --> 00:12:51,416 And what changed over the time, especially when we entered as Dickx, the US market, introducing the European, very neutral, very distributed model. 140 00:12:51,416 --> 00:13:00,939 This changed the landscape in the US as well to a very positive development because immediately the services of the exchange, they became available in 20 different data 141 00:13:00,939 --> 00:13:05,920 centers across New York and New Jersey and not just available in the data center of one operator. 142 00:13:05,920 --> 00:13:08,841 And this inspired the entire community 143 00:13:08,841 --> 00:13:20,096 we published last year in light of our 10 years anniversary celebration in New York, a study that showed how big the market now grew in comparison to the LA market, for 144 00:13:20,096 --> 00:13:24,667 instance, where we see the non-neutral uh model only. 145 00:13:24,667 --> 00:13:31,070 In fact, by 400%, New York has been growing faster than the LA. 146 00:13:31,070 --> 00:13:38,293 uh And this also uh encouraged the business side of data center investments to 147 00:13:38,464 --> 00:13:47,410 become very focused on those markets and we see a huge increase in investments on the data center side in the great area of New York. 148 00:13:47,410 --> 00:13:52,384 Just as one example, same is valid for Dallas or Chicago where we do also operate. 149 00:13:52,384 --> 00:14:00,859 So this was also a very strong element in our global expansion introducing this data center distributed and highly neutral model. 150 00:14:01,890 --> 00:14:04,715 So let's talk a little bit more about peering. 151 00:14:04,836 --> 00:14:12,270 What is the case for peering cost, performance, resilience, and where do companies leave money on the table? 152 00:14:13,833 --> 00:14:14,433 All right. 153 00:14:14,433 --> 00:14:20,357 um So as I mentioned, Peering stays for rightly said uh from you. 154 00:14:20,357 --> 00:14:31,385 uh It's extremely uh robust, uh very secure because of the direct exchange and uh you can have great redundancy. 155 00:14:31,385 --> 00:14:38,530 uh I will talk about performance just in a sec and then uh where the companies I believe they do live money on the table. 156 00:14:38,530 --> 00:14:43,641 But if we talk about uh resilience and today resilience is extremely important. 157 00:14:43,641 --> 00:14:44,001 uh... 158 00:14:44,001 --> 00:14:55,086 them separate done to see the model i just described perfectly matches these demands on the market uh being present in let's take example of new york again twenty different data 159 00:14:55,086 --> 00:15:06,030 centers so a company can have to different different physical set ups with two different data center operators one in jersey one in manhattan probably being connected to one 160 00:15:06,030 --> 00:15:13,011 distributed integrated platform for traffic exchange this makes it very robust and very resilient 161 00:15:13,011 --> 00:15:16,684 Now talk about the performance, the latency talk, right? 162 00:15:16,684 --> 00:15:18,916 And I say latency is the new currency. 163 00:15:18,916 --> 00:15:23,170 Every single millisecond counts today for very good reasons. 164 00:15:23,170 --> 00:15:24,831 We have real-time applications. 165 00:15:24,831 --> 00:15:26,613 They deserve exactly this performance. 166 00:15:26,613 --> 00:15:38,593 uh And this is the main advantage of peering today in comparison to other services like transit, upstream, uh IP transit as a service, or different other type of interconnections 167 00:15:38,593 --> 00:15:40,385 because uh it is... 168 00:15:40,385 --> 00:15:40,876 uh 169 00:15:40,876 --> 00:15:43,627 highly performable and performance is everything today. 170 00:15:43,627 --> 00:15:52,870 uh it's, uh I'll give you an example, when Netflix produced the show of Mike Tyson, we remember what happened. 171 00:15:52,870 --> 00:16:01,852 There uh has been a big outage at the beginning of not 65 milliseconds, but more than 60 seconds. 172 00:16:01,852 --> 00:16:03,253 And this is huge, right? 173 00:16:03,253 --> 00:16:04,709 At the beginning of the show. 174 00:16:04,709 --> 00:16:08,532 A lot of millions of users have been disappointed. 175 00:16:08,833 --> 00:16:12,736 This shows us how important performance on IP is today. 176 00:16:12,809 --> 00:16:15,598 It has a direct impact on business and reputation. 177 00:16:15,598 --> 00:16:21,744 uh This is one of the main advantages of peering, the high performance and the great security. 178 00:16:21,744 --> 00:16:24,386 Where companies might leave money on the table? 179 00:16:24,566 --> 00:16:32,553 Using uh upstream services only, IP Transit only, because in a lot of cases it's... 180 00:16:32,563 --> 00:16:42,179 is more expensive than peering and doesn't provide the direct controllability on the traffic flows companies need to have for their performance uh control. 181 00:16:42,640 --> 00:16:58,120 And also in general, when companies, do not compare different options on the market, they do not rethink how they can change their digital strategy to uh the way their data is 182 00:16:58,120 --> 00:17:02,389 stored, their data is processed or 183 00:17:02,389 --> 00:17:04,860 how they plan their further expansions. 184 00:17:04,860 --> 00:17:12,314 If they stuck with one data center only, uh is uh obviously very often the case that they leave money on the table. 185 00:17:12,314 --> 00:17:26,062 Very long contracts is another topic here because nowadays the market dynamics are so, uh I would say, active that if a company just thinks that I will stay five years long with 186 00:17:26,062 --> 00:17:31,865 this operator only, probably for the operator this is a great deal, but for the company it's not because 187 00:17:31,883 --> 00:17:42,251 prices they change more dynamically these days and I recommend to have this dynamic in the contractual negotiation as well and just create a bigger redundancy. 188 00:17:42,251 --> 00:17:43,962 We are not afraid of this as DxC. 189 00:17:43,962 --> 00:17:47,305 I know a lot of networks that are connected to DxC New York. 190 00:17:47,305 --> 00:17:57,193 They're also connected to NYX, the New York Internet Exchange operated by Tally House and this is great because this creates a huge redundancy and they can load down their traffic 191 00:17:57,193 --> 00:17:58,043 better. 192 00:17:59,816 --> 00:18:02,608 So let's talk about the future of the internet. 193 00:18:02,608 --> 00:18:10,873 Fast forward five to 10 years with AI everywhere, immersive apps, machines, talking to machines. 194 00:18:11,914 --> 00:18:16,367 What has changed in how the world interconnects? 195 00:18:16,960 --> 00:18:22,820 It will change the foundation of how the Internet is built today. 196 00:18:23,140 --> 00:18:27,280 It creates already the demand for the so-called new Internet. 197 00:18:27,280 --> 00:18:28,520 We need the new Internet. 198 00:18:28,520 --> 00:18:31,100 Internet that is highly distributed. 199 00:18:31,100 --> 00:18:42,800 Talking about data centers everywhere, we're talking about exchanges everywhere, we're talking about fiber, 5G, 5G Advanced, 6G everywhere, Leo satellite capabilities. 200 00:18:43,260 --> 00:18:45,784 This is really the... 201 00:18:45,790 --> 00:18:56,186 the requirement uh by AI everywhere, by robotics everywhere, by automation everywhere in real time. 202 00:18:56,186 --> 00:19:01,569 I want to explain this briefly for your audience using the example of a robot. 203 00:19:02,430 --> 00:19:07,443 We know robotics uh as a topic is extremely important today. 204 00:19:07,443 --> 00:19:13,184 We see already a lot of examples in the manufacturing center, but also 205 00:19:13,184 --> 00:19:20,764 different type of of robots engaged in surgeries in the health care sector of the robots engaged in in uh... 206 00:19:20,764 --> 00:19:21,064 uh... 207 00:19:21,064 --> 00:19:21,524 uh... 208 00:19:21,524 --> 00:19:23,624 in in in hospitals uh... 209 00:19:23,624 --> 00:19:24,784 in in uh... 210 00:19:24,784 --> 00:19:35,064 in households helping people right and and they are there it's not a kind of a all it might happen one day think it's they're already there you can buy them online starting by 211 00:19:35,064 --> 00:19:41,224 twelve thousand dollars and they will become better all the time and they will require 212 00:19:41,843 --> 00:19:56,101 a real time communication because you cannot uh build a robot that is completely isolated and disconnected from cloud, from the agents, the AI agents, from the internet, so to say. 213 00:19:56,101 --> 00:20:01,663 So not everything can be in the body of robot on-prem, so to say. 214 00:20:01,683 --> 00:20:06,806 The robot must be online every single millisecond. 215 00:20:06,806 --> 00:20:10,664 Now, what type of digital infrastructure does this require? 216 00:20:10,664 --> 00:20:13,485 And the robot here stays just as one example. 217 00:20:13,485 --> 00:20:23,488 You can take also examples of autonomous driving, the uh cars outside there, or other IOTs. 218 00:20:23,488 --> 00:20:29,269 But I think taking the robot is a good example to display the needs for different digital infrastructure. 219 00:20:29,289 --> 00:20:38,332 We at Dickix, we did analysis with professors for uh how our uh brains function. 220 00:20:39,344 --> 00:20:41,325 They gave us very interesting information. 221 00:20:41,325 --> 00:20:49,069 We asked them what is the speed our human brains require for three different sensors. 222 00:20:49,070 --> 00:20:53,612 The tactile information to feel cold, warm, soft heart. 223 00:20:53,612 --> 00:20:56,913 The visual information and the audio information. 224 00:20:57,014 --> 00:21:05,078 And they gave us free numbers in terms of latency because they have done measurements on how our human being brain functions. 225 00:21:05,078 --> 00:21:08,274 Our brain on a good day without hangover 226 00:21:08,274 --> 00:21:11,186 and not uh being sick, right? 227 00:21:11,186 --> 00:21:17,004 It requires 18 milliseconds and even better for the tactile information. 228 00:21:17,004 --> 00:21:23,273 12 milliseconds are better for the visual and one millisecond or better for the audio. 229 00:21:23,273 --> 00:21:35,860 So now if we want to have one day the immersive internet application world, the robots, they function in real time, they must act as us, as human beings. 230 00:21:35,860 --> 00:21:49,551 to feel very natural right otherwise will not be satisfied with the services so they must hear us properly they must see us properly and they must function in their uh other uh 231 00:21:49,551 --> 00:22:02,801 features properly and this will require infrastructure that meets the lowest latency as i mentioned the one millisecond because if a robot cannot see cannot hear uh it's probably 232 00:22:02,801 --> 00:22:05,907 not a good idea to send this guy out right 233 00:22:05,907 --> 00:22:18,553 So the infrastructure needs to be as close as possible to the applications because one millisecond means 50, 60 miles in physics because of all speed of light. 234 00:22:18,553 --> 00:22:31,519 Now, if the agent that manages this robot or different agents that manage the audio sensors of the robot are far away, 235 00:22:31,895 --> 00:22:34,416 This will not function not function properly. 236 00:22:34,416 --> 00:22:48,340 So the digital infrastructure data centers transmission technology fiber Internet exchanges uh Must be deployed everywhere in the deep edge This will not replace the big 237 00:22:48,340 --> 00:22:49,471 aggregation hubs. 238 00:22:49,471 --> 00:22:57,253 We do have nowadays like the hubs in New York Miami LA Bay Area Dallas, Chicago 239 00:22:57,395 --> 00:22:59,176 This will be additionally needed. 240 00:22:59,176 --> 00:23:06,752 So we'll see a more distributed, meshed infrastructure world supporting AI everywhere. 241 00:23:06,752 --> 00:23:08,723 And DIGX is working already on this. 242 00:23:08,723 --> 00:23:20,613 We have first prototypes designed for the deep edge to be deployed next to manufacturing facilities, next to um highway crossroads, shopping malls, business parks. 243 00:23:20,613 --> 00:23:24,155 This is where we need the infrastructure of the future as well. 244 00:23:24,155 --> 00:23:30,543 next to the existing big digital capitals I just mentioned. 245 00:23:31,676 --> 00:23:38,416 So what do business leaders need to do to be, I guess, deep edge ready, right? 246 00:23:39,102 --> 00:23:53,148 Depending on their business model, they need to accept new type of usage of digital infrastructure to uh become more interested in what I just described in peering in 247 00:23:53,148 --> 00:24:04,243 different close user groups to be more present in the deep edge to support digital infrastructure development next to their manufacturing facilities. 248 00:24:04,243 --> 00:24:06,665 for instance let's take the example of uh... 249 00:24:06,665 --> 00:24:18,897 real-time fraud detection that is extremely important for the banks though the the finance industry this needs to be done indian strong on site where the the users are uh... 250 00:24:18,897 --> 00:24:21,213 as close as possible to the uh... 251 00:24:21,213 --> 00:24:30,889 as as as close as possible to the a t as close possible to the branch of the bank et cetera to the only banking where the users are not all users they sit on the east or west 252 00:24:30,889 --> 00:24:31,487 coast 253 00:24:31,487 --> 00:24:34,507 uh We have people, are between them. 254 00:24:34,507 --> 00:24:44,631 So the need for this different uh understanding of how infrastructure should be used, but this infrastructure needs to be built as well. 255 00:24:44,631 --> 00:24:50,092 So we do have in the United States 15 states without a single Internet exchange. 256 00:24:50,092 --> 00:24:54,594 We do have areas without a single carrier neutral data center. 257 00:24:54,594 --> 00:24:57,735 This needs to oh be changed. 258 00:24:57,735 --> 00:24:59,075 We need support from... 259 00:24:59,075 --> 00:24:59,667 uh 260 00:24:59,667 --> 00:25:02,548 from the public administration if it comes to uh... 261 00:25:02,548 --> 00:25:07,829 permissions if it comes to even the financial contribution we need more uh... 262 00:25:07,829 --> 00:25:18,172 capital capital of getting interested in this area and i think now there's a great idea to invest in this area because this is the area of the future this is where we will 263 00:25:18,172 --> 00:25:28,115 experience a world of undreamed off applications people will will fell in love with and they will use it i'll just give you an example of of uh... 264 00:25:28,115 --> 00:25:29,105 immersive 265 00:25:29,105 --> 00:25:31,796 immersive experience on a football match. 266 00:25:31,796 --> 00:25:45,904 uh Just sitting with a new type of glasses, let's take the Apple's ProVision Pro on your couch in your living room, enjoying in a fully immersive uh way, uh an incredible lifetime 267 00:25:45,904 --> 00:25:52,128 life, real life, uh real time experience uh with your favorite football team. 268 00:25:52,128 --> 00:25:58,770 Because you are unable to buy a ticket, it's either too expensive or you don't want to do it because you are afraid 269 00:25:58,770 --> 00:26:02,892 of too many people in one stadium or you're just disenabled. 270 00:26:02,892 --> 00:26:07,225 lot of people they cannot do it but they can enjoy this. 271 00:26:07,225 --> 00:26:21,262 are just at the beginning of an amazing era that will be only possible to happen if we understand that the underlying digital infrastructure is the must-have critical element to 272 00:26:21,262 --> 00:26:22,013 build it. 273 00:26:22,013 --> 00:26:26,905 And this is where it's very, very, very worth to invest in. 274 00:26:27,708 --> 00:26:29,059 Well said, well said. 275 00:26:29,059 --> 00:26:33,831 Let's talk about AI inference versus training. 276 00:26:33,831 --> 00:26:43,726 Where in the AI lifecycle does low latency, high capacity, distributed interconnection create real competitive advantage? 277 00:26:45,055 --> 00:26:50,867 uh I recall my friend the robot again here giving this example. 278 00:26:50,867 --> 00:26:53,297 This is exactly thanks for the question. 279 00:26:53,297 --> 00:27:00,960 So as I said the robots the cars the autonomous cars just these two examples to make it simple. 280 00:27:00,960 --> 00:27:07,702 oh Let's take the the fraud detection in real time as well approaching the finance sector very important. 281 00:27:07,702 --> 00:27:11,935 So these three examples they require real time interconnection. 282 00:27:11,935 --> 00:27:13,139 We're talking about 283 00:27:13,139 --> 00:27:15,601 one, two, three milliseconds reaction. 284 00:27:15,601 --> 00:27:17,841 And I said, what does this mean? 285 00:27:17,882 --> 00:27:36,112 This means a hundred hundred fifty miles where radius of the the robot of the car of the hospital where the surgery happens or or the finance fraud detection installations are at. 286 00:27:36,112 --> 00:27:41,105 So this is the distance the agents, the AI agents. 287 00:27:41,105 --> 00:27:46,277 And millions of those we will see in the future will need on the inferencing side. 288 00:27:46,277 --> 00:27:49,498 This is the main difference between inference and training. 289 00:27:49,498 --> 00:27:54,260 Training is a play about power and pipes, right? 290 00:27:54,320 --> 00:27:57,521 Huge power density and big pipes. 291 00:27:57,541 --> 00:28:05,784 Training is not that sensitive if it comes to latency and it's not that sensitive if it comes to certain outages even. 292 00:28:05,824 --> 00:28:11,236 Yes, this make it slower, but doesn't affect it in terms of quality, right? 293 00:28:11,484 --> 00:28:17,018 But talking about the inference, the world changes completely to the opposite. 294 00:28:17,018 --> 00:28:19,910 It's not that power intense. 295 00:28:19,910 --> 00:28:24,003 It's still power needed, but it's not that power intense. 296 00:28:24,003 --> 00:28:38,452 And uh it's not really that pipe intense, but it's extremely latency sensitive and extremely security sensitive. 297 00:28:38,492 --> 00:28:41,434 This is very, very, very important. 298 00:28:41,758 --> 00:28:52,483 Now, why I'm saying this, because if you have a 2-millisecond uh outage, the car doesn't know what to do, the car can create an accident, the robot will fall apart, right? 299 00:28:52,483 --> 00:28:58,286 uh The fraud detection will not function and people will lose their money. 300 00:28:58,286 --> 00:29:03,498 We're talking about critical reaction based on data exchange. 301 00:29:03,498 --> 00:29:08,582 And this is the difference why inference for 302 00:29:08,582 --> 00:29:12,612 A state like Cannes cannot happen out of New York and out of Miami. 303 00:29:12,612 --> 00:29:22,966 It has to be as close as possible where the robots are, the manufacturing facilities are, the cars are driving, and the people they enjoy real-time applications. 304 00:29:23,066 --> 00:29:30,588 And inferencing, by the way, will be the dominant part of AI just in three to five years from now. 305 00:29:30,588 --> 00:29:37,127 As I heard from a couple of friends from the data science industry, the investments today 306 00:29:37,127 --> 00:29:45,344 are by 70 % focused on the training side of building data centers for training and 30 % inferencing. 307 00:29:45,344 --> 00:29:49,778 This will change to the opposite just within a couple of years only. 308 00:29:49,919 --> 00:29:54,723 Because what do you do with millions and billions of agents? 309 00:29:54,723 --> 00:29:59,808 You train them because you want to invite them to work. 310 00:29:59,808 --> 00:30:02,410 We need to send them uh the streets. 311 00:30:02,410 --> 00:30:02,928 oh 312 00:30:02,928 --> 00:30:14,334 like FBI does, An agent that has been trained by FBI doesn't stay in school only he or she they need to be active on the street, right? 313 00:30:14,334 --> 00:30:18,787 And this is exactly what will happen with the inferencing side of the EIA. 314 00:30:18,787 --> 00:30:26,891 We will need to get the agents on the streets literally in the manufacturing facilities and around football stadiums. 315 00:30:26,891 --> 00:30:32,824 So this is what we will see and this is the main difference between inferencing and training. 316 00:30:34,935 --> 00:30:41,780 You have a bold vision to stand up an internet exchange in space. 317 00:30:41,920 --> 00:30:43,622 So I want to talk about that. 318 00:30:43,622 --> 00:30:46,184 What would it take to make this happen? 319 00:30:46,184 --> 00:30:51,267 And also what business value would that unlock for enterprise? 320 00:30:52,882 --> 00:31:06,166 This is one of my uh favorite projects, probably the most favorite nowadays, as this is my next dream, approaching the stars and introducing DickxSpaceIX. 321 00:31:06,166 --> 00:31:10,307 By the way, there is a landing page for uh the valid audience. 322 00:31:10,307 --> 00:31:18,569 If someone wants to visit uh the page for more information, is space-ix.net. 323 00:31:18,569 --> 00:31:20,760 Doesn't have anything to do with SpaceX. 324 00:31:20,760 --> 00:31:22,224 Please do not mix it up. 325 00:31:22,224 --> 00:31:26,498 It's space-ix, that's for spaceinternetexchange.net. 326 00:31:26,498 --> 00:31:28,219 A lot of information about this. 327 00:31:28,219 --> 00:31:30,161 No, how this will happen? 328 00:31:30,161 --> 00:31:40,739 It will happen and we'll make it happen because we see a demand that is already partially there. 329 00:31:40,739 --> 00:31:51,123 If we look at the landscape uh with uh LEO satellite operators, uh low Earth orbit satellite operators, we do have already three carriers in space. 330 00:31:51,123 --> 00:31:56,025 This is Starlink obviously, have Kuiper, the Amazon project and we do have Rivara. 331 00:31:56,025 --> 00:32:06,388 So they are three carriers with own satellites in space running their networks already in space serving on Earth customers. 332 00:32:06,789 --> 00:32:13,811 Now I learned that first caches from content operators, they have been installed on one of those satellites. 333 00:32:13,932 --> 00:32:15,772 What does this tell us? 334 00:32:15,772 --> 00:32:21,144 That content will also approach in one or another way the space as well. 335 00:32:21,202 --> 00:32:24,344 For one simple reason, it makes perfect sense. 336 00:32:24,344 --> 00:32:36,380 Because in this case, the satellite operators, the Leo network operators, they will be able to serve their customers on earth with much better latency if the content comes 337 00:32:36,380 --> 00:32:39,592 directly from space to their devices on earth. 338 00:32:39,592 --> 00:32:43,314 This is logical, right? 339 00:32:43,934 --> 00:32:50,920 And having carriers outside there, having content outside there, what is the missing element? 340 00:32:50,920 --> 00:32:52,650 bringing all those guys together. 341 00:32:52,650 --> 00:32:58,972 is the Internet Space Exchange as this happens on Earth with us for 30 years now. 342 00:32:58,972 --> 00:33:13,697 So this is the logic we follow and we started first discussions with operators and I asked one of those, is there a chance to have a, Dick explains the Internet Exchange router, one 343 00:33:13,697 --> 00:33:18,098 of your satellites and we start interconnecting and delivering 344 00:33:18,140 --> 00:33:18,760 better service? 345 00:33:18,760 --> 00:33:28,913 Is there a chance to put a Dicic sticker on this router so people, uh enthusiasts, they can see it from where it says yes we will do but this will be the most expensive sticker 346 00:33:28,913 --> 00:33:30,754 you have ever produced. 347 00:33:31,634 --> 00:33:34,995 Yeah so that's the vision but why this is important. 348 00:33:34,995 --> 00:33:38,656 I mentioned latency and then the latency is not the one thing only. 349 00:33:38,656 --> 00:33:44,818 It's about also avoiding the digital divide across the globe. 350 00:33:45,018 --> 00:33:49,542 We do have a lot of areas and the is not an exception here. 351 00:33:49,663 --> 00:33:52,265 We do have lot of areas in the US as well. 352 00:33:52,265 --> 00:33:56,068 But also in countries like Brazil, look at the Amazonas area. 353 00:33:56,068 --> 00:34:03,636 uh Look at countries like Indonesia or Philippines with thousands over thousands of islands. 354 00:34:03,636 --> 00:34:07,359 Look at countries, huge countries like India, Africa. 355 00:34:07,359 --> 00:34:09,401 uh 356 00:34:09,401 --> 00:34:16,546 with different types of landscape, very difficult landscapes sometimes to deploy fiber on the ground. 357 00:34:16,606 --> 00:34:22,650 All these regions, the people are living in those regions, they are disadvantaged today. 358 00:34:22,931 --> 00:34:37,949 And ensuring that they will be connected to the internet uh in an affordable way and with uh appropriate quality of digital services, bandwidth, speed, performance. 359 00:34:37,949 --> 00:34:43,732 of 250 megabit plus this will change their life to a much better one. 360 00:34:43,732 --> 00:34:51,457 uh This is given today for me because digital services are part of our quality of life today. 361 00:34:51,457 --> 00:35:04,225 Without proper access to digital services, uh people cannot enjoy education sometimes, cannot enjoy better type of healthcare services, customer... 362 00:35:04,225 --> 00:35:04,809 uh 363 00:35:04,809 --> 00:35:14,752 patient care, remote patient care, virtual doctor visits, they are uh isolated from even uh finding a better job, right? 364 00:35:14,752 --> 00:35:23,184 I think we need to take care of enabling proper high quality access of digital services everywhere. 365 00:35:23,184 --> 00:35:34,437 And this will become essential for enterprises, for automotive, that the cars are the connected cars and autonomous cars, they do not stop because they do not have internet. 366 00:35:34,658 --> 00:35:46,989 middle of nowhere that robotics functions everywhere disregarding where the manufacturing facility is and fraud detection is available to everyone so the criminals have no chance 367 00:35:46,989 --> 00:35:48,301 to steal money 368 00:35:50,543 --> 00:35:55,428 I just love your examples and your metaphors there that makes things so clear. 369 00:35:55,428 --> 00:35:56,969 I'm learning so much. 370 00:35:57,090 --> 00:36:10,282 I want to talk now about, um I guess you've referenced before emerging energy ideas such as diffusion and other next wave power sources that, you know, as AI workloads really 371 00:36:10,282 --> 00:36:11,043 explode. 372 00:36:11,043 --> 00:36:14,296 So what do we need to know about this? 373 00:36:15,589 --> 00:36:18,790 I'm a big fan of the fusion technology. 374 00:36:18,830 --> 00:36:21,291 This will solve our current problem. 375 00:36:21,291 --> 00:36:30,455 uh Problem with some certain power sources that are not clean enough and they're bad for our environment. 376 00:36:30,775 --> 00:36:33,436 But also they are not available everywhere. 377 00:36:34,057 --> 00:36:37,318 And power uh will be needed everywhere. 378 00:36:37,378 --> 00:36:39,299 We describe the need for training. 379 00:36:39,299 --> 00:36:43,080 We describe the need for even inferencing everywhere. 380 00:36:43,080 --> 00:36:45,681 We describe the need for power in space. 381 00:36:46,307 --> 00:36:47,898 Also very important topic. 382 00:36:47,898 --> 00:37:01,406 um I learned recently during uh a visit I made to Seattle that uh they are visiting one of those companies working on the first fusion reactors. 383 00:37:01,406 --> 00:37:07,109 are uh 37 companies actually around the globe. 384 00:37:07,109 --> 00:37:13,457 Two thirds are, by the way, uh they based in the US. 385 00:37:13,457 --> 00:37:19,159 that they're working on these generators and this or reactors. 386 00:37:19,159 --> 00:37:26,581 These reactors, will change our world in terms of uh also power usage. 387 00:37:26,581 --> 00:37:33,172 Because they uh have the size uh like a human being, right? 388 00:37:33,172 --> 00:37:42,945 They're not big, they're not huge, they can be deployed much easier and they deliver great power density from a couple of hundreds of... 389 00:37:44,172 --> 00:37:56,253 kilowatts up to hundreds of megawatts and even a gigawatt, bigger size then but still unbelievable small comparing this to the power capabilities. 390 00:37:56,573 --> 00:37:58,693 And this will definitely change the world. 391 00:37:58,693 --> 00:38:05,433 They predict to have a market ready version by 2030 to 2035. 392 00:38:05,733 --> 00:38:12,164 So we are getting closer and closer and as we know the latest predictions on the so-called 393 00:38:12,164 --> 00:38:21,564 general AI, this is the human being AI, are also around 30, 35, 2030, 2035. 394 00:38:22,013 --> 00:38:25,957 We see uh it's a kind of a lucky situation, lucky match. 395 00:38:25,957 --> 00:38:37,067 We will get the new source of power uh needed around the globe, needed everywhere, needed in the edge, in the small village uh and in space as well. 396 00:38:37,067 --> 00:38:38,228 uh 397 00:38:39,113 --> 00:38:42,986 in uh match with the development of the AI side. 398 00:38:42,986 --> 00:38:52,704 And this will create the next era of economical industrialization, the era of um AI everywhere. 399 00:38:55,109 --> 00:38:59,796 I'm curious, last question, because I could talk to you for hours and hours. 400 00:39:00,258 --> 00:39:04,555 What does the future workforce look like in this world that you're painting? 401 00:39:05,569 --> 00:39:07,651 It's more, it be more specialized. 402 00:39:07,651 --> 00:39:13,194 I'm not the one who uh believes that AI will destroy jobs. 403 00:39:13,355 --> 00:39:22,451 Yes, we will see a need for a reskill of people and upskill of people. 404 00:39:22,451 --> 00:39:25,103 And I strongly believe in these two directions. 405 00:39:25,103 --> 00:39:32,069 um Reskill people and as a leader, as a manager of a company, 406 00:39:32,069 --> 00:39:44,044 and upscale people because if they know the business you're doing well they will be perfect uh further contribution to this business if you help them to be reskilled and 407 00:39:44,044 --> 00:39:52,277 upskilled and use the AI for the stupid processes they can be done by the agent and not by real human beings. 408 00:39:52,577 --> 00:40:00,974 about talent in general uh we uh need a completely new type of talent 409 00:40:00,974 --> 00:40:12,888 support talent development around the globe everywhere that is more knowledgeable about AI it's not knowledgeable about digital industry in general because this will be essential 410 00:40:12,888 --> 00:40:15,188 for so many different job profiles. 411 00:40:15,188 --> 00:40:27,796 uh Knowledgeable about software I think it's extremely important to support schools and even even preliminary schools uh how they can teach the kids, the students uh 412 00:40:27,796 --> 00:40:28,216 uh... 413 00:40:28,216 --> 00:40:36,631 on on on called on coding on general awareness about the digital infrastructure and processing a i as well uh... 414 00:40:36,631 --> 00:40:43,696 and and we as they kickstarted an initiative just support this in our field of the field of interconnection uh... 415 00:40:43,696 --> 00:40:44,726 i call it uh... 416 00:40:44,726 --> 00:40:51,490 we call it the interconnection academies of global program the the initial start has been done with the uh... 417 00:40:51,490 --> 00:40:54,722 the political university in barcelona the u p f 418 00:40:54,740 --> 00:40:59,423 And we have first modules available already uh online. 419 00:40:59,423 --> 00:41:12,823 Interconnection-academy.net is the landing page where professionals, students, professionals, newbies, even people with more uh seniority and experience uh working for a 420 00:41:12,823 --> 00:41:23,664 bank, working for a telco operator, disregarding, they can pass different modules and classes and get a blockchain-based certificate on their interconnection skills. 421 00:41:23,664 --> 00:41:35,387 with the approach to have a program resulting into a uh publicly recognized um and accepted certificate for an interconnection uh expert. 422 00:41:35,387 --> 00:41:42,789 A similar thing would what companies like Microsoft or Cisco they have been built in the past for their certified engineers. 423 00:41:42,789 --> 00:41:48,917 uh With this we want to support the next generation of talents and experts for digital infrastructure businesses. 424 00:41:48,917 --> 00:41:53,180 I want to invite a lot of partners because it's a very open program. 425 00:41:53,180 --> 00:42:05,458 uh to be part of it, contribute content and enjoy the content, distribute this content further and we are in touch with a lot of universities already around the globe to start 426 00:42:05,759 --> 00:42:09,344 the local uh distribution of this program as well. 427 00:42:11,577 --> 00:42:14,225 How can we reach Deakin's? 428 00:42:14,368 --> 00:42:18,180 Do you have any, like an open jobs page for careers? 429 00:42:18,678 --> 00:42:21,471 Absolutely, we hire. 430 00:42:21,471 --> 00:42:26,086 Yes, we do in all the different markets and there is a global website. 431 00:42:26,086 --> 00:42:27,898 It's easy to remember. 432 00:42:27,898 --> 00:42:32,343 It's de-cix.net. 433 00:42:32,343 --> 00:42:37,188 I hope everybody in the audience understands that I will not share my personal mobile number. 434 00:42:37,188 --> 00:42:41,292 ah But please use the website first. 435 00:42:42,717 --> 00:42:44,399 Oh, Ivo, this has been a pleasure. 436 00:42:44,399 --> 00:42:46,919 I want to thank you so much for coming on the show. 437 00:42:46,919 --> 00:42:51,314 I know I learned a lot and I just can't wait to see you at the next event. 438 00:42:51,622 --> 00:42:53,953 Same here, thank you so much for having me. 439 00:42:54,657 --> 00:42:55,761 You take care. 440 00:42:56,474 --> 00:42:57,509 You too. 441 00:42:57,738 --> 00:42:58,580 Thanks. 442 00:42:59,609 --> 00:43:02,460 Okay, I'm gonna stop recording, don't leave just yet.