The ultra-low-power DX-M2, leveraging Samsung Foundry’s 2nm process, is designed for real-time inference of models up to 20 billion parameters, consuming under 5W, and is expected to accelerate the adoption of the 2nm AI semiconductor ecosystem.

SEOUL, South Korea, Aug. 20, 2025 — DEEPX, an on-device AI semiconductor company led by CEO Lokwon Kim, has announced the signing of a 2nm process development agreement with Samsung Foundry and its design partner Gaonchips for its next-generation generative AI chip, the DX-M2.

Under the terms of this contract, DEEPX becomes the first commercial client for Samsung Foundry’s advanced 2nm Gate-All-Around (GAA) process specifically for an AI chip. The company will commence silicon fabrication for the DX-M2 processor, which is engineered for ultra-low-power generative AI inference on devices. Multi-Project Wafer (MPW) prototype production is slated for the first half of 2026, with mass production targeted for 2027.

This initiative is anticipated to play a crucial role in expanding South Korea’s high-value semiconductor industry and fostering the global development of a 2nm AI semiconductor ecosystem.

While mass production is scheduled for 2027, securing Samsung Foundry’s 2nm GAA process at this early stage was a strategic imperative for DEEPX. The company recognized that achieving truly practical on-device generative AI is nearly impossible with existing process nodes due to the immense computational demands and strict power limitations involved.

The superior power-to-performance ratio of the 2nm process—offering approximately twice the power efficiency of the 5nm process utilized in DEEPX’s current DX-M1—is not merely an improvement, but a fundamental requirement to overcome this technological barrier. This decision underscores DEEPX’s commitment to delivering a future-proof solution poised to define the next era of AI, rather than pursuing incremental gains.

DEEPX’s existing 5nm-based vision AI chip, the DX-M1, is currently in mass production for applications in robotics, smart cameras, and factory automation, and is experiencing global customer adoption. The DX-M2 extends the company’s portfolio to generative and multimodal AI, forming the core of its strategy to lead the transition towards a hyper-intelligent, AI-driven society.

Presently, generative AI largely relies on expensive, energy-intensive data centers. The DX-M2 chip is designed to offer an alternative, an ultra-low-power, no-fee on-device AI solution that provides significantly higher power efficiency than data centers while substantially reducing carbon emissions.

The chip can perform real-time inference for models up to 20 billion parameters at a rate of 20 to 30 tokens per second (TPS), all while consuming 5 watts or less of power. This represents a monumental leap in power efficiency, particularly given that global competitors are currently running smaller 10-billion-parameter models at around 10 TPS with a power consumption of 10 to 20 watts. This capability enables expert-level generative AI models to operate independently on devices such as robots, home appliances, and laptops, even in environments with tight power and thermal constraints.

The DX-M2 is set to deliver the world’s first practical on-device solution capable of real-time processing of DeepSeek and LLaMA 4 Scout—20 billion-parameter models that achieve near-AGI performance comparable to 100 billion-class models through a Mixture of Experts (MoE) architecture—as well as other advanced generative AI models in real time.

DEEPX began designing the DX-M2’s architecture in early 2024 and had a demonstration-ready prototype by early 2025. The company aims to set new benchmarks in computational performance, intelligence, and power efficiency for on-device AI.

“The DX-M2 is the core platform that will facilitate the widespread adoption and industrialization of generative AI,” stated Lokwon Kim, CEO of DEEPX. “By combining 2nm process efficiency with our AI architecture, we aim to eliminate the cost and energy barriers of today’s AI and create a world where anyone can benefit freely and sustainably from this technology.”

About DEEPX

DEEPX is a leading on-device AI semiconductor company headquartered in South Korea. Its product portfolio includes high-performance, low-power AI processors optimized for edge applications in robotics, smart cities, industrial automation, and consumer electronics. For more information, visit www.deepx.ai.

DEEPX PR team
Ella Lee
lah@deepx.ai