新闻中心
Weekly News | Supply chain trends in Semiconductor industry #218
发布日期:2026-06-29


01

Company TrendJune 23

Kuaishou Chip Unit Lingchuan Technology Achieves Commercial Breakthrough in In-House AI Compute Chips

Kuaishou has spun off its internal chip division into an independent entity, Lingchuan Technology, which has completed a Series A+ funding round raising several hundred million RMB, marking a full transition of its self-developed chip business toward independent market-oriented operations. Lingchuan Technology was formally established as an independent company in March 2024, jointly controlled by Kuaishou and a Beijing AI fund, focusing on the development of dedicated compute chips for video processing and multimodal large models. Its flagship mature product SL200 has shipped nearly 100,000 units, covering 99.7% of Kuaishou’s live-streaming transcoding workloads and supporting video traffic for approximately 700 million platform users. The chip is also supplied to external customers including Alibaba Cloud, Baidu Cloud, and Bilibili. The SL200 has reportedly won international video encoding competitions for three consecutive years, delivering higher compression efficiency than comparable Nvidia solutions while reducing compute energy consumption by approximately 30%. A next-generation fully domestic 3D-stacked chip completed tape-out in April 2026. The new funding will be used for product iteration, capacity expansion of existing chips, and overseas market development. Leveraging its internet-scale application scenarios, Lingchuan Technology is also deploying hardware solutions including AI acceleration cards and integrated compute appliances, targeting use cases in intelligent computing, security, and automotive applications. The company is emerging as a key domestic supplier in China’s vertical video compute chip segment.


Comment: Kuaishou’s decision to spin off its chip business reflects a broader trend of internet platform companies moving toward in-house compute self-sufficiency, aligning with similar strategies pursued by Alibaba T-Head and Baidu Kunlun. This confirms that the development of proprietary application-specific chips has become an industry-wide direction. In the past, short-video and live-streaming platforms were highly dependent on foreign general-purpose GPUs, facing elevated procurement and operational costs as well as supply constraints. The development of in-house ASIC chips enables closer optimization for video transcoding and AI inference workloads, achieving both cost reduction and improved supply chain autonomy. Unlike highly competitive general-purpose AI compute markets, Lingchuan focuses on vertical video workloads and defines chip architecture directly from real-world application scenarios, forming a differentiated competitive position. The transition from internal deployment to external commercialization further expands revenue potential for internet-based chip developers and supports the development of a more integrated domestic AI hardware-software ecosystem. Kuaishou’s chip initiative is also expected to stimulate demand across upstream and downstream segments, including wafer fabrication, packaging, and server infrastructure, thereby accelerating the localization of China’s multimodal AI compute stack.


02
Company Trend(June 24
SK Hynix Yongin Fourth Fab Completion Timeline Brought Forward by 10 Years: HBM Capacity “Arms Race” Enters Accelerated Phase

South Korean authorities have confirmed that the completion timeline for SK Hynix’s fourth wafer fab in the Yongin semiconductor cluster has been significantly advanced from 2044 to 2034, a full 10-year acceleration. The Yongin cluster is planned to include four wafer fabs and represents one of South Korea’s largest semiconductor investment projects in recent years. Previously, SK Hynix had already brought forward the timeline for the cluster’s first fab, which is now scheduled to begin production in February 2027, three months earlier than originally planned. According to The Elec, SK Hynix has disclosed to key suppliers a plan to double its DRAM capacity — from approximately 550,000 wafers per month currently to 1 million wafers per month by 2030. The expansion will be heavily concentrated in the Yongin cluster, with Phase 1 alone expected to add about 360,000 wafers per month of capacity in the first half of 2030.


Samsung Electronics is simultaneously pushing forward its own Yongin cluster development. The parallel expansion by both Korean memory giants is driven by a single factor: explosive demand for HBM (High Bandwidth Memory). SK Hynix currently holds over 50% of the global HBM market, yet its capacity still struggles to keep pace with AI-driven compute demand. Citi estimates that the global DRAM market will face an approximate 5% supply deficit in 2026, with shortages concentrated in HBM. UBS notes that some customers have already secured up to eight quarters of demand visibility with equipment suppliers, signaling a stronger-than-expected capital equipment cycle. While markets had already anticipated SK Hynix’s capacity expansion, the magnitude of this 10-year acceleration significantly exceeds prior industry expectations.

Comment: SK Hynix’s decision to bring forward the completion timeline of its fourth fab in the Yongin semiconductor cluster by 10 years marks a clear acceleration in the expansion cycle of global memory leaders, shifting from a gradual capacity build-out to an aggressive supply expansion phase. This move is driven by a structural supply-demand imbalance in the HBM market, where the current gap is estimated at approximately 40%–50%, and the combined HBM capacity of major suppliers for 2026 has already been fully pre-sold. With AI model training requiring 5–10 times more compute intensity than traditional workloads, HBM has effectively evolved from an optional component into a critical “oxygen supply” layer for AI compute systems.


This expansion is an inevitable result of the global AI compute “arms race” extending from GPU chips into the underlying memory layer. The Yongin cluster, as SK Hynix’s future production core, is expected to integrate EUV lithography, TSV (Through-Silicon Via) processes, and advanced packaging lines, with a focus on HBM4 and next-generation products. While the expansion will help alleviate medium-term supply constraints, physical limitations in cleanroom construction, equipment delivery timelines, and yield ramp-up imply that meaningful incremental supply will remain limited over the next 2–3 years.

03
Market Trend(June 25
Global Analog Chip “Price Hike Wave”: Leading Vendors Raise Prices for the Third Time as Supply Chain Restructuring Accelerates
Since 2026, the global analog semiconductor market has entered a new round of price increases, with leading vendors repeatedly raising prices and the magnitude of hikes continuing to expand. Texas Instruments (TI) issued its third price increase notice on May 8, 2026, announcing a comprehensive price adjustment effective July 1. Certain products, including digital isolators, isolated gate drivers, and power management ICs, will see price increases ranging from 15% to 85%. Previously, TI implemented two rounds of price adjustments in August 2025 (10%–30%) and April 2026 (15%–85%). Analog Devices (ADI), the world’s second-largest analog chipmaker, also implemented price adjustments across its full product portfolio starting February 1, 2026. Commercial-grade products increased by 10%–15%, industrial-grade products by approximately 15%, and military-grade products by up to 30%. NXP also announced price increases effective June 1, citing inflationary cost pressures in raw materials, energy, labor, and logistics.

This round of price increases is not an isolated event but the result of a structural contraction in mature-node capacity combined with a synchronized recovery in downstream demand. According to TrendForce, TSMC and Samsung continue to reduce 8-inch mature-node capacity. Global 8-inch capacity is expected to decline by 2.4% YoY in 2026, while utilization rates have rebounded to nearly 90%. Meanwhile, AI data center expansion is driving strong demand for power management ICs and signal chain products. Inventories in industrial and automotive sectors have returned to healthy levels, supporting robust restocking demand. TI’s analog segment revenue grew 22% YoY in Q1 2026, while NXP’s industrial and IoT business increased by 24% YoY, both exceeding market expectations.

Comment: This round of analog chip price increases marks a structural shift in the industry cycle from inventory destocking to a tight supply-demand equilibrium, with underlying drivers extending far beyond simple cost transmission. Price increases are driven by three structural forces: 1) Continuous contraction of 8-inch wafer capacity at the upstream level, as TSMC and Samsung shift resources toward advanced nodes and AI chips, resulting in constrained supply of mature-node capacity that analog semiconductors rely on. 2) The surge in AI computing demand is driving strong growth in analog devices such as power management and signal chain products, significantly increasing analog chip value content per vehicle and per server. 3) Rising prices of key raw materials such as gold, silver, and copper, combined with increasing global energy and logistics costs, are forming a dual-driven structure of cost-push and demand-pull dynamics.


Successive price increases by leading players such as TI and ADI reflect a structural shift in pricing strategy—from a “market share defense war” based on price-for-volume over the past four years to a return to value creation centered on profitability. This shift creates three major opportunities for Chinese analog chip manufacturers: 1) Pricing dividend: following price increases by global peers, domestic players can follow suit and rapidly restore gross margins. 2) Order migration: extended lead times from overseas suppliers are accelerating adoption of domestic solutions across automotive, industrial, and computing sectors. 3) Domestic substitution window: current localization rate in analog chips is approximately 20%–30%, reaching about 50% in power management, while high-end segments such as precision signal chain and data conversion still rely on imports. This price cycle provides a penetration opportunity for technologically advanced domestic vendors.

04
Company Trend(June 26
Onsemi to Acquire Synaptics for Nearly USD 7bn in All-Stock Deal: Targeting the “Physical AI” Frontier

Onsemi announced that it will acquire Synaptics in an all-stock transaction valued at approximately USD 7bn, marking the largest acquisition in the company’s history.


The deal is structured as a share swap, with each Synaptics share exchanged for 1.35 Onsemi common shares, representing a 19% premium to the target’s 10-day average price. Upon completion, Synaptics shareholders are expected to hold approximately 12% of the combined entity. The transaction is expected to close by mid-2027, subject to shareholder and regulatory approvals across multiple jurisdictions. The acquisition is intended to strengthen Onsemi’s position in the emerging “physical AI” sector. The integration combines Synaptics’ edge AI processors, wireless connectivity, and human–machine interface product portfolio with Onsemi’s capabilities in sensing and power semiconductors, aiming to establish an end-to-end stack covering sensing, connectivity, and compute. The combined platform is expected to add approximately USD 30bn in potential market opportunity by 2030, expanding the total addressable market to around USD 243bn. From a financial perspective, the transaction is expected to be accretive to earnings per share approximately 18 months after closing, with projected annual cost and revenue synergies of around USD 200mn. Following the announcement, Onsemi’s shares declined approximately 6% in after-hours trading. The deal comes amid a broader wave of semiconductor and technology companies accelerating M&A activity to strengthen AI-related hardware and software capabilities.

Comment: This acquisition is a landmark move by a global power semiconductor vendor to strengthen its position in edge physical AI, reflecting a shift in industry competition from cloud-based computing power toward intelligent endpoint systems such as automotive platforms and industrial robots. Through this deal, onsemi aims to fill gaps in edge computing and wireless connectivity, while integrating its strengths in power semiconductors and sensing with AI compute capabilities. This will help the company build differentiated system-level solutions and align with the broader trend of AI moving deeper into physical hardware. At present, peers including Qualcomm, TI, and STMicroelectronics are also accelerating M&A and integration in related fields, indicating that the industry has entered a new phase of consolidation. However, the transaction still faces multiple uncertainties. The all-stock structure will dilute existing shareholders, while integration across the two companies’ customer bases and product portfolios could prove challenging. In addition, the approval process is expected to last more than a year, making it difficult for synergies to materialize in the short term. The market’s reaction, reflected in the decline in onsemi’s share price, also shows investor concern over the high acquisition cost. For Chinese companies, the rapid expansion of full-stack capabilities by overseas giants through M&A will pressure domestic power semiconductor and edge AI chip companies to accelerate product coordination and supply chain integration, in order to build integrated hardware-software competitiveness in domestic physical AI.

05
Market Trend(June 22
RMB 11.448bn Capital Injection Completed as Domestic 300mm Silicon Wafer Expansion Accelerates

Recently, National Silicon Industry Group announced that it and its major shareholder Shanghai Guosheng Group have completed a total RMB 11.448bn capital increase in its core subsidiary Zing Semiconductor Corporation. The full amount will be invested in a capacity upgrade project for high-end 12-inch semiconductor silicon wafers, marking the formal implementation of this RMB 10bn-level capital increase.


The capital injection consists of two funding channels. The listed company will contribute equity interests in three of its silicon wafer business entities, valued at RMB 7.448bn, while Guosheng Group will provide RMB 4bn in cash. After the capital increase, Zing Semiconductor’s registered capital will expand significantly, while the listed company will continue to maintain absolute control. Zing Semiconductor is a core domestic mass-production platform for 300mm silicon wafers, with current monthly capacity of 850,000 wafers. Supported by the new funding, the company will expand production bases in both Shanghai and Taiyuan, increasing capacity for polished wafers and epitaxial wafers.


At present, the global high-end large-diameter silicon wafer market remains dominated by overseas leaders, while domestic wafer fabs continue to face a significant supply gap. Demand for 12-inch silicon wafers is being driven higher by AI computing and automotive chips. This large-scale capital injection will accelerate process iteration and customer qualification for domestic silicon wafers, further improve local material self-sufficiency, and support coordinated localization across upstream and downstream segments including crystal growth furnaces and high-purity consumables. The move will strengthen the security foundation of China’s upstream integrated circuit supply chain and further narrow the capacity and technology gap with international leaders.

06
Company Trend(June 25
Domestic Nanoimprint Lithography System Officially Delivered, Core Performance Surpasses Comparable Canon Equipment

Recently, Prinano’s self-developed PL-AS vacuum pneumatic wafer-level nanoimprint lithography system completed commercial delivery. The equipment has been deployed on a photonic chip production line in Shenzhen and passed mass-production validation on 8-inch wafers, marking China’s breakthrough from Japan’s long-standing monopoly and a technological leap in the nanoimprint lithography sector.


Nanoimprint lithography is a core patterning technology for photonic chips and LiDAR devices. Previously, the global market for mass-production nanoimprint equipment had been exclusively dominated by Canon. The domestically developed system adopts a full-area vacuum pneumatic surface-pressure architecture, creating a generational advantage over Canon’s step-and-repeat small-field imprinting approach. The domestic equipment achieves resolution below 10nm, outperforming Canon’s 14nm limit. Its full-wafer pressure uniformity error is below 0.5%, while residual layer thickness variation is controlled within 2nm, helping avoid alignment errors and yield losses caused by Canon’s multi-step imprinting process.


In terms of mass-production efficiency, the domestic system can complete full-wafer imprinting in a single step, more than doubling throughput compared with Canon’s equipment. The overall process bypasses DUV lithography, reducing photonic chip manufacturing costs to only one-tenth of traditional optical lithography solutions, while significantly lowering equipment operation, maintenance, and consumables expenses. The entire equipment system, photoresist, and template process have achieved domestic self-reliance and controllability, supporting mass production of silicon photonics and OPA automotive LiDAR chips.


Industry analysts noted that the deployment of this equipment fills a key gap in China’s micro- and nano-fabrication equipment sector and establishes a differentiated breakthrough path. It can help ease supply constraints in high-end optical lithography equipment, accelerate cost reduction and production scaling across China’s high-speed optical module and automotive optical sensing supply chains, and provide a domestic solution capable of directly competing with Canon in the global nanoimprint lithography market.

07
Company Trend(June 25
CXMT Q1 Revenue Surges 719%, SemiAnalysis Expects Company to Overtake Micron as World’s Third-Largest DRAM Supplier by End-2026

U.S. semiconductor research firm SemiAnalysis said in an in-depth report that ChangXin Memory Technologies (CXMT), China’s leading DRAM manufacturer, is expected to overtake Micron by the end of 2026 and become the world’s third-largest DRAM supplier. CXMT reported revenue of RMB 50.8bn in the first quarter of 2026, up 719.13% YoY. SemiAnalysis expects the company’s full-year revenue to exceed USD 50bn. Its profitability was also remarkable, with a profit margin of approximately 70% in the first quarter. In 2025, CXMT achieved annual profitability for the first time, posting net profit of RMB 1.875bn and a gross margin of 41.02%. In terms of capacity, CXMT is expected to reach monthly production capacity of 350,000 12-inch-equivalent wafers by the end of 2026 and increase this to 500,000 wafers by 2028. Its share of the global DRAM market is projected to rise from 11% in 2025 to approximately 17%. Data released by Counterpoint Research on the same day showed that CXMT’s DRAM market share had increased from 3% last year to 8% in the first quarter of this year, while the overall DRAM market expanded 80% QoQ and 260% YoY.


At present, CXMT’s biggest weakness remains in HBM, where its capacity is only around 5,000 wafers per month, and mass production of HBM3 8-hi still faces yield challenges. However, SemiAnalysis expects CXMT’s HBM wafer capacity to reach 100,000 wafers per month by 2028, accounting for 12% of global capacity. CXMT plans to raise RMB 29.5bn through an IPO on the STAR Market, with approximately 70% of the proceeds to be used for wafer manufacturing expansion and DRAM technology upgrades.

了解最新消息

如果你想先了解我们的最新消息,可以即刻订阅
×
即刻订阅
欢迎您订阅泓明网站邮件!


只需填写您的E-MAIL地址和其他信息,您就会收到泓明为您订制的相关新闻、服务、政策、产业等信息内容。


愿泓明能成为您的好伙伴!


您的姓名: *
您的公司: *
联系电话:
Email: *
地址:
订阅内容: *
您想订阅那种信息
  • 公司动态
  • 行业前沿
  • 政策法规
  • 产业洞察
城市:
省:
国家: