Internet Computer pitches sovereign cloud model as AI reshapes software infrastructure

Growing concerns over data sovereignty, geopolitical tensions and the rapid adoption of artificial intelligence are prompting businesses and governments to reconsider how their digital infrastructure is hosted and controlled.

Internet Computer founder Dominic Williams is advocating a different approach to cloud computing, arguing that decentralised infrastructure could give organisations greater control over their software, data and online services.

The Internet Computer (ICP) is designed to extend the internet’s role beyond connecting websites and applications. Its network can also host and run software, including its data, through a decentralised infrastructure built around software units known as canisters.

Williams describes the concept as a new kind of cloud built using the architecture of the internet itself. The proposition is that applications can operate across a distributed network rather than depending entirely on conventional cloud providers and their centrally managed infrastructure.

The discussion comes as digital sovereignty becomes a growing concern for governments and organisations. Questions around where data is stored, which laws apply to it, who can access it and whether services can be disrupted by external decisions are increasingly relevant to cloud infrastructure choices.

Artificial intelligence is adding another dimension to the debate. Much of today’s cloud infrastructure was designed around human developers, system administrators and conventional software deployment. Williams argues that the next generation of infrastructure needs to accommodate AI agents that can build, manage and operate applications with less human intervention.

One of the proposed developments is the introduction of cloud engines on the Internet Computer. These are intended to provide an environment where developers and businesses can install, operate and manage software through a more integrated cloud platform.

Another concept being promoted is Open SaaS, which aims to offer open alternatives to conventional Software-as-a-Service (SaaS) products. The idea is to make applications available for installation on a cloud engine in a way comparable to installing an app on a smartphone.

These applications are expected to be developed as what Williams calls ‘AIware’, software with artificial intelligence built into its core functions.

An example outlined by Williams involves an HR team handling requests through a plain-language instruction. An employee could ask an AI-powered system to identify overlapping paid time off (PTO) requests within the Customer Success department, approve the remaining requests and prepare emails asking affected employees to change their dates.

The proposed system would handle several connected tasks through a single instruction, reducing the need for staff to navigate different software tools and complete each action manually.

Williams expects startups to be among the early adopters once cloud engines become generally available. He believes smaller businesses will have a strong incentive to use the technology to operate more efficiently and reduce their dependence on conventional software infrastructure.

He also anticipates network effects as applications hosted on different cloud engines become interoperable, subject to access permissions. This could allow businesses to connect software services across different environments, a model Williams compares with the interoperability that helped drive the early growth of the internet.

The proposition, however, faces practical questions around adoption, data protection, regulatory compliance and the management of sensitive business information. Decentralised hosting does not automatically resolve data residency or privacy requirements, particularly where organisations must comply with specific national laws. Businesses will need clear assurances about data access, governance and operational responsibility before moving critical systems to a new infrastructure model.

The Internet Computer’s underlying architecture supports canisters that store data and execute applications across a network of independently operated nodes. Its governance system also provides a mechanism for decisions about network development and operation.

The wider ambition is to make cloud infrastructure more open, interoperable and suited to AI-driven services. Whether the Internet Computer can attract mainstream businesses and compete with established cloud providers will depend on the availability, performance, cost and reliability of its proposed services.

For now, the cloud engine and Open SaaS concepts offer a glimpse of how Williams believes software infrastructure could evolve as AI becomes a routine part of building and operating digital services.