Next Step Fusion News Q3 2026

News & Updates

• MDPWe signed an agreement with the EPFL Swiss Plasma Center (SPC) to test the first prototype of our Modular Diagnostics Platform on the TCV tokamak. The prototype combines ECE radiometry and profile reflectometry, and testing is planned for 2027.
• MDPWe started assembling the first MDP prototype unit — the ECE radiometer and profile reflectometer. The necessary components are arriving in Luxembourg, where the unit will be assembled and tested before it goes to the planned tests on the TCV and DIII-D tokamaks that we announced earlier.
• THORWe started working on THOR testing units. THOR is a novel graphene-based Hall sensor that promises to be a compact, FPP-relevant solution for magnetic diagnostics. We received the actual sensors from our Polish partner, Łukasiewicz – Institute of Microelectronics and Photonics, and began developing units ready for testing on MCF devices.
• DANTEWe published a blog post on the DANTE design point — a compact spherical-tokamak neutron source for medical isotopes and materials testing. We lead the design activities as a core team member: target equilibria and scenarios, magnetic system layout, VDE simulations, heating requirements, and neutron and tritium cycle assessment.
• NSFsimNSFsim v2.0.0 is out, with numerous fixes and improvements. Highlights: bit-exact checkpoint and restart — a finished step can be saved and the calculation branched from it or replayed exactly, which makes long runs and controller what-if studies from a common state practical; a breakdown and burn-through module that extends the simulated scope to the very start of the pulse; a 3D passive-structure current solver coupled to the 2D equilibrium; and a fixed-boundary mode with a corrected Shafranov shift, with the option to switch from free to fixed boundary during a calculation.
• HPCWe received approval to use MeluXina, Luxembourg’s national supercomputer. We will use it for:
  • High-fidelity simulations — running ASCOT5 and other computationally intensive codes to improve accuracy and enable more advanced optimizations.
  • Synthetic data generation — building high-quality synthetic datasets to train physics surrogate models. Extending the KARHU model to more devices is the first to benefit, with more fast models in the pipeline.
  • Reinforcement learning for scenario development — training RL-based control models to advance discharge scenario development by coupling magnetic and kinetic control tasks.
• Genesis MissionThe project “Foundation Model Platform for Fusion Energy: A Multi-Modal, Multi-Device Learning Framework” led by Prof. Egemen Kolemen (Princeton University and PPPL), for which we provided a letter of interest, has been awarded under the U.S. DOE Genesis Mission. We plan to contribute private-sector expertise in plasma simulation, tokamak modeling and diagnostic data workflows.
• Alpha AllianceWe joined the Alpha Alliance of Proxima Fusion. We technically announced it in early October (Q4), but we are happy to include it in the Q3 report. We firmly believe this is the beginning of a great collaboration on building the Alpha stellarator and Proxima’s future fusion power plants.
• Nuclear FusionA paper on estimating global energy fluxes during COMPASS disruptions from magnetic measurements, co-authored by our Ekaterina Matveeva, has been accepted in Nuclear Fusion.
• NeurIPS 2026Two papers by our Dmitry Sorokin and colleagues were accepted to NeurIPS 2026 workshops: “Dynamic Plasma Shape Control Robust to Diagnostic Dropout” (RL4XS, spotlight) and “Reconstruction of Unobserved Core Electron Density in Tokamak Plasmas with a Neural Operator Transport Surrogate and Uncertainty Quantification” (Sim2Sci).
• EventsIn Q3 2026 we participated and presented at industry events and workshops including the UK Fusion Forum in London, SOFT 2026 in Aix-en-Provence (as a sponsor), FusionX:Europe in Grenoble (where we had a booth and our CPO Alexei Zhurba took part in a panel), DIII-D Industry Day, the 18th Integrated Modelling Expert Group (IMEG) meeting at ITER, ICDDPS-7 in Kiel, and more.
• Events PageWe launched a dedicated Events page that lists all the events in which we participate. We also publish a monthly LinkedIn post with the events we attend that month — see an example.

Current Opportunities

• We are actively looking for diagnostics relevant to fusion power plants (FPPs), especially novel, breakthrough solutions — as well as supply chain companies and labs, to discuss how we can leverage their experience and solutions in building the Modular Diagnostics Platform. Get in touch.
• We invite all private and public MCF device builders to start working on their diagnostics together with us. Reach out — let's build something great.

  

Useful Links

• DANTE: Hitting the Ground Running on a Compact Fusion Neutron Source Design — a blog post on the conceptual design point of a compact spherical-tokamak neutron source, where we acted as the design and engineering core team
• Estimating global energy fluxes during COMPASS disruptions via magnetic measurements — accepted in Nuclear Fusion, with our Ekaterina Matveeva among the co-authors
• Dynamic Plasma Shape Control Robust to Diagnostic Dropout — our paper accepted as a spotlight at the NeurIPS 2026 RL4XS workshop
• Reconstruction of Unobserved Core Electron Density in Tokamak Plasmas with a Neural Operator Transport Surrogate and Uncertainty Quantification — our paper accepted at the NeurIPS 2026 Sim2Sci workshop