Comments
Transcript
SCOTT MALCOMSON: Unless something like quantum computing transforms the world of information processing, this trend of computing power information processing becoming more and more energy intensive is going to increase. Information is becoming terrestrially bound and controlled by nation-states. And as long as everyone made assumptions that globalization and borderless expansion and international supply chains and so on would just get more and more intense, and there wouldn't be these kinds of political competitions, it was not that big a deal.
DEE SMITH: Hello. I'm Dee Smith. Today I'm going to talk with Scott Malcolmson. Scott has a background in journalism, diplomacy, technology, and foreign policy. And he's an expert on the internet and its development from its earliest days to today and where it may be going. And today we're going to talk about 5G. 5G is a very hot subject. And like many hot subjects, it's a very hyped subject. But we're going to try to get behind the hype and understand some of the things that may not be the most advantageous. Scott, good to see you.
SCOTT MALCOMSON: Good to see you, too.
DEE SMITH: So today I wanted to talk to you about 5G, which is an interesting topic and a topic you've fooled around in quite a bit and become an expert on. And there are lots of interesting things. It's a very complex situation. I would like to talk about several of those, one being why it's different in the US than it is everywhere else, but also about energy use. Most of us don't even think about in 4G, when we do an internet search or a Google search or pull up a map on our phone, how much energy that processing uses. But it's really quite a lot, isn't it?
SCOTT MALCOMSON: Yeah. Well, it's hard to estimate. But one number out there is that roughly 70% of the cost of a phone is in the energy that it uses, whether manufacturing, charging it, operating the network, and all of that. On a broad definition, information and communication technologies use something like on the order of over 2% of human generated carbon. So 2% of the human generated energy use today is-- at least 2% is from what are called ICTs, or information and communication technologies, which includes computers and the networks and the phones and all of that.
For comparison, aviation uses about 2%. That's aviation on a broad definition around the world. So it's very likely-- it's more or less inevitable-- that the ICT share, of which cell phones are a big part, will continue to grow. That said, there are differences of opinion about the exact amount. One of the more prominent, though also more pessimistic, researchers in the field is a guy named Anders Andrae, who works for Huawei in Norway.
And his most recent study, which just came out in January of this year, he said that he thought that ICT was going to-- by 2030, would probably account for between 8% and 20% of human energy use globally, which is a gigantic number. If you think that aviation is about 2% of the total now, and it runs about 12% of human transportation carbon use, that if you get ICT up to anywhere near 8%, it's a fantastic growth, quadrupling of the energy use for the computer systems that we have. The proliferation of cell phone use in some countries like China-- most people access the internet through apps that are on their phones. And so the phone share of that is likely to increase, although again, there's controversy among scholars about exactly what uses more energy.
DEE SMITH: And if we were to take the 20% figure, that is an order of magnitude--
SCOTT MALCOMSON: It's kind of an apocalyptic figure.
DEE SMITH: It's an apocalyptic future--
SCOTT MALCOMSON: Yeah, it really is.
DEE SMITH: --for many reasons. So one of the real issues with this power use issue is the cooling of the server farms, these hyper server farms they have now. And that has really inspired some innovative solutions, it sounds like.
SCOTT MALCOMSON: The general way that the networks, which include or will include 5G and includes mobile phones, generally speaking, you have a signal that goes from a sensor-- it might or might not be a phone-- that goes through intermediary processing systems, and that eventually ends up via the internet at what they call mega centers or gigantic data centers, basically, which are collections of servers. And on the one hand, the more computer processing gets done at large, large mega servers, the more efficient it is because there's less downtime. Things aren't plugged in when they're not being used, which by the way, accounts for a huge portion of the energy in these systems.
The industry is motivated to make the big servers more efficient and so on. So on the one hand, there's an energy savings. On the other hand, the cooling costs go up. And those big systems are very hard to cool. And they're also hard to-- or it's very important to keep them secure. So they are beginning to be sited, in the case of China, in Guizhou province, which was a very poor province-- still is a very poor province-- in the mountains, which has an abundant supply of cool, fresh water that comes down the mountains. Temperatures are cool generally.
So they are keeping-- that's the designated province for China's big, big, mega servers, led by Alibaba, which is sort of the Amazon of China building servers there. Apple also is building a server there. At the other end of the water cooling system is a really interesting project that Microsoft has been doing, where they have a mega server that's operating underwater off the Orkney Islands, which are part of and to the north of Scotland. So it's actually in the ocean. And water is being used to cool it there.
DEE SMITH: It's really a challenge in many ways because it exacerbates the problem of the north-south divide. This seems to really make it more favorable. The conditions are more favorable for countries in cooler climates, which for most of the inhabitable world, means the north. And it's really unfavorable for people in warmer countries, particularly as climate change takes off. Are you concerned about that?
SCOTT MALCOMSON: Yeah. There's already a digital divide. And this aspect of the technology is likely to make it worse because as you say, the currently wealthier countries tend to also have cooler climates. In the United States' case and Canada, we have a relatively temperate climate and a lot of water, both freshwater and saltwater, and a lot of sediment along the coasts.
In the hotter part of the world, it's just exponentially more expensive to cool all the equipment. And so it's less likely to be placed there because it's just simply more expensive per kilowatt hour. And it also means that development that depends on these systems and depends on the rapidity of computing, which is really the key question, will tend to be located along coasts, which already account for the majority of world production and population. Paradoxically or ironically or deservedly, it also means that these systems will tend to be located in coastal areas where rising sea levels would do the most damage.
DEE SMITH: So one of the things that is interesting, I think, is the extent to which the intensive use of processing power for things like Bitcoin is driving the-- not to mention various kinds of video processing and so forth-- but Bitcoin is an example. I think a lot of people don't even know what Bitcoin really is. They know it functionally. But more people understand it mathematically than understand how Bitcoin is mined. And that is something that requires huge amounts of energy. Can you go through just a little bit about how that works?
SCOTT MALCOMSON: Well, Bitcoin mining-- in a way, it makes sense that it's called mining just because it requires computation. And the more computation you perform, the more you're able to mine, in a nutshell, which is why Bitcoin companies took off in places like Iceland, which are not obvious places for Bitcoin mining. I don't know what is obvious for Bitcoin mining, but to have it in an area where you would be able to cool servers better, basically.
And part of the reaction of the Chinese government against Bitcoin-- there are many reasons, but one reason was because once it started going, it was using a lot of energy. And energy was subsidized anyway. So you'd end up having government, which is to say, in a sense, citizen's money going to subsidize the production of Bitcoins, which made no sense and was one of the reasons why China basically cut down and more or less eliminated Bitcoin mining.
DEE SMITH: I've seen predictions that cryptocurrencies already take up half of a percent of energy use. And when they move into these places that are cooler or have more readily available electricity, the price of electricity goes up, simply because the demand is so great. And then the Bitcoin mining facilities, so to speak, are always moving, finding the cheapest source of energy.
SCOTT MALCOMSON: It connects to the whole issue of electricity and computing power as being attached to the land in a way that you wouldn't expect, as having an environmental effect that you wouldn't necessarily anticipate. It doesn't look like it's-- you don't see pollutants coming out of the tailpipe. But essentially, information does not want to be green. And it's only likely, really, to get worse for a variety of obvious reasons but also because computing basically depends on the processing information on silicon chips.
And the speed of silicon chips or the rate of the increase in speed of silicon chip computing is going down pretty steadily and is anticipated to continue to go down. And what that means is that it used to be you could make one chip move information around more quickly over time. Once that rate of increase goes down, then you have to use two chips to get the same growth in your computing power. And so each of those uses-- two chips use the energy of two chips, simply put.
So the less processing power increases on a chip, the more power you need to get the same amount of computing or to grow the amount of computing. The growth has to come from increased power usage, basically. And unless something like quantum computing transforms the world of information processing, this trend of computing power information processing becoming more and more energy intensive is going to increase, which is why Anders Andrae estimated it at between 8% to 20%.
But of course, the range of that estimate answers the other question, which is, how predictable is this? And it's not very predictable. The only thing that's really predictable, I think, is that it's going to increase.
DEE SMITH: The usage is going to increase.
SCOTT MALCOMSON: That the energy intensity of computing is going to go up.
DEE SMITH: What does that mean about the practicality of future developments like global 5G? Leaving aside all the other questions, how do you deal with that, that energy curve that's just skyrocketing?
SCOTT MALCOMSON: Well, I think geographically, higher speed computing power is going to be in particular geographic pockets. And they'll be pretty much where you'd expect them to be. They'll be in wealthier countries along the coasts, in the cities. And you'll get an increase in what could be considered inequity in terms of processing power between certain processing-rich, if you will, parts of the world and processing-poor parts of the world. And so I think that's the single biggest impact of this.
And 5G is a question of definition. In essence, 5G is just like a code word for increased processing power from sensors to servers and back to sensors. That's the simplest way to put it. It tends to be used to express two related but pretty different things, one of which is faster speeds on phones, and the other, which is the use of sensor networks to send information to servers and back. So when you read that 5G is essential to having driverless cars, that doesn't have anything to do with phones.
That doesn't mean each car's going to have a cell phone. What it means is that each of these cars has to be able to ping signals back to a server at a very, very high speed in order to avoid hitting the lamp post. And the 5G with phones is a different issue. It's the same technology, in a way, but it's unnecessarily confusing between the two of them.
DEE SMITH: So that's an interesting question, the question of driverless cars and the amount of data that it requires or will require to actually make those systems operational, which is staggering, because each one of those vehicles have to be transmitting and receiving data all the time that relates to all the other vehicles on the road. And that's going up and down the system. So that's the kind of thing that most people don't think about when they think about the rise in energy utilization. But I want to talk about another aspect of 5G, which is that these systems being rolled out around the world are different. And this relates to something that's fundamental that also explains why Huawei has such a prominent place or position in the 5G market. Can you explain how that happened?
SCOTT MALCOMSON: So when AT&T was broken up, there were a number of different companies that were competing for the mobile phone market and investing in R&D to develop different kinds of phones. And they did it according to different standards, one of which was CDMA. And older people will remember these different standards, the little things that used to come up on your phone.
So at that time, Europe, being Europe, decided to take a single standard and run with it. And the US, being the United States, decided to have different large companies fight over different standards and see who would win. Well, in the end, what happened was that the standard that the Europeans were using won. And meanwhile, that was the standard that had been adopted in Asia. And China in particular had decided to make the same bet. So they were on the same mobile phone standard in the 2000s.
We were meanwhile sorting out these different standards. So that's the first point. The second point at the same time was that through the breakup of AT&T and various other things that had to do with the management of the telecommunications industry, existing American companies were doing really, really poorly, like for example, Motorola, which was a huge innovator in cell phone technology. The clam phone, the flip phone, and so on was losing more than $4 billion, if I'm remembering correctly, sometime in the mid-2000s over a period of a few years.
And so Motorola, AT&T, these other companies, which had been American companies leading the way in this kind of technology, essentially sold off in parts. So part of Motorola was sold to Google and then was sold to Lenovo in China. Bell Labs was sold to Nokia in Finland. Lucent, which was a child of AT&T around the time of the earlier breakup, was also bought by Nokia in Finland.
And so now in the world, the main makers of this technology are Nokia, Ericsson and Sweden, Samsung in South Korea, and Huawei in China. Huawei deserves a lot of credit for what it's done. All four of these companies were essentially national champions. Samsung is obviously this gigantic chaebol company in South Korea. So they, using this older standard, had been able to innovate within it and expanded aggressively.
By 2005, Huawei, which is a relatively new company, was having more than half of its income outside of China. While it is a Chinese company, it's been an international company, really, from more or less the beginning of its existence. So the result of all that was that American companies, with the partial exception of Qualcomm, were no longer the leading edge of developing this kind of phone technology. And that was a huge problem for the American telecoms industry.
And as long as everyone made assumptions that globalization and borderless expansion and international supply chains and so on would just get more and more intense, and there wouldn't be these kinds of political competitions, it was not that big a deal. But in the situation we've been in for the last couple of years, where the US and China are in intense conflict over trade, and especially the kinds of technology that Huawei builds, the US was really just left behind. Unfortunately, this is being repeated in terms of radio spectrum because mobile phones communicate with their-- to put it simply, mobile phones communicate with computers and back via radio spectrum. That's how the signal gets transferred.
And the Europeans and Asia have once again essentially agreed on a single radio standard for developing 5G. It's between 3,400 and 3,800 megahertz. The US-- four companies, ATT, Verizon, Sprint, and T-Mobile-- are going about it differently, rather as they did the first time around with cell phone design or cell phone standards, where T-Mobile is at the 600 megahertz spectrum. Sprint is more active in a lower-mid spectrum. AT&T and Verizon are focused on the extremely high spectrum, which is called millimeter wave, which is very, very fast.
So you've got the main four companies on different parts of the spectrum. But meanwhile, that middle band around 3.5, or 3,500 megahertz, is where everybody but the US is developing its 5G standards. And it's building technology to that band of spectrum. And to this day, although President Trump said earlier this year, I think in April, that we have to win this 5G war and so on, that middle band spectrum is still not being auctioned. It's not being made available for 5G development. And so personally, as an American, I hope that we can get that band opened up so that US companies can develop technology for the band that's being used or will be used in pretty much every other market in the world for 5G, not just