Virtium Unveils US-Made AI Chip to Secure Critical Tech Supply Chains

RANCHO SANTA MARGARITA, CA – January 05, 2026 – Virtium, a long-time manufacturer of industrial-grade hardware, today announced a significant strategic move into the burgeoning Edge AI market with the launch of its VMX-004 AI Accelerator Module. Assembled in Southern California, the module is being touted as the industry’s first “Made in the USA” AI accelerator, a designation that positions it to address pressing demands for supply chain security in critical sectors like aerospace, defense, and industrial automation.

Built on the MemryX MX3 AI inference processor, the new M.2 module is engineered for high-performance Vision AI workloads, delivering 24 TOPS (Tera Operations Per Second) of compute power while maintaining exceptionally low energy consumption. This launch not only expands Virtium's product portfolio but also signals a deeper commitment to providing mission-critical solutions that align with national strategic interests.

Securing the Edge with Domestic Manufacturing

The most significant differentiator for the VMX-004 is its domestic origin. In an era marked by global supply chain disruptions and heightened geopolitical tensions, the “Made in the USA” label is more than a marketing point; it is a strategic asset. By designing and assembling the module in its Southern California facility, Virtium directly addresses the growing demand from government agencies and defense contractors for secure, traceable, and reliable technology components.

The VMX-004 is fully compliant with the Trade Agreements Act (TAA), making it eligible for procurement by U.S. government agencies under GSA Schedules. This compliance is critical for federal contracts and simplifies the acquisition process for public sector clients. Furthermore, its domestic production aligns with the principles of the Defense Federal Acquisition Regulation Supplement (DFARS), which increasingly emphasizes the use of U.S.-made components to enhance national security and reduce reliance on foreign suppliers. For defense applications ranging from autonomous drones to battlefield surveillance systems, a secure and resilient supply chain is paramount to prevent the introduction of counterfeit parts and ensure operational integrity.

This focus on domestic production offers tangible benefits beyond compliance. It promises shorter lead times, greater quality control, and increased resilience against international trade volatility. As industries seek to de-risk their operations, the availability of a high-performance, domestically produced AI accelerator provides a compelling alternative to components sourced from overseas.

Redefining Performance with Innovative Architecture

Beyond its strategic sourcing, the Virtium VMX-004 introduces notable technical innovations designed to tackle the core challenges of Edge AI: latency, power, and heat. The module’s performance is anchored by the MemryX MX3 AI inference processor, which features a unique architecture that runs AI models entirely from on-chip SRAM. This design choice eliminates the need for external DRAM for model execution, a common bottleneck in conventional AI hardware.

By avoiding the round-trip data transfers to external memory, the VMX-004 achieves ultra-low latency, a critical requirement for real-time applications such as robotic guidance, high-speed quality inspection, and autonomous vehicle navigation. This architectural advantage also drastically reduces power consumption and heat generation. Eliminating external DRAM simplifies the system design, lowers the bill of materials, and enables the module to operate reliably in compact, fanless enclosures often found in harsh industrial environments.

Delivering 24 TOPS of compute, the accelerator provides ample power for complex vision models. The processor’s support for floating-point activations further enhances its capability, enabling superior model accuracy compared to many fixed-point solutions, which is vital for nuanced tasks in medical imaging or advanced surveillance. The module's adherence to the standard PCIe M.2 2280 form factor ensures seamless integration into a wide array of ARM and x86 motherboards, allowing OEMs and solution developers to accelerate their time-to-market without undertaking complex hardware redesigns.

From Memory Modules to Mission-Critical AI

The launch of the VMX-004 represents a deliberate and strategic evolution for Virtium. With over 25 years of experience designing and manufacturing high-reliability industrial SSDs and DDR memory modules, the company has built a reputation for delivering robust products with long lifecycles, tailored for demanding markets like telecommunications, networking, and industrial OEMs. This legacy provides a foundation of trust and expertise in quality control, supply chain management, and long-term customer support that is essential for mission-critical deployments.

Virtium’s expansion into the Edge AI hardware space is a calculated pivot toward one of technology's highest-growth sectors. The partnership with U.S.-based MemryX is a key component of this strategy, combining MemryX’s specialized AI processor technology with Virtium’s proven industrial hardware expertise. This collaboration allows Virtium to offer a solution that is not only technologically advanced but also ruggedized and reliable enough for the challenging environments its customers operate in.

As Keith Kressin, CEO of MemryX, noted, “The collaboration with Virtium reflects our shared commitment to simplifying AI deployment at the edge. By combining MemryX’s low-power inference technology with Virtium’s proven industrial-grade expertise, we’re delivering a solution that brings real, immediate value to customers.”

Powering the Next Wave of Industrial Vision

The VMX-004 is purpose-built to meet the specific needs of high-performance Vision AI across a spectrum of demanding industries. Its industrial-temperature operating range of -40°C to +85°C ensures dependable performance whether deployed in an unconditioned factory, an outdoor surveillance camera, or an avionics bay. This robustness, combined with its low-latency performance, directly addresses key industry challenges.

In industrial automation, the module can power real-time quality control systems on fast-moving production lines or provide the inferencing capability for collaborative robots. For transportation and agriculture, its efficiency and durability make it ideal for on-vehicle systems in smart tractors or fleet management solutions. In the medical field, it can accelerate diagnostic imaging analysis on portable devices, while in security and surveillance, it enables complex, multi-stream video analytics to be performed locally, reducing bandwidth needs and improving response times.

Engineering samples of the Virtium VMX-004 AI Accelerator Module are available now, with qualification samples and full-scale production scheduled for the first quarter of 2026. “Virtium’s new AI accelerator module, VMX-004, expands our Edge AI portfolio with a U.S.-made solution engineered for high-performance vision workloads,” said Dave Beasley, Executive Vice President at Virtium. “This new module positions Virtium at the forefront of emerging Vision AI opportunities.”