> For the complete documentation index, see [llms.txt](https://whitepaper.aitech.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://whitepaper.aitech.io/products/editor.md).

# Compute Layer

The Compute Marketplace is the infrastructure layer of AITECH Cloud Network (ACN), designed to provide enterprise-grade access to high-performance compute through a distributed and flexible model. It enables organizations to run AI workloads without dependence on centralized hyperscalers, offering improved cost efficiency, accessibility, and control over resource utilization.

***

#### **7.1 Overview**

The Compute Marketplace operates as a distributed compute network, aggregating resources from multiple providers to deliver scalable GPU access on demand.

* **Distributed Infrastructure Model**\
  Compute is sourced from a network of providers rather than a single centralized entity
* **On-Demand GPU Access**\
  Users can access high-performance GPUs based on workload requirements
* **Flexible Deployment**\
  Supports a wide range of AI and compute-intensive applications

| Feature                | Description                           |
| ---------------------- | ------------------------------------- |
| Infrastructure Model   | Distributed across multiple providers |
| Access Type            | On-demand & Reserved GPU provisioning |
| Workload Support       | AI, HPC, rendering                    |
| Deployment Flexibility | Scalable across use cases             |

***

#### **7.2 Infrastructure Design**

The infrastructure combines dedicated high-performance computing (HPC) environments with a broader global supply network.

* **HPC Data Centers**\
  Purpose-built facilities optimized for high-performance workloads
* **Global Compute Providers**\
  Integration with external providers to expand capacity and availability
* **Individual Providers**\
  Availability of supplying compute on the marketplace through individual supply

| Component          | Role                                               |
| ------------------ | -------------------------------------------------- |
| HPC Data Centers   | Core infrastructure for high-performance workloads |
| External Providers | Expand compute supply and geographic reach         |
| Allocation System  | Individual Preference                              |

***

#### **7.3 Pricing Model**

The Compute Marketplace is designed to provide transparent and predictable pricing structures, addressing one of the primary challenges of traditional cloud platforms.

* **Credit-Based System**\
  Purchase platform credits using ACN, FIAT and Stable coins
* **Pay-as-You-Go Model**\
  Users are charged only for the compute they actively use from their platform credits balance
* **Reservation Model**\
  Users can reserve on long term commitments using the platform credits
* **No Idle Billing**\
  Charges are not incurred for inactive resources

| Pricing Model      | Benefit                                          |
| ------------------ | ------------------------------------------------ |
| Pay-as-you-go      | Cost aligned with actual usage                   |
| Credit-based usage | Simplified budgeting and access                  |
| Reservation Model  | Reserve on long term commitments at lower prices |
| No idle charges    | Eliminates unnecessary costs                     |

***

#### **7.4 Key Capabilities**

The platform supports a broad range of compute-intensive workloads required by modern enterprises:

* **AI & ML Training**\
  Large-scale model training and fine-tuning
* **Inference Workloads**\
  Real-time and batch inference processing
* **Rendering**\
  High-performance rendering for media and design applications

| Capability    | Use Case                           |
| ------------- | ---------------------------------- |
| AI Training   | Model development and optimization |
| Inference     | Production-level AI deployment     |
| **Rendering** | Visual processing                  |

***

#### **7.5 Competitive Positioning**

The Compute Marketplace is positioned as an alternative to traditional hyperscalers by addressing key inefficiencies in cost, access, and flexibility.

| Factor            | Hyperscalers (AWS / Azure / GCP) | ACN Compute Marketplace  |
| ----------------- | -------------------------------- | ------------------------ |
| Pricing Structure | Complex, variable billing        | Transparent, usage-based |
| Cost Efficiency   | High overhead costs              | Optimized pricing model  |
| Vendor Lock-In    | High level of lock in            | Flexible access          |

***

#### **7.6 Summary**

The Compute Marketplace provides:

* Distributed access to high-performance compute
* Flexible and predictable pricing models
* Scalable infrastructure for enterprise workloads
* Reduced dependency on centralized cloud providers

It serves as a foundational infrastructure layer within AITECH Cloud Network, enabling organizations to efficiently access the compute resources required to support modern AI and high-performance applications.