Unigen Corporation, established in 1991, initially focused on developing memory and storage modules. Over time, the company has evolved to provide Electronic Manufacturing Services (EMS) to its growing industrial customers. By leveraging its manufacturing facilities to produce OEM products, Unigen has expanded its reach into new sectors, including medical devices and defense/aerospace.
This ongoing expansion led the way for Unigen’s entry into edge artificial intelligence, as many of its customers began deploying edge solutions. Given its established expertise in memory, storage, and high-reliability manufacturing, Unigen decided to incorporate edge AI accelerators into its product offerings.
Currently, Unigen’s AI accelerator modules include Biscotti and Poptart. Biscotti is designed in an E1.S form factor, while Poptart comes in an E3.S form factor. These devices enable the company’s AI accelerators to be integrated into standard server drive bays or edge devices.
“We know our customers want to feel confident that the AI solution they deploy today will continue to meet their needs for years to come,” says Paul W. Heng, founder and CEO of Unigen. “That’s why we always include plug-and-play and hot swap capabilities. These features reduce the risk of system downtime and are easily upgradable, so customers can scale quickly.”
Unigen particularly designed these AI modules to support a wide range of deployment environments, from compact edge gateways to edge AI servers and even full racks for on-premises AI in data centers. The design approach emphasizes ease of integration and scalability.
Both Biscotti and Poptart come with Hailo-8 deep learning processors. The Biscotti E1.S module integrates two Hailo-8 AI accelerators on a small, 9.5 mm-thick card, along with a PCIe switch to interface them with host systems. The larger Poptart E3.S module is designed for higher density and performance. Even the Poptart module has two Hailo-8 processors (52 TOPS AI performance) with more board area and thermal headroom.
Looking into the future, Unigen is expanding its edge AI portfolio with new products across visual and generative AI domains. “In terms of growth opportunities for visual AI, we see opportunities in video management systems (VMS), crime prevention, medical, and agriculture industries,” says Heng.
Additionally, in healthcare, medical imaging diagnostics, and patient monitoring at the edge could benefit from visual AI accelerators, which will provide real-time AI insights into hospitals or remote clinics. In agriculture, edge AI modules could power smart farming solutions such as crop monitoring, automated harvesting systems, and livestock monitoring, all while operating on-site (often in bandwidth and connectivity-constrained environments).
In parallel, the upcoming generative AI offerings aim to bring the power of AI content generation and large language models (LLMs) to customer premises. These will focus on enabling small and medium businesses and educational institutions that often have use cases for AI assistants or content generation.
“The new solutions will serve emerging application spaces by allowing customers to add AI to their premises without changing the existing architecture of their facilities. This is achieved because our inference solutions follow Open Compute Project (OCP) specifications,” says Heng.
By aligning to OCP standards, Unigen ensures interoperability. For example, an OCP-compliant edge AI server or module from Unigen can work with other OCP-compliant racks, chassis, and management frameworks out-of-the-box.
“We always prioritize performance-per-watt and performance-per-dollar,” Heng adds. “This allows us to strike the best balance of efficiency vs value for our customers.”
To support its AI ambitions, Unigen has been upgrading its manufacturing footprint across the United States and internationally. The drive to onshore high-tech manufacturing in the United States has led Unigen to ramp up domestic production capacity. The company has added more production lines (including efficient U-shaped assembly lines for higher throughput) and invested in surface-mount and automated test equipment to meet the customer demand.
Unigen has also expanded into Southeast Asia with a new manufacturing site in Malaysia. The move was a response to customer feedback and a strategy to increase geographical diversification. The Malaysian facility is a state-of-the-art plant with advanced manufacturing lines and automation.
By establishing a presence in Malaysia, Unigen has strengthened its business continuity, mitigating regional disruptions and optimizing logistics. An additional strategic benefit of this location is its proximity to Coraza Systems, a sheet metal fabrication and tooling company within Unigen Global Corporation. “Coraza is only 30 minutes away from the new Malaysian facility, allowing us to offer vertical integration services in a co-located area,” says Heng.
Unigen’s dual-pronged strategy of investing in both the OEM and EMS segments remains as relevant as ever. The AI modules and servers have seen strong growth over the past several years, enabling them to expand their product line to serve emerging application spaces for visual and generative AI.
“AI adoption is still in its early days, and the opportunities ahead are immense,” Heng concludes.
Related
Article Topics
AI accelerators | AI/ML | edge AI | electronic manufacturing | generative AI | Unigen
