Work email*
Company*
Website
Full name*
Job title*
Country / region Select an option United Kingdom Ireland Germany France Netherlands Sweden United States Canada Australia Singapore Japan UAE India Other
Organisation type Select an option Enterprise Startup / scale-up AI lab / research group Public sector / government Cloud / managed service provider Other
Current setup / provider
Job pattern Select an option One-off training run Recurring training / retraining Always-on inference / serving Bursty / experimental Batch / scheduled jobs
Software stack Select an option CUDA / NVIDIA only ROCm / AMD acceptable Not sure / no preference
Cooling preference Select an option No preference Air-cooled Direct-to-chip liquid Immersion
Budget (monthly) Select an option Under $10k / month $10k–$50k / month $50k–$250k / month $250k–$1M / month $1M+ / month
Scale (number of GPUs)* Select an option 1–8 GPUs (single node) 8–64 64–256 256–1,024 1,024+ / not sure
Deployment unit Select an option Individual GPUs / instances Bare-metal nodes (8-GPU HGX) Rack-scale (NVL72 / NVL) Not sure — advise me
Contract length Select an option Weekly / monthly 1–3 months 3–12 months 1 year+
Data residency / sovereignty Select an option No requirement EU data residency EU-incorporated sovereign (CLOUD Act-free) UK data residency US only Air-gapped / on-prem
Compliance needs Select an option None / not sure SOC 2 Type II ISO 27001 HIPAA (BAA required) GDPR / DPA FedRAMP
GPU model preference Select an option No preference — recommend for me NVIDIA H100 NVIDIA H200 NVIDIA B200 NVIDIA B300 NVIDIA GB200 NVL72 NVIDIA GB300 NVL72 NVIDIA GH200 NVIDIA Vera Rubin (2026+) NVIDIA A100 NVIDIA L40S AMD MI300X AMD MI325X AMD MI355X Other / not sure
Exact quantity
Primary workload* Select an option LLM / foundation model pre-training Fine-tuning / post-training (LoRA, RLHF) Real-time inference / serving Batch / offline inference HPC / scientific simulation Mixed / general purpose
Target start date* Select an option Immediately (0–2 weeks) Within 1 month 1–3 months 3–6 months 6 months+ / just exploring
Interconnect requirement Select an option Single node — NVLink (up to 8 GPUs) Multi-node — InfiniBand NDR 400G Multi-node — InfiniBand XDR 800G Multi-node — Ethernet / RoCEv2 Not sure — advise me
Storage type Select an option Parallel file system (WEKA / VAST / DDN) Local NVMe scratch S3-compatible object Standard network file system Not sure
Storage capacity
Interruptible workloads OK? Select an option Yes — we checkpoint frequently No — needs guaranteed uptime Not sure
Commitment model* Select an option On-demand Short-term reserved (weekly / monthly) 1-year reserved 2–3 year reserved
Evaluating other providers? Select an option Not yet Yes — comparing a few Already have quotes Just exploring the market
Additional notes