CONNECT WITH US
May 29, 09:32
Analysis: ASIC market tightens as capacity becomes key battleground for cloud chips
Cloud service providers' demand for application-specific integrated circuits, or ASICs, is increasingly locked in as advanced process nodes, advanced packaging, and component supply tighten worldwide. For readers across global tech markets, the shift means access to manufacturing capacity, not just chip design, is becoming the main determinant of who can supply the next wave of AI hardware.
Generative AI, HPC, and large data centers are raising demand for chips with higher power efficiency, stronger thermal control, and denser packaging, making advanced packaging a more strategic part of the semiconductor supply chain. In China, panel-level packaging (PLP) is gaining traction for its larger format, higher output, and lower-cost potential.
Airoha Technology is sharpening its strategy around global networking and edge AI chips, aiming to expand its customer base and strengthen its competitive moat. The MediaTek subsidiary is targeting infrastructure, audio, and positioning markets with products already shipping, in production, or advancing through development milestones that could matter to users and operators worldwide.
Foxconn chairman Young Liu said the company is preparing to break ground on its advanced packaging plant in France while continuing to explore sovereign-AI opportunities in Africa, as the electronics manufacturer looks beyond AI servers for its next phase of growth.
Arm EVP and Chief Commercial Officer Will Abbey visited QBit Semiconductor's Taiwan headquarters on May 27, signaling deeper cooperation that could shape future chip design for edge AI, physical AI, and quantum-computing cybersecurity. The visit highlights how platform alliances are increasingly influencing next-generation computing architectures for global technology markets.
Genius Electronic Optical (GSEO) held its shareholders meeting on May 29, with chairman Jones Chen and president Ying-Li Guo in attendance. GSEO reported standalone revenue of NT$20.89 billion (approx. US$665.7 million) in 2025, up 13% from NT$18.496 billion in 2024. Consolidated revenue reached NT$24.989 billion, increasing 8% from NT$23.187 billion in 2024. Earnings per share (EPS) came in at NT$32.84, slightly lower than the previous year's NT$38.35.
Unimicron said it will push advanced substrate and AI system-board upgrades in 2026 to capture accelerating demand for AI-related ASICs and switches, aiming for record revenue that would exceed its 2022 peak. The IC substrate and printed circuit board maker announced the strategy at its annual shareholders' meeting at the end of May, citing a shift in demand away from GPUs and CPUs toward ASICs, switches, and high-performance computing applications.
As AI moves from cloud environments into factories and physical systems, semiconductor design is being reshaped by new demands in speed, energy efficiency, and on-site learning. At a recent system-semiconductor seminar in South Korea, Seong-jun Jang, a research center director at the Korea Electronics Technology Institute (KETI), outlined four key architectural directions for future AI chips aimed at supporting industrial "physical AI."
AI demand for high-performance chips is driving up orders for semiconductor silicon capacitors, or Si-Caps, with Samsung Electro-Mechanics and several Taiwanese memory makers emerging as key names to watch. Supply-chain sources say the shift could reshape who captures future demand for advanced packaging.
C2i Semiconductors has advanced its power management technology with the tape-out of a chip designed for AI infrastructure, while also securing fresh venture backing to expand development and scale production. The developments underscore growing investor interest in addressing power delivery bottlenecks in data centers, driven by accelerating AI workloads.
EU seeks to spur government buying of homegrown chips
May 29, 10:40
The European Commission wants governments to buy chips from EU startups as Brussels seeks to reduce the bloc's reliance on US and East Asian suppliers, Reuters reported, citing a document it has seen.
Samsung Electronics has started shipping samples of its latest high-bandwidth memory (HBM) chip, the 12-layer HBM4E, marking the industry's first delivery of the next-generation AI memory product, according to the company. The move underscores intensifying competition among memory suppliers as demand accelerates for higher-performance components used in AI computing systems.