What will happen to the technology industry in an age of expensive energy?
The technology industry has grown up in an age of cheap and abundant energy, and that has shaped, deeply and fundamentally, the way it sees the world, what it chooses to make, and how it designs what it does. You have to think only of the short lifespan of the devices, the fact that they are discarded, not upgraded, when technology moves on; or the emerging service designs based on the world of the cloud; and always on, on demand access. But the age of cheap and abundant energy is coming to a close. It is about to become scarcer and more expensive. How does the technology energy need to respond?
It’s worth, quickly, making the case about energy, for not everyone believes this. John Michael Greer, in his book The Ecotechnic Future, uses the analogy of a group of field mice who find one day that someone has dumped a huge pile of grain in a corner of their field. They eat it and multiply, until the grain starts to run out.
The fieldmice’s grain is the same as our discovery of cheap fossil fuels – as the moment of peak oil approaches, and as reports of ‘peak coal’ gain credibility. Of course, humans are smarter than mice, and there is a strong view that we’ll be inventive enough to solve such problems. Greer is more sceptical:
Industrial civilization had its day in the sun [he writes] because, in a world where fossil fuel could be had for the digging or drilling, the industrial mode of production was more efficient than its rivals, and enabled the communities that embraced it to prosper at the expense of those that did not.
And the research suggests that technology delays the moment of impact, it doesn’t avert it. (I’ve written about this before). Oil itself is getting tougher to find and extract, as the Deepwater Horizon disaster showed. Renewables may help, but not at the scale necessary over the next decade, or even two. And so on. Even the International Energy Agency, which reports to governments and largely draws on figures for official reserves, sees a rapid decline in oil production from conventional sources, and estimates of large increases in prices are increasingly widespread. Set against that the global energy consumption of the internet is estimated at 3-5%, but growing very rapidly. The entire global transport industry takes around 25%, by way of comparison.
Making computers and phones is highly energy intensive
This consumption, though, is only half the story. The production of computers and module devices is one of the most energy-intensive in the world, as Low Tech Magazine has explained. All of the precision engineering, in large complex fabrication plants, means that the energy required to make each device (known as the “embedded energy”) is huge. There’s not enough work been done on this, but Low Tech magazine calculated that the energy required to make just the memory chip in a laptop exceeded the amount of energy consumed by the computer in its entire lifetime, typically three years. Each successive generation of fabrication plants is significantly more energy intensive than the last.
There are other hazards here. In the near-term, routine power cuts are likely even in places where they have been the exception, unless there is implausibly rapid reduction in demand for power. And sitting behind this is a resources threat: many of the materials (opens pdf) needed to make high-tech devices are scarce and found in areas which are difficult to access, for political or geographical reasons. Some are simply running out.
Of course, the industry has been responding. The big ICT companies are finding ways to reduce the energy consumption of servers. Intel and other chip companies are looking to make energy efficient chips. Apple has been greened, apparently slightly reluctantly. But reducing the sustainability impact of the existing business model will not have sufficient impact. The tech industry may have to rethink itself from the bottom up.
In what follows, I will try to sketch out what a sustainable technology sector might look like in an age of expensive energy.
- Consumers will realise that ICT costs them money, because they’ll see it in their bills. High energy costs will also affect the costs of equipment, because the level of embedded energy is so high. And transport costs will also rise. For all of these reasons, the notion that the device shouldn’t be disposable, and the relative attractiveness of machines and services which have been redesigned to manage energy far better, may reach the consumer mindset long before it’s properly understood in the boardroom.
- Batteries and storage will be better. In general, any improvements in battery performance over the years since Alessandro Volta invented it have been eaten by increased demands placed on them by more complex devices. (In a different sector, this is particularly true of the electric car). Batteries are also particularly toxic when they are disposed of. Expect to see renewed interest in battery technology, perhaps coupled with elements of self-powering (the ‘proof of concept’ OLPC 2, for example, comes with its own solar cell.)
- Trickle. The notion of rapid on-demand downloads of large files is inherently energy intensive, both for machine and server. Expect to see more emphasis on sending large data files slowly, and doing it overnight, when energy is cheaper. One modest early consequence might be the bundling of updates from social media companies rather than constant and immediate refreshes. (Think about the economics of storage heaters instead of central heating). Whisper it not to Google, but the whole notion of the cloud, certainly as its currently conceived, may prove to be a blip rather than a fundamental reorganisation of the industry.
- Local. Whenever energy costs go up, local becomes more important, as it does when resilience becomes more important than efficiency. Expect local providers to move beyond access to service design and management. Large global providers may have to become more distributed, rather than relying on smaller numbers of large server centres to deliver a continental footprint.
- Materials will need to be less exotic. Expect to see innovation to reduce the industry’s dependence on rare earth and other hard-to-find elements, and towards more common materials. There are already some examples of this in the poorer countries of the South.
The re-usable device
But the biggest story here is about moving beyond the disposable device towards the re-usable device. The notion that you throw away something when it needs an upgrade is an inherently 20th century idea. If you think of a computer or mobile as a bundle of highly expensive energy, in an age of energy scarcity it becomes an absurd one. And while this is slightly technical, recycling such products is unwise, because it means that the energy used to make them is lost. Re-use is the only viable sustainable option.
What this needs, therefore, is equipment which can be repaired, renovated, reconfigured, or rebuilt, when a change needs to be made. The metaphor, if you like, is Lego: parts which can be removed, replaced, or extended. It requires designers to rethink the entire production process and service model. I think this is a good thing. The technology industry was built by hackers. The change forced on it by high energy costs will be a return to its roots. And it will provoke some real, socially useful innovation.
This post is based on a presentation I made at WPP’s Stream 2010 conference in September. The photo at the top of the post is from into mobile, and is used with thanks.
Lloyd Walker, on the apf list, sent me an interesting email in response to this post. With his permission I’m re-posting this here:
“Worth more conversation. Here’s some thoughts.
Reusable device:
What components of a tech products do you think can have service life extended if they were swappable? This is a real challenge IMO when you start to look at the bill of materials and parts. The problem is that most of the underlying technology is on a Moore’s law improvement curve. Processor, motherboard, RAM, battery, video card, storage, battery… these things just aren’t worth reusing or replacing in whole or part because the total system becomes outdated by increases in performances and advancing specifications and features of *software*. These physical components aren’t really wearing out. The video screen may be worth keeping but even those are advancing in performance and may include energy improvements that make switching to new screens better for the environment (crt to lcd as an example).
It seems to me one of the only way to increase service life is to slow down the advancing performance requirements of software. I have no idea how to do that – I’m just thinking about the system dynamics of your topic. I would pleased to be using most versions of my hardware and software from 1998 with just a larger hard drive, faster processor, and faster broadband connection. I would pleased to lock in a device, form factor, ergonomics and set-up that is tuned to me and have all the applications just perform better each year.
Perhaps cloud computing will help with these issues by offloading OS and application processing to the cloud and allowing consumer devices to have longer service life. That of course is not the trend we are currently seeing with laptops, tablets and smart phones.
Energy units vs QOL units
Something I ponder is the environmental cost of personal computing and the value it provides vs the replacement cost of that value / quality of life (QOL) by another mode. I’m quite bullish about tech in this regard. Once a video game, application, or media file is created its energy unit cost to deliver is quite small (relative to physical goods). The joy produced and QOL provided by that small unit cost can in many cases be quite large. As a gross example: it seems to me that teens these days are much more interested in the status and style and freedoms provided by smart phones than teens in 1950’s to 1980s. In my younger days it was all about CARS. Obviously the environmental impact and energy cost of adding a car to the system is far greater than adding another smart phone. Yet I get a sense that most young people get similar joy, social connectedness and QOL from ICT as we did from cars. More tangible examples including ebooks and video conferencing clearly win over the physical alternatives even with ICT embedded costs. All this to suggest that if energy becomes really expensive we may see an even greater shift to ICT because it is more energy efficient than almost all physical alternatives.
The potential of creating high QOL experiences for the maximum number of people at reduced energy per person (relative to other products and services) is hopeful. Of course with reduced unit cost the experience becomes more affordable and thus consumable by a larger audience. So given expanding population, and a growing addressable market the net effect is more energy consumption by ICT. Ahh the frustrating beauty of systems.”
I think you raise some good points about the energy ( and you could have mentioned water usage at fabs too) use of server farms and chip-fabs. Obviously we have to pull back on the disposable society and examine the embedded energy of our objects. Perhaps we could get an embedded energy label one day. But this same process must be applied to all our industries.
But the computer industry strikes me as different in some ways from conventional industry: 1) it changes faster 2) there is much fundamental scientific and engineering work left to do to reduce the wattage of chips, and progress is well under way. 3) the ubiquitous change to mobile tech is driving the consumer end right now. If we were talking about the car biz, it would be the equivalent of having 50% light electric cars on the road. Plus, computing makes other processes more efficient, and so an increase in computing might actually save total energy elsewhere.
On networking and cloud computing, I have to say you are making assumptions that don’t follow. Sure it takes some computing power per packet on a network, and more packets means more power used. But packets aren’t physical objects, and don’t have irreducible physical properties like the laws of thermodynamics. The quantum rules followed are much less understood and the resultant engineering is far from optimized. I think routers and switches actually move more packets for a lot less input than they used to, though I haven’t any reference to prove it (I work for a switch company). Cloud computing is more efficient on almost every level so its here to stay.
cheers
John Fisher