Debbie Ding, AutoCAD modelling for The Library of Pulau Saigon
20 March 2019
Weeklies • so-far reads
The Stack: On Software
and Sovereignty

Smart Grid: Ouroboros

The Cloud layer, just above the Earth layer in The Stack, makes epic, rapidly expanding energy demands (the total carbon footprint of the world’s data centers has already surpassed that of the airline industry and is presumed to triple by 2020), and so risk is not hard to find. Data centers are located to mitigate cost and uncertainty, away from likely natural disasters, in proximity of cheap or clean energy sources, diverse power grid interconnects, favorable land use zoning, and inexpensive intermediate bandwidth, for example. Because there is no planetary computational economy that is not first a planetary energy economy, the limits to growth for The Stack are not only Moore’s law and Shannon’s law (accelerating the speed of processors and squeezing more information into existing channels) but also, and perhaps foremost, to secure the energy necessary to power those data centers, smart cities, homes, cars, roads, smart objects, and phones, as well as the real costs (or benefits) of doing so at the expense of other infrastructure, like new roads and buildings. In principle, there is a potentially virtuous correlation at work for innovation across computation and energy sectors, and the gamble on that potential is another generative accident of the Earth layer. The continuing growth of The Stack and the computationally intensive transformation of energy sourcing and distribution infrastructures on which it depends likely cannot occur without one another. The architectures of new energy-information networks, including so-called smart grids, require pervasive computational systems to realize necessary gains in the timely, efficient, and equitable distribution of megawatts of energy across networks. For this, each point in the grid that might produce or consume energy, which is in principle pretty much anywhere and anyone, must not only be able to store or transmit that energy, but also to calculate and communicate its activities to platforms that steer the whole. All electrons must pass through the angelic regime of recording and optimization, but today such grids are slow to come online. Their politics are filled with inertia and gamesmanship, and, moreover, the underlying physics is uncooperative; electrons do not work “like bits” no matter what your smart city consultants are saying. At the same time, The Stack itself depends on new energy grids to feed and undergird its growth. It requires a conjoined-twin energy-information network that can generate, calculate, and allocate those usable electrons point-to-point. Absent a radical relaxation of energy scarcity by renewable sources, the finely grained electron sorting between points of production and consumption must be realized at global scale or the growth of planetary-scale computation will hit physical energy limits and will stall. A more scalable grid of electrons needs to be wrapped inside and around The Stack’s Earth layer. In short, planetary-scale computation needs smart grids to grow, and for smart grids to grow, they need more ubiquitous computation. The computational future of energy and the infrastructural program of computation form such a coil, one end feeding on the other like Ouroboros, the ancient symbol of a snake eating its own tail.

Whether or not the risks associated with the energy costs of Stack infrastructure will outpace the efficiencies provided by calculative technologies as they become pervasive across industrial sectors is unknown, and probably unknowable at the moment. Prognostications vary from measured good news to very bad news. According to a Greenpeace report on cloud computing and climate change, the electricity consumed by cloud computing globally will increase from 632 billion kilowatt-hours (0.6 terawatts) in 2007 to 1,963 billion kilowatt-hours (1.9 terawatts) by 2020, and the associated carbon dioxide equivalent emissions would reach 1.034 megatons (currently the world economy’s total energy appetite is roughly 15 terawatts). If imagined as an emergent nation-state, the Cloud would be today the fifth largest consumer of electricity, ahead of India, Germany, Canada, France, Brazil, and the United Kingdom. But even this doesn’t fully capture the climatic and ecological impact of planetary computation.

The extraction of mineral resources to manufacture and dispose of devices and hardware can be extremely destructive in its own right. For their part, data centers are estimated to represent only 20 percent of the information computing technologies (ICT) sector’s total footprint by 2020, while telecoms infrastructure, PCs and peripherals, will represent much more. While the shift toward Cloud platforms represents efficiencies over “dead tree media” and perhaps will ultimately reduce reliance on air travel, it also enables an exponential growth in data flotsam, such as search histories, redundant personal media, legally mandated trails of sales receipts, unfiltered spam, backups of the trails of spam and search, and so on, all of which need to be stored somewhere, and it is now stored online in various databases here and there. Sometime in the future, this Cloud landfill of postcontemporary data junk may provide new insights for digital humanities as to the real nature of global discourse (or existential clues for a future artificial intelligence, itself born of spam perhaps, seeking out the meaning of its origin), but until then, it is just more carbon debt. It is estimated that the electricity required to send the trillions of spam e-mails worldwide each year is equivalent to powering 2 million American homes and generates the same amount of greenhouse gases as 3 mil- lion cars. But the Cloud layer is not uniform, and how it affects the Earth layer depends on where it is buried. It matters where data centers are located because the available energy mix ranges from dirty fossil fuels to robust renewables sources (Hong Kong hosts one of the dirtiest clouds, while Iceland and Sweden are among the cleanest, and so Iceland has made Cloud hosting a key part of its national industrial strategy). In the United States, many of the largest data centers (such as Google’s in Lenoir, North Carolina, Yahoo’s in La Vista, Nebraska, Microsoft’s in Chicago and Apple’s in Apple, North Carolina) are, as of this writing, all at least 50 percent coal powered. The incentives to introduce greater energy efficiencies in data centers and to reduce operating costs thereby are considerable, and they represent critical competitive advantages for different Cloud platform players, as will be discussed further in the following chapter. Nevertheless, without significant gains in energy and carbon efficiency (and cost), it may be too expensive for the Cloud to “grow” beyond certain thresholds. If so, then less assured paths of innovation will ensue, many of which may further distort infrastructural access between the global North and South.

Benjamin H. Bratton's work spans Philosophy, Art, Design and Computer Science. He is Professor of Visual Arts and Director of the Center for Design and Geopolitics at the University of California, San Diego. He is Program Director of The New Normal programme at Strelka Institute of Media, Architecture and Design in Moscow. He is also a Professor of Digital Design at The European Graduate School and Visiting Faculty at SCI_Arc (The Southern California Institute of Architecture).

In The Stack: On Software and Sovereignty (MIT Press, 2016. 503 pages) Bratton outlines a new theory for the age of global computation and algorithmic governance. He proposes that different genres of planetary scale computation -smart grids, cloud platforms, mobile apps, smart cities, the Internet of Things, automation- can be seen not as so many species evolving on their own, but as forming a coherent whole: an accidental megastructure that is both a computational infrastructure and a new governing architecture. The book plots an expansive interdisciplinary design brief for The Stack-to-Come. See

so-far reads is a companion column to so-far's current biannual Issue that presents extracts of recommended further reading. The Stack: On Software and Sovereignty by Benjamin H. Bratton presents a philosophy of technology, design and globalisation in an era of "planetary-scale computing". It can be purchased on Amazon.