Paul van Gerven is an editor at Bits&Chips.

2 November

Chip shrinking won’t help reduce greenhouse emissions, as ASML claims. The gains in energy efficiency do not nearly compensate for the increase in carbon output resulting from the huge business opportunities that shrinking creates.

Digitalization, powered by ever more powerful and energy-efficient chips, will enable a 15 percent reduction in global greenhouse gas emissions by 2030, according to ASML CEO Peter Wennink. “There’s a whole list of areas where digital technology will help reduce emissions. Our vision is to develop lithography tools that will help our customers produce very-high-performance chips, which will increase in energy efficiency by a factor of three every two years,” Wennink said at ASML’s Investor Day 2021.

Like any self-respecting multinational these days, ASML is revving up the green ambitions. Environmental, social and governance (ESG) issues have become an integral part of his company’s strategy, Wennink said. “Our people talk about it. Our customers want it, our suppliers want it, our shareholders want it and the community wants it.”

I don’t want to be cynical about this. It’s a massive improvement over the typical corporate attitude towards climate change only a few years ago. However, it’s simply not true that shrinking chip structures is conducive to reducing carbon emissions. Quite the opposite, in fact.

Wennink’s line of reasoning is only true on face value, when evaluated across a relatively short time span. For example, switching over to next-gen IC technology allows data centers to operate more energy efficiently. But because the improved chips give more bang for the buck, they also drive down costs for data services, which in turn results in their increased adoption. While it’s true that the energy consumed by the world’s data centers didn’t grow nearly as much as the amount of computing they do, in the end, the availability of better, cheaper and more energy-efficient chips results in more carbon emissions, not less.

This example can be generalized to what I’d like to call the sustainability paradox of Moore’s Law. The semiconductor industry has doubled the energy efficiency of its products every couple of years for decades – an incredible feat that no other technology could ever hope to match – yet it has been responsible for massive increases in carbon emissions. The gains in energy efficiency do not nearly compensate for the emission increases resulting from the huge business opportunities that shrinking has created and will continue to create.

You could argue that these emissions mostly belong on the ledger of the semiconductor industry’s customers, or of their customer’s customers. But make no mistake, the semiconductor industry is a dirty business. Research has shown that “chip manufacturing, as opposed to hardware use and energy consumption, accounts for most of the carbon output attributable to hardware systems.”

It’s not getting any better, either. According to an analysis by Imec, greenhouse gas emissions per wafer increase 2.5x going from the 28nm to the 2nm node. ASML projects that wafer capacity for advanced logic manufacturing will grow by 9.9 percent (CAGR) in 2020-2025. Like ASML, semiconductor manufacturers have set sustainability targets, but it’s hard to conceive how meaningful emission reductions can be achieved when the market is growing that strongly.

Now, of course, chips also enable solutions that reduce carbon emissions. Electric cars need power and battery management, smart building technologies reduce energy consumption of homes and offices and smart factories manufacture more efficiently. As far as I can tell, the 15 percent reduction in greenhouse gas emissions mentioned by Wennink was calculated by researchers from Ericsson, who tallied the emission reduction potential over seven categories of ‘smart’ ICT solutions.

I’m sure that potential is real, but in our current economic system, there’s just no way chip technology will drive an overall cut in emissions. All the energy saved will be offset by the data centers we’ll keep building, the gadgets we’ll crave once they’ve become affordable for the masses (VR headsets, anyone?) and the new smartphones we’ll buy every two years because we need that cool new feature – and we have Moore’s Law to thank for it. Indeed, according to an ‘expected scenario,’ electricity consumption for ICT (consumer devices, networking and data centers) will grow from 5 percent of global demand in 2015 to 20 percent in 2030.

So, clearly, lithography-enabled chip shrink won’t be saving the planet. The seemingly endless cycle of Moore’s Law powers the world’s continuing digitalization, which may help to save energy in some areas but overall puts additional strain on our planet’s resources and ecosystems. Mr. Wennink would do well to tone down the green rhetoric.