Conversation with Dean Nelson
Good day everyone and welcome to another episode of the Momenta Edge Podcast, this is Ed Maguire, Insights Partner, and with us today we have a special guest, this is Dean Nelson who amongst many accomplishments is Chairman and Founder of Infrastructure Masons. Dean has a deep career, 30 years-plus in tech and business, he’s involved in a lot of interesting things, we met last year at part of an event that Dean was organizing, we got to talk and he’s really got some fascinating views on technology, data centers, and the infrastructure of the future of the information economy. So, we’ll just dive right into it, Dean it’s a pleasure to talk to you.
Excellent, thanks Ed, I’ve been looking forward to doing this with you for a while, since we talked!
I’d love to start with some context and understand a bit of your background, what attracted you to technology, and what are some of the formative experiences you’ve gone through that have really shaped your view of the world?
I think the punchline to this whole thing is, I did not choose this career, it found me. I happened into it, I stumbled upon it, and I’ll tell you a quick story around it; in High School I aced all my electronics classes, it was just natural, I really enjoyed it. When I graduated, I thought, okay I’m a smart guy, I will go out and get a job and everything will be fine, I didn’t decide to go to college etc. So, I went and joined a group called Manpower, they had me do all the jobs no-one else wanted to do, so I spent a year doing things, making no money, that were just not fun and I realized, ‘What am I doing?’
So, I went back to talk to my dad, and he said, ‘You’re really good at electronics, maybe you should explore that side’. So, I went to a trade school, I went to the Brian Phoenix and graduated in two years, then Sun Microsystems came to the campus and hired half my graduating class. So, I started in the Silicon Valley on my 21stbirthday, July 10th, 1989, I had no idea what Silicon Valley was, I never had touched anything when it came to hardware; I knew electronics etc., they taught me that, but I learned everything on the job when it came to it.
Going into the Bay Area, I’ve done 30 years now in the Bay Area with just some incredible companies, Sun Microsystems I did 17 years there, and I believe I went to the University of Sun. I learned technology, I learned business, I learned strategy, I learned management, because of the experience there, it was a great environment. In the middle I also went to a startup and learned a ton there for about three years, called Allegro Networks, then I left when Oracle acquired Sun and went over to eBay where I ran a global infrastructure. Then I took a six-month sabbatical, then joined Uber about three years ago, and I’ve just recently left there. I actually made the change to be out of the corporate world, and back into strategic advisory and consulting side on my 30-year Silicon Valley anniversary, that was July 10ththis year.
So, it’s been a really interesting ride, and watching the technology from the beginning, at least 30 years of it in the Silicon Valley, it’s been absolutely incredible.
I think what’s really interesting about the term infrastructure is it defines that technology has become foundational to our society, to our economy, and I’d love to get your perspective from when you started; how people were thinking about technology, and the way you saw the skillsets and the awareness of technology evolve through your career, because being in Sun I think that’s just about as close to Ground Zero of the real information revolution as you’re going to get.
Yes, they put the dot in .com, and if you look at the boom that happened before 2000, before the bust when the bubble burst, it was incredible to watch this more than a decade-worth of growth in an emerging space. I remember my first web browser, Mosaic, and I remember my first email, all these things that happened in ’89, ’90, and ’91, it was all starting and being a part of that. And now when you fast-forward, it’s all common, everyone expects it.
Here’s an interesting thing, I started in Infrastructure Masons which is a professional association, it’s about individuals and it’s the builders of the digital age. These are the people that do this underlying digital infrastructure work, that makes the Internet of Everything happen; without that foundation the world doesn’t actually operate, and 99 percent of the world has no idea that this exists, they just expect it to work. So, what we wanted to do was bring together that group of professionals that drive over $150 billion worth of infrastructure around the world, I’m talking digital infrastructure, data centers, network, hardware, the software that manages these things and brings them together in a common environment where you don’t have sales and everything else. We leave our companies at the door, and we share experiences, they build that network, and they connect, grow and give back.
So, that’s the community we started. But if I look at today versus 30 years ago, and I look at the university systems, any of the colleges, etc., where they’re sending students today is not into the foundational side, everybody wants to create the next path, they want to do the next start-up, but there is so much need and so much growth underneath the foundation that makes that happen, there’s an awareness problem. Thirty years later people are happening into the same job as I did, so we want to change that. So, as an industry and as an association we are creating an industry awareness campaign to help colleges, universities and that, to understand what they need to teach so they can attract it onto our field.
It’s interesting to me that 30 years later the same problem exists, people have to happen into this industry. That’s a problem because like baby boomers and everything else, all the retirements are coming, we’re going to have a talent shortage, and all of us fighting for the same people in all these different areas, and the infrastructure is not slowing down. We are building stuff everywhere in the world, I’ve never seen this rate of growth of capacity going on right now, it’s explosive.
It’s so interesting. You’ve touched on this broad array of skillsets, I’d love to hear your views on some of the key skills that are involved in building out infrastructure, because my conversations with you and some of the other members of the Infrastructure Masons, what has struck me is just how comprehensive and how multi-disciplinary the necessary skills are required, to be able to build these giant datacenters that have to be operating 24/7. It’s just an enormous task.
You know, I think it’s under-appreciated! Everyone expects it to work, but they don’t really understand how much work goes in to make it function, and they only realize it once something goes out, when there’s the blackout on the East Coast, and the lost power, all of a sudden the Stock Exchange gets impacted. When you have a global outage and you’re trying to pay for something, you can’t, or you can’t log into Facebook, and just the basics around it you realize there’s so much underneath there that has to be built, so, the disciplines you’re talking about.
First-off, I do some charity work in India, and one of the things I was really proud of was I went back there earlier this year, and I taught a digital infrastructure one-on-one course to all the seniors, juniors, and teachers of the school my mom and I built about six years ago. What I did is I started with my phone, I was in the corner, I took a picture of us like a selfie of the room, and then I did a WeChat of that picture to somebody else in the room. So, I asked all these people in the room, ‘What happened?’ they’re raising their hands, ‘You took a picture and then sent it’, and I said, ‘But how did it work?’ We proceeded to now follow what happened on the phone, when it went over the ether into a cell tower, the cell tower to the fiber, the fiber to an internet exchange, the exchange to an edge. The edge goes back to a core data center, it hits the server, the front-end server it gets stored on there and gets processed, and then is transmitted back the other direction. Very interesting. I said, ‘Now, follow that path, think about it. Every step in there are thousands and thousands of jobs all over the world, and its only growing because all demand that’s being generated out there will only put the need for more capacity in all the places where people are. So, it’s a great career and a really interesting problem to solve’.
So, you talk about the disciplines, think about it, a data center engineer, well there’s engineering, design, construction operations. How about a network engineer who does stuff in the LAM, the folks that are interconnecting the metro, the ones that do the global WAN aspects of it, so there’s subsidy cables of fiber under the Ocean that connects everything. Then you look at the hardware that’s in there that does compute, everything from standard compute to storage, into machine learning, all the stuff that’s going on with AI. Then you start to sample the software that has all their stuff function. All those are different disciplines.
Then you’ve got the business side too, the financial analyst saying, ‘What’s the TCO in infrastructure and deployment, what investments do we make?’ To the sales and systems engineers who go back and make these systems work, from the partner side into the people that buy it. So, there’s literally hundreds of thousands of jobs all over the world on the foundational infrastructure. And here’s the thing that I believe is a risk that people haven’t really seen; today in core datacenters in all these major regions like Ashburn, Virginia, the Northern Virgin Market is one of the largest, if not the largest datacenter markets in the world. Then you’ve got the same thing replicated in different parts of the US, all over Europe, and are starting to emerge in huge places than APAK. Then you’ve got emerging things like Latvia and Africa, and they’re building up more infrastructure so that the other half of the population that’s not connected, will get connected.
Then you have technology like 5G, and machine learning, augmented reality, virtual reality, etc. that will start to now boost it again. My prediction is that the edge market will dominate, the wave will be bigger than all of the core datacenter deployment that have every boom deployed. So, how do you fulfil that demand, and how do you have the talent when you’ve got this extremely rich, robust, and growing segment based on the world’s population’s consumption of data, and processing? We need people in it. And Infrastructure Masons when we start talking about all these advisory council members, they have the same challenge, I have a hard time finding the actual people.
Absolutely, and I was struck when I spoke to members of your group, just the absolute depth and breadth of expertise. It’s unique because you have an understanding of real estate, and property, and citing and power, not to mention of course the understanding of the technology in business as you’ve mentioned. Where do you feel… maybe we start in the US, what are some of the educational areas, or shortcomings you see that you think may be hampering the development of some of the skills that are going to be needed?
We have these academic institutions that are well-turned machines, you look at what their charter is, great curriculum that allows students to get the skills to be able to learn the job, simple as that. But it takes years for them to develop the programs curriculums, and degree programs, that can be associated to those jobs. I think what’s happened is, the technology industry, and what’s happened in digital infrastructure, is adapting and changing so fast that what I think the schools are missing is the picture, the bigger picture of what is digital infrastructure, how do the pieces fit together, and then what are the programs that will align to them.
On the flipside, first up, what jobs are being created, and then what jobs are in demand. So, think of a heat map, but what we lack today is we don’t have a taxonomy or a job ladder about all these different careers. Then from there if we had a job ladder we could create heat out which companies are hiring, and in which areas, and that can tie right back into the programs that these colleges, universities, and even trade schools to say, ‘Okay, I can see that demand’, because the schools just want the same thing. ‘What do you need filled? I’ll build the programs to do it’, they’re very good at it, right, but they’re lacking that demand. What is it?
And so there’s a misalignment there that we see, and it’s not one thing, it’s not an Ivy League school, or a university, or a community college, or a certification program, it’s all the above; they all need to have insights into what this underlying foundation is, so they can create the programs that create the jobs, that land their students in it, in rewarding lucrative careers.
Who do you think would be the appropriate party to help create this taxonomy, and drive this awareness as you’ve articulated? Is it going to be companies do you think, or do we need some government involvement? I think with the challenge of a lot of people in the public sector are very well-meaning, but they’re just not that up to speed on technology, there’s a pretty steep learning curve.
Yeah, I look at this as a supply and demand problem again. I do believe the ones that are hiring should help create the taxonomy, the job ladder of what do they need, it should come from that side, the people who are going to hire. Then that should align to the professionals who create programs that will help align for those needs. So, for example, I’ll give you a really simple use case; one of the hyper-scalers right now is deploying 40MW a month, 40MW a month of new capacity. So, when they go into a region, just like they’re building our data centers that are anywhere from 10 to 20MW each, and whether it’s their own or with their partner etc., they have to build more capacity. So, when they go into a region, they literally hire every electrician to go in, so they get pulled off all the other projects, they don’t have enough electricians. Then the same thing when it comes to mechanical engineers, then architects and others, of how do you put these projects and turn them into final things.
So, when you look at capacity that’s coming in, I think there’s a demand from all these people that need to hire, and then there’s the mapping to what they’re hiring, and then to the programs who have to come out of those schools. So, I think there’s a joint thing, for example, there’s a number of hyper-scalers which are working together to create this taxonomy, and we’re participating in that, that job ladder, so that it can be better informed to schools. You think of the Department of Labor and others that are there, I think they can help enable that and support that, but they can’t lead it because they have the same learning curve. They really need to say who’s hiring and what are they hiring, and we need that funneled in that direction, so the people who do education and academia for a living can do what they do best, align programs to a demand.
We’ve had some other discussions previously about the lack of flexibility, the undergraduate system for instance. We’ve got associates degrees and master’s degrees, and the two- and four-year programs that are the pre-requisites for a lot of jobs. But on the flipside the actual skill-sets people use on the job tend to be acquired in a much more dynamic way. Are there some discussions that you’ve had, or thoughts you have, about how it might be possible to craft a more adaptive approach to upskilling people, and being able to put them into the roles, and fill that demand more rapidly?
I don’t have all the answers for this of course, none of us do, but we do see there’s trends and opportunities emerging, so I’ll just give you a few thoughts around it. A trend; Google and Apple, and a few other large companies are no longer requiring a four-year degree to be hired. They’re looking at the potential of the employee, and they expect that when they come in they’re going to have to train them. So, imagine a High School student that’s done a bunch of CS work, and is able to code and get to that certain level, then they can come in and get a job and be mentored by the people there, and learn skills that will be more applicable than they would have at school. So, the trend is changing a bit.
I look at that as, in the past all these trades they used to have apprenticeships, and what did the apprentices do? They came in and were learning from masters, those masters are passing down those skills. There’s a similar thing here with these different companies where internships when they come in, they’re freshmen, they’re usually not given this granted access, but sophomores of juniors and seniors are, because when they come in now, they’re given real projects.
In a previous life in my previous companies I was hiring interns, bringing them in we put them on production projects. They go back to school and are all excited because they did something, for example at Uber, or at Google, or at Apple, and they’re doing things that are really out there, true real projects. Then they come back and are excited for the next one, but they learn real experience and ask the right questions when they go back to school, for their next semester and term. Then they go in and they go to another internship program, and ultimately that’s a recruiting ground for a trial by the companies to say, ‘Great, these people fit in the culture, they have the capability, the potential, and we want to hire them’, and they convert them into full-time employees.
But I will tell you, there’s one challenge in all that, I do think the internship program is incredibly important, but the corporate culture, or the corporate structure in a lot of companies today is based on headcount. So, one of the limitations that happens is when you say from an accounting standpoint, or a finance standpoint, ‘Oh, you have a hundred heads, you can hire a hundred people’, ‘Okay, great, well I’m going to do software development engineering on this one, I’ve ten in each’. ‘Well, I need to have more bang for my buck, I’m going to go hire senior people’, and so it shuts out a lot of the recent college grads and others as opportunities. So, instead you look at it as, ‘Well, I can hire ten college students for… and still have five senior people’. I can do 15 hires and get more capabilities, and they could mentor them, but there’s an economic problem of the companies and the way they’re structured from a finance standpoint.
So, that has been an interesting thing, it’s almost like a forcing function, the executives have to drive down to say, ‘You know what, calls hiring is a priority. I would say I saw this at eBay where John Donahoe when he did e-Bank, when PayPal and StubHub media were all together, he set a program in there to say, ‘We will grant you budget, headcount budget, to ensure there’s going to be 20 percent new college hires in the company. It’s that level program which really will drive this innovation to give opportunity for these students to come in’. I love that because I went over to UIUC with him, he was the executive sponsor and I was a campus captain, and we were there trying to recruit more people to come into eBay. It was incredible, the talent that came out of that, because we were given the leeway to go back and hire more people than we would have if we stuck to a specific headcount budget.
So, there’s some business elements of this one where there can be some strategic moves by executives at companies that can enthuse so much of this creativity and innovation, by having a good balance of senior mid-level and entry level engineering and management talent.
That’s got to be a big shift for a number of companies. I think you really put your finger on this issue of culture, and it’s funny because I think of Oracle which at the time, they acquired Solaris, or Sun, they were not necessarily considered that forward-thinking in many ways! But they’ve been quite active in recruiting on campuses recently.
I’d love to just go back to some of your specialization, and your experience, and get a bit of perspective on what does it take to build a hyper-scale data center, I mean since you started you’d stick a bunch of servers in a room, and you’d have to buy double what you needed, so you’d have some fail-over. But so much has changed in that time, but the demands have gotten so much greater. What are some of the challenges and processes involved with deploying? I don’t think even the term datacenter was in the com and lexicon 20-25 years ago, so this is a new thing, and I would love to get your perspective on how its evolved, and what goes into creating one of these massive information factories, or information utilities.
These data centers yeah! That’s a really broad question but let me break it down into some very simple things. Back in 1989-’90, think about it, you had the wiring closets, you had server rooms, you had a few access things in there and they housed together. Then as the companies kept growing, growing, and growing, they started building these larger configurations of systems. I remember CoLOS where you would have desktops or deskside servers sitting in a rack somewhere, then it started migrating towards you got mainframes and everything else, then you got distribute computing to replace it, then you had these hyper-scaled-type growths happening. So, for example, when AWS started 2005-’06, maybe ’07 range, they started looking at how do you take this capacity and build-up lots of it. So, a decade later it took for cloud to get adopted.
Well this was grid computing before, then it was load sharing facilities etc., then it turned into cloud, but all of it has the same underlying elements. I think what I learned in the last 10 years is, there is a recipe and a model of how you go out and build hyperscale infrastructure, and a lot of the same common themes, not that I have all the answers to it, but there are common themes across many of these different hyperscale companies. So, the first one is, whether you’re on cloud or on premise with the data centers you build, you’re going to create a region, and that region is basically a configuration of zones, and those zones are really a configuration of sets of hardware. So, for example in a previous life we would put three zones in a region, each of those zones had 480 cabinets, and it had a five-year life.
I would roll equipment in, fill up those 480 cabinets of a mixture of different types of equipment, and then three years later I would be refreshing that equipment in place by just pulling out a whole rack, and rolling another one in, and I would increase 1.5 or 2x of the actual power drawing computer pace capacity, but I would do it in the same zone. Then, when I go to the end of that zone, I would replace the entire zone with a new one. So, we would do these 5 to 8 MW chunks of anywhere from 15 to 35,000 servers per zone. There’s a recipe here of how you design the power, the space, the cooling, the compute, the network fabric that goes into that zone itself, the Metrolink that connects all those zones, and then the actual WAN that connects these global regions together.
So, that recipe whether its cloud, AWS, Azure, GCP, Oracles Cloud, or its on-premise like what happens at Facebook, Uber, and Apple, they follow a very similar model. It’s all about building hyper-scale standardized deployments that are rack and roll capable, that you can say I can drive down the cost and increase the efficiency. And by the way the magic that happens here is, when the software is at a point where it is more resilient, you can now start to remove the redundancy in the datacenters, and the hardware deployments. Does that make sense?
It does, and that was my next question, which was how you can ensure this resiliency, I think you’ve just make that point, but earlier we were talking about grids going down, and interestingly we’ve just had a reminder here in New York with a blackout, we had another one 16 years ago, but with any essential service, any use essential infrastructure, you have severe repercussions of failure. What are some of the considerations that go into making infrastructure truly resilient from no knows, and unknown knowns?
Yeah, it really goes down to that architecture again. The resiliency, if you are designing for a failure, and when I say failure, you can say my lowest common denominator is a rack full of equipment. That entire rack should be able to fail. Then my next level is a block of those inside of the zone, and my next one for there is an entire zone, and then from there it’s an entire region which is filled with free zones.
So, think of it as, I’m going have a compartmentalized failure, lowest common denominator in the rack, then a group of racks, then there an entire zone, and then a region. If I design the architecture to say I will have faults like that, how do I dynamically basically recover from those faults? So, architecture principles have come into that, from a technology standpoint; I will have three zones, no more than two milliseconds away from each other, so that for example if I’m going to go right to a database, I’m writing the three parallel databases, and one in each zone, so if I lose a zone I don’t lose any data. Then if I lose two zones, I still don’t lose data, but I have a risk and I will fail over an entire region at that point.
So, you design the principles to say, I expect failure, I plan for failure. If you do that, imagine if I had one datacenter with everything in it, I’d have to have all this power, cooling, and network redundancy, really-really high level of what they call tier 4 deployments. But if I now have three datacenters what are working in parallel, they’re working together, I can have a whole one fail, I can make each of those less resilient from the physical infrastructure. I will tell you there are some of the hyper-scalers that have really taken this to the next level, they have single-corded servers, they have no cooling – it’s all using outside air, they don’t have any UPSs, and some have straight street power. Why? Because they’re on one grid with one zone, a different power grid with another zone, and they can have open sale and it’s okay, it will drop, the others will cover, it will recover, and then they can still distribute it across the three. That architectural strategy is what enables hyperscale and drives the unit cost down to be extremely efficient for the PCL.
And it’s pretty amazing some of the locations in the world where we’re starting to see some datacenters and opportunity.
Yeah, and the emerging markets right now, what’s mature? Europe; Europe and America have probably been in this space the longest when it comes to digital infrastructure deployments, then you’ve got all of the emerging things in China and India. China is going to build more datacenters this year than the entire rest of the world combined. One country. Then you look at India; India is bigger, it’s hugeand there’s so much need there because guess what? There’s another billion people who are continuously coming on and consuming more, and so that is playing out all over the world, and the places that have infrastructure already built, just like you have road systems, transportation, electrical grids and water systems, you have digital infrastructure there, like the US, Europe, and other parts of the world, it already is built for it.
But the ones that aren’t, there’s a huge amount of investment and growth underneath to build that foundation to enable energy, to enable connectivity, to enable compute and storage, to get back up to where the rest of the world is at. Like I said earlier, half of the population is not even really participating in the digital age yet, so 3.5-4 billion people, wow!
Massive opportunity. You made a comment earlier that I would love to expand on, which is the role of edge computing. What do you see driving adoption of edge computing, at least from used cases? I’d like to get into some of the technology challenges and opportunities that come from expanding your logic, and architecture, to incorporate edge computing.
Again, I keep talking about some merging space. My prediction again is the edge deployments even though they’re going to be smaller, there’s going to be so many more of them, and so much demand it will actually be larger than all of the core infrastructure over the last decade, so edge will dwarf what’s going on in core. It’s not meaning the core like these big datacenters are going to go away, it’s all going to grow, the tide is going to lift all boats in this. It’s just that today… I’m going to give a use case, gaming – Fortnite; anybody that’s doing virtual reality, when they have these games that’s built into the phones that have AI capability, I’m talking about chips in the new phones that are coming out, they’re going to be driving more data which requires low-latency performance. Imagine, if you’ve got a fully immersive virtual reality experience like oculus glasses on etc., which are fully covering your eyes, you’re going to be doing 240 frames a second. So, if you don’t deliver 240 frames a second and people are looking around, their equilibrium will get messed up, they’ll pass out, or they’ll throw up.
Pokémon Go, when it first came out it was this global amazing thing to watch all these people running around the city, chasing the virtual Pokémon all over the place! But the demand that came from it, why – because all these people wanted to participate. What happened was, they needed to find 40,000 servers, the ones which were serving us, 40,000 servers, that happened overnight. Why? Because the world’s population said, ‘Pokémon Go, let’s do it’. That’s with the technology, which was available then, the network that was available then. So, imagine if we have 5G open up, and all these devices have 100 times more bandwidth, and they have phones that are able to operate at low latency, high performance-type things, the demand is going to come in. What that’s foreseeing is that the core datacenter like having a phone that will connect back to an exchange, it goes back to a datacenter to do processing, will not be enough. It will be the people with all these devices, which is all of us, with the demand, because every time you get a phone you realize you’re using all the space and the processing power, so you get another one and they think, ‘Oh, I’ll never use that’, you use it all, it’s there – and then you multiply that times the population.
That’s going to drive more data, more real-time need and low latency expectations, so that to me is what’s really going to push the edge deployments. When I say edge, there’s a Micro-Edge and then there’s the edge that would serve a city, but Micro-Edge in every building, everywhere across every major city. Why? Because it doesn’t matter if you’re in Ohio or New York City, you have people who are still wanting this experience, the concentration thing. Well, there’s going to be an edge need in all of them and that’s a huge emerging market. So, how do you deploy out there? There’s a last mile element on this, I’ve been working on it, it’s really-really interesting, there may be other players that don’t participate today in this space that can. Why? Because they are the last mile, and it’s not the telecoms and the clouds etc., it’s all these other people that own real estate and are in places that want to participate, because guess what? I’m on this corner chasing Pokémon, or doing a virtual reality, or augmented reality experience, and I need to have my compute within that same block. Where do I get it? They are not there.
And when you bring that up I immediately think of what happens when we move into a really broader-based adoption of autonomous cars and mobility as a service, how do you see that affecting the need for infrastructure and edge computing, and some of the considerations that we’ve got to be looking at right now for this to become a reality?
I learned something the past couple of years that changed my mind. Think of autonomous vehicles, and everyone is thinking, ‘Oh there’s so much data generated on autonomous vehicles, and so much need of that data to be off-loaded. We’ve never seen that much data, terabytes of stuff coming off of every car’. The reality is this, they’re autonomous so they can, if everything works the way they’re designed, accommodate 95 percent of the things they need to do within the car. This isn’t like I’m streaming data back and forth to these different things, I have small snippets of data that are going back. But if I’m in an autonomous vehicle driving the street, and they’ve got LiDAR that’s doing the 3D modelling in real-time, they’ve got cameras that are watching and are doing, ‘This is a cat’, ‘This is a telephone pole’, ‘This is a person walking across the street’, but it’s doing that processing within itself. So, 95 percent of that data is going to be the same thing, and now we get hundreds of cars driving that same street, the data is going to get turned away, ‘I don’t need it’, what you need is a delta. ‘I just saw something that’s different, that I couldn’t understand’, and then these snippets.
So, imagine if there’s an incident and you need to be able to get the previous 90 seconds, and the next 30 seconds after the incident. Why? So, you can go back and interpret what happened. But incidents are what are really mission critical, that will go for cellular some other close proximity Wi-Fi to get it out, but it’s not a massive amount of data. There will be a large volume of these cars, and more stuff will be streamed, and they’ll start to do a mesh with each other, so data will go back and forth, but it’s not huge amounts of data that’s going in.
I think there’s a misnomer that autonomous vehicles are going to eat the edge when it comes to bandwidth, I think there’s going to be lots and lots of devices out there that will eat that bandwidth, but it’s not necessarily autonomous vehicles. They’re going to participate in it, but like I said, the data that’s on there the majority of it you don’t need, the big stuff is video, the next one is Lidar, and the one after that is the sensors and the cars, so you want a snippet of all that for an incident, but not 24-hours a day, at terabytes a day.
Yes, that makes a lot of sense, but when you were talking about augmented reality which Pokémon really was the first breakout augmented reality application, are there other potentially use cases, or demand catalysts, or traffic catalysts you see that might also have a huge impact on surprising people, in terms of the need to build out capacity quickly?
The thing that’s emerging, and again there’s more and more of this popping up, smart cities, everybody’s talked about that, IoT all the devices around there, but really, I think the one that’s emerging is still gaming. Gaming does have a need for high bandwidth, low latency, near real-time experiences, and that will drive the need. Then as soon as 5G opens up you’ll have allof these creative minds coming up with new ways to use it; did anybody envision a Pokémon Go a year before it actually happened! Did they expect it to blow-up that much? No, but how many more of those are going to come, because they’ve got a hundred more lanes of bandwidth, and their performance from a latency standpoint even better. So, again that will drive more capacity at the edge that’s needed, versus at the core.
We don’t even know what’s coming out, the trends like I’ve said, really is gaming and virtual reality/augmented reality, and most likely to use the large mesh networks that a mission-critical thing will need that. But it’s the sum of all those parts that’s going to increase the bandwidth consumption, and the compute needs of the edge, not any single one.
Let’s look out over the next decade, how do you see the market evolving, are there any constraints or opportunities you see emerging up ahead as we start to see these continual forces of demand, and expansion, driving need for more and more capacity and bandwidth?
To me it comes down to concentration of people. One of the coolest things I saw at Uber was Uber Elevate, so imagine, we’ve built out the infrastructure with roads for people to connect, and highways and freeways etc., to be able to get more people to more places. Then we realize we need to have more space, so we created high-rises, those high-rises now concentrate more people, yet we have one dimension of roads – it’s horizontal. So, now you look at people drones that are going to happen, this is Uber Elevate, you’re going to be able to call this thing, and instead of doing a two-hour commute back and forth from San Francisco to Saratoga, its 15 minutes. So, the concentration of people into cities at these peak times, you’re going to have all these different levels, just like a high-rise you’ll be able to have seven layers of people drones going back and forth, they call them vertical takeoff and landing devices. They start at one point, go up, go over, go down, people get off, people get on, and they continue to repeat this, that’s the simplest autonomy problem to solve, and it’s also controlled airspace etc.
So, that’s going to happen all over, and then that’s going to concentrate more people in these cities. It’s also going to now expand where people can live. So, if I can live two hours away if I had to drive it, and be 20 minutes by Uber Elevate, then it affects real estate, but it will now have more concentration of people in the cities, that will increase demand. So, really its high population of where people are that will drive edge even more, and this new technology which is now allowing us to get more people and faster, and make their lives more efficient, they’re all going to be driving more demand.
That’s pretty incredible, just thinking of the implications are pretty mind-boggling. I’d love to ask, broadly looking forward, what are you most optimistic about, and what are some considerations or concerns that keep you up at night?
I’d say what I’m most optimistic about is, we’re living in such an incredible time right now. Think about it, we’re about to experience the Jetsons, it was on television, George Jetson flying back and forth from his job, we’re going to have that with these Uber Elevate type things, that’s going to continue to go, whether it’s Uber Eats or Amazon deliveries, I think our delight with technology and our efficiencies is going to continue to go up, and up, and up.
The thing that keeps me up at night I think is the underlying infrastructure to feed that demand, is one, but I also look at it from the sustainability standpoint, we’re going to have more, and more, and more consumption, and if we don’t get our arms around this, the global environmental impacts of this, because we’re going to need more power, we’re going to need more compute, we’re going to need more connectivity, we’re going to have to do it in a sustainable way. At Infrastructure Masons we have core pillars that we care about:
- Industry awareness. We need people to understand that this industry exists, and that there’s rewarding and lucrative careers in it.
- We need an education system, an education focus to help drive more people into it.
- We need to have more diversity, and right now we’re focusing on gender, and continue to expand that out to other under-represented groups, because less than 10 percent of our workforce in the infrastructure space, digital infrastructure, are female.
- How do we as an industry who have a responsibility as we build out tens of billions of dollars-worth of infrastructure, and drive these things that are going to be 10 to 20-year assets, how do we do it sustainably? How do we drive where there’s an economically beneficial deployment of infrastructure that’s 100 percent renewable?
These are the kind of things that we have to do as leaders across the world, and how we build up sustainable infrastructure. That’s what keeps me up at night.
That’s many life’s work ahead, that’s really an amazing opportunity ahead, its great to hear that you and Infrastructure Masons are pursuing this. Hopefully people listening to this will have some more awareness and direct the young and promising students into the industry.
One final question that I always like to ask, which is whether you have a recommendation of a good resource or book that you might be able to share for our listeners?
Ah, three books come to mind for me because I’ve just read them, and they’re amazing. One which you have featured on here which is, Julie Albright’s Left to Their Own Devices.
I couldn’t agree more, she’s amazing.
Yes, and she’s on our board for Infrastructure Masons, and it’s such an amazing perspective for us to really understand. That book I think is worth reading, especially for baby boomers or Gen X etc., do you understand this new workforce with all this capability, how they operate? Because you can’t apply to this generation the old principles of how you do it, they’re digital natives and they move around, they care about purpose and cause, they’re looking for a community, they’re looking for a place to be able to belong. So, understanding that is a critical piece I think for everybody. If you’re going to manage it you need to understand the companies have to adapt this generation, whether its millennials or Gen Z, to be effective for them and for the companies.
Another one for me is, AI Superpowers. This is a book written by the top AI researcher in the world, Kai-Fu Lee, and he talks about the battle between Silicon Valley and China for AI dominance. It is fascinating, the amount of data that’s in there, and when you look at where China was three years ago, just three years ago, and where they are today; they were laggards. And now they are literally submitting more than half of the actual white paper into the AI congress. That is huge, the amount of investment that a country is doing. So, I think it’s really important for people to understand both the pros and cons about AI, because it will be here no matter what, and what does it mean to the world.
Then the last one was written by John Chambers, it’s called Connecting the Dots, I think it is one of the best business books out there. I happened to be at dinner with John when he was signing these at the Bloom Aspire Conference, he was a mentor of John Donohoe, the CEO of eBay Inc, and now is the CEO of ServiceNow, and John Donahoe was a mentor of mine. So, I went over and said, ‘I just want to say thank you because you’re a mentor of a mentor of mine, it was just an amazing little moment. But his insights of running Cisco and his philosophy of how he approaches it, and what he did with business playbooks, they are really impressive. That book, Connecting the Dots, is something I believe every business professional and technology professional should read, it will give you that full perspective of what it means to operate in this digital infrastructure world.
Wow, those are great recommendations.
Dean this has been a fascinating conversation, and I actually have to put in my own recommendation, a little plug which I know you probably wouldn’t do, but your daughter is a fabulous singer, and her band Citizen Queen, I had the pleasure of hearing at Madison Square Garden. I’m going to put that in the show notes that I think everybody should check them out. They’re a wonderful-wonderful acapella group of very talented young women, I wanted to just put that in there!
Thank you for that yeah, I’m a proud dad, I’m biased but they’re good!
It was really terrific.
This has been another podcast for Momenta Edge, we’ve been speaking with Dean Nelson who is founder and chairman of the Infrastructure Masons. Again, it’s Ed Maguire, your insights partner at Momenta Partners.
Dean, thanks again, it’s been just an absolute pleasure.
Great talking to you, thanks Ed.