May packaging substrate shipment value hit a record high, while Intel EMiB-T advances competition in advanced packaging

Overview

This report focuses on semiconductor packaging substrates and progress in advanced packaging technology within Japan's electronic components sector. The core messages include: May monthly shipment value for packaging substrates reached a record high, with both pricing and area increasing; meanwhile, Intel disclosed technical details of EMiB-T at ECTC, showing improved power delivery capability in its interposer-free 2.5D packaging route, which may strengthen competition with TSMC in quad-die and HBM4-related advanced packaging.

Core views

The report believes demand for packaging substrates has recovered from the April slowdown and reached a new high, and Rubin packaging shipments may ramp materially in the summer. On technology roadmaps, the view is gaining traction that Rubin Ultra may shift from four dies to two dies, while some also believe Feynman may adopt a 3D stacked structure with two modules, each containing two dies. Due to the rising complexity of large interposers and HBM4 routing, the report believes leading packaging players using interposers may rely more on a 3.5D route that combines 3D stacking with 2.5D packaging before visibility on large interposers improves.

Analysis framework

The report mainly combines monthly packaging substrate statistics from Japan's production dynamics survey, changes in ASP and area, and Intel's EMiB-T technical paper presented at ECTC to assess supply-demand trends and the competitive landscape in advanced packaging technology.

Methodology notes

• industry monthly trackingBreakdown of packaging substrate shipment value, area, and ASP

  Observe demand strength and product mix changes simultaneously through value, area, and price per unit area.

  The report uses May packaging substrate shipment value, shipment volume in m², and average price per m² to judge that the industry recovered from the April slowdown and set a new high.

• technology roadmap assessmentAdvanced packaging power delivery and ecosystem competition analysis

  Compare TSMC and Intel in advanced packaging in terms of power delivery, thermal management, CPO, and 3D stacking capabilities.

  The report views EMiB-T's TSVs, power mesh, and MIM capacitors as evidence that Intel is improving package-level power delivery issues, and focuses on whether it can build an ecosystem similar to TSMC's 3DFabric Alliance.

Asset mapping & comparison

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

• packaging substrates

  The core industry asset and monthly statistical subject of this report.

  Strengths

  May shipment value, shipment area, and average selling price all increased year over year, with shipment value hitting a record high.

  Weaknesses

  There was a slowdown in April, and it remains to be seen whether demand momentum can continue in subsequent months.

  Comparison

  May shipment value of ￥27.8bn exceeded the previous high of ￥27.3bn in March; average selling price of ￥1.356mn/m² was above the previous high of ￥1.226mn/m² in April.

  Risks

  If the pull-in schedule for AI accelerator packaging is delayed, or if customer technology roadmaps change, subsequent shipment momentum could be affected.

• Intel EMiB-T

  An advanced packaging technology route and a potential competitive variable.

  Strengths

  Improves package-level power delivery through TSVs, a power mesh, and MIM capacitors, while not relying on an interposer.

  Weaknesses

  The report still relies mainly on technical papers and mass production preparation judgments; commercial ramp-up and customer adoption still need verification.

  Comparison

  Compared with traditional EMiB, the paper says EMiB-T improves power delivery network voltage drop by 68%-80%; compared with TSMC, Intel still needs to build a similar ecosystem.

  Risks

  Mass production progress, ecosystem partners, thermal management, CPO, and 3D stacking capabilities could all affect the pace of catch-up.

• TSMC advanced packaging

  The incumbent leader and benchmark in advanced packaging competition.

  Strengths

  The report believes TSMC has strong competitive advantages in package-level power delivery and thermal management and benefits from the 3DFabric Alliance ecosystem.

  Weaknesses

  The combination of larger interposers with HBM4 may face technical challenges.

  Comparison

  Intel may improve power delivery issues through EMiB-T and catch up quickly; TSMC may temporarily rely on a 3.5D route combining 3D stacking with 2.5D.

  Risks

  If Intel accelerates ecosystem development and customer adoption, competitive pressure on TSMC in quad-die advanced packaging may increase.

Key data

• May packaging substrate shipment value￥27.8bnUp 36% year over year, exceeding the previous record high of ￥27.3bn in March.
• May packaging substrate shipment area+10% y-yShipment volume measured in m² increased year over year.
• May average selling price￥1.356mn/m²Up 23% year over year, above the previous high of ￥1.226mn/m² in April.
• Intel EMiB-T mass production preparation2026The report expects Intel to complete mass production preparation in 2026 for products with 8x reticle size and four dies.
• EMiB-T packaging substrate size120x120mmThe report says this package can integrate up to 12 HBM4.
• EMiB-T power delivery network improvement68%-80%Intel's paper says that compared with EMiB, EMiB-T improves voltage drop in the package-level power delivery network by 68%-80%.

Impact & implications

If packaging substrate shipments continue to remain at elevated levels, demand visibility for the advanced AI accelerator packaging chain will improve. On the technology side, Intel EMiB-T's improvement in power delivery issues may make it more competitive in the interposer-free 2.5D packaging route and put catch-up pressure on TSMC's existing 3DFabric ecosystem. At the same time, the technical difficulty of combining large interposers with HBM4 may push the industry toward interim adoption of a 3.5D packaging route.

Risks

• Packaging architectures for AI accelerators such as Rubin Ultra and Feynman may continue to change, affecting the demand structure for packaging substrates.
• There are technical challenges in combining large interposers with HBM4, which may slow progress on some advanced packaging routes.
• Intel EMiB-T still needs to prove mass production, customer adoption, and ecosystem build-out progress.
• Monthly statistics are volatile, and May's record high does not necessarily mean subsequent months will continue rising.

What to watch

• Whether Rubin packaging shipments begin to ramp materially in the summer, or by around August at the latest.
• Whether subsequent packaging substrate shipment value, shipment area, and average selling price continue May's strength.
• Whether Rubin Ultra shifts from a four-die architecture to a two-die architecture, and whether Feynman adopts a structure with two 3D stacked modules.
• Progress in Intel EMiB-T adoption among accelerator customers such as Google.
• Expansion of partnerships by TSMC and Intel across power delivery, thermal management, CPO, and 3D stacking ecosystems.