The GPU cloud for the Middle East, Asia & Africa
H100, H200, B200 and MI300X capacity from vetted data centers in 16 metros. One request, firm quotes from operators in-region, one contract.
| GPU | Metro | Config | Status | $/GPU-hr |
|---|
$ hasoub quote --gpu H100:64 --region riyadh --term 12mo --ib → matching operators in SA (PDPL in-country)… ✓ 3 operators can deliver 64× H100 SXM by Nov 03 OPERATOR FACILITY FABRIC $/GPU-HR MONTHLY op-ksa-1 Tier III IB NDR 3.2T 2.01 $93,907 op-ksa-2 Tier IV IB NDR 3.2T 2.09 $97,645 op-ksa-3 Tier III RoCE v2 400G 1.96 $91,571 Quote valid 7 days · hasoub reserve --quote q_7f3k2
# request.yaml · submit with: hasoub request -f request.yaml resources: accelerators: H200:128 interconnect: infiniband storage: 200TB shared placement: regions: [dubai, abu-dhabi] # stay in UAE residency: strict term: 12mo start: 2026-11-15 budget: max_per_gpu_hr: 2.90 orchestration: slurm # or k8s, bare-metal
One layer over every operator in the region
You describe the workload once. We find the capacity, negotiate the price and hold the contract, wherever the GPUs physically sit.
Sovereign by design
Pick a jurisdiction and we only quote facilities inside it. Residency terms go into the contract.
UAE · KSA · India · AfricaBelow market median
Every list rate is benchmarked against the public on-demand median for the same GPU.
H100 from $2.72 on-demandQuotes in 48 hours
Firm, side-by-side quotes with SLAs, facility tier and fabric from operators that can deliver.
Same-day for on-demandOne contract, one invoice
Mix operators and regions under a single agreement, billed monthly in USD or AED.
From 1 GPU to 1,024Priced below the market median
Every rate is benchmarked against the public on-demand median for the same GPU. Filter by GPU, region and term, then turn any row into a quote request with one click.
| GPU | Location | Configuration | Term | Availability | Hasoub $/GPU-hr | vs market median |
|---|
Hasoub rates are indicative USD per GPU-hour, excluding VAT and egress, and are confirmed in each quote. Market median is the public on-demand median across providers tracked by GetDeploying on 14 September 2026. Reserved-term rates are compared against the same on-demand median. Index date: 4 October 2026.
Sovereign by default
Training data and inference traffic stay inside the jurisdiction you choose. We match each request to operators that meet the local residency rules.
Round-trip time from Dubai
Median round-trip time from Dubai data centers, in milliseconds. Hatched bars are estimates.
Measured: Abu Dhabi, Doha, Chennai, Singapore, Jakarta, Tokyo, Seoul and Johannesburg from Azure inter-region P50 (UAE North, July 2026); Mumbai and Hyderabad from AWS me-central-1 P50; Riyadh from Zenlayer's Dubai backbone. Jeddah, Nairobi, Lagos and Cairo are estimates for direct subsea routes; public-internet paths that detour via Europe can run 200 ms or more. Each quote states the tested latency for the facility offered.
In-country hosting in Dubai and Abu Dhabi for government, finance and healthcare workloads.
TDRA · DIFC DP Law · ADGM DPRCapacity in Riyadh and Jeddah for workloads that must stay in the Kingdom.
PDPL · NCA ECCMumbai, Chennai and Hyderabad clusters for teams serving Indian users and data.
DPDP Act 2023Johannesburg, Nairobi, Lagos and Cairo for low-latency inference across the continent.
POPIA · NDPA · Kenya DPAFrom request to running cluster
Send one request
GPU model, quantity, interconnect, region and term. It takes about two minutes.
~2 minCompare firm quotes
We go to operators in your chosen jurisdiction and return side-by-side quotes with SLAs and residency terms.
Within 48 hDeploy and scale
Sign one contract with us, get one invoice, and add capacity in other regions through the same account.
On-demand: same dayGet quotes from operators in-region
No commitment to submit. Online requests go up to 1,024 GPUs and $50M in contract value. Anything above $10M goes straight to our enterprise team.
Online cap: $50M per request
Estimate uses the Hasoub list rate for the selected GPU, metro and term, at 730 hours a month. On-demand is shown for one month. Volume above 256 GPUs is quoted lower.
Have idle GPUs in the region?
List capacity with Hasoub and reach enterprise and AI-lab demand across the Gulf, India, Asia and Africa. We handle sourcing, contracting and first-line support.
- No listing fee
- Demand matched to your jurisdiction and certifications
- You set floor prices and minimum terms
Common questions
Are these prices bookable?
They are indicative from-rates from operators we work with. Submit a request and we return firm quotes with availability, SLAs and contract terms.
Who operates the hardware?
Vetted data-center and cloud operators in each region. Your quote names the operator, facility tier and certifications before you sign.
Can I keep data in-country?
Yes. Pick a jurisdiction and we only quote facilities inside it. Residency and sovereignty requirements go into the contract.
What is the minimum order?
One GPU on-demand. Reserved clusters start at one 8-GPU node. Online requests go up to 1,024 GPUs or $50M in contract value; larger builds are handled directly by our team, with a 1-month minimum on most operators.
How is billing handled?
One contract and one monthly invoice from Hasoub in USD or AED, whichever operators you use.