Skip to main content

Mythic

Category: AI Chips / Semiconductors

A semiconductor company developing high-performance analog compute-in-memory (CiM) processors designed to provide high-end AI inference at the edge with extreme power efficiency. Mythic was founded in 2012. The company is led by Taner Ozcelik. Based in Redwood City, USA. Team size: 50-100. Total funding raised: $289.8M. Latest round: Series C ($125M, Dec 2025) - Analog AI chips. Key investors include ["DCVC","Valor Equity Partners","New Enterprise Associates (NEA)","SoftBank Ventures Asia","Lockheed Martin Ventures","BlackRock"].

Founded
2012
Headquarters
Redwood City, USA
Team size
50-100
Total funding
$289.8M

Value proposition

Delivers up to 10x-100x better power efficiency and cost-effectiveness compared to traditional digital GPUs and DSPs by eliminating the 'memory wall' through analog matrix processing.

Products and solutions

["M1076 Analog Matrix Processor (AMP)","M1108 Analog Matrix Processor (AMP)","MM1076 M.2 M Key Card (Edge AI acceleration)","MP10304 Quad AMP PCIe Card (Server-grade inference)","Mythic ACE™ (Analog Compute Engine) architecture","M2000 Series (Next-generation high-performance processors)"]

Unique value

Utilizes analog computing principles (Ohm’s Law and Kirchhoff’s Law) to perform matrix multiplication directly inside flash memory cells, effectively merging storage and computation.

Target customer

Original Equipment Manufacturers (OEMs), system integrators, and developers in the fields of robotics, aerospace, defense, and industrial automation.

Industries served

["Aerospace & Defense","Robotics & Autonomous Systems","Smart Cities & Security","Industrial IoT","Consumer Electronics (AR/VR)","Healthcare (Medical Imaging)"]

Technology advantage

Eliminates the energy-intensive data movement between processor and external memory (von Neumann bottleneck), allowing complex AI models like YOLOv5 to run on high-resolution video at just 3-4 Watts.

How they differentiate

Mythic utilizes a unique Analog Compute-in-Memory (CiM) architecture. Unlike digital competitors that move data between memory and processors (the 'von Neumann bottleneck'), Mythic performs matrix multiplication directly inside flash memory cells using analog currents, resulting in 10x-100x better power efficiency and significantly lower latency for high-end edge AI.

Main competitors

["Hailo","Blaize","NVIDIA (Jetson)","Syntiant"]

Key partnerships

["Lockheed Martin (Defense applications and deployment)","Codasip (Integration of RISC-V cores for processor control)","Micron Technology (Strategic investment and memory ecosystem collaboration)","Hewlett Packard Enterprise (HPE) (Edge-to-cloud infrastructure)","Valor Equity Partners & DCVC (Strategic investment and growth scaling)"]

Notable customers

["Lockheed Martin","U.S. Air Force","Various Robotics OEMs","Industrial Security Providers"]

Major milestones

["Founded in 2012 to commercialize analog compute-in-memory technology.","Launched the M1076 Analog Matrix Processor (AMP) in 2021.","Successfully secured $13M in 'rescue' funding in March 2023 to restart operations.","Appointed former NVIDIA executive Taner Ozcelik as CEO in June 2024.","Raised $125M Series C in December 2025 to scale production and challenge NVIDIA's edge dominance."]

Growth metrics

Estimated revenue of $6.4M - $7.5M for 2025; transitioned from near-insolvency in 2023 to a major growth stage in 2025.

Market positioning

High-performance Edge AI inference specialist targeting power-constrained autonomous systems and high-resolution computer vision.

Geographic focus

North America (Defense and Aerospace), Asia-Pacific (Industrial Automation and Robotics), and Europe (Automotive).

Patents and IP

Holds over 100 patents related to flash-based analog compute-in-memory, device physics, and processor architecture (including key IP for the Mythic ACE engine).

About Taner Ozcelik

Dr. Taner Ozcelik is a veteran semiconductor executive with over 30 years of experience. He is best known for founding and leading NVIDIA's automotive division, where he grew the business into a multi-hundred-million-dollar unit. He previously served as SVP & GM of the Image Sensor Group at ON Semiconductor, and held senior leadership roles at Luminar Technologies and Hesai Technology.

Official website: