AI infrastructure
Private and scalable
Airon combines AI with iron in a bare-metal infrastructure that ensures you the capability to leverage AI in your business. Our solution provides private non-shared AI resources, guaranteeing performance and the safety of your confidential data and proprietary models. Allocate, accelerate, and control with Airon.
Bare metal is the new future-proof standard
Capability without complexity
Guarantee AI capability by allocating necessary resources
Security with zero-trust
Retain full control over sensitive and proprietary data with private infrastructure
Performance with bare metal
Bare metal has an unbeatable price-performance ratio compared to the cloud
AI deployment free from obstacles
Compute resources
Ready-to-use AI infrastructure. Pre-train or fine-tune large generative models with high-performance NVIDIA® GPU resources. Guarantee capabilities with 1- to 3-year contract
Sustainable by default
Learn why the nordic climate and clean electricity makes Airon sustainable by default
Privacy and security
Learn why we own and operate the HPC data centers where the AI infrastructure is deployed
Software platform
Airon easy-to-use platform is supporting your organization entire AI lifecycle and simplifies managment, visibility and control over the infrastructure resources.
Storage capacity
Performant storage and scalable persistent distributed block and object storage (S3). The storage is non-shared and connects directly to your compute resources
Supported by the global leader in Artificial Intelligence
Our compute GPU resources is provided by NVIDIA® and runs in our safe and sustainable HPC datacenters in Europe. See available resources and read more about or technical architecture
Press release
Introducing Airon — private bare-metal AI infrastructure powered by NVIDIA
San Jose, California, 19 March 2024. Introducing Airon — private bare-metal AI infrastructure powered by NVIDIA®. Airon provides private non-shared AI resources, guaranteeing performance and the safety of your confidential data and proprietary models. Allocate, accelerate and control with Airon.
Key Features
Capability without complexity or lead time: Demand for AI resources outstrips supply. Guarantee AI capability by allocating necessary resources.
Capability without lead time or complexity: Demand for AI resources outstrips supply. Guarantee AI capability by allocating necessary resources.
Security with zero-trust: Retain full control over sensitive and proprietary data with private infrastructure.
Performance with bare metal: Bare metal servers offer an unbeatable price-performance ratio compared to the cloud.
Availability and Pricing
The first batch of 512 GPUs (NVIDIA® HGX™ H100) is released and available to allocate on a first-come, first-served basis. Pricing starts from $3.6 an hour on a 12-month commitment. For more information, visit https://airon.ai.
The infrastructure is deployed during Q2 and Q3, to HPC data centers in the Nordics, Europe. Next-generation Blackwell B200 revealed at NVIDIA GTC, will be available later this year and are open for pre-booking.
For media inquiries, please contact:
Robert Lidberg <[email protected]>, +46 73 530 50 94 (PST)
Resources
NVIDIA® HGX B200
The new Blackwell GPU architecture, available later this year!
NVIDIA® GH200
New Grace Hopper™ 200 Superchip available for reserveration!
NVIDIA® HGX™ H100
Unprecedented performance and scalability with 8x H100 SXM5.
NVIDIA® L40S
The latest Ada Lovelace Architecture 2-8x L40S SXM or PCI-E.
NVIDIA® RTX A6000
Powered by the NVIDIA Ampere, 48 GB GDDR6 Memory.
NVIDIA® RTX A5000
Powered by the NVIDIA Ampere, 24 GB GDDR6 Memory.
Iron powered by nature
Airon is engineered for large-scale workloads that requires large amounts of energy, this is facilitated by our self-operated HPC data centers located in the Nordic region of Europe. Leveraging domestic renewable energy sources like hydro, wind, and solar power, we maximize sustainability while capitalizing on the naturally cold climate for efficient cooling, ensuring a stable infrastructure for AI operations. This is Sustainable by default.
Answers
Here are a few questions and answers. If you're unable to locate the information you need, don't hesitate to contact our sales team below.
Technical
architecture
European HPC data centers
We own and operate the data centers where the infrastructure is deployed
Performant and secure
Optimal alternative for both on-premise and cloud deployment
Isolated non-shared infrastructure
Your compute and storage resources are directly connected with each other
Storage solutions
Large high speed NVMe disks and distributed block and object (S3) persistent storage
3.2Tb/s InfiniBand ingress speed
Run multiple bare metal GPU resources that is interlinked with fast InfiniBand
SLA and reliable technical support
We offer optional service level agreements and technical support
10 Gbit ethernet egress speed
Transfer outbound data fast and without caps