Semifive's 8nm eMRAM Edge AI SoC Tape-Out Unlocks 2B Parameter Inference for Humanoid Robotics

BrunoSan Robotics Intelligence · AI-generated concept image

← All Robotics News

🎯 MILESTONE 🤖 HUMANOID

Semifive's 8nm eMRAM Edge AI SoC Tape-Out Unlocks 2B Parameter Inference for Humanoid Robotics

BrunoSan Robotics Intelligence May 08, 2026 1 source Intel Score 1.000 Tone: SHARP

TL;DR

Semifive and Icy Tech have achieved a successful tape-out of an 8nm eMRAM-based Edge AI SoC, enabling on-device inference for models up to 2 billion parameters, a critical development for ultra-low-power edge AI applications such as humanoid robots, with initial commercialization targeted for Asia.

Semifive and Icy Tech achieved successful tape-out of an 8nm eMRAM-based Edge AI SoC.The SoC supports on-device inference for AI models up to 2 billion parameters.Target markets include ultra-low-power edge AI, AI PCs, and humanoid robots.First commercialization is targeted for Asia.

The successful tape-out of an 8nm eMRAM-based Edge AI SoC capable of 2 billion parameter inference represents a foundational step for next-generation, ultra-low-power autonomous systems, particularly in humanoid robotics.

What happened

On May 7, 2026, Semifive, a South Korean custom silicon solution provider, in collaboration with Icy Tech, announced the successful tape-out of an 8nm embedded MRAM (eMRAM)-based Edge AI System-on-Chip (SoC). This achievement enables on-device inference for AI models containing up to 2 billion (2B) parameters, specifically targeting ultra-low-power edge AI markets, including AI PCs and humanoid robots. First commercialization is targeted for Asia.

Why this matters — the mechanism

This tape-out represents a significant technical milestone for on-device AI processing. The integration of 8nm eMRAM provides non-volatility and high-speed memory access within a compact, power-efficient footprint, crucial for edge applications where power budgets are constrained. Enabling inference for models up to 2 billion parameters on-device directly addresses the computational demands of increasingly complex AI workloads without reliance on cloud connectivity, reducing latency and enhancing privacy. For humanoid robotics, this translates to more sophisticated, real-time decision-making capabilities, improved perception, and advanced motor control directly at the robot's edge, critical for dynamic and unstructured environments. This development positions Semifive and Icy Tech as key enablers for next-generation autonomous systems requiring high-performance, low-power AI at the point of action. Cross-verified across 1 independent sources · Intel Score 1.000/1.000 — computed from signal velocity, source diversity, and robotics event significance.

What to watch next

Monitor Semifive and Icy Tech for announcements regarding the availability of development kits or initial product integrations, particularly within the Asian market. Key indicators will include specific customer partnerships or pilot deployments in the humanoid robotics sector. As of 2026-05-08T05:32:01Z, the focus shifts to the validation of this SoC in real-world applications and its subsequent market penetration, with potential updates at upcoming industry events like Automatica 2026 or IROS 2026.

• techbuzz.ai: Report on Semifive and Icy Tech's successful tape-out of 8nm eMRAM-based Edge AI SoC — https://www.techbuzz.ai/press-release/PRNewsWire/PRNewsWire-https%3A%2F%2Fwww.prnewswire.com%2Fnews-releases%2Fsemifive-and-icy-tech-achieve-successful-tape-out-of-8nm-emram-based-edge-ai-soc-targeting-first-commercialization-in-asia-302765052.html

This article does not constitute investment or operational advice.

✓ Cross-Verified Intelligence

1.000 / 1.000

Sources1

Burst Score3.0

Novelty Score1.0

First Sourcetechbuzz.ai

Source TierA

Event Typemilestone

Segmenthumanoid

Verified At2026-05-08T05:32:01Z

Sources

→

techbuzz.ai

Report on Semifive and Icy Tech's successful tape-out of 8nm eMRAM-based Edge AI SoC

techbuzz.ai

Frequently Asked

What did Semifive and Icy Tech achieve with their recent tape-out?

Semifive and Icy Tech successfully taped out an 8nm eMRAM-based Edge AI SoC capable of on-device inference for AI models up to 2 billion parameters, targeting ultra-low-power edge AI markets.

What does Semifive's 8nm eMRAM Edge AI SoC mean for humanoid robotics?

This achievement enables humanoid robots to perform more complex, real-time AI inference directly on-device, reducing latency, improving autonomy, and enhancing capabilities in perception and control within power-constrained environments.

Edge AISoCeMRAMHumanoid RoboticsOn-device AISemiconductorAI InferenceUltra-low-power

Access the Full Robotics Intelligence API

Real-time signals, entity graphs, funding flows, and novelty alerts — queryable via MCP.

→ Access Robotics MCP API

This article does not constitute investment or operational advice.