SEMIFIVE Ships 500mm² AI Inference Chip on Samsung 4nm Process

SEMIFIVE Ships 500mm² AI Inference Chip on Samsung 4nm Process

SEMIFIVE has commenced mass production of an AI inference accelerator for HyperAccel, marking the Korean ASIC provider's first large-scale manufacturing project using Samsung Foundry's 4-nanometer process node. The chip, designed for data center workloads including large language models, measures more than 500 square millimeters—a size that presents significant challenges in power management, thermal control, and manufacturing yield.

The company delivered what it describes as a complete turnkey solution, handling front-end design, verification, packaging, testing, and volume manufacturing. HyperAccel's initial production contract is expected to generate follow-on purchase orders as the customer expands its service offerings, according to details first reported by SEMIFIVE.

Why it matters

The successful production of a 500mm² die on an advanced node demonstrates execution capability that matters in a market shifting toward custom silicon. As hyperscalers and AI infrastructure providers increasingly pursue purpose-built accelerators rather than off-the-shelf GPUs for inference workloads, ASIC design houses that can navigate the complexity of large dies on leading-edge processes gain competitive positioning. SEMIFIVE's ability to convert design wins into recurring manufacturing revenue—rather than one-time engineering fees—signals a business model transition that investors and partners will watch closely.

Accelerating production ramp

The HyperAccel project represents SEMIFIVE's third mass-production program to reach volume manufacturing in the past year. The company previously shipped Hanwha Vision's Wisenet 9 security camera chip in the third quarter of 2025 and a high-performance computing chip for a Japanese customer in the second quarter of this year.

This production cadence has driven measurable financial momentum. SEMIFIVE secured 42.3 billion Korean won in new mass-production orders during the first half of this year—nearly double the 21.2 billion won booked for all of 2025. Quarterly order intake jumped 71 percent from 15.6 billion won in the first quarter to 26.7 billion won in the second quarter, with international orders comprising 45 percent of second-quarter bookings.

Portfolio diversification across markets

The company has expanded its customer base across security, high-performance computing, and now data center AI inference. This diversification across application domains and technical complexity levels provides revenue stability beyond single-customer concentration. SEMIFIVE emphasized that its order quality has improved, supported by stable original equipment manufacturer volumes with secured end customers and high-value data center ASIC projects.

SEMIFIVE's CEO noted that demand for ASIC accelerators in data centers is increasing as the AI industry shifts focus from training to inference. The executive highlighted the company's execution capabilities in advanced processes and its ability to combine customer architecture with comprehensive turnkey services to deliver volume production of large-area dies.

The company stated it aims to maintain the production momentum demonstrated in the first half through the remainder of the year. Details were announced by SEMIFIVE in a September 8, 2026 statement.