SEMIFIVE Strengthens Position in AI ASIC Market with New NPU Deals

SEMIFIVE Soars in the AI ASIC Market

In a significant stride towards cementing its dominance in the AI semiconductor realm, SEMIFIVE has recently announced the signing of a high-value turnkey design contract. Valued at KRW 18 billion (approximately USD 12.5 million), this pact is a continuation of SEMIFIVE's collaboration with a prominent AI semiconductor fabless firm. The project will revolve around creating a high-performance AI neural processing unit (NPU) leveraging Samsung Foundry's cutting-edge 4nm (SF4X) process technology. This win not only reaffirms SEMIFIVE's position in the rapidly expanding AI semiconductor sector but also enhances its diverse portfolio of AI chip design projects.

The company's prowess in NPU design and mass production has been consistently demonstrated through various successful projects. These include notable initiatives such as the XCENA CXL-based 4nm project, and the HyperAccel LPU-based project optimized for large language model (LLM) inference. Such consecutive wins underscore SEMIFIVE's technical sophistication and execution expertise in the high-performance AI semiconductor industry.

The newly acquired project aims to bring to life a next-generation AI NPU tailored for on-premise environments, which allows extensive AI computations to occur locally without dependence on external networks. The chip is designed for immediate processing of vast data inputs, particularly relevant for high-resolution Vision AI and LLM inference. Advanced memory interfaces will be incorporated to eradicate data bottlenecks, enabling SEMIFIVE to achieve an extraordinary level of efficiency in Power, Performance, and Area (PPA). The goal is to deliver an unprecedented AI accelerator that merges high performance with reduced power consumption.

A significant advantage for SEMIFIVE is its proprietary System-on-Chip (SoC) design automation system, coupled with comprehensive engineering solutions that greatly enhance the efficiency of ASIC development. The company assumes full accountability for the entire development lifecycle—ranging from chip specifications and logical design to packaging, software development, prototyping, and mass production. This holistic approach permits clients to concentrate on their core AI architecture and competitive service offerings without diverting attention to the intricacies of chip development.

Remarkably, SEMIFIVE is also overseeing large-die turnkey projects exceeding 800 mm² and holds vital high-speed interface design capabilities essential for high-performance AI silicon. This, paired with the firm’s rich experience in managing global semiconductor supply chains, positions SEMIFIVE as a highly reliable partner for the commercialization of intricate AI ASICs.

Brandon Cho, CEO and co-founder of SEMIFIVE, articulated that this latest contract is a testament to the company’s accumulated technical acumen and custom design capabilities, marking a substantial competitive advantage in the marketplace. He stated, "Through enhanced collaboration with customers that possess innovative AI architectures, we aim to maximize synergies in the development of high-performance AI chips, solidifying our position as leaders in the custom semiconductor domain as specialists in AI ASICs."

In parallel to these developments, SEMIFIVE is accelerating its global outreach by diversifying its client base across key regions, including the United States, China, Japan, India, and Europe. Building on its domestic projects within the AI semiconductor landscape, the company is poised to establish itself as a pivotal partner within the global AI ecosystem and a front-runner in the realm of custom semiconductor design, particularly as the demand for high-performance AI ASICs surges exponentially.

With these strategic initiatives, SEMIFIVE is not just participating in the AI semiconductor boom but is actively shaping its trajectory towards becoming a leader in this transformative industry.