CSC – IT Center for Science manages LUMI’s day-to-day service from Kajaani, Finland, but it does not allocate every resource or set all access policy alone. CSC hosts the HPE Cray EX system, produces and develops the service, and supports projects that receive allocations through EuroHPC or participating countries. LUMI’s “Europe’s biggest” label needs a date: the official EuroHPC listing accessed in 2026 places it 11th globally in the June 2026 Top500 ranking, while CSC reported ninth globally in June 2025.
What LUMI is
LUMI is a EuroHPC pre-exascale supercomputer and shared research infrastructure in Kajaani, Finland. It is an HPE Cray EX system intended for several classes of work rather than a single-purpose artificial-intelligence appliance. EuroHPC lists deep learning and other AI workloads, traditional large-scale simulations, and massive data analytics among its uses.
The machine is divided into four service partitions:
| Partition | Primary purpose |
|---|---|
| LUMI-G | GPU-accelerated computing |
| LUMI-C | CPU-based computing |
| LUMI-D | Data analytics |
| LUMI-K | Container-cloud services |
The hardware documentation describes an HPE Slingshot 11 interconnect for the hardware partitions other than LUMI-K. This design lets a project select a partition that fits its algorithms, scaling pattern and data movement instead of treating all LUMI capacity as interchangeable.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware match#1 Best Overall
What CSC actually manages
CSC identifies itself as LUMI’s service producer. Its responsibilities include hosting the machine in Kajaani, producing and developing the LUMI Supercomputer service, operating the user-facing environment and supporting allocated scientific use.
Hosting and service production
CSC provides the facility and operational service through which approved projects run jobs. Users can run their own code, install Linux-compatible applications, use CSC’s software collection and build parallel programs with MPI or OpenMP. GPU development options include HIP, OpenMP offload and OpenACC.
Support and training
CSC documents a batch-queue service for parallel workflows and offers training at beginner, intermediate and advanced levels. Linux and programming experience—such as Fortran, C, C++ or Python—is useful, but the training path is designed to help users build those skills.
Rank #2
Responsibilities that remain with users
CSC’s service description does not make the operator responsible for every project obligation. Customers remain responsible for their data, backups and information security. That distinction matters when planning a project: receiving compute time does not transfer ownership or governance of the project’s data to CSC.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteWho decides access and allocations?
LUMI is not an unrestricted public sign-up service. You need a LUMI account and a project with resources allocated to it. The route depends on your country and on whether the request is made through a national allocation process or a EuroHPC call.
- Identify your allocation route. Check the national organization serving your country or the applicable EuroHPC call. LUMI’s access guide distinguishes resource allocators and EuroHPC allocations before and after 15 April 2026.
- Submit or join an approved project. A resource allocator creates the project and invites the principal investigator and project members.
- Receive an account and allocation. Once the project is approved, members use the account and the project’s assigned compute and storage resources.
- Use the appropriate partition. Select GPU, CPU, data-analytics or container-cloud capacity according to the workload and the project’s allocation terms.
CSC’s policy says some academic use cases covered by its free-of-charge policy are free. That is not a blanket promise that all academic, commercial or other use costs nothing. Where a service package applies, accounting can be expressed in CPU-core-hours or GPU-hours for compute and TB-hours for storage. Check the live national and EuroHPC instructions before applying because eligibility, calls and accounting rules can change.
How large is LUMI?
Keep performance figures tied to their metric and date. EuroHPC’s current system listing, accessed in 2026, reports the following:
| Measure | Official figure and qualification |
|---|---|
| Sustained performance | 386 petaflops (EuroHPC current listing) |
| Peak performance | 539.13 petaflops (EuroHPC current listing) |
| Storage | 117 petabytes across flash, Lustre and Ceph storage (EuroHPC current listing) |
| Top500 position | 11th globally in the June 2026 ranking (EuroHPC listing) |
These figures should not be treated as the same measurement. Sustained and peak performance describe different capabilities, and Top500 rankings use the HPL benchmark. CSC’s 10 June 2025 release reported 379.7 petaflops on HPL and ninth place in that June 2025 list. The two snapshots differ in date and reporting context.
Free tools Windows power users keep installed
One-click scans. No signup required.
Is LUMI still Europe’s fastest or biggest supercomputer?
There is no timeless yes-or-no answer. “Europe’s biggest” is a framing used for LUMI, while rankings change with the release date, benchmark and whether a list covers EuroHPC systems or the global field. The official EuroHPC listing available in 2026 places LUMI at number 11 worldwide in the June 2026 Top500 ranking. CSC’s June 2025 announcement placed it ninth worldwide in the June 2025 list.
When comparing LUMI with another system, use four checks:
- the ranking date and benchmark, especially Top500 HPL versus another measure;
- workload fit, such as GPU AI, CPU simulation, data analytics or cloud containers;
- eligibility and the national or EuroHPC route for obtaining an allocation; and
- storage capacity and input/output requirements, not just floating-point performance.
What has LUMI delivered so far?
In its 2025 release, CSC reported more than 3,000 scientific projects, nearly 7,000 users across Europe, 140 million GPU-hours and over 3,800 million CPU-core-hours. Those are CSC-reported totals for that reporting period, not a live counter or a guarantee of present-day usage.
The figures illustrate LUMI’s role as shared infrastructure: capacity is divided among projects and countries rather than sold as on-demand retail computing. The machine supports research teams that could not economically operate an equivalent system themselves.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Best Value
What is the LUMI AI Factory?
The LUMI AI Factory is already using the existing LUMI system and combines AI-focused computing, datasets, training and expert support. It is a programme around the infrastructure, not a separate claim that every planned capability is operating today.
Planned successors
CSC says LUMI-AI is planned for the second half of 2027 and LUMI-IQ for autumn 2027. CSC projects that LUMI-AI will provide ten times the AI capacity and nearly twice the HPC capability of current LUMI. Those are host projections for a planned system, not measured operating results.
In a 10 June 2025 CSC release, Managing Director Kimmo Koski said: “The upcoming exascale LUMI-AI supercomputer and quantum computer LUMI-IQ will form the most advanced public quantum-accelerated HPC+AI research infrastructure in the world.” That is Koski’s forward-looking assessment, not an independently established ranking.
Quick Recap
What a prospective user should prepare
- A research or development project that fits an eligible national or EuroHPC allocation route.
- A principal investigator and participating members for the project application.
- An estimate of GPU-hours or CPU-core-hours, storage in TB-hours and expected data movement.
- Code that can run in Linux and, where relevant, uses MPI, OpenMP, HIP, OpenMP offload or OpenACC.
- A plan for data ownership, backups, security and any sensitive-data restrictions, which remain project responsibilities.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
