[ Announcement 001 ]gpuisrael.com
SovereignGPU computelaunching in Israel_
GPU Israel is building a dedicated GPU cloud on the NVIDIA Blackwell platform, including B300 and RTX PRO 6000, hosted in a Tier III data center in Israel. Capacity will be available by the hour, starting from a single GPU. Your models, your data and your jurisdiction remain under your control.
System readiness0%
[01] Our intent
Advanced compute.
Hosted in Israel.
Israel is home to a world-class AI ecosystem, yet access to the latest generation of GPUs within the country remains limited. GPU Israel intends to close that gap.
01
Blackwell, in-country
The latest NVIDIA architecture, deployed locally for low latency and in-country support.
02
Flexible engagement
No minimum cluster size and no long-term commitment required to begin. Hourly, weekly, monthly and long-term terms.
03
Security by design
Dedicated hardware under Israeli jurisdiction, designed for private models and AI agents that handle sensitive data.
[02] Infrastructure
The fleet
B300
NVIDIA HGX 8× B300 for model training and large-scale inference
288 GBHBM3e per GPU
RTX PRO 6000
Server edition in 1× and 2× configurations for fine-tuning and inference
96 GBGDDR7 per GPU
Network100 / 400G
StorageNVMe
FacilityTier III
LocationIsrael
[03] Security architecture
Four layers
of protection.
Private models and AI agents operate on an organisation's most sensitive data. The infrastructure beneath them should remain under the organisation's control and within national borders.
L-01
JurisdictionHardware and data are governed by Israeli law, with no cross-border data transfer.
L-02
FacilityTier III data center with redundant power and cooling, and controlled physical access.
L-03
HardwareDedicated servers. Workloads run on hardware that is not shared with other customers.
L-04
DataModels and datasets remain in Israel, supporting private inference and AI agent deployments.
[04] Clients
Built for AI teams
AI startupsResearch institutionsMid-market enterprises
Model trainingFine-tuningPrivate inferenceAI agentsLocally hosted LLMsRendering
[05] Launch roadmap
Fleet procurement
Data center integration
Private beta
General availability
[ Early access ]
Reserve
initial capacity
Organisations that register early will receive priority allocation when the platform becomes available.