GPU Cloud
Managed GPU instances, but difficult for small suppliers to participate.
NURA AIExchange Network
Open Compute. Open Models. Open Participation.
An open AI compute marketplace anchored by modular infrastructure in Abu Dhabi, designed for public access, independent GPU participation and token-coordinated network growth.
4 MW
Phase-I site power target
20 MW
Expansion power application
40,000 m²
Approximate land area
256-512
Initial high-end GPU fleet
Market Gap
AI compute is abundant, but access is fragmented.
Cloud capacity exists, open-source models exist and idle GPUs exist. What is missing is a trusted market that connects all three through one product path.
GPU Marketplace
Independent host rentals, but no strong consumer AI demand engine.
AI Apps
Open or closed model access, but users cannot earn or provide compute.
DePIN Compute
Protocol incentives, often with weak user experience and weak SLA.
NURA is the missing bridge: one account for AI, dedicated GPU rental, host listing, staking, credits and settlement.
Brand Architecture
NURA and AIX stay distinct.
NURA represents the Al Ain AI campus, managed GPU cloud, customer portal, service quality and commercial operations. AIX represents the open network: staking, host collateral, routing incentives, usage settlement, node rewards and decentralized governance.
NURA
Data Center & Managed Compute
- Al Ain modular AI campus
- Managed GPU cloud and API service
- Enterprise contracts, SLA and billing
- Owned H100/H200/B200 capacity
AIX
Token & Decentralized AI Compute Network
- Host staking and collateral
- Independent GPU node onboarding
- Verified usage, settlement and rewards
- Network reputation, governance and slashing
Demand Engine
NURA creates recurring workload for the AIX network.
Every chat request, image generation, voice task, video job and AI agent workflow becomes paid demand. The routing layer sends each request to the most suitable endpoint: company-owned data center capacity, enterprise partner nodes or community GPUs.
NURA Workspace
Route: warm endpoint
User prompt
Build a bilingual product launch plan for a creator economy platform in the GCC.
NURA Routing
Best endpoint selected by SLA, model readiness, price, trust tier and queue depth.
Ask anything...
1. Use AIOne account routes requests to the right product mode.
2. Route workloadScheduler selects NURA, partner or community capacity.
3. Settle usageCredits, rewards and host income are calculated automatically.
Supply Tiers
From owned GPUs to open nodes.
AIX does not pretend that all compute is equal. The scheduler routes by task type, trust tier, GPU memory, model cache, uptime, latency, price, security rating and user preference.
| Tier | Supplier type | Best workloads | Trust model |
|---|---|---|---|
| S | NURA data center | Enterprise APIs, private endpoints, H200/B200 workloads | High SLA, full control |
| A | Enterprise / colocation partners | Longer jobs, batch inference, dedicated rentals | Certified hardware and contracts |
| B | Prosumer 4090/5090 nodes | Images, TTS, small LLMs, agents and economy batch | Stake, rating and sandboxing |
| C | Experimental spot nodes | Cheap interruptible tasks and test workloads | Low SLA, low price |
Demand Layer
Users, developers, agents
Chat, images, voice, video, notebooks and APIs.
NURA Portal
One account, many AI modes
Billing, credits, model catalog and API access.
AIX Router
Policy, matching, sandboxing, telemetry and settlement.
VRAMModel fit
SLATrust tier
PriceLive quote
S Tier
NURA owned data center
H200/B200, private endpoints and enterprise APIs.
A/B Tier
Partners and prosumer hosts
4090/5090 nodes, studios, colocations and verified rentals.
Settlement
Escrow, proofs and payout
Rewards only after verified runtime and clean teardown.
Control Plane
Routing, sandboxing, billing and verified settlement are the moat.
External hosts run the AIX Node Agent to verify hardware, start isolated jobs, report telemetry and upload usage proofs. Customers receive temporary, revocable access to isolated environments. Hosts receive settlement only after verified uptime and usage.
User / API
Request, rental or deployment order enters the marketplace.
AIX Router
Matching, pricing, policy and load balancing select the best endpoint.
Host Agent
Containers, model images, logs and telemetry are managed automatically.
Settlement
Escrow, usage proof, reputation and host payout close the loop.
Al Ain, Abu Dhabi
A modular AI campus built for scale.
A 200m x 200m square parcel planned around modular AI halls, liquid-ready cooling, power yard, transformer area and clear expansion paths.
200m x 200m
Plot geometry
4 ha
Site area
4 MW
Phase-I capacity
20 MW
End-state capacity
Infrastructure Economics
Power, cooling and fleet mix define service quality.
The customer-facing offer is packaged around reliable GPU-hours, reserved capacity, private endpoints and clear service tiers rather than exposing the underlying operating inputs.
Liquid-ready from day one
Support air/hybrid zones for RTX PRO, L40S and lower-density systems, plus direct-to-chip liquid cooling for H200/B200 and future high-density racks.
Transparent customer pricing
Published GPU-hour plans include infrastructure, cooling, network, platform access and operating support in one service price.
Right GPU for each market
H100/H200 support enterprise capacity; B200 handles premium liquid-cooled demand; RTX PRO, L40S and 4090/5090 nodes support creative, voice and small-model workloads.
AIX Token
AIX coordinates access, trust and rewards.
AIX is designed as a utility and coordination token for AI compute access, not as a tokenized profit right or a claim on land, power contracts, GPUs or dividends.
User staking
Receive NURA Compute Credits, priority tiers or access benefits where permitted.
Host collateral
List machines, receive high-value jobs and secure customer trust.
Verified-work rewards
Rewards are tied to measurable contribution: GPU-hours, uptime, completed jobs and low dispute rates.
Users
Buy compute credits, stake for access benefits and consume NURA services.
AIX
Staking, trust, routing priority and verified-work rewards.
GPU Hosts
Stake collateral, run verified workloads and receive payout after proof.
NURA OpCo
Bills usage, operates platform, support, API services and marketplace controls.
Treasury
Supports GPU expansion, insurance, liquidity, audits and ecosystem grants.
Developers
Publish models, endpoints, tools and agents that create more compute demand.
Proposed Genesis Allocation
1,000,000,000 AIX planning supply.
This is an indicative planning proposal for legal and investor review. The allocation favors real users, real hosts and real capacity growth over short-term insider liquidity.
- 20% Community users
- 20% Node / host rewards
- 15% GPU expansion treasury
- 15% Strategic investors
- 15% Protocol treasury
- 10% CHG / team / advisors
- 5% Initial liquidity
| Reward pool | Indicative share | Purpose |
|---|---|---|
| Host operators | 60% | Verified GPU-hours, uptime, successful rentals and SLA score. |
| User staking / credits | 20% | Demand, API usage and long-term platform participation. |
| Developers / model providers | 10% | Model endpoints, tools, agents and ecosystem integrations. |
| Insurance / security reserve | 10% | Disputes, slashing appeals, audits and liquidity support. |
Customer Compute Pricing
Clear GPU-hour prices for customer workloads.
NURA packages capacity as customer-facing GPU-hour plans with platform routing, billing, access control, monitoring and support included.
| Service plan | Capacity reference | Customer GPU-hour price | Availability target | Included service | Best fit |
|---|---|---|---|---|---|
| Launch Pool | 256 GPUs | US$3.90/hr | 65% | Shared API, model catalog, usage billing and support. | NURA MVP usage, developers and initial API customers. |
| Base Phase-I | 512 GPUs | US$4.23/hr | 70% | Reserved GPU-hours, private endpoints and monitoring. | Studios, AI teams and recurring production workloads. |
| Expanded Phase-I | 1,024 GPUs | US$5.53/hr | 75% | Dedicated pools, priority routing and managed deployment. | Enterprise AI services, large campaigns and premium demand. |
Indicative public service pricing. Final customer quotes may vary by GPU class, reserved duration, privacy requirements, support level and regional availability.
Roadmap
A staged path to launch.
The plan proves utility, revenue and operations before expanding power draw or token exposure.
Legal, power and design lock
Confirm site rights, utility capacity, connection pathway, cooling strategy, SPV structure, token counsel and GPU procurement list.
NURA MVP
Launch website, account system, payments, API billing, model catalog, host listing prototype and first owned GPU cluster.
4 MW Phase-I build
Commission modular halls, liquid-ready cooling, network core, monitoring, security and initial 256-512 GPU fleet.
Open host beta
Release AIX Node Agent, onboard controlled 4090/5090 and enterprise hosts, and test staking, escrow, slashing and payout.
AIX launch readiness
Publish whitepaper, complete legal review, finalize token utility, build liquidity plan, activate rewards and scale capacity.
Documents