GPU access, secure hosting and orchestration, designed to keep national AI workloads in-country.
One national facility, 2,000 GPUs planned, in-country.
Your cloud should not sit abroad.
When data, models and compute sit on foreign infrastructure, control sits there too. For government and national enterprises, that is worth settling before AI becomes core to operations.
MintAI keeps data, models and compute in-country under your governance, so you adopt AI at scale without giving up control.
What the stack is built to provide.
Facility
Secure, in-country environments that host sensitive workloads under your governance, isolated and auditable throughout.
Compute
On-demand GPU capacity when you need it, with reserved access for sustained, heavy workloads.
Platform
Orchestration that schedules and balances workloads, scaling from a first production job to organization-wide adoption.
Models
Models promoted into production securely, without leaving national borders.
The footprint behind it.
One national facility, designed to grow with your workloads.
- 2,000
- GPUs planned, in-country
- 36 EFLOPS
- Peak FP4 performance, designed for
- 5 MW
- Facility power, designed for
- 99.99%
- Availability, designed for
What you can run here
Seven ways to put sovereign compute to work, from a single GPU to a dedicated facility.
GPU-as-a-Service
NVIDIA H200, B200, B300
GPU compute on demand or reserved, in-country.
Access current-generation accelerators without buying or racking hardware. Designed to run as on-demand instances, reserved capacity, or a dedicated single-tenant cluster, so a first job and a national training run share the same platform. Every workload stays inside Qatar, under your governance.
- GPU models
- NVIDIA H200, B200, B300
- Deployment
- On-demand, reserved, dedicated cluster
- Isolation
- Single-tenant
- Interconnect
- NVLink and 400G InfiniBand
- Scale
- 1 GPU to 1,024-GPU cluster
HPC-as-a-Service
Supercomputing capacity for training and simulation.
A managed high-performance cluster built for large-scale training, modeling, and scientific simulation. Designed around a high-bandwidth fabric and parallel storage so jobs scale across nodes without starving on I/O. Scheduled and shared through Slurm, with capacity reserved to your organization.
- Cluster scale
- Up to 256 nodes
- Interconnect
- InfiniBand 400G
- Parallel storage
- Lustre, 10 PB usable
- Scheduler
- Slurm
- Use cases
- Training, simulation, modeling
Whitespace and colocation
Rack, cage, or suite in a sovereign facility.
Host your own hardware in a secure, in-country data center with room to grow. Designed for high-density AI workloads with air and direct liquid cooling on the same floor. You keep the equipment and the control, and the facility keeps it running.
- Space unit
- Rack, cage, suite
- Power density
- High density
- Cooling
- Air and direct liquid
- Tier
- Tier III+
- PUE
- ~1.3
- Redundancy
- N+1
Private AI cloud
A dedicated cloud region that is only yours.
A single-tenant environment with isolated network, storage, and compute inside Qatar. Designed to run under your governance and identity controls, so the region behaves like your own cloud rather than a shared platform. No other tenant touches the hardware or the traffic.
- Tenancy
- Single-tenant, dedicated
- Network
- Isolated, private
- Governance
- Customer-controlled
- Region
- In-country
- Access
- Private link and dedicated VPN
Financing
Fund compute on terms that fit the mandate.
Match the commercial model to how your organization actually budgets, from capital purchase to fully managed operating spend. Designed to support reserved multi-year terms, leasing, and build-to-suit facilities, including sovereign financing structures. The aim is capability now, without forcing the balance sheet into one shape.
- CapEx, OpEx, leasing, build-to-suit
- Models
- 1, 3, and 5 year
- Reserved terms
- Available
- Sovereign financing
- Designed to fit national procurement
- Structure
Managed services and SLA
06We run the platform so your teams run workloads.
Operations, monitoring, and support handled by the people who built the environment. Designed for 24/7 coverage with tiered support and a clear path from an alert to a named human. You get an availability commitment in writing, not best effort.
- SLA uptime
- 99.99%
- Support tiers
- Standard, priority, dedicated
- Monitoring
- 24/7 operations
- Escalation
- To on-call human engineers
- Response
- 15-minute critical response
Security and compliance
Data stays in Qatar, provably.
Residency, encryption, and access control built into the platform rather than bolted on. Designed to meet recognized security and audit standards, with every action logged for review. Sovereignty is a property you can verify, not a promise you have to take on trust.
- Data residency
- In-country
- Certifications
- ISO 27001 and SOC 2, in progress
- Encryption
- At rest and in transit
- Access control
- Role-based, least privilege
- Audit
- Full action logging
Whatever you run here, the data, the models, and the machines never leave Qatar.
Build AI capability that remains under your control.
Start with a conversation. We come prepared with a practical view of where AI fits.