🔥Stalecollected in 5m

Intel EMIB Clients H2, Billions Revenue

Intel EMIB Clients H2, Billions Revenue
PostLinkedIn
🔥Read original on 36氪

💡Intel's EMIB packaging to unlock billions—key for AI chip supply chain shifts

⚡ 30-Second TL;DR

What Changed

First EMIB packaging customers expected in H2 2024

Why It Matters

Strengthens Intel's foundry services amid AI chip demand boom. Positions Intel to challenge TSMC in advanced packaging for multi-die AI processors. Signals growing external adoption of Intel's packaging tech.

What To Do Next

Contact Intel Foundry Services to inquire about EMIB design kits for AI accelerator prototypes.

Who should care:Enterprise & Security Teams

🧠 Deep Insight

Web-grounded analysis with 8 cited sources.

🔑 Enhanced Key Takeaways

  • EMIB technology has been in high-volume production since 2017 for server, network, and HPC products, demonstrating proven manufacturability and reliability across multiple Intel product generations[5]
  • Industry-standard EDA tool support from Ansys, Cadence, Siemens, and Synopsys was qualified for EMIB designs as of early 2025, removing a critical barrier to third-party customer adoption[2]
  • Amkor Technology established a strategic partnership with Intel to implement EMIB assembly processes at facilities in Korea, Portugal, and Arizona, creating alternative manufacturing sources and expanding global packaging capacity[4]

🛠️ Technical Deep Dive

  • EMIB uses small embedded silicon bridges positioned directly in package substrate cavities, held with adhesive, followed by dielectric and metal build-up layers with via drilling and plating[3]
  • EMIB-T variant integrates through-silicon vias (TSVs) into the bridge architecture to support high-bandwidth interfaces including HBM4 and Universal Chiplet Interconnect Express (UCIe)[1]
  • EMIB bridges are thinned to less than 75 micrometers prior to assembly, minimizing thermal-induced mechanical strain while maintaining reliability comparable to conventional organic packages[3]
  • The technology combines 2.5D lateral high-speed interconnects with Foveros 3D vertical stacking (Co-EMIB architecture) to enable multi-chiplet designs exceeding reticle size limits with optimized power, performance, and area characteristics[1][7]
  • Individual die-to-die links can be customized with optimized driver/receiver circuitry, supporting high datarate signaling with low metal RC delays and minimal latency[3]

🔮 Future ImplicationsAI analysis grounded in cited sources

EMIB adoption by foundry customers will accelerate heterogeneous chiplet integration for AI and HPC workloads
EDA tool qualification and Amkor's manufacturing partnerships remove technical and supply-chain barriers that previously limited third-party customer access to the technology[2][4]
EMIB-T will enable higher-density interconnects for next-generation memory and compute integration
TSV-based architecture support for HBM4 and UCIe standards positions the technology for advanced memory-compute co-packaging in data-intensive applications[1]

Timeline

2017
EMIB enters high-volume production for Intel server, network, and HPC products
2025-01
Ansys, Cadence, Siemens, and Synopsys qualify EDA tool support for EMIB chiplet designs
2025
Intel Foundry Direct Connect highlights EMIB-T as key advanced packaging technology for chiplet roadmaps
2026-01
Amkor-Intel strategic partnership announced to expand EMIB assembly capacity in Korea, Portugal, and Arizona
📰

Weekly AI Recap

Read this week's curated digest of top AI events →

👉Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪