日本30家企业抱团，能否改写半导体封测格局？

Note: The English version is provided below.

摘要

日本半导体产业链长期被“前重后轻”所困扰：熊本迎来台积电、北海道筹建 Rapidus，前道晶圆制造一时风光，后道封装测试却散落在九州、四国的地方工厂，八成是员工不足百人的中小企业。

如今，这股分散的力量开始合拢。4 月成立的“J-OSAT”联盟把 AOI ELECTRONICS、Amkor Technology Japan 等约 30 家企业、共 1.5 万名从业者纳入同一平台，首次共享产能、客户结构与成本数据。

按照计划，2025 年夏季起，联盟将围绕自动化、数据互联、人才培养、设备升级、联合采购五个分组会展开实质合作，并准备向政府申请 100 亿—200 亿日元补贴，用于引进新一代封装与测试设备。

会长澄田诚给出的测算简洁明了：如果 30 家企业共用平台、集中采购，后道整体成本可降 20%，而所需投入仅为前道晶圆厂一次扩产的百分之一。对汽车、工业芯片这类仍停留在 28 纳米及以上节点的日本主力产品而言，成本每降一分，便多一分留在本土的订单。

J-OSAT 的出现让人想起上世纪 80 年代日本 VLSI 联盟的翻版，但背景已大不相同。那时是为了追赶美国 DRAM，现在则是为了不被台积电、日月光、长电科技甩出赛道。

后道技术虽不像 EUV 光刻那般烧钱，却同样被规模与资本裹挟：全球前五名 OSAT 市占率超过八成，留给“小而散”日本厂商的缝隙越来越小。联盟能否真正弥合这道裂缝，关键不在口号，而在三步。

补贴能否落地。日本政府若只愿象征性出资，设备更新仍是一句空话；参照海外动辄补贴七成的先例，至少需拿出百亿日元级别的真金白银。
数据共享能否突破商业机密藩篱。后道工厂历来为 IDM 保密，联盟若无法建立可信的防火墙，联合采购与产能互补就无从谈起。
人才流动能否打破地域壁垒。九州、四国的中小企业常年面临“招不到、留不住”工程师的困境，若联盟不能建立统一培训与轮岗机制，自动化升级仍会卡在“人”这一关。

站在中国封测产业的角度来看此事件，短期内日本J-OSAT联盟对中国封测产业冲击有限，但需关注其在特定领域（如车规芯片）的竞争力提升。长期来看，日本J-OSAT联盟的成败将为中国提供“分散产业如何集约化”的观察样本，但中国封测的核心竞争力仍在于规模、速度与全产业链配套能力。

Can Japan's Consortium of 30 Companies Reshape the Semiconductor Packaging and Testing Landscape?

Summary

Japan's semiconductor supply chain has long been plagued by an "imbalanced focus," with excessive emphasis on front-end processes. While Kumamoto welcomes TSMC and Hokkaido prepares for Rapidus, making front-end wafer manufacturing the spotlight, back-end packaging and testing remain scattered across small local factories in Kyushu and Shikoku—80% of which are small and medium-sized enterprises (SMEs) with fewer than 100 employees.

Now, these fragmented forces are beginning to unite. The "J-OSAT" alliance, established in April, brings together around 30 companies, including AOI ELECTRONICS and Amkor Technology Japan, totaling 15,000 employees, under a single platform for the first time to share capacity, customer structures, and cost data.

According to the plan, starting in the summer of 2025, the alliance will launch substantive collaboration across five working groups: automation, data connectivity, talent development, equipment upgrades, and joint procurement. It also plans to apply for government subsidies of 10–20 billion yen to fund the introduction of next-generation packaging and testing equipment.

Makoto Sumida, the alliance's chairman, offers a straightforward calculation: if the 30 companies share a platform and engage in centralized procurement, overall back-end costs could be reduced by 20%, with the required investment being just 1% of a single front-end wafer fab expansion. For Japan's key products like automotive and industrial chips, which still rely on 28nm and above nodes, every cost reduction means more orders staying domestically.

Commentary

The emergence of J-OSAT evokes memories of Japan's VLSI consortium in the 1980s, but the context is vastly different. Back then, the goal was to catch up with U.S. DRAM; today, it is to avoid being left behind by TSMC, ASE, and JCET.

Though back-end technology is not as capital-intensive as EUV lithography, it is equally constrained by scale and capital: the top five global OSAT players hold over 80% of the market share, leaving shrinking room for Japan's "small and scattered" players. Whether the alliance can truly bridge this gap depends not on slogans but on three critical steps.

First, whether subsidies materialize. If the Japanese government offers only symbolic funding, equipment upgrades will remain an empty promise. Following overseas precedents where subsidies cover up to 70%, at least tens of billions of yen in real investment are needed.

Second, whether data sharing can overcome commercial secrecy barriers. Back-end factories have long been bound by IDM confidentiality. Without a trusted firewall, joint procurement and capacity sharing will be impossible.

Third, whether talent mobility can break geographical barriers. SMEs in Kyushu and Shikoku chronically struggle to attract and retain engineers. Without a unified training and rotation system, automation upgrades will still be hindered by the "human factor."

From the perspective of China's packaging and testing industry, the short-term impact of Japan's J-OSAT alliance is limited. However, attention should be paid to its potential competitiveness gains in specific areas (e.g., automotive-grade chips). In the long run, the success or failure of J-OSAT will provide China with a case study on "how to consolidate a fragmented industry." Yet, China's core competitiveness in packaging and testing still lies in scale, speed, and full supply chain integration.