In early 2024, the Netherlands’ domain registry (SIDN), which maintains .NL, announced it was moving part of its critical work to Amazon Web Services (AWS), Amazon’s cloud computing environment. Framed as routine modernisation, this choice triggered parliamentary motions and a year of debate about who really runs this critical part of the ‘Dutch’ internet.
In a new open access paper in Information, Communication & Society, ARTICLE 19’s interim Director of Digital Dr. Corinne Cath outlines the human rights implications of this growing reliance on Big Tech’s clouds. Internet governance, the work of keeping the internet running, sits with a distributed set of public-interest stewards: registries, standards bodies, network operators, amongst others. As more of these stewards move their ‘core operations’ onto the clouds operated by Amazon, Microsoft, or Google, the conditions for free expression and access to information online increasingly depend on the products, prices, and policies of these three Big Tech firms.
She uses this recent Dutch case study to introduce ‘cloud drift’: the slow structural transformation that happens when stewards of critical internet infrastructure migrate to the ‘production environments’ of Amazon, Microsoft, or Google. Even partial migration, or just preparing for it, can quietly reshape internet governance to the detriment of their ability to facilitate freedom of expression and access to information.
The paper traces three patterns in how moving to hyperscaler clouds reshapes internet governance organisations from within, and what that means for ARTICLE 19’s mission to protect freedom of expression and access to information online.
People: Cloud-specific expertise displaces mission-focused knowledge. Long-time engineers who knew their networks intimately and could ‘just pick up the phone’ to peers or other actors in the internet ecosystem to keep the internet running leave or get sidelined.
Systems: Control over ‘production environments’ transfers to proprietary clouds. Foundational decisions about compute, pricing, and service availability sit with the hyperscaler, not the steward.
Purpose: Commercial logics displace public-interest mandates. Stewards of critical infrastructure begin to act like software vendors, chasing scale rather than the public mission that justified their existence.
This pattern likely extends well beyond internet registries. Universities, media organisations, governments, civil society: any institution whose authority rests on operational independence and trusted expertise risks trading that authority for the perceived convenience and cost savings of hyperscaler clouds.
The encouraging news is that cloud drift is not destiny. In December 2025, DNS Belgium announced it would leave AWS for European providers by 2027 for geopolitical reasons. Registries can change course. But it requires them to see their cloud choices are governance choices, not just technical ones. Treating them as such is the first step in resisting drift, for registries and for the universities, ministries, and media organisations considering the cloud.