We are a company that likes to be able to create the opportunities for people to connect with each other, and connection can be the absolute mundane, or the absolute profound. It is the daily experiences that people have in their lives. And one of the first things is about how to share experience. So, about 20 years ago, maybe less than that, in the developed world people used to communicate by text and voice. And if someone had an experience they wanted to share with somebody they would just write it in text. Here is an example: “You would not believe the acrobats I am watching right now at the circus”. You might call somebody, or you might text that and that would communicate the emotion, the whole experience of what that moment was like for you.
That is a 2G network. As you go to 3G networks, you can start to add photographs to explain that experience. With LTE (Long Term Evolution) you can start to show video, you can do it live, you can pre-record it, send it over YouTube, things like that. You start to get more of an enriched experience to get a stronger feeling of being there. And that is one of our aims around what connectivity means and why we care about the progression of networks around the world. Finally, we are hoping that, ultimately, this leads to concepts of virtual and augmented reality or rather than showing a video from a particular perspective, you can have any perspective. You can feel that space and time, that were once separate, are now able to be transversed very easily. You can go to any place or any time.
That is why we care about networks getting better and better over time, because each stage requires more and more connectivity. But there is another aspect of why we care about this, which is completely orthogonal, which is around creating community. We really, deeply care about community existing everywhere. That can be in cities, in developed markets; it can be in rural communities of 400 people, in an emerging market. Each one of those communities can be focused on employment, health, education, entertainment, financial and government services and applications. These are just a few, but ultimately we care about every community around the world being connected because we think this is the best and most scalable way for those kinds of experiences to be shared.
So this is a little cartoon map, what we focus on are a range of technologies and business models for thinking about how connectivity can happen. And what it takes to connect people in cities is very different from trying to connect people in rural communities. And I will try to get into that a little bit, but we believe there is no silver bullet technology that would exist to connect everybody in the world. You have to look at the specific use cases, the specific demographics, not in excruciating resolution; there may be five or ten different technologies or business models that are required. We do think there is more than one. I am not going to get into all of those today, but that is just one of the aspects of our philosophy. What I would like to get into is some of the data that goes behind some of the work that we did. So, the real interesting insight, at least for me is being able to see a World Wide Network of 2 billion people active on a monthly basis. This reflects our network, this is 2 billion monthly active people from the perspective of telecommunications. The yellow parts, that you can see, for example in the United States, parts of Europe, Korea, South Korea and Japan, Australia; those represent LTE networks. You know, peak capacities range from 100 MB to, hopefully soon, 1 GB of speed. Then, at the bottom you see 2G connectivity, and large parts of the world, southern parts of the world are still very much 2G and 3G connections. There is roughly 25 years of span of technology development between 2G and 4G, and roughly, in terms of peak data rates, 1000 times of speed differences between the most connected users and the least connected users. So, while we focus on the unconnected to the connected populations, we still want to be cognizant of the fact that three orders of magnitude of speed differences can exist between somebody in the United States and somebody in India.
Let us pay attention to India for a second. This map was actually generated in June of 2016 and I would like you to pay attention to India, because over 9-month period, subsequent to when this image was generated, something pretty miraculous happened. To me it represents an example of a market success. So just watch India, and I am going to switch to what happened in March of 2017, of this year. It switched. It went completely 4G in a matter of 9 months, based on the entry of a new competitor in the market named Reliance GIA and they have connected over 100 million people in a matter of months, providing affordable LTE access to the entire country. So I think this is to me an example of really positive, rapid change. Let us take away the specifics of who is involved and what they did and things like that, but just to take away the example that rapid change can happen if there is determination to get it done. I think an example, a counterpoint, are perhaps market failures. There have been conversations around public market failures, and I would like to point out private failures as well. This question around rural acts, if we look at sub-Saharan Africa, the cost of backhaul access is one of the primary drivers for why rural access is not sustainable for private enterprises in sub-Saharan Africa. In the United States, in California, I can get 1 GB of access to my home, I can routinely get a peak capacity of 1 GB and it costs me $40 per month to do so. Now, this is not quite apples-to-apples, but 1 GB of access anywhere in sub-Saharan Africa costs anywhere today between $500,000 and $2.5 million. That is an enormous gap that exists, that is not driven by any regulatory environment, that is just the state of competition today.
I am going to switch gears and talk about a question that we wanted to answer at a fundamental level, which is where are people located? We wanted to understand, from our connectivity datasets, could we look at a difference map between where people are located to understand where unconnected people were. And so Columbia University (the Earth Institute), Jeffrey Sachs is here, has created these wonderful maps worldwide. And, there is a database called Grided Population of the World and they are at Version 4 and in some cases the resolution of these census maps goes to 1 square km, where in each one of those pixels at 1 km around the globe, you can find out what population density is. And so we were really interested in this as a perspective to understand, could you do network planning? Could you understand where resources are? Can you think about what the cost of connectivity would be for the unconnected people of the world?
And this is an example in in Naivasha, Kenya, and the data from CIESIN Columbia is on the right. Those are 1 km square pixels. So, from a network planning perspective, you wanted to think about what is the cost of connectivity to connect this community. You wanted to do it via means other than satellite, it is very hard to predict. If you look at the image on the left, which is a satellite image, your brain is able to parse and see that there is human development there at much higher resolution than the image on the right. And so we asked ourselves this natural question: could we train computers, using machine learning, to identify human artefacts in that picture, to be able to understand population density at a much higher resolution that is done in the image on the right? So we actually collaborated with Columbia University on this because they have these global census databases, this is a collaboration with the World Bank and local governments, to see if we could combine the best of machine learning of images on the left, with the census databases on the right, to get much better resolution, to be able to understand some of the sort of systematic problems of thinking about what would it take to do connectivity. And so this is an example from the World Bank, their database on census information called WorldPop, and then this is what we were able to obtain in collaboration together. So the resolution went from 1 km pixels to, roughly, 10-meter pixels. So, in aerial resolution, it improved by a factor of a 100 thousand. And because we have a lot of compute infrastructure in our data centers, we were able to process 15.6 billion images worldwide and we have now released publically with Columbia, in partnership, and the World Bank, 23 countries at a practically unprecedented resolution. And it has been really pleasing for us to get emails from the Red Cross and Unicef and other aid organisations, saying thank you very much for proving these databases because now we can find communities that we did not even know existed and think about what does aid, what does connectivity, look like for those communities.
So why do we care about this? Because it helps to think about how we can do modelling of demand and where people have no coverage, and things like that. To be frank, almost all of the public information in datasets around coverage and connectivity are overly optimistic, I truly believe. So we have looked at both the population mapping as well as our own mapping of where our 2 billion users are, to try to come up with statistics around what connectivity looks like in terms of coverage and things like that. And so these are some of the discoverings. Roughly 80% of people today have some form of 3G or 4G coverage, that is miraculous. At the same time, roughly 18% have no coverage whatsoever, this represents 1.8 billion people. This is obviously in some overlap with the 1.5 billion that are being discussed today, but it is, I think, a significant challenge to think about. What is it going to take to do that? And then, only 3% have 2G only at this point.
So what we did with that is we can start to build a model of what it would take to connect all of those people. 1.4 billion people. And, when I say “we”, I mean the sort of “we of the world”, of what it would take in principle to do this. So, if you go down the column on the left in terms of what is your assumption on how much data someone needs, I heard Nicholas talk about 100 MB per month. We start at a larger number but you know, roughly at this point, somewhere between 2000 books a month, to 4 high definition videos, that is sort of the bounding on what 2 GB per months would be. And you look at what the adoption rate is, ranging form 10% up to 100%. If you wanted to connect all of those people it would take something between 3.6 terabits a second, maybe to 54 terabits per second, that is the range. So, rather than anchor on a particular number, the point is that any of these numbers are less than 1% of what global Internet demand will be by 2020. It is a very, very small number. And if you factor in devices and everything that is required to consume and create content on those networks, it is still, I think, a very reasonable number. So, I do not think this has been discussed before, money is not the issue, and technology is not the issue. It only comes down to who has the will to do it.
We wanted to share this because we believe that we wanted to use data to support the argument that this doesn’t have to be a big problem to solve. This is another perspective; we looked at sub-Saharan Africa to understand where are people relative to current infrastructure. So we segmented, based on population, distance-to-towers and tried to do a sort of cumulative count of where people are. In the example on the left, 2 km from 4G, all the way up to 50+ km from 2G. This is to me one of those slides that sort of points to why, in some cases, there are currently market failures in trying to connect to people on the right side. Every incremental dollar that a telecommunications company will receive, logically and rationally, should go to people on the left. It is just so much easier to connect someone who is 2 km or 5 km from an existing tower, than it is to try to think about connecting a rural community of a few hundred people 50+ km from existing infrastructure. And while there are only 45 million people that are greater than 30 km from an existing piece of infrastructure, we think it is an imperative to think about what would it take to connect those communities. So at Facebook we spend a lot of time thinking about what it would take, and right now we believe that what it does involve is satellite. And we are invested in our own efforts around high altitude platforms, sort of a crazy bet to see if we can make satellites that can sort of persist in the atmosphere as well. So this is one example of that, this is an aircraft that we built, it runs on the power of about five blow-dryers but has the wing span of a Boeing 737 and this is it, it is a fully autonomous aircraft that can fly itself and this is it landing with no wheels, it has Kevlar skids, on a sand patch in the Arizona desert. I wanted to show you this just as an indication of where we are in terms of the state of the technology, we have a long way to go in terms of technology development but we are pretty excited about the progress we have made so far. We have also worked on communication systems; we have built a terrestrial link using a portion of spectrum called the e-band and demonstrated over 20 Gbps per second over 13 km. So this is about one one-thousandth capacity that I talked about in the previous slides, just one link. We have since doubled that rate and then we flew it in a plane, ultimately it will go into our drone system to demonstrate 16 Gbps. So, roughly a thousand aircrafts circling worldwide could meet those global needs to connect all the unconnected. And I think the point here is that with adequate amounts of access to spectrum and technology advancements, the belief is that these kinds of capabilities are imminently doable. And so with that I just wanted to close. I think ultimately connecting communities has to be a priority and we want to be part of the solution with this community.