
1. Technology overview
Mythic makes analog compute-in-memory AI processors, operating as Mythic, Inc. with offices in Palo Alto and Austin in the US, Bangalore in India and Hannover in Germany. It addresses the power and efficiency limits of running AI inference on conventional digital processors.
The Analog Processing Unit stores model weights in flash memory and computes in analog; the site claims up to 100x efficiency. Products include the M1 for edge, camera and drone applications and an APU chiplet platform that scales from 1 to over 1,024 chiplets for enterprise LLMs. The page lists Lockheed Martin as a defense partner.
2. Highlighted features & specifications
- Analog Processing Unit: weights stored in flash and processed in analog.
- M1: processor aimed at edge, camera and drone applications.
- Chiplet platform: scales from 1 to over 1,024 chiplets, up to 120 TOPS for enterprise LLMs.
- Vanguard line: listed with 24 analog cores and a chiplet platform.
- ADAS platform: designed to be automotive-grade, with Videantis technology.
3. Potential application
Camera and drone makers can evaluate the M1 for on-device vision with limited power. Automotive electronics suppliers can look at the ADAS platform for in-vehicle perception.
4. Benefits
Analog compute-in-memory is presented as delivering up to 100x efficiency over digital approaches, which targets power-limited edge systems. The chiplet platform scales from a single chiplet to more than 1,024 for larger workloads. The acquisition of Videantis adds automotive-grade processor experience, with silicon cited in 30 million production vehicles.
5. Market trends & opportunities
Power and cost of AI inference are driving interest in compute-in-memory chips for cameras, drones and vehicles. Vietnamese and ASEAN camera, drone and automotive electronics firms can track analog AI processors as an energy-efficient option for edge products.
6. Other information
Legal entity Mythic, Inc.; offices in Palo Alto, Austin, Bangalore and Hannover. Acquired Videantis in January 2026. Investors shown on the site include NEA, DCVC, SoftBank (SBVA) and Future Ventures. Founding year not stated.
One platform connecting startups, corporations, research institutions, regions and investors











Playbook & Handbook · February 6, 2026
Winning Accelerator Playbook 2026 by BambuUP
EDTECH: LEVERAGING AI-POWERED TECHNOLOGIES TO TRANSFORM THE LEARNING EXPERIENCE








































