High-performance compute for frontier AI
Bare-metal nodes, InfiniBand/RoCE clusters and single GPUs for training, inference and simulation. Aerospace engineering for your next model.
- Nodes
- Multi-node clusters
- Individual GPUs
- Guided training
Choose your computing power
Prices in USD, taxes excluded where applicable. Availability and price are re-checked when you order.
Answers before you begin.
Cloud GPUs, pricing and access: what you need to know to prepare your project.
Contact01What is EMT Compute?
EMT Compute is the GPU cloud by EMT Aerospace Technologies S.r.l.: rent single GPUs, full nodes and clusters in minutes, pay only for the time you actually use from your credit, and manage everything from the console or API. It includes guided AI model training and a scientific simulation lab.
02Which GPUs can I find?
From RTX 3090, 4090 and 5090 to A100, H100, H200 and B200, in data centers around the world. The catalog shows only the machines that are free right now and updates itself: a rented GPU disappears and reappears as soon as it is available again.
03How do pricing and credit work?
You top up credit in US dollars while paying in your own currency. At launch, the amount for the maximum duration you chose is reserved, but you only pay for the time actually used: on termination, the unused portion is immediately available again. Credit never expires.
04How are VAT, taxes and invoices handled?
At checkout you see taxes and the total right away: VAT, GST or sales tax is calculated automatically based on your country and any business VAT number, with reverse charge where applicable. You receive a receipt and an invoice by email for every top-up.
05How do I access a GPU?
Choose a GPU, a template (for example PyTorch or CUDA) and a duration, add your public SSH key and confirm. Within minutes the console shows the SSH command, ports, logs and real-time GPU, memory, CPU and disk usage. Your private key stays on your computer.
06I’m not an expert: can I train an AI model?
Yes. Guided training asks what you want to build, estimates data, memory, time and cost on a single GPU, node or cluster, and launches the best-suited resource in one click. Experts get every parameter plus downloadable recipes for distributed training.
07Which simulations can I run?
The simulation lab offers codes for fluid dynamics, molecular dynamics, quantum chemistry, high-energy physics, quantum computing, AI-driven physics, structural analysis, robotics and rendering, with levels for students, universities and frontier labs, cost estimates and ready-to-run commands.
08Can I automate rentals?
Yes: the EMT APIs expose the catalog, quotes, orders, metrics and transactions. Keys have permissions, an expiry date and a maximum budget; orders use an idempotency key, so a retry never creates a duplicate rental.
09How do storage and backups work?
The instance disk is temporary and is erased on termination: save results and checkpoints to your own systems before shutting down. Persistent volumes and managed backups will be enabled once their testing is complete.
10Do the GPUs belong to EMT?
Computing resources come from external providers. EMT provides its own brand, price list, workspace and ordering flow. The location and specifications of each machine are those reported by the provider in the catalog.
11From which countries can I use EMT Compute?
From almost anywhere in the world, paying in your own currency by card, Apple Pay, Google Pay or local payment methods. Countries and parties subject to international sanctions are excluded.
12How do I get support?
Email info@emtaerospacetechnologies.com with your customer code and, if needed, the order reference. For payment questions you can also use the support contact shown on your receipt.