MemryX
About
About MemryX
MemryX’s MX3 accelerator uses an at-memory dataflow architecture that distributes neural-network operations across 192 compute engines. A single chip provides approximately 6 TFLOPS while operating at roughly 0.5–3 W, and as many as 16 devices can be cascaded for larger models or multiple concurrent video streams. The compiler accepts established models without requiring retraining or specialised quantisation work.
Cascade products package those chips into PCIe cards, USB modules, M.2 devices and a Raspberry Pi HAT+, allowing developers to add inference to x86, Arm or RISC-V hosts. The CPU continues to run the robot or application software while MemryX executes vision models for detection, inspection, safety and navigation. This division is intended to keep power and host utilisation predictable as camera count or model complexity increases.
Funding
- 2025-03-27 - Series B - $44M - MemryX raises USD 44 million Series B for MX3 and MX4
MemryX raised USD 44 million in Series B financing to scale MX3 production and complete the design of its next-generation MX4 accelerator. - source: prnewswire.com
Investors
- Arm IoT Fund - Participant - VC - 1 funding event - first seen 2025-03-27
- eLab Ventures - Participant - VC - 1 funding event - first seen 2025-03-27
- HarbourVest - Participant - Unknown - 1 funding event - first seen 2025-03-27
- M Ventures - Participant - VC - 1 funding event - first seen 2025-03-27
- Motus Ventures - Participant - VC - 1 funding event - first seen 2025-03-27
- NEOM Investment Fund (NIF) - Participant - VC - 1 funding event - first seen 2025-03-27
Team
- Joe Faris - VP of Sales & Marketing - Faris leads sales and marketing for MemryX edge-AI accelerator products across industrial, video, and smart-vision markets.
- Ross Jatou - President & CEO - Jatou leads MemryX commercialization of edge-AI accelerator products for embedded vision and industrial AI workloads.
- Wei Lu - CTO & Co-Founder - Lu defines MemryX product architecture for at-memory AI accelerators and embedded edge-AI processing.
- Mohammed Zidan - VP of Architecture - Zidan leads architecture work for MemryX edge-AI accelerator products and embedded AI processing.
Timeline
Events
- 2026-07-09 - Partnership - Ebttikar and MemryX target Saudi computer-vision deployments
Ebttikar contributed local systems-integration and customer-delivery capability. MemryX supplied production-proven edge-AI inference accelerator technology. The agreement allowed expansion into qualified opportunities elsewhere in the Middle East.
Source - 2026-06-22 - Product Launch - MemryX expands Cascade edge-AI accelerator platform
Cascade 100P combines 16 MX3 processors for high-density edge-server and multi-camera AI deployments. The expanded product line extends MemryX inference hardware from Raspberry Pi and USB systems into server-class edge deployments.
Source - 2025-12-26 - Market Signal - MX4 roadmap introduces distributed asynchronous dataflow design
The roadmap introduced distributed asynchronous dataflow processing. MX4 extended MemryX's architecture from edge deployments toward data-centre workloads. The design direction built on production experience from the MX3 platform.
Source - 2025-12-19 - Partnership - Macnica becomes MemryX distribution and solutions partner in Japan
Macnica added MemryX hardware to its Japanese distribution and solutions portfolio. The partnership combined accelerator technology with local engineering and customer support. Target deployments included industrial, vision and other edge-AI workloads.
Source - 2025-09-25 - Market Signal - MemryX announces Riyadh office and local engineering teams
The office extended MemryX's operating presence into Saudi Arabia. Riyadh-based leaders were assigned to business development and strategic partnerships. Local development and customer-engineering teams were planned to support government, enterprise and academic users.
Source - 2025-05-06 - Partnership - MemryX joins Saudi National Semiconductor Hub
The partnership established a route for increased development activity in Saudi Arabia. MemryX planned local customer support and broader regional reach. The hub relationship supported deployment of edge-AI semiconductor technology.
Source - 2025-05-05 - Partnership - NEOM Investment Fund partners with MemryX on edge AI
The collaboration targeted responsive and privacy-preserving digital infrastructure. MemryX technology was selected for energy-efficient edge inference. The partnership connected accelerator development with planned deployment across NEOM use cases.
Source - 2025-04-21 - Award - MX3 M.2 wins 2025 Edge AI Vision product award
The award recognized the MX3 M.2 in the Edge AI Computers and Boards category. The module averaged under three watts while processing multiple-camera computer-vision workloads. Its M.2 form factor supported integration into existing platforms with limited thermal overhead.
Source - 2025-03-27 - Funding - MemryX raises USD 44 million Series B for MX3 and MX4
The round received support from new and existing investors. Capital was allocated to ramping production of the MX3 accelerator. The financing also supported completion of the MX4 design.
Source - 2025-01-06 - Partnership - Variscite modules pair with MemryX acceleration for industrial edge AI
The collaboration paired MemryX accelerator technology with Variscite system-on-modules. The combined platform targeted higher performance and efficiency in edge applications. Industrial users gained a pre-integrated path for local AI inference.
Source - 2024-11-07 - Product Launch - MemryX opens Developer Hub 1.0 to the public
The hub became freely accessible to developers building prototypes or production systems. Automated tools supported source models without requiring a fixed model zoo. Examples and tutorials covered users with different levels of deployment experience.
Source - 2024-10-01 - Product Launch - MX3 accelerator reaches production availability
Production availability covered both standalone silicon and an M.2 accelerator module. The MX3 platform targeted efficient inference for edge applications. The release moved the accelerator from sampling and demonstrations into customer production.
Source - 2024-01-11 - Demonstration - MX3 runs inside Lenovo ThinkCentre neo Ultra at CES
The system combined CPU, GPU and a discrete MemryX neural-processing unit. The configuration supported up to 40 TFLOPs for AI-centric PC workloads. The live integration demonstrated MX3 operation inside a commercial desktop platform.
Source - 2023-12-21 - Demonstration - Production-ready MX3 demonstrations announced for CES 2024
The demonstrations covered intelligent video monitoring and next-generation AI PC computing. MemryX also planned smart-city and other edge-AI use cases. The CES programme presented MX3 performance, efficiency and versatility to prospective users.
Source - 2023-10-03 - Partnership - Edge Impulse tools connect to MemryX AI accelerators
The integration linked Edge Impulse model-development workflows with MemryX accelerators. Developers could move from model creation to hardware testing through a connected toolchain. The collaboration targeted faster development of production edge-AI applications.
Source - 2023-03-16 - Partnership - Nota.ai software integrates with MemryX edge-AI chips
Nota.ai's platform could search, compress and deploy models to MemryX hardware. The integration targeted transportation, robotics, industrial and IoT applications. The collaboration shortened the workflow from trained model to edge deployment.
Source - 2023-03-01 - Partnership - Cachengo selects MemryX processors for modular edge servers
The combined platform targeted local computer-vision and AI workloads. Cachengo's modular servers could scale to eight parallel MemryX processors. The partnership covered industrial automation, smart infrastructure and other edge-AI applications.
Source
Relations
Relationships
- Telit Cinterion
Edge AI Tool Integration - MemryX's MX3 accelerator and Developer Hub integrate with Telit Cinterion's deviceWISE AI and VI Studio.
- Ebttikar Technology Company
Systems Integration Partner - Ebttikar and MemryX partnered on computer-vision edge-AI deployments in Saudi Arabia.
- Macnica
Distribution Partner - Macnica became a MemryX distribution and solutions partner in Japan.
- National Semiconductor Hub of Saudi Arabia
Ecosystem Partner - MemryX joined Saudi Arabia's National Semiconductor Hub.
- NEOM Investment Fund
Strategic Partner - NEOM Investment Fund partnered with MemryX on energy-efficient edge-AI development and deployment.
- Variscite
System On Module Partner - MemryX and Variscite paired AI accelerators with industrial system-on-modules.
- Lenovo
Integration Partner - MemryX and Lenovo demonstrated MX3 integration in the ThinkCentre neo Ultra.
- Edge Impulse
Development Platform Partner - MemryX and Edge Impulse integrated their AI development tools.
- Nota.ai
Software Partner - MemryX and Nota.ai integrated model optimization and edge-AI deployment workflows.
Context
Market context
MX3 / Cascade
Comparable profiles
- Ambarella
stack adjacency - Edge AI Compute Supplier · Computer Vision Acceleration
- Qualcomm
stack adjacency - Edge AI Compute Supplier · Computer Vision Acceleration
- Intel
stack adjacency - Edge AI Compute Supplier · Computer Vision Acceleration
- Premio
stack adjacency - Edge AI Compute Supplier
- D-Robotics
stack adjacency - Edge AI Compute Supplier
- ADLINK Technology
stack adjacency - Edge AI Compute Supplier
- Fuzhou Rockchip Electronics
stack adjacency - Edge AI Compute Supplier
- Huixi Intelligence
stack adjacency - Edge AI Compute Supplier
- SiMa.ai
stack adjacency - Edge AI Compute Supplier
- Chipforge
stack adjacency - Edge AI Compute Supplier
- One Stop Systems
stack adjacency - Edge AI Compute Supplier
- Acrab
stack adjacency - Edge AI Compute Supplier
Shared workflows
- Horizon Robotics
workflow adjacency - Robot Perception Compute
- Advanced Micro Devices
workflow adjacency - Robot Perception Compute
- Ultralytics
workflow adjacency - Robot Perception Compute
- Inbolt
workflow adjacency - Industrial Machine Vision
- Square Robot
workflow adjacency - Industrial Machine Vision
- IDS Imaging
workflow adjacency - Industrial Machine Vision
- Teledyne Technologies
workflow adjacency - Industrial Machine Vision
- Bucket Robotics
workflow adjacency - Industrial Machine Vision
- Allus AI
workflow adjacency - Industrial Machine Vision
- Cerrion
workflow adjacency - Industrial Machine Vision
- Optifye.ai
workflow adjacency - Industrial Machine Vision
- Photonium
workflow adjacency - Industrial Machine Vision
Context tags
- Edge AI and robotics compute layer - Stack Layer - MemryX supplies edge AI acceleration hardware.
- Computer vision acceleration - Capability - Cascade products target demanding multi-camera AI applications.
- Robot perception compute - Workflow - The accelerators support edge inference for perception and machine-vision workloads.
- Industrial machine vision - Workflow - The product line is relevant to industrial machine-vision deployments.