A recent Organization for Economic Co-operation and Development (OECD) policy paper on competition in cloud computing frames the sector as fragile, with an imminent threat of anticompetitive behavior. As the paper notes, 70-80% of the global cloud share is controlled by Microsoft, AWS (Amazon Web Services), and Google. This paints the picture of a concentrated market.
It isn’t, however, the full story. The cloud market continues to experience shifting market shares, meaning that entrenched incumbents may not be entrenched for long. For instance, AWS’ market share dipped from 45% in 2019 to 39% in 2021. Despite the continued potential for such rapid shifts, coupled with trends toward declining prices and fast-paced innovation, the paper’s authors still raise concerns about the potential for future anticompetitive behavior, flag how it may come about, and raise options for enforcers and policymakers to address it.
Are they jumping the gun? Let’s consider the evidence.
The OECD’s competition concerns are rooted in a host of different practices within the cloud-computing industry that are purportedly or potentially anticompetitive. They are concerned, for example, about switching and lock-ins. Contractual impediments can prevent companies from switching from incumbent cloud services to newer, more innovative ones. Proprietary APIs that restrict data portability and interoperability with external services or alternative application programming interfaces (APIs) also create exit barriers for consumers by complicating transfers of their data. High egress fees may also raise barriers to switching cloud providers and ostensibly hurt competition.
The paper claims that a lack of common data standards hampers multi-cloud capabilities. And pricing models by big firms that offer large scale discounts to new customers supposedly resemble loyalty rebates with which smaller firms may not be able compete. These could harm consumers in the long run by reducing the competitive pressure that these big firms face to keep prices relative to quality low.
With a mix of high market concentration and possibly anticompetitive practices, it makes sense that the OECD is concerned. But at least for now and into the foreseeable future, these concerns are wildly overstated. As things stand, the cloud-computing space is fiercely competitive. And many apparently anticompetitive practices have pragmatic explanations that warrant exploring.
For one, switching and lock-in fears seem overstated. Although data formatting and transferability might not be standardized between companies, this does not necessarily mean that anticompetitive behavior is taking place. Switching IT systems always comes with costs, and the cost of migrating company databases between IT systems before cloud computing was far greater than it is today. The fact that these costs exist also does not necessarily suggest there are excessively high barriers.
For example, multi-cloud data homing is mainstream. As a report by Flexera noted: “In practice, businesses adopt multi-cloud and hybrid-cloud strategies… 70 % of cloud-using companies rely on multiple providers.” This indicates that concerns about switching and lock-in have not yet manifested in real market data. Instead, the industry standard of using multiple providers is a sign of healthy competition.
“Above cost” egress fees are also certainly a concern for keeping consumers “locked in.” But we must again consider market conditions. None of the three major cloud services charge anything for data ingress (uploading data into the cloud), even though accepting and storing incoming data comes with costs to them. Instead, cloud providers recoup the costs of moving data through egress fees. It should be noted that these fees are applied not only when a customer switches providers, but also when data is moved within the same service.
Regardless, the OECD makes recommendations to address these purported problems. They suggest that companies should be forced to standardize data to enable portability and reduce switching costs. But instead of facilitating competition, forced standardization can freeze infrastructure by discouraging experimentation with different standards. This could create a lowest-common-denominator blueprint that would slow, rather than advance, the race of emerging services.
Competition officials should also note that, far from needing a top-down mandate, the industry is already coalescing around bottom-up, voluntary frameworks that facilitate interoperability without sacrificing experimentation between different standards. For instance, OpenTelemetry is an open-source, cloud-provider-neutral collection of APIs and other tools for generating, instrumenting, collecting, and exporting telemetry data in order to analyze software behavior and performance. Similarly, the Cloud Neutral Computing Foundation (CNCF) offers a burgeoning catalogue of open-source, vendor-neutral “cloud-native” platforms, technologies, and architectures. These allow for different systems to interoperate in a secure, manageable, resilient, sustainable, and observable way.
Unlike these bottom-up arrangements, top-down rules come with heavy tradeoffs. On the investment front, the three largest providers alone poured roughly $200 billion into global data-center buildouts between 2022 and 2024. This not only indicates a fiercely innovative and competitive industry, but it also shows that incentives need to be high for these companies to continue to invest and innovate. Mandates that privilege “one-size-fits-all” hardware could also destroy incentives to design custom silicon (GPUs, TPUs, AI chips) that are currently pushing performance and cost frontiers.
History tells frightening tales of the tradeoffs of preemptive regulation in tech. In 1996, the Telecommunications Act required companies to provide “equal access” to their network infrastructure. This hampered innovation in telecommunications infrastructure, as the fruits of any innovation by one company would result in free-riding by rival companies. The EU’s data-localization standards have also had catastrophic tradeoffs. These strict rules and restrictions around international storage, transfer, and data processing have crippled Europe’s cloud sector. Europeans bear the costs through worse products and less national productivity, leaving their firms less internationally competitive.
It is therefore important to remember what would be lost with preemptive regulation. Cloud computing has demolished entry barriers for startups. But layered compliance obligations risk re-erecting them. Furthermore, the costs of regulation are more easily borne by larger players who benefit from economies of scale, as these let them spread out costs over larger volumes of business. Ironically, this means that cumbersome new rules could entrench the dominance of the corporate giants they target.
Meanwhile, the real prices of infrastructure as a service (IaaS)—a cloud-computing model that allows users to rent and access computing resources like servers, storage, networking, and virtualization on-demand—have been dropping by roughly 20% a year. This is evidence that monopoly price gouging is not currently a problem.
Finally, there is a geopolitical dimension. If the OECD’s interventionist script is copied wholesale by domestic competition-enforcement agencies, policymakers, or tech-market regulators in various jurisdictions, regulatory fragmentation will widen. All the while, Chinese companies operating under looser regimes are extending their global reach.
Fortunately, there is a better way. Existing antitrust remedies are largely sufficient to deal with cloud computing. And nothing about cloud computing in particular merits costly, broad, and sweeping preemptive regulation. Instead, these issues should be dealt with case-by-case, while enforcement officials continue to prosecute specific companies that engage in anticompetitive practices.
Government agencies, enforcers and regulators tasked with monitoring entire markets for compliance with ex-ante restrictions on potentially procompetitive business practices may also find themselves left with fewer resources to vigorously prosecute wrongdoers who do harm to consumers and competition.
Instead of enforced industry standards, the OECD needs to prioritize what promotes consumer welfare most: unfettered competition. Regulators must resist the urge to pass sweeping regulations that hurt innovation and look to existing antitrust law around tying and bundling, monopolization, and other offenses to address anticompetitive cloud-computing practices as they emerge.
Competitive and rapidly evolving high-tech markets have a track record of producing better, more creative solutions to address the problems of the day than leviathan regulators. Competition enforcers should focus on preserving the conditions for businesses to freely compete, innovate, and experiment. It is not their job to redesign a better economy through proscriptive and costly rules that could burden rather than promote innovation and competition.
