📊 Full opportunity report: EuroHPC. The compute substrate. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
EuroHPC’s compute infrastructure underpins Europe’s AI initiatives, supporting mid-sized models but facing structural limits for frontier-scale training. The €20B AI Gigafactory plan aims to address these gaps, with ongoing procurement and policy developments shaping the future.
EuroHPC’s compute infrastructure underpins Europe’s current AI projects, supporting mid-sized model training but is not yet sufficient for frontier-scale AI training, as confirmed by recent assessments. The €20 billion InvestAI Facility’s planned AI Gigafactories are intended to address these limitations, with ongoing procurement processes expected to finalize in summer 2026.
The EuroHPC Joint Undertaking (JU) manages Europe’s supercomputing resources, including 19 AI Factories and flagship systems like JUPITER, LUMI, and Leonardo, which support a range of AI projects. These systems enable training of models up to approximately 70 billion parameters, such as Apertus on Alps, demonstrating operational viability for mid-sized models.
However, structural challenges remain. The current compute substrate is insufficient for training frontier models exceeding hundreds of billions of parameters, which the €20 billion InvestAI framework aims to address through the development of up to five AI Gigafactories. These large-scale facilities are designed to support trillion-parameter models but are still in planning and procurement phases, with selection processes ongoing through 2026.
Additionally, the infrastructure exhibits heterogeneity and geographical concentration. The flagship systems are primarily located in wealthier member states—Germany, Italy, Spain, and France—raising concerns about structural inequality and capacity disparities across Europe. The reliance on heterogeneous hardware and software ecosystems, including CUDA and ROCm, further complicates software optimization and interoperability for AI development.
EuroHPC.
The compute
substrate.
€10 billion AI Factories + €20 billion AI Gigafactories. 19 AI Factories + 13 Antennas. JUPITER #4, LUMI #9, Leonardo #10. Federation Platform shipped April 15. The compute substrate underlying every project in the seven-essay framework — and the three structural complications the framework didn’t address directly.
This is the eighth standalone essay in the European sovereign-LLM track and the first Tier 2 expansion piece. The prior seven essays documented six institutional answers plus the integrative synthesis framework. Every one of those projects depends operationally on the EuroHPC compute substrate or a national-equivalent. Apertus trained on Alps (10,752 GH200 superchips, 4,096 GPUs). OpenEuroLLM allocated millions of GPU hours across multiple EuroHPC systems. Minerva trained on Leonardo. AMÁLIA on Deucalion. Mistral on commercial cloud + ASML strategic-investor partnership. Aleph Alpha historically on alpha ONE + now Schwarz Group STACKIT + €11B Berlin DC. The compute substrate is the unifying infrastructure question the seven-essay framework didn’t address directly. Summer 2026 is the operational moment when the substrate’s strategic positioning is determined.
Two tiers. One scale gap.
The EU policy framework operates two structurally distinct programmatic tiers. The bifurcation explicitly acknowledges that current AI Factory tier infrastructure is insufficient for frontier-class model training. The AI Gigafactory framework is the EU policy framework’s operational response to the structural capability gap Finding 1 from the synthesis essay surfaces empirically.
Two Channel SXM2 Expansion Board Builts for Data Center GPUs Featuring Advanced 300G Cooling Solution Servers GPU Accelerators Board
Engineered for, the SXM2 two GPU expansion baseboard 300G supports two SXM2 GPUs ( V100) with integrated NVLink…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Six flagships. Six chromatic cross-references.
The flagship EuroHPC systems crystallize the substrate underlying the seven-essay framework. Three rank in the global TOP500 top 10. Two are exascale (one operational, one deploying 2026). All six are project-cross-referenced in the seven-essay framework. The chromatic register of each system maps to its project cross-reference.
30B+ trained
LUMI users
training
Factory
2026
70B
Supercomputing Frontiers: 4th Asian Conference, SCFA 2018, Singapore, March 26-29, 2018, Proceedings (Lecture Notes in Computer Science Book 10776)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Three cohorts. 21 European countries.
The AI Factory selection has expanded rapidly through December 2024 – October 2025 across three cohorts. 13 AI Factory Antennas in 7 EU Member States plus 6 partner countries complete the framework. The Antennas are the institutional infrastructure connecting Apertus (Switzerland) and other partner-country projects to the EuroHPC framework.
Large-Scale AI Engineering: Design, Train, and Optimize Foundation Models on NVIDIA GPU Clusters
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Three complications. Three policy gaps.
The compute substrate analysis surfaces three structurally distinct complications. These are not criticisms of EuroHPC — they are the operational realities the strategic discourse should integrate. The Federation Platform partially addresses the first; the AI Factory Antennas framework partially addresses the second; the AI Gigafactory framework explicitly addresses the third.
ENTERPRISE LLM DEVELOPMENT WITH NVIDIA NEMO FRAMEWORK: TRAIN, FINE-TUNE, AND DEPLOY CUSTOM MODELS WITH LORA, NEMO CURATOR, AND DISTRIBUTED GPU ACCELERATION AT SCALE
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Summer 2026. Three deadlines simultaneously.
The June 2026 AI Gigafactory selection process, the August 2 EU AI Act enforcement window, and the Q4 2026 EuroHPC Federation Platform second release all converge in summer 2026. This is the operational moment when the European sovereign-AI compute substrate’s strategic positioning is determined for the 2027-2029 horizon.
4 weeks ago
from now
moment
from now
from now
months
from now
The work is real across the EuroHPC framework. Substantial infrastructure built. 19 AI Factories operational or in deployment. 13 Antennas connecting smaller member states. EuroHPC Federation Platform shipped April 15, 2026. Apertus 70B operationally demonstrates Alps-tier training. The structural complications are also real. Heterogeneity hidden cost. Geographical concentration. Scale-tier bifurcation. Both can be true at once. Summer 2026 is the operational moment when the European sovereign-AI compute substrate’s strategic positioning is determined.
Implications of EuroHPC Infrastructure for European AI Leadership
The current EuroHPC compute substrate supports Europe’s AI development at the mid-sized model level but faces significant structural limitations for scaling to frontier AI training. Addressing these challenges through the AI Gigafactory framework is critical for Europe’s strategic position in global AI competitiveness. The concentration of flagship systems in wealthier nations and hardware heterogeneity could influence the equitable distribution of AI capabilities across Europe, impacting policy and investment decisions in the coming months.
European Supercomputing and AI Investment Landscape
Since its creation in 2018, the EuroHPC JU has coordinated Europe’s supercomputing efforts, with a €10 billion investment from 2021-2027, including support for AI Factories and flagship systems like JUPITER, LUMI, and Leonardo. These systems have enabled the training of models up to approximately 70 billion parameters, demonstrating operational support for current AI projects.
The €20 billion InvestAI Facility, announced as part of the EU’s broader AI strategy, aims to fund up to five AI Gigafactories capable of training trillion-parameter models. The selection process for these facilities is ongoing, with final decisions expected by summer 2026. The EU AI Act enforcement window opens in August 2026, making this a critical period for infrastructure scaling and policy implementation.
Previous assessments highlight that the existing infrastructure is adequate for mid-sized models but insufficient for frontier-scale training, necessitating the development of dedicated gigafactories. The institutional architecture includes regional AI Factories, national gateways, and large-scale AI Gigafactories, all dependent on the EuroHPC compute substrate.
“The EuroHPC infrastructure framework is the operational backbone for European AI projects but reveals critical structural limitations for frontier model training.”
— Thorsten Meyer
Unresolved Challenges in AI Infrastructure Scaling
While the planning and procurement processes for AI Gigafactories are progressing, specific details about the locations, capacities, and hardware configurations remain unconfirmed. The extent to which these new facilities will fully address the current structural limitations, especially regarding hardware heterogeneity and geographic concentration, is still uncertain. Additionally, the impact of upcoming policy enforcement, such as the EU AI Act, on infrastructure deployment and AI development remains to be seen.
Upcoming Procurement and Policy Milestones in 2026
The AI Gigafactory selection process will continue through summer 2026, with final decisions expected by then. The European Commission’s enforcement of the EU AI Act in August 2026 will also influence the strategic deployment of AI infrastructure. Stakeholders will closely monitor procurement outcomes, hardware choices, and policy impacts to gauge Europe’s capacity to develop frontier AI models and address existing structural challenges.
Key Questions
What is the current capacity of Europe’s EuroHPC infrastructure for AI training?
Europe’s EuroHPC infrastructure supports training models up to approximately 70 billion parameters, sufficient for mid-sized AI models but not for frontier-scale models exceeding hundreds of billions of parameters.
What are the main challenges facing Europe’s AI compute infrastructure?
The primary challenges include insufficient capacity for frontier models, hardware heterogeneity leading to software complexity, and geographical concentration of flagship systems in wealthier member states, which may exacerbate structural inequalities.
How will the €20 billion InvestAI Facility impact Europe’s AI capabilities?
The InvestAI Facility aims to fund up to five AI Gigafactories capable of training trillion-parameter models, addressing current capacity gaps and enabling Europe to compete in frontier AI development.
When will the final decisions on AI Gigafactory locations be made?
The selection process is ongoing, with final decisions expected by summer 2026, ahead of the EU AI Act enforcement in August 2026.
What is the significance of hardware heterogeneity in Europe’s AI infrastructure?
Hardware heterogeneity, including different architectures and software ecosystems like CUDA and ROCm, increases software complexity and optimization overhead, potentially hindering scalable AI development across Europe.
Source: ThorstenMeyerAI.com
