Packaging Substrate Shipments Up 20%; 3.5D Packaging Competition Heats Up

Overview

This report updates the latest monthly data on Japan's semiconductor packaging industry, focusing on packaging substrate shipment volumes and pricing trends, while outlining the competitive landscape for next-generation 3.5D packaging technology. The report notes that although April shipments declined from the historical peak in March, they still achieved 20% YoY growth, with unit prices continuing to rise. Driven by AI chip iterations, demand for Rubin processor packaging is expected to surge in summer 2026. Meanwhile, the technological race between TSMC and Intel in the 3.5D packaging space has quietly begun.

Core views

Rising Demand and Prices: According to METI’s April Production Dynamics Statistics, packaging substrate (rigid module substrate) shipments reached JPY 23.3 billion, up 20% YoY. Although this represents a sequential decline from the March peak of JPY 27.3 billion, the drop was primarily due to factors elsewhere in the supply chain. More notably, the average price per square meter rose 19% YoY and slightly MoM to a record high of JPY 1.226 million, indicating a structural price increase driven by a higher mix of high-end products. Timing of Rubin Processor Ramp-Up: The report attributes part of the MoM decline in April shipments to delays in server production for the Rubin processor. Recalling the high area-based shipments during the full rollout of Blackwell packaging in April 2025, the institution expects the full volume ramp-up of Rubin packaging to begin in summer 2026, starting as early as June and no later than August, triggering a new wave of shipment peaks. 3.5D Packaging Technology Competition: Industry focus is shifting toward 3.5D packaging technology, which is expected to enter mass production in 2028. At its shareholder meeting, TSMC revealed plans to launch a panel-level packaging (CoPoS) pilot line in June 2026, followed by mass production in 2028 after validation; it is also preparing to advance mass production readiness for glass-core substrates with partners like Ibiden over the next two to three years. In contrast, Intel’s EMIB-T technology is securing more order commitments due to lower uncertainty; reportedly, Google has ordered at least 3 million TPUs using this technology, and NVIDIA is also considering it.

Analysis framework

The report employs a typical analytical framework combining 'high-frequency data tracking' with 'technology roadmap projection.' First, by interpreting government-released monthly production statistics, it breaks down shipments into volume (area) and unit price (price per square meter) to determine whether industry momentum is driven by volume expansion or price increases. Second, by aligning with downstream key customers' product iteration cycles (e.g., AI chipmakers transitioning from Blackwell to Rubin), it cross-validates the causes of short-term fluctuations and forecasts future ramp-up timing. Finally, by reviewing announcements from leading foundries (TSMC, Intel) and supply chain rumors, it compares the maturity and customer adoption of different technology routes (CoPoS vs. EMIB-T) to assess the medium- to long-term competitive landscape.

Methodology notes

• Industry/Sector Analysis FrameworkVolume-price decomposition

  Decomposing revenue or shipment metrics into 'volume' and 'price' dimensions for analysis

  By separately observing changes in substrate shipment area (volume) and unit price per square meter (price), the report finds that while volume grew only marginally by 1%, prices surged 19%, leading to the conclusion that industry growth is primarily driven by a shift toward high-end products rather than mere scale expansion.

• Industry/Sector Analysis FrameworkAdoption S-Curve

  The adoption process of new technologies from introduction to maturity

  The report analyzes the evolution path of 3.5D packaging technology—from piloting in 2026 to mass production in 2028—suggesting that the technology is at a critical juncture transitioning from the introduction phase to the growth phase, and investors should monitor the pace of its penetration rate ramp-up.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Ibiden

  TSMC glass-core substrate partner; potential beneficiary

  Strengths

  Collaborating with TSMC on glass-core substrate development; leading technology reserves

  Weaknesses

  Current capacity insufficient to meet large-scale demand such as Google's 3mn TPU order

  Comparison

  More deeply aligned with TSMC's advanced packaging roadmap compared to other substrate makers

  Risks

  Capacity expansion plans remain unclear; difficult to directly reflect benefits from large orders in the near term

Key data

• April Packaging Substrate ShipmentsJPY 23.3 bnUp 20% YoY; down sequentially from JPY 27.3 bn in March
• April Shipment Area+1% YoYMarginal area growth indicates growth is not volume-driven
• Average Price per Square MeterJPY 1.226 mnUp 19% YoY, reaching a record high
• Expected Rubin Packaging Ramp-UpSummer 2026Full rollout expected between June (earliest) and August (latest)
• Expected 3.5D Packaging Mass Production2028TSMC's CoPoS and glass-core substrates expected to enter mass production by then

Impact & implications

For upstream substrate manufacturers, near-term performance will be affected by the pace of the Rubin ramp-up, but the trend of rising prices supports earnings quality. Over the medium to long term, the battle over 3.5D packaging technology routes will reshape the supply chain landscape: if TSMC successfully brings CoPoS to mass production, it will solidify its back-end process alliance advantage; if Intel captures market share with early EMIB-T orders, it could force more design houses to build independent back-end supply chains or rely on Intel's platform. Furthermore, given that 3.5D packaging involves complex power delivery and thermal management, joint R&D ties between substrate makers and chip designers will deepen, allowing leading firms with high technical barriers to command greater premiums.

Risks

• Short-term shipment volatility due to delays in Rubin processor server production
• 3.5D packaging technology mass production progress falling short of expectations
• Technical transition challenges for design houses shifting from TSMC to Intel platforms

What to watch

• Actual ramp-up trajectory of Rubin processor packaging shipments from June to August 2026
• Whether substrate makers like Ibiden announce capacity expansion plans targeting 3.5D packaging
• Validation progress of TSMC's CoPoS pilot line and development status of glass-core substrates