Issue 9 / Nature

December 07, 2019
A microchip.

An early microchip prototype, developed in 1958 at Texas Instruments by Tom Yeargan and Jack Kilby. Courtesy of Heritage Auctions.

Seeing Carbon Through Silicon

Anne Pasek

Finding a future for the planet in the history of the microchip.

Climate action today is increasingly a question of exponents. Merely reducing greenhouse gases won’t cut it; according to the UN Intergovernmental Panel on Climate Change, emissions must be halved in ten years and halved again in subsequent decades if we are to avoid the worst effects of global warming. Similarly, renewable energy needs to do more than just increase its market share; it must spread exponentially, replacing fossil fuel energy sources within fifty years. 

These mandates represent an unprecedentedly rapid transition in the nature of our energy grids, transportation, housing, and all of the related patterns and habits that make up our daily lives. Decarbonization, if it is adequate to the climate math, must be both incredibly ambitious and incredibly disruptive.

Changes on this scale are difficult to imagine. To complicate matters, history offers scant examples for reference. Accordingly, the task of charting a pathway through decarbonization is in large part also a question of stretching our metaphorical imagination to reframe the possible. This is difficult, creative, and necessary work. It is also fraught with hazards.

In the global orbit of Silicon Valley thought, where disruption is a word with more positive cachet, one analogy is gaining momentum: we should think about carbon like we think about computers. The story of the microprocessor, after all, is a story of exponential growth curves and adoption rates: per Moore’s Law, the density of semiconductors has doubled every two years, making computers cheaper, smaller, and more powerful — a win-win that fueled the digital revolution. Couldn’t renewable energy follow a similar path? 

This idea is at the root of a sweeping policy proposal currently circulating in both UN climate conferences and Davos event halls: a global “Carbon Law,” styled after Moore’s Law, that sets a roadmap for exponential climate action. Unlike most views of climate change, this future is surprisingly optimistic. Carbon Law proponents point out that renewable energy, although currently representing only about 2 percent of global electricity generation, has already followed an exponential growth curve in its short history. They expect this trend to continue, given the proper incentives from governments and investment from industry. With the right social alignment, as they see it, the technology will simply take over.

The risk here, as with any framing comparison, is that the metaphor will not hold. Stories of digital disruption have long been sources of prediction, optimism, and analogy — as well as sites of dangerous fantasies. As a framework for energy transition, the Carbon Law can do more harm than good if it imparts the wrong lessons, provides false comfort, or seeks to mobilize the wrong people. 

Metaphors matter: nothing less than the future of the planet is at stake. And interrogating the charisma of exponential thinking suggests that the Carbon Law is unlikely to help make that future a fair and habitable one. Insofar as silicon’s history helps us understand carbon’s future, its lessons are the opposite of those circulating at Davos. The story of silicon doesn’t teach us to sit back, relax, and let technology save us. On the contrary: its real lesson is the power of purposeful struggle within systems of constraint.

Legislating Moore’s Law

Exponential growth is remarkable wherever you find it, and the steady gains in chip densities that began in the late 1960s remains a defining standard for rapid technological advancement. Moore’s Law, however, is not a law of physics. It took considerable social effort and material happenstance to make the growth curve hold. 

In 1965, Gordon E. Moore was the director of research at Fairchild Semiconductor, a pioneering firm that helped create Silicon Valley. In a magazine article, he observed that the number of components in integrated circuits had grown exponentially over the past seven years and would likely continue on this trajectory a decade hence. This prediction could have easily faltered. At the time, shrinking transistor sizes seemed to many engineers to invite disaster through unneeded complexity and melted components. Miniaturization, moreover, was an imperative unique to military contracts that needed chips small enough to fit onto rockets, while researchers were quite content to have room-sized computers.

It took considerable barn-raising by key figures in the industry, as well as hefty military spending, to make Moore’s prediction into a de facto law by changing industry R&D allocations and targets. The ensuing rates of growth were formally ratified in national and international industry roadmaps in the 1990s, essentially securing Moore’s Law as a group-fulfilling prophecy. 

From the start, a combination of peer organizing and institutional mandates propelled Moore’s supposition into a standard. The technology did not simply take over. 

Yet this is not to say that exponential doubling is a purely socially constructed outcome. The unique properties of silicon supported the otherwise unlikely win-win of densification and miniaturization. The price, size, and processing power of chips are tightly correlated; smaller silicon circuits require less energy to power, produce less heat, and can process at faster speeds. 

Writ large as an industry-wide trend, this fact about silicon’s thermo-electric properties led to the massive popularization of cheaper, smaller, and more powerful digital devices. But this isn’t true of electrical grids, photovoltaics, or other forms of technology that don’t experience consistent density scaling across their components. If, for example, Moore’s Law applied to air travel, a New York City-to-Stockholm flight today would hold 120 million passengers and take eight minutes. 

However, even within the scalar logics of silicon, the predictive success of Moore’s Law today is widely acknowledged to be over. Microchips are heating up, R&D costs threaten to outpace density gains, and, as engineers parse the design challenges of nanometer circuits, they may simply be running out of atoms. Future prospects for densification are multiple and uncertain. Rather than relying on exponential growth in processing capacities, software designers are increasingly depending on gains in efficiency first developed for mobile applications — seeking as an industry, if somewhat belatedly, to do more with less. 

Surfing the Waves of the World Spirit 

Techno-optimism is easy when exponential growth holds. Proponents of the Carbon Law largely see technology — both digital and electric — in this register. As a result, exponential decarbonization appears to them as little more than a technological coordination problem. It requires innovation and cooperation on the part of politicians and green tech companies, but asks very little from citizens. Its model of power is predicated on the agency of executives and devices, not political mobilization by large numbers of people.

Two men sit at the center of the Carbon Law’s development and distribution, and their backgrounds help explain some of its political character: Johan Rockström, director of the Stockholm Resilience Center, and Johan Falk, the director of Intel’s Stockholm IoT Ignition Lab until he quit to work with Rockström. At Intel, Falk’s mandate was to promote the spread of smart networks to new industries — to spread narratives of exponential growth and scalar disruption. Rockström’s center, on the other hand, develops and disseminates “resilience thinking” across high-profile climate talks and a C-suite executive education program. Together, they have targeted political and corporate elites with a simple message: the climate crisis can’t be solved without exponential thinking, which requires elite “accelerators” to reenact the roadmap and feedback loops that propelled Moore’s Law forward. 

Strip away the Silicon Valley language from this proposal and the details are themselves common enough to most mainstream climate governance plans: price carbon, make cuts across multiple sectors, increase energy efficiency, and fund renewable R&D. What’s unusual here is the emotional register of the plan and the apolitical certainty of its promise — factors that can’t be disentangled from tech industry tropes. For instance, Falk’s “Exponential Roadmap Report” argues that decarbonization “is nothing short of a global economic transformation. But transformation appears assured through revolutions driven by digitalisation. Harvesting this power will help drive unstoppable momentum.” Similarly, Rockström predicts that under the Carbon Law, “big masses [will] simply surf along a sustainable journey without knowing that they’re doing it.” 

In short, the technology will do the work. Exponential growth curves will continue along an unchanging trajectory, as if by natural law. Existing social arrangements, fossil fuel interests, and the economic and environmental justice barriers to the energy transition will cede to the power of elite leadership and digital disruption. In turn, our carbon footprints, seemingly without effort, will just shrink and shrink.

Creativity Within Constraints

This top-down, technologically determined future ignores all the ways in which energy transitions aren’t just a question of market shares, but of the social pressures and material constraints that cut across them. Decarbonization will demand more than just a different kind of technology curve, accelerating sharply into the horizon. It will very likely require abrasive changes to well-worn cultural norms, the structure of cities and trade, and perhaps even the valorization of economic growth in its broadest terms. It will be conflictual, classed, and expensive.

Technology alone proves to be a poor analytic for these kinds of social changes. Moreover, as demonstrated by recent waves of popular opposition to climate policy, market fixes without considerations for equity are politically disastrous. People, infrastructure, and culture don’t fit into industry roadmaps or silicon wafers. They contain differences and resistances that can’t be universally scaled.

If Moore’s Law is to be a useful story through which to approach this future, it will be for all the reasons its green proponents currently ignore. The history of the microprocessor revolution is ultimately about the immensity of effort that goes into maintaining the dream of exponential growth — and its inevitable collapse. Moore’s Law was neither a socially constructed prophecy nor a materially determined outcome. It was a period of coordinated action within specific material parameters that have now passed. It leaves us facing a technological future that will require creativity within new constraints.

Rockström and Falk are correct that time is short and the need to muster political and technological resources is great. Where they are wrong is the assumption that a better future will arrive on our desks without a fight. Instead, it will require a public that can stand up and push.

Anne Pasek is a postdoctoral fellow in Transitions in Energy, Culture, and Society at the University of Alberta.

This piece appears in Logic's issue 9, "Nature". To order the issue, head on over to our store. To receive future issues, subscribe.