Brane Technologies

|

|

AI on the Edge — Without Compromise

Run AI exactly where the data is created. Brane Boards bring low-power, high-performance inference to the edge—whether beside a satellite antenna, coordinating robots in the field, or powering real-time control in industrial systems.

See Edge Configurations
Brane Board Rev.2

AI at the edge where data lives

The Brane Board Rev.2 combines AMD AI CPU with integrated GPU, Kalray Coolidge processor, and Versa AI FPGA on a single board. Run Small Language Models and edge AI workloads locally with predictable performance, low latency, and energy efficiency. M.2 connectors support third-party accelerators for maximum flexibility.

Flexible Processing Power

AMD AI CPU, Kalray Coolidge, and Versa AI FPGA work together seamlessly. The Brane SDK helps developers write and port code across all processing units without complexity.

Edge-Optimized Architecture

Designed for deployment where data is generated—reduce latency, improve privacy, and minimize bandwidth by processing locally instead of cloud round-trips.

Expandable Platform

M.2 connectors enable integration of third-party accelerators and storage, while the Brane SDK provides unified development across all processing units.

Technology

The Brane Board Rev.2 integrates multiple processing units on a single board for edge AI applications. AMD AI CPU with integrated GPU, Kalray Coolidge processor, and Versa AI FPGA work together through the Brane SDK for optimal performance and power efficiency.

AMD AI CPU

Integrated GPU
& Management

Integrated GPU

AI Inference
& Graphics

Kalray Coolidge

Predictable
Data Processing

Versa AI FPGA

Reconfigurable
Acceleration

M.2 Expansion

Third-Party
Accelerators

Integrated Edge AI Platform
  • All-in-one design: Four processing units on a single board eliminate complex multi-board configurations and reduce power consumption
  • AMD AI CPU with integrated GPU: Handles general compute, system management, and AI inference workloads.
  • Kalray Coolidge processor: Provides predictable, deterministic processing for real-time data streams and edge applications.
  • Versa AI FPGA: Reconfigurable logic for custom algorithms, hardware IP integration, and ultra-low latency processing.
  • M.2 expansion slots: Support third-party accelerators and high-speed storage for maximum flexibility.

Edge-optimized: Compact form factor and efficient power design make it ideal for deployment where data is generated—labs, field sites, and edge computing applications.

AMD AI CPU (Main Processor)

High-performance CPU with integrated GPU for AI inference and system management.

Kalray Coolidge (Data Processor)

Predictable parallel processing for real-time data streams and deterministic workloads.

Versa AI FPGA (Custom Logic)

Reconfigurable hardware for custom algorithms and ultra-low latency processing.

M.2 Expansion (Flexibility)

Support for third-party accelerators and high-speed storage solutions.

Use Cases

Edge AI Inference

Small Language Models (SLMs) Real-time object detection Local voice processing

Field Deployment

Remote monitoring stations Industrial IoT gateways Laboratory equipment control

Private Processing

On-premise data analysis Medical device processing Compliance-sensitive AI

Deployment Environments

Research laboratories Manufacturing floors Remote field sites Healthcare facilities Transportation hubs Smart buildings
Board Rev.2 Benefits

Data Locality

Run where the data is – eliminate network bottlenecks
Process at source to reduce bandwidth requirements
Keep sensitive data on-premises and secure

Power Efficiency

Low power consumption for edge environments
Optimized compute mix keeps thermals in check
Right-sized processing without datacenter overhead

Real-Time Response

Ultra-low latency – no cloud round trips
Instant inference for time-critical applications
Always available, even with intermittent connectivity
↓ Network bandwidth ↓ Power consumption ↓ Latency ↑ Data privacy
Edge AI Platforms

Brane Edge AI Platforms

Single-board computers with flexible AI acceleration for edge applications. Multiple compute architectures work together for data processing, inference, and custom workloads.

SnapCore Platform

EOL
Brane SnapCore edge AI board

First-generation edge AI platform that pioneered multi-accelerator integration in a single-board form factor. While no longer in production, it established the foundation for our current PowerEdge architecture.

Key Specifications

  • CPU: Qualcomm Snapdragon processor
  • DPU: Kalray Bostan acceleration
  • FPGA: Intel Arria 10 for custom compute
  • Form factor: Single board computer
  • Status: End of life, PowerEdge available
No Longer Available

PowerEdge Platform

Current
Brane PowerEdge edge AI board with CPU, Kalray DPU, and FPGA

Next-generation edge AI platform optimized for real-time inference and field deployment. Combines AMD AI processing with Kalray’s deterministic many-core architecture and configurable FPGA acceleration.

Key Specifications

  • CPU: AMD AI CPU with integrated GPU
  • DPU: Kalray Coolidge processor
  • FPGA: Versa AI FPGA for custom pipelines
  • Storage: Dual M.2 connectors
  • Optimized for edge AI inference
Shop PowerEdge Ask Questions

Need custom configurations or volume pricing? Our team can help design the right PowerEdge solution for your edge AI deployment.

Contact Us

PowerEdge Platform — FAQ

Key questions about the Brane Board Rev.2 and how it brings flexible AI acceleration to edge deployments.

What is the PowerEdge Platform?

A single-board edge AI computer that integrates AMD AI CPU, Kalray Coolidge processor, and Versa AI FPGA. It runs Small Language Models and AI workloads locally with predictable performance and low latency for edge applications.

Why multiple processors on one board?

Each processor excels at different tasks: AMD AI CPU for general compute and inference, Kalray Coolidge for data processing with predictable timing, and Versa AI FPGA for custom algorithms. Together they provide flexible acceleration without cloud dependencies.

What types of AI models can it run?

Small Language Models (SLMs), computer vision models, edge inference workloads, and real-time processing pipelines. The integrated GPU handles inference while the DPU and FPGA accelerate data preparation and custom algorithms.

How does it compare to cloud AI services?

Data stays local for privacy and security, with no network latency or bandwidth costs. Predictable performance and costs versus variable cloud pricing. Ideal for applications requiring real-time response or operating in remote locations.

What development tools are available?

The Brane SDK provides unified development across all processors using Gradle and IntelliJ. Pre-configured drivers, toolchains, and AI plugins help developers write and port code without managing complex multi-architecture setups.

What are typical deployment scenarios?

Edge inference (manufacturing, healthcare), field deployments (remote monitoring, IoT gateways), and private processing (compliance-sensitive AI). M.2 expansion supports additional accelerators for specialized workloads.

How do I get started?

Describe your AI workload and deployment constraints. We’ll help map your application to the right processors and provide development support through the Brane SDK. Talk to an engineer.

Assistance & Integration

Need help deploying Brane Boards at the edge?

Brane Board render, dark PCB angled on white background

From first bring-up to production deployment, our engineers help you deploy AI where data is created: beside antennas, in the field, or inside industrial systems. Hardware sizing, SDK orchestration, and proof-of-value—covered.

  • Linux bring-up, drivers, and Brane SDK runner setup
  • Pipeline mapping across CPU / DPU / FPGA with perf/watt tuning
  • I/O integration: sensors, radios, storage, and enclosure/power envelopes
  • Observability, updates, and proof-of-value engagement
Talk to an Engineer Request Dev Access