Recommended Free Tools
Nebius has announced a phased AI factory in Lappeenranta, Finland, designed for up to 310 megawatts (MW), with the first customer capacity expected in 2027. The project is real; the often-repeated $10 billion price tag is not a budget Nebius has disclosed. The city describes the development as a multibillion-euro project, but the available announcements do not itemize its total cost.
What Nebius has announced
On March 31, 2026, Nebius said it would build an AI factory in Lappeenranta, in Finland’s South Karelia region. The planned campus is in the Pajarila district, on an industrial site of about 41 hectares (roughly 100 acres), and is being developed with Finnish data-center developer Polarnode. The project is to be built in phases, with first capacity expected to become available to customers during 2027. These are company and city plans, not evidence that the full site will be operating by then. (Nebius announcement; City of Lappeenranta project page.)
Nebius describes the fully deployed facility as one of Europe’s largest dedicated AI factories. That comparison depends on how size is measured—such as power capacity, IT load, accelerator count or campus area—and should be read as the company’s characterization of its planned end state, not a measure of capacity available at launch.
Is the $10 billion figure official?
No $10 billion project budget appears in Nebius’s announcement. The City of Lappeenranta calls the investment multibillion-euro, while Data Center Knowledge uses $10 billion in its headline and coverage. On the available disclosures, $10 billion is a reported estimate, not a clearly itemized commitment by Nebius. The sources do not establish exactly what the estimate includes: it could refer to a broader, phased build and equipment, but that scope is not confirmed. (Data Center Knowledge.)
#1 Best Overall
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
That distinction matters to investors and anyone comparing data-center projects. A building-and-site budget is not necessarily comparable to a figure that also counts substations, cooling, networking, servers and GPUs bought over several years. Nebius has not published a complete Lappeenranta capital-expenditure schedule in the cited materials. Treat the city’s multibillion-euro description and the reported $10 billion estimate as different levels of detail, not interchangeable figures.
What 310 MW does—and does not—tell you
MW measures power, or the rate at which a site can draw or use electricity. Nebius’s announcement does not provide a technical breakdown establishing whether 310 MW means grid capacity, connected capacity, total facility load or IT load. It is best understood as the project’s stated maximum planned capacity—not a count of GPUs or a promise that all of the power will be usable by customers at once.
- Grid or contracted power describes electricity planned or secured for a site; it does not prove that the connection is complete.
- Connected power means infrastructure can receive electricity, but a facility may still need construction, equipment and commissioning.
- IT load is the electricity used by computing equipment. The facility also needs power for cooling, networking, lighting and other systems.
- Installed accelerator capacity depends on the servers, GPUs, networking and storage actually deployed.
- Usable customer capacity is the portion available for sale after accounting for redundancy, maintenance and internal workloads.
Consequently, 310 MW cannot be translated directly into GPU count, revenue, floor space or operating capacity. Nebius’s Q1 2026 shareholder letter set a company target of more than 3 GW of contracted power by the end of 2026; that is a contracted-power target across the business, not 3 GW of energized, revenue-producing compute. (Nebius Q1 2026 shareholder letter.)
Why the site matters to Nebius
Nebius is building a distributed AI-cloud footprint, with capacity in owned facilities and colocations. Lappeenranta would be a major addition in Europe, alongside the company’s existing Finnish operation and planned facilities elsewhere. In its March announcement, Nebius said it had secured more than 750 MW of contracted power in EMEA across owned sites and colocations. That figure describes contracted power, not operating compute.
Rank #2
- Powered by Radeon AI PRO R9700 - Supercharge you workflow with the cutting-edge RDNA 4 Architecture and 2nd-gen AI Accelerators.
- 32GB GDDR6 with 256-bit memory bus - Tackle larger, more complex projects without limits.
- PCIe Gen 5 - Unlock lightning-fast data transfers with PCIe Gen 5 support.
- GIGABYTE TURBO Fan Cooling System - Indented metal cover and blower fan increase airflow intake, while the vapor chamber, all copper heat sink, and metal frame offer efficient heat dissipation. Optimized airflow design allows for easy multi-GPU scalability.
- Double Ball Bearing Fan - Delivers superior heat resistance and rotational efficiency for better performance and a longer lifespan compared to conventional sleeve fans.
| Location | Disclosed plan or status |
|---|---|
| Mäntsälä, Finland | Nebius says the site has expanded to approximately 75 MW and includes an operational NVIDIA GB300 NVL72 deployment. |
| Lappeenranta, Finland | Up to 310 MW planned, with first customer capacity expected in 2027. |
| Lille, France | Facility planned at up to 240 MW, according to Nebius. |
| Independence, Missouri | Planned gigawatt-scale U.S. AI factory; a specific capacity is not stated in the cited announcement. |
| Pennsylvania | Nebius reported up to 1.2 GW of power and land secured, with delivery planned in phases beginning in 2027. |
The site figures and status descriptions are company disclosures; they are not all equivalent measures of commissioned capacity. The Pennsylvania plan was included in Nebius’s Q1 2026 results. (Nebius Q1 2026 results.)
For the business, more European capacity could support customers seeking regional compute, geographic redundancy or a European location for workloads. Nebius’s strategy is to sell more than power or a building: it positions itself as a full-stack AI-cloud provider, combining accelerator infrastructure with cloud services and software for training and inference. A large campus only becomes a successful cloud business if customers use and pay for the capacity.
Hardware, cooling and the reasons for Finland
Nebius says its AI factories are intended for training and inference workloads using NVIDIA Blackwell and Vera Rubin platforms. It also said it intended to offer NVIDIA Vera Rubin NVL72 later in 2026. Those company-wide platform statements do not confirm which systems, or how many, will be installed at Lappeenranta. The hardware available when each phase opens will depend on procurement, delivery schedules and the pace of construction.
Accelerators are only one part of a high-density AI facility. Racks, networking, storage, electrical distribution, cooling and redundancy all have to work together. Nebius says Lappeenranta will use closed-loop liquid cooling designed to avoid relying on local water intake for server cooling and to minimize water consumption. That is a company description of the design, not independently verified operating performance.
Rank #3
- [Local AI Inference & 70B Model Ready] Equipped with the AMD Ryzen 7 PRO 8845HS processor, NEXUS is engineered for heavy local AI workloads. With a full-size GPU bay, it runs 70B LLMs natively without an internet connection. Ideal for AI developers and tech enthusiasts who need private environment for coding and model testing.
- [132TB Mass Storage with ZFS Integrity] Features a hybrid storage architecture (3×NVMe + 4×3.5" HDD) supporting up to 132TB. Utilizing the enterprise-grade ZFS file system and ECC memory, it prevents data corruption and bit rot—a must-have for professional photographers and video editors safeguarding 4K/8K RAW footage.
- [OpenClaw-Driven Automation Workflow] The built-in OpenClaw execution layer allows complex automated tasks to be processed locally. Even when offline, your backup schedules and AI file organization continue seamlessly. Say goodbye to monthly cloud subscriptions and high latency.
- [Dual 10GbE & USB4 Ultra-Connectivity] Experience server-class speeds with dual 10GbE ports and a 40Gbps USB4 interface. It enables multi-user real-time collaboration on large project files directly from the NAS, ensuring zero-lag editing for creative studios and production teams.
- [Open-Source ZimaOS for Total Privacy] Running on the fully open-source ZimaOS, NEXUS ensures your data stays physically on-premise with no backdoors. It acts as a "Digital Fortress" for privacy-conscious families and small businesses who demand absolute data sovereignty.
Nebius points to Finland’s low-carbon electricity mix, industrial land, established data-center ecosystem and its experience operating at Mäntsälä. The company also cites local cooperation and access to engineering, construction and university talent. These are relevant location factors, but “low-carbon” electricity does not by itself establish the project’s emissions profile, hourly power matching, grid impacts or wider local environmental effects.
Jobs and local effects are projections
Nebius projects up to 700 skilled construction jobs during the build, about 100 permanent positions once the facility is operational, and hundreds of indirect operations-and-maintenance jobs. These are forecasts, not realized employment figures. The city describes a phased project on a 41-hectare site, with first-phase completion planned for 2027. The project’s actual local effects will depend on how much is built, when phases proceed and how many roles are filled locally. (City of Lappeenranta project page.)
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How the build could be financed—and why that matters
Nebius disclosed substantial corporate fundraising in 2026, but that is not the same as a dedicated financing package for Lappeenranta. Its 2025 annual report said it raised more than $6 billion through additional equity and convertible-debt financing in Q1 2026 and was evaluating asset-backed financing and other capital sources. The company’s newsroom lists a March 2026 private offering of convertible senior notes with approximately $4.3 billion in aggregate gross proceeds. These disclosures do not specify how much, if any, is allocated to the Finnish project. (Nebius 2025 Form 20-F; Nebius newsroom.)
Different funding sources carry different trade-offs. Equity can dilute existing shareholders; debt adds repayment and interest obligations; customer prepayments can help match investment to demand but depend on winning customers and meeting delivery terms. Asset-backed or project financing can connect funding to infrastructure, but the available materials do not establish a Lappeenranta-specific structure.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Rank #4
- Professional AI & Creator Workstation: AMD Radeon AI PRO R9700 GPU with 32GB GDDR6 is engineered for AI development, professional content creation, and compute-intensive workloads.
- Massive 32GB Memory Capacity: 32GB of GDDR6 memory on a 256-bit bus provides ample bandwidth for large AI models, 8K video editing, and complex 3D rendering.
- Advanced RDNA 4 with AI Accelerators: 64 Compute Units with 3rd Gen Ray Tracing and dedicated 2nd Gen AI Accelerators for groundbreaking AI performance and visual computing.
- Professional Blower Cooling: Efficient single blower design exhausts heat directly out of the chassis, ideal for multi-GPU workstation and server configurations.
- Enterprise-Grade Thermal Solution: Vapor chamber heatsink with industrial Honeywell PTM7950 thermal interface material ensures reliable cooling under sustained professional loads.
The risks that will determine whether 310 MW becomes a business
- Grid and schedule: The full plan depends on connection timing, firm power, substations, transmission upgrades, permits and construction sequencing. The announcement does not say how much of the 310 MW is already connected or when each phase will be energized.
- Accelerator supply and economics: Delays, hardware transitions, purchase costs, rack density, cooling constraints and networking bottlenecks can affect both launch timing and the useful life of equipment.
- Customer demand and utilization: Nebius must fill capacity at prices that support the investment. Its annual report warns that inadequate demand could leave expansion capacity underutilized and harm expected returns.
- Capital intensity: Higher build costs, financing expenses, falling contract prices or faster GPU replacement can weaken returns. A completed campus can still be commercially underused.
- Competition: Nebius competes for AI workloads with hyperscalers and specialized GPU-cloud providers. Its results will depend on customer commitments, utilization, service quality and its ability to offer more than scarce power.
- Local impacts: Electricity sourcing, grid costs, backup generation, noise, construction traffic, heat rejection, water use and permitting are distinct questions. Positive project descriptions do not establish that these impacts are negligible.
For investors, the most useful milestones are not just whether construction starts. Watch for disclosures on the amount of power secured and connected, phase-by-phase capacity and delivery dates, customer commitments, GPU procurement, project funding and utilization once systems are available. Those details will show whether the planned power is turning into sellable compute.
What the announcement means for investors and cloud buyers
For investors, Lappeenranta is evidence of Nebius’s ambition to scale European AI infrastructure, not proof of a $10 billion expenditure or future returns. The company’s economics will depend on how much it spends, how quickly capacity becomes usable, what it costs to finance and operate, and whether customers keep it occupied.
For enterprise buyers, the announcement is not an availability or pricing offer. Nebius has not published a reliable public price for future Lappeenranta capacity in the cited sources. Actual commercial terms would depend on accelerator generation, reservation length, networking, storage, geography and contract size. Buyers considering GPU capacity should seek current availability and terms directly from providers rather than infer them from a campus power figure.
Quick Recap
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.
