Specifications
Product
Virtium VMX-004 AI Accelerator Module
AI Processor
MemryX MX3 inference processor
AI Performance
24 TOPS
Form Factor
PCIe M.2 2280
Power Profile
Low power consumption for an industrial M.2 accelerator
Host Compatibility
Arm and x86 carrier boards and motherboards with an M.2 2280 socket
Manufacturing
Designed and assembled in Southern California, USA
Target Workloads
Vision AI inference
Target Markets
Aerospace and defence, industrial automation, transportation, agriculture, medical, surveillance
Partner
MemryX (US inference silicon)
Sample Status
Samples available
Showcase
CES 2026, MemryX suite
Overview
The Virtium VMX-004 is an industrially rated AI accelerator module that puts a MemryX MX3 inference processor onto a standard PCIe M.2 2280 card. It delivers 24 TOPS while keeping power consumption low enough for fanless and conduction-cooled industrial enclosures.
Virtium designs and assembles the module in Southern California — the company positions it as the industry's first 'Made in the USA' AI accelerator module, a procurement consideration for defence, aerospace and government-adjacent programmes. Because it conforms to the M.2 2280 standard, it drops into most Arm and x86 carrier boards without a host redesign.
Key Benefits
Standard M.2 2280 footprint avoids carrier-board changes; 24 TOPS of vision inference at industrial power levels; domestic US design and assembly simplifies defence and aerospace sourcing; works across Arm and x86 host platforms with the MemryX software stack.
Applications
Aerospace and defence vision systems, industrial machine vision and inspection, intelligent transportation, agricultural perception, medical imaging and physical-security surveillance.
Request a Quote — VIRTIUM VMX-004 — MADE-IN-USA M.2 AI ACCELERATOR MODULE ON MEMRYX MX3
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →