> For the complete documentation index, see [llms.txt](https://deltaengine.gitbook.io/deltaengine-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://deltaengine.gitbook.io/deltaengine-docs/our-technologies/delta-engine.md).

# Delta Engine

Delta Engine is the decentralized compute layer powering the DeltaEngine platform. It provides seamless access to scalable GPU resources, intelligently managed across a global network of decentralized providers. Delta Engine ensures affordable, efficient, and reliable performance for your AI workloads, simplifying infrastructure management.

***

### Delta Engine vs. Traditional Cloud Providers

| Feature                   | Traditional Cloud Providers  | Delta Engine                  |
| ------------------------- | ---------------------------- | ----------------------------- |
| Pricing                   | Premium, fixed pricing       | Affordable, pay-per-use       |
| Scalability               | Automated but costly         | Automated and efficient       |
| Vendor Lock-in            | High                         | None, vendor-agnostic         |
| Infrastructure Management | Fully managed, costly        | Fully managed, cost-effective |
| Geographic Flexibility    | Limited regions, higher fees | Global availability           |
| Reliability               | High                         | High, intelligent failover    |

Delta Engine gives you the convenience of traditional clouds combined with the flexibility and affordability of decentralized compute resources.

***

### Key Benefits

**Cost Efficiency**\
Access GPU compute at significantly lower costs compared to traditional providers.

**Automatic Scalability**\
Automatically adjust GPU resources based on real-time workload demands.

**Reliable Performance**\
Intelligent workload management ensures steady performance and minimal downtime.

**Simple Integration**\
Easily integrate Delta Engine within Delta Lab, Delta Flow, or your own software through our comprehensive SDK.

***

### How it Works

Delta Engine aggregates GPUs from decentralized providers into one unified platform. It dynamically scales resources, optimally distributes workloads, and handles infrastructure seamlessly, enabling you to focus solely on your AI projects.

#### Step 1: Resource Aggregation

Delta Engine securely connects and aggregates GPU resources from decentralized compute partners worldwide, creating a unified, elastic pool of computing power accessible via a single, simplified control plane.

#### Step 2: Dynamic Auto-scaling

As demand fluctuates, Delta Engine automatically leases additional GPU resources and releases unused capacity. This dynamic scaling ensures optimal resource utilization, cost efficiency, and performance alignment with your workload demands.

#### Step 3: Intelligent Load Balancing

AI workloads are intelligently routed to the optimal available GPUs based on criteria such as latency, cost, and performance. Delta Engine continually optimizes workload distribution, providing consistent and efficient performance.

#### Step 4: Seamless Integration

Delta Engine integrates directly with Delta Lab and Delta Flow, automatically managing infrastructure complexities. Additionally, users can integrate Delta Engine directly into their own applications through a powerful SDK.