Green Data: Embedding sustainability in IT infrastructure
It is well-known that the world is going through a data revolution. The amount of data being created, consumed, processed and stored is growing exponentially. There will be more data created in the next three years than in the entire history of humanity to this point.
But the world is also undergoing — and needs to undergo — a sustainability revolution. This is driven by many factors — not least the staggering pace at which human activity is emitting carbon into the atmosphere. We continue to spew 162 million tons of man-made global warming pollution into the atmosphere every day as if it were an open sewer.
In many ways, the data and sustainability revolutions complement each other. So many sectors — from transportation to energy to healthcare — have the capacity to be transformed for the better and made radically more sustainable by the intelligent use of data. We see that across our own portfolio. But there are also emerging tensions — namely the potential growth of information and communications technology (ICT) related emissions.
Precise numbers are hard to calculate but reasonable estimates are that about 2% of global electricity is used in operating ICT infrastructure. This number then goes up significantly when we factor in the embedded carbon that goes into building the servers, chips and networks that make up the infrastructure — and this embedded carbon will continue to drive emissions even as datacentres themselves transition to renewable energy. Even today, it all adds up to a similar carbon impact to that of the aviation industry.
Looking ahead, ICT emissions get harder to predict. If the exponential growth of data leads to a corresponding growth in CO2, then we are all in trouble. Data analysis, video streaming, picking up a sensor data from a corn field or inputting a sales record — all have carbon costs. Moreover, new types of digital workloads — those that rely on unstructured data, are used for machine learning (ML) and other emerging use cases like augmented reality — require orders of magnitude greater resources. Some studies have talked about worst case scenarios where the ICT industry reaches up to 20% of global energy demand. If that happens then we’ll be in a situation where IT-related emissions are catching-up with those of global agriculture or transportation. Even at much less alarming levels, the potential impact is unsustainable.
There is also hope.
ICT-related energy demand has remained flat over the last few years — largely thanks to technological innovations like the advent of cloud computing and the continuation of Moore’s Law, which have had a tremendous impact on efficiency. But as the data revolution continues to accelerate and we start to reach the limits of current technologies, we will need new approaches — across hardware and software — to help contain the growth of ICT-related emissions. The range of outcomes is wide.
Luckily, we believe that much of the technology we need to mitigate the risks is available or under development today. We see a shift towards increasingly more sustainable architectures where efficiently written workloads are run on optimal hardware, in the right place and at the right time to achieve performant, low-cost and sustainable infrastructures.
It is for this reason that we recently convened a summit on Zoom led by Al Gore, our chairman with academics, growth stage companies and ecosystem partners to look at the risks and opportunities from the drive towards sustainability in IT infrastructure.
Coming out of that session and our research, we see opportunity across four core themes:
· Accelerating shift to the cloud: much of the efficiency gains of the last decade have been driven by a shift to the cloud. Investments in efficiency by the hyperscalers combined with much higher utilisation mean that for the vast majority of enterprises, the CO2 cost of being in the cloud will be radically lower than on-premise on a like-for-like basis, perhaps by as much as 90%. Enterprises are seeing this and accelerating their cloud transformations. We are interested in technologies that will accelerate the trend and allow enterprises to move more of their workloads and faster to the cloud.
· Better visibility and automation for CIOs and IT in the cloud: up to 20%-30% of workloads run at any given time are “zombie,” meaning that capacity does not serve any useful purpose. Whilst the cloud can be effective in driving up efficiency, the “infinite availability” mindset — unless coupled with strong observability and CloudOps — can mean that “I know half my cloud spend is totally wasted, I just don’t know which half,” in the words of one global CIO we spoke with. We are interested in tools that will help CIOs know and act on the state of their infrastructure capacity.
· Developer tools for more efficient workloads: we believe there will be more tools for software engineers and data scientists to take control of and optimise the resource consumption of their own workloads themselves. They might be case specific tools such as those within machine learning which allow for better trade-offs between model performance and cost — or generic tools such as more efficient database design that allow developers to craft more performant and more efficient workloads. We are also interested in DevOps tools, such as infrastructure-as-code, which empower more developers to interact programmatically with infrastructure in the agile development process, understand it better and make better and more informed trade-offs around resource utilisation.
· More efficient hardware: across the infrastructure stack — but particularly in storage and networking — we see opportunities for new software and hardware designs to supercharge performance, allowing us to achieve more with less. These systems may interact with on-premises infrastructure or act as an overlay in cloud environments. For instance, within storage we have seen the ability for newer architectures to decrease the amount of power needed by an order of magnitude or more. Beyond this, the growth of specialist chips from TPUs to FPGAs will create greater opportunity to match the right workload to the right hardware and drive efficiency. As hardware SKUs proliferate, so will the tools needed to help enterprises match the right underlying hardware to their workloads.
We anticipate a revolution in the way in which CIOs and CTOs perceive sustainability.
Sadly, we know that today, the issue of sustainability is not top-of-mind for most CIOs or CTOs. They are regularly beset by competing priorities, and the need to digitise their companies at a rapid pace, often without the technical skills in-house to tackle the challenges they face. But we are convinced this is going to change rapidly. No part of our society and economy is going to be left untouched by the sustainability revolution — and that very much includes IT infrastructure.
Large corporations are realizing that sustainability is not a nice-to-have. The CEOs of the largest companies are evaluating their sustainability footprint, and the largest tech companies on the planet — Apple, Amazon, Microsoft — are leading the way. The pressure from shareholders, CEOs and boards is going to flow down rapidly to individual functional groups — including those of the CIO and CTO. We further see the rise of the Developer Ethics as a key trend — particularly within the ML community — and believe it will encompass questions of resource efficiency and sustainability.
Prescient CIOs will see that technologies they need to manage their carbon emissions are those that will drive efficiency and help them deliver other parts of their agenda — managing cost and improving agility.
We believe companies working on the four trends we outline are on the right side of history and placed to experience a sustainability tailwind to go with their general market pull related to digital transformation. But we go beyond that — we think companies which see the future and embed sustainability into their culture, product roadmap and go-to-market will have a fundamental competitive advantage — they will be where their customers and employees are headed.
We at Generation are always excited to meet, share perspectives with and invest in growth stage companies which are shaping the world for the better — and where doing so provides a sustainable competitive advantage. We think the opportunities driven by the data and sustainability revolutions in driving a more efficient ICT infrastructure stack are enormous; we are excited about what we are seeing and what we expect to see in the months and years ahead.
