OpenAI for Nations: How This Soft Power Lever Translates Into National Impact

It had been anticipated, and it is now official: SAP Sovereign Cloud will be available within AWS’s European Sovereign Cloud.

The German software maker is not saying much more for now, except that Cloud ERP (formerly S/4HANA Public Cloud) and the BTP platform will be the first building blocks on offer.

This “European sovereign cloud” will initially be located in Germany. The same will apply to the infrastructure that will underpin another SAP-led initiative: OpenAI for Germany. It is planned to roll it out in 2026. The broad promise, in essence: to give the German public sector access to OpenAI technologies on an Azure-based infrastructure managed by Delos Cloud (SAP’s subsidiary). In this framework, Delos Cloud should scale its capacity to 4,000 GPUs. Depending on demand, SAP would use both its own infrastructure and third-party resources to scale up, with the prospect of opening to other markets.

A program ‘in coordination with the U.S. government’

The initiative sits within the OpenAI for Countries program, itself a component of the Stargate project. Launched in May 2025, it translates into formal collaborations with governments. Notably for the development of data center infrastructures. But also, among other things, for the deployment of ChatGPT in the economy and the creation of funds aimed at startups.

Read also: AI Safety Index 2025: a troubling assessment of AI safety

The OpenAI for Countries program is conducted “in coordination with the U.S. government”; notably its “export controls authorities,” to “provide a clear alternative to authoritarian versions of AI.”

Stargate expands to Norway and the United Kingdom

The first formal collaboration OpenAI publicly referenced consisted of a memorandum of understanding. Signed in July with the United Kingdom, it paved the way for the “British Stargate,” announced in September. Under this framework, the company Nscale is expected to raise its capacity to 8,000 GPUs in the first quarter of 2026, then up to 31,000 thereafter. The aim is to enable local deployment of OpenAI models for regulated industries or critical public services.

On the ground, the government has already used the OpenAI API to develop a chatbot that helps small businesses obtain advice on business rules and the aid available to them. There is also OpenAI integrated into Humphrey and Consult. These tools are used respectively to ease administrative burdens in the public sector and to triage public responses to consultations.

Stargate has also spread to Norway. Nscale is involved there as well, with a local partner (the Aker group, a player in shipbuilding and engineering). The project will include a 500 MW site in the northern part of the country. It should be operational in 2026, starting with 100,000 GPUs and scaling to 250,000 in the long term. Norway’s AI ecosystem (startups and research) will have priority access. The surplus will be made available to the public and to the private sector in the United Kingdom, Scandinavia, and northern Europe.

Before Germany, Greece, and Estonia

Before OpenAI for Germany, there was OpenAI for Greece. The memorandum of understanding was signed in early September with the Greek government, the Onassis Foundation, and the Endeavor Greece entrepreneur community. It comprises two components. On one hand, a Greek AI Accelerator program intended to provide startups with OpenAI credits, technical mentorship, and connections to investors. On the other hand, an initiative to integrate AI into secondary education.

This initiative will rely on ChatGPT Edu, an offering already adopted at Harvard, at Oxford… and at ESCP, since October 2024. The pilot program initially concentrates on high school teachers. It is overseen by the Onassis Foundation and other local actors. OpenAI aims to contribute to creating training resources while providing technical assistance.

A similar accord had been signed in February 2025 in Estonia. The experiments were to begin in the September term, initially in 11th and 12th grade classes.

An “economic project” for the EU

In July, OpenAI announced its intention to sign the EU’s code of good practices for AI used generally as a general-purpose AI, provided that its then-version is officially approved by the AI Office.

Read also: Why OpenAI declares an ‘red alert’ against Google

In parallel, the OpenAI for Countries European Rollout project was launched. “Over the summer and autumn, we will meet with governments, businesses, and regional stakeholders,” explained Sam Altman’s company. And it recalled having applied to join consortia participating in the AI Gigafactories initiative.

OpenAI has ideas for the EU… and laid them out, among other things, through an “economic project.” It calls for augmenting the AI Factories with an AI Compute Scaling Plan to at least triple, by 2030, the available computing capacity. Among its other recommendations:

  • Invest broadly in specialized education programs, such as an “AI Erasmus”
  • Create a one-billion-euro fund for pilot projects that clearly demonstrate the societal or economic value of AI
  • Mobilize, by 2030, 10,000 “AI ambassadors” (educators, influencers, entrepreneurs)
  • Create an “Innovative Award for Responsible AI” modeled on the Sakharov Prize
  • Incorporate AI into Safer Internet Day
  • Fund an initiative supporting the development of AI applications designed with young people in mind, particularly ages 13-17
Dawn Liphardt

Dawn Liphardt

I'm Dawn Liphardt, the founder and lead writer of this publication. With a background in philosophy and a deep interest in the social impact of technology, I started this platform to explore how innovation shapes — and sometimes disrupts — the world we live in. My work focuses on critical, human-centered storytelling at the frontier of artificial intelligence and emerging tech.