The United Kingdom Atomic Energy Authority and the U.S. Department of Energy’s Princeton Plasma Physics Laboratory signed a Joint Declaration of Intent on September 14, 2026, to explore linking UKAEA’s SUNRISE AI-computing platform with PPPL’s STELLAR-AI platform. The declaration builds on a memorandum of understanding the institutions signed in June covering cooperation in advanced computing and artificial intelligence.
The September increment is a more specific federation plan rather than the beginning of UK-U.S. cooperation itself. UKAEA and PPPL describe a system in which researchers could use experimental data from UKAEA’s MAST Upgrade and PPPL’s NSTX-U facilities, train models using results from both facilities and move models or computing jobs between the two computing environments. The public material still describes that work as proposed or under exploration rather than as an operating technical federation.
The consequential question is therefore not simply whether the laboratories can combine more computing capacity. It is whether data, models, workloads and research practices can move meaningfully between distinct facilities. The participating institutions say researchers will examine moving computing jobs between platforms that use different hardware. Making the wider federation useful would also depend on practical arrangements for data access, interoperability and governance, and on whether models developed with one facility’s data remain useful when evaluated against another facility.
That distinction matters because a federation could create a way to test models across institutional and experimental boundaries rather than only within one facility’s computing environment. But the reviewed announcements do not report a completed cross-facility validation, a demonstrated increase in predictive accuracy or an implemented shared-resource governance system. The institutions’ statements that the federation could improve models, shorten design cycles, support digital twins or accelerate fusion development remain projected benefits.
The alternative is that the declaration produces substantial institutional coordination but little functional technical federation. That outcome would look different from a system in which researchers actually obtain authorized access to shared data, transfer reproducible workloads, run models across both environments and report cross-facility evaluations. Executed access arrangements, interoperable data and software interfaces, successful workload transfers, held-out results across MAST Upgrade and NSTX-U, and operating rules for shared resources would show whether the declaration becomes durable scientific infrastructure.