Unigen’s Cupcake Edge AI Server is a reliable, high-performance, low-latency, and energy-efficient platform designed for machine learning and AI inference. Housed in a compact, rugged enclosure, Cupcake integrates flexible I/O interfaces and expansion capabilities, enabling seamless video capture and signal processing via its Power-over-Ethernet (PoE) ports. It then efficiently delivers processed data to clients over wired or wireless networks. Supporting neural networks from leading ISV providers, Cupcake offers a highly customizable solution for a wide range of applications. Cupcake excels in industries such as delivery, transportation, retail, agriculture, healthcare, manufacturing, hospitality, and security.
Features
- Rugged Visual Analysis at the Edge
- Up to 60 TOPS in a Compact Edge AI Server at under 20W (CPU + AI)
- Multiple OS and AI Vendors Available
- Vertically Integrated Design and Manufacturing
- Design to PCB to SMT to Modules to Enclosure/Chassis on Prem.
- Validation, Qualification and Complete testing in-house
- AI Hardware, Software and ISV relationships to create Unigen solutions
- SDKs, Custom Neural Networks, Cloud Extensibility
Benefits
AI That Works Where Your Business Actually Operates
Most AI infrastructure lives in a server room or the cloud, far removed from where decisions actually need to be made. The Cupcake changes that. Its compact and rugged enclosure is built to operate in demanding environments where a traditional server would never survive. It captures live video feeds and multiple types of sensor signals through its built in Power over Ethernet ports, processes that data locally using on board AI, and delivers actionable results in real time. Whether it is monitoring equipment on a production line, analyzing foot traffic in a retail environment, or processing security feeds at a remote location, the Cupcake puts AI intelligence exactly where it is needed most.
Scalable Power When the Job Demands More
The Cupcake is designed to grow with the complexity of your workloads. For organizations already operating Poundcake inference servers or Coding Servers, the Cupcake can accept a single Amaretti E1.S AI module directly from either platform, instantly inheriting the same 60 TOPS of computing power and 32GB of on-board 32GB memory that drives Unigen’s flagship products. This means a small edge deployment can be upgraded to handle significantly more demanding AI tasks, computer vision, large language model inference, or multi stream video analysis, without replacing the unit or adding new infrastructure. One module swap transforms the Cupcake from a capable edge device into an AI compute node.
Seamless Integration in Your Environment
The Cupcake is built to integrate into your existing environment without disruption. It supports neural network frameworks from all leading software providers, meaning the AI models your team already relies on will run on the Cupcake without modification. Data processed at the edge is delivered seamlessly to clients over both wired and wireless networks, fitting naturally into however your organization already moves and consumes information. For teams that need a custom solution, the Cupcake’s flexible I/O and expansion capabilities make it a strong foundation to build on. It is not a locked down appliance; it is a platform that adapts to your workflow rather than forcing your workflow to adapt to it.
Specifications
| System | ||
| CPU | Intel® Atom® x6414RE Processor SoC | |
| Memory | 4GB/8GB/16GB/32GB DDR4 SO-DIMM | |
| Graphics | Intel® UHD Graphics | |
| Storage | SATA M.2 64GB Standard | |
| Ethernet | GbE x 1, GbE With POE+ (60W Max) x 4 | |
| Audio | Line in x1, Line out x 1, Mic in x 1 | |
| Display | HDMI 2.0b x 1 | |
| I/O Placements | ||
| Power | 12V DC-IN connector and power button / additional 60 watts available | |
| Reset | Reset button | |
| Debug port | USB to UART Micro B | |
| COM | D-Sub 9-pin connector x selectable RS232/RS485 | |
| CAN | 4-Pin terminal block header | |
| Serial Ports | 10-Pin terminal block header (I2C x 1, SPI x 1, PWM/GPIO x 2) | |
| Others | ||
| Power | 12V DC 10A | |
| Security | TPM 2.0 | |
| OS Support | Linux, Microsoft | |
| Dimension | 170mm x 158mm x 49.5mm | |
| Operating Temperature | 0°C ~ 50°C |
Ordering Information
| Part Number | Note |
| UCT-10X-XXX-XX-XXXX | Multiple configuration options are available. Please contact Sales@unigen.com for availability. |