Despite every government across the world talking up a digital revolution, few have truly grasped what it means to support an economy that is being fundamentally reorganised by technology. One of the most important elements is the shift away from in-house servers, software and tools, towards cloud-based services on a grand scale.
Software services increasingly underpin vast swathes of economic activity, not only increasing the productivity and efficiency of companies but also augmenting their capability and enabling new innovations to be delivered at scale. In this post, we argue that this new breed of services should be thought of as economic infrastructure; and that this new infrastructure deserves a dedicated policy approach.
Supporting efficient infrastructure is a top priority for promoting productivity and economic growth. But all too often the ambition remains stuck in a bygone era. Two decades into the 21st century, we’re still focused on the infrastructure of the last economic era – roads, railways and even fibre – rather than the cloud, software and data infrastructure that will underwrite the new one. Politicians and policymakers seem more alive to some of the immediate consumer-facing issues surrounding tech companies – such as data privacy and online content regulation – than on the strategic implications.
Where politicians have recognised the importance of these new forms of infrastructure, attention tends to be narrowly focused on security or mitigating structural risks – think Huawei and 5G, or the European Banking (EBA) Authority's caution over EU banks using cloud services based outside the European Economic Area (EEA).
Instead, to take full advantage of the internet era, governments must go beyond this limited mindset and adopt a more rounded perspective, taking into account both the risks and the opportunities presented by the new infrastructure. This post sets out three key observations that leaders must grasp:
The scale of the phenomenon: Internet infrastructure services are both widespread and critical to businesses’ core functions, with significant volumes of economic activity dependent on them;
How this sort of infrastructure differs from the old: while new technologies play a comparable role in the economy to those that came before, their nature means that there are some important differences between new infrastructure businesses and traditional utilities;
The challenges and opportunities for economic policy: as well as resilience, security, and access, it’s equally important to consider efficiency, value, innovation and diffusion if we’re to maximise the potential benefits of the new infrastructure for driving productivity.
Everything as a service, as infrastructure
To grasp the opportunities of this new world, the first task for policymakers must be to better understand the nature of the software services that now underpin whole sectors of the economy, and the economic significance of these businesses.
As we set out in A New Deal for Big Tech, infrastructure businesses are based on selling services that are more attractive than building, owning and operating the underlying assets, allowing users to benefit from economies of scale without incurring the corresponding costs. In many cases this is akin to how people already get electricity, water or phone services from utility companies without having to build their own power plants, reservoirs or telecoms networks. A classic example might be businesses using cloud services like Amazon Web Services to store their data, rather than a traditional IT set up of equipment managed on-premises (indeed, Jeff Bezos has compared Amazon Web Services to power utilities.[_] Figure 1 outlines the shift from traditional IT models to a service-based ecosystem:
The shift from traditional IT models to a service-based ecosystem
The same basic principles apply throughout the stack, ranging from cloud computing, web hosting and software development, to core technical functions such as payments and business services such as accounting which pre-date the Internet but can now be done cheaply and efficiently using cloud-based software. The standard picture of IaaS, PaaS and SaaS tools also continues to evolve, as illustrated by StackShare, a website that collates software tools used by companies. Business needs such as warehousing, advertising, logistics, office space and even human intelligence are now available as scalable, cloud-based “Anything as a Service” (XaaS) tools. And as we highlighted in Transforming Government for the 21st Century, governments should also think of digital identities, open data and public APIs as enabling infrastructure for the internet era.
The extent to which businesses are dependent on these technology services varies according to the business, and the service provided. For some services, such as analytics services or productivity software, failure would be disruptive but not fatal – there would be costs in terms of lost information and time, but the business’s survival is not likely to be at stake. However some infrastructure services, such as payments or cloud storage, can be thought of as critical infrastructure – any significant or prolonged failure of one of these could undermine the business’s ability to carry out its activities to a potentially existential degree, while possibly triggering a domino effect throughout the rest of the economy. It’s also important to note that many infrastructure businesses higher up the stack are themselves dependent on infrastructure businesses lower down the stack.
Crucially, the scale and potential of these services is significant. The global cloud computing market is forecast to be worth over $400bn by 2020; WordPress is estimated to power 30% of websites, Stripe processes hundreds of billions of dollars in online transactions annually, GitHub hosts over 100 million code repositories and Amazon Web Services has a market share of at least 33% in the cloud infrastructure market – more than the next four firms (Microsoft, Google, Alibaba and Tencent) combined. At that scale, some of these companies may now be systemically important to the economy, with 26% of EU enterprises (and 41.9% of UK businesses) having used cloud computing services in 2018. Indeed, figures from the UK, Germany and the Netherlands show that the most intensive 'digital using industries' now account for the bulk of their productivity growth, while 'digital producing industries' play the same role in the US – perhaps reflecting not only the strength of the US tech sector but how others benefit from it.
The potential for technology to radically transform startups and entrepreneurship can be seen in the rise of the ‘zero-stack start-up’. This is an inversion of the ‘full-stack’ concept, which describes a person, team or organisation which able to do everything required to deliver a product or service. With many necessary competencies now commoditised and delivered as a service over the internet, technology has massively reduced the barriers to entry for many new businesses. By stripping away the need for companies to create, own and manage their own infrastructure, the energy and resource required just to get a company up and running has been massively reduced, while firms also benefit from a dedicated team iteratively improving the service for all users. As such, founders can be more productive by focusing much more strictly on their core proposition, rather than wasting time and money on building up common backend functions. In turn, the ability to operate at scale is no longer a sufficient differentiator in business activity, weakening incumbents’ advantage and mitigating the gap between leading and laggard firms by forcing them to compete on quality and value-for-money rather than scale alone.
How is this new class of infrastructure different from what came before?
Internet-era economic infrastructure shares many characteristics with traditional hard infrastructure: relatively high upfront investment costs with comparatively low marginal costs; network effects meaning that services become more valuable the more users they have; high potential spillovers (new use cases emerging for tools that weren’t envisaged in their initial design), with the concomitant risk that the market may underprovide these goods.
But on the other hand, technology means that there are some significant differences: crucially, given that these services are delivered online, marginal costs associated to supply new users are even lower than for traditional infrastructure such as utilities networks. There are no physical barriers to developing a new service, such as land ownership and permissions. And, for now at least, most cloud sectors are relatively free of regulatory constraints. These factors should mean that cloud markets remain relatively contestable, if not competitive.
However, the lack of geographic limitations also means that the market is global – so if dominance is established, there are no limits. This factor also means that governments can lack the necessary levers to regulate or influence the major cloud providers, should they need to do so.
Kushida, Murray and Zysman summarise this comparison by proposing that these cloud services should be thought of as a dynamically configured utilities, combining “the scalable, always available, pay for use and – economic critical – attributes of traditional utility models with the configurable, technologically differentiated and non-geographically bound aspects required by a highly competitive services market”.
Challenges and opportunities for policy
The significance and complexity of this shift mean that policymakers need to step back and take a holistic view of its implications, both positive and negative. In the case of traditional infrastructure, its contribution to productivity is well understood. Governments and regulators therefore feel justified in looking closely at the security, resilience, quality and efficiency of the country’s infrastructure, and taking steps where necessary to improve it. In the UK, the National Infrastructure Commission has been established to provide a long-term assessment of the country’s infrastructure needs. In some cases, it’s considered appropriate to invest public money in infrastructure.
In the case of the new infrastructure, governments have so far focused on minimising risks at the very bottom of the stack: data centres are included in the government’s definition of critical national infrastructure; the EU’s 2016 Network and Information Systems Directive set a framework for digital services providers to improve the physical and cyber-security of their systems. There will inevitably be trade-offs between ensuring the security of IEI and maximising its efficiency – this has cropped up in the EBA's cautionary guidance for European banks regarding cloud services where data is stored outside the EEA, of which an overly zealous reading might restrict banks' options and increase their costs compared to global competitors.
It is also evident in the French and German proposal for Europe to create its own cloud service to protect ‘data sovereignty’. Again, the security benefits of developing cloud capacity that is subject to closer European regulatory control are likely to be outweighed by the inefficiency of redirecting funds to duplicate non-EU services that companies will still use anyway.
In contrast, there has been very little attention paid to maximising the potential benefits of internet-era infrastructure (although, in the UK, Liberal Democrat and Conservative proposals to allow cloud computing spend to be counted towards R&D tax credits at least acknowledge its importance). Likewise, although competition in digital markets is a high-profile topic, it is almost exclusively approached from the perspective of consumer welfare, with little attention paid to whether markets in essential internet infrastructure are working well from the perspective of business customers.
In order to avoid missing an increasingly important factor of productivity, all these questions should be looked at, together with the issue of whether there are some technology infrastructures that the market is unlikely to provide or where society has a greater stake (e.g. identity), and that therefore the government should facilitate.
The internet era is firmly here. This blog introduces the economic infrastructure underpinning today’s economy, highlights how our historical toolkit has limited use in this new age, and sketches the contours of the challenges and opportunities for policymakers. Fundamentally, this new class of infrastructure has significant potential to improve productivity, generate competition and improve business dynamism. But in order to realise these benefits, political debate must catch up to the internet era. This means governments must wrap their heads around this new world, understand how it differs from traditional infrastructure, and step up to maximise the opportunities ahead.
To that end, some of the issues we’ll be considering going forward are:
How can governments improve their understanding of these services and their role in the economy?
How should states trade off security and regulatory concerns about internet-era infrastructure against the opportunities for innovation and improved efficiency?
Are there steps for governments to take to maximise the productivity benefits of these services across the economy and in the third sector?
What digital infrastructure should government provide, not only in the public sector but for all society?