From Packaging to Platform: ECTC and IEDM Lead the Era of System-Level Integration
Original Articles By SemiVision Research (IEEE ECTC, IEEE VLSI , IEEE IEDM ,TSMC, Nvidia, Bosch,Google ,Chipletz,ASE,Tyndall ,Intel.AIMPhotonics,AMAT,AMD,Samsung,Imec ,KLA,Adeia,Marvell ,Georgia Tech)
As the global electronics industry moves into a new wave driven by AI and high-performance computing (HPC), advanced packaging technologies are quietly becoming the foundational pillars that support the compute infrastructure of the future. From accelerating inference of large language models (LLMs) to meeting the data center’s growing demand for low-power, high-speed transmission, this transformation is no longer about isolated innovation—it is a full-scale system-level integration race.
This is precisely why the annual IEEE Electronic Components and Technology Conference (ECTC) has become the most iconic and influential event in the global semiconductor and advanced packaging industry. Marking its 75th anniversary this year, ECTC once again demonstrated its irreplaceable position through record-breaking participation, paper submissions, and corporate engagement—solidifying its role as a premier platform for trend forecasting, ecosystem dialogue, and technological foresight.
Whether you are in chip design, materials R&D, advanced packaging, system integration, or AI applications, ECTC is a must-attend event. It brings together technology leaders from the world’s top companies and research institutions, offering a rare opportunity for international perspective and collaboration. Attendees are placed directly at the frontier of innovation, gaining insight into the technologies shaping tomorrow.
In an era where AI and heterogeneous integration are redrawing the landscape of the electronics industry, ECTC is the global stage that every technologist should engage with. The future is already here—and transformation starts now.
As we enter the age of AI-driven heterogeneous integration, the bottlenecks of advanced semiconductor manufacturing are no longer confined to process scaling. Instead, breakthroughs must come from co-optimization at the system level. Methodologies such as Design-Technology Co-Optimization (DTCO) and System-Technology Co-Optimization (STCO) have already become central pillars in process development. However, in the realm of advanced packaging, the challenge goes beyond packaging itself—it is a matter of system-level integration.
Take TSMC’s latest SoW-X architecture as an example. It represents more than an evolution of packaging platforms—it is a convergence of thermal management, power delivery, SerDes interconnect, and photonic-electronic integration. The future of packaging will not be led by OSATs alone; it will require collaboration among system integrators, IC designers, materials innovators, and test equipment suppliers.
Moreover, as the Optical Engine becomes a key solution for AI/HPC data centers, the complexity of design is escalating rapidly. Innovations in EDA tools are now required to co-optimize PICs (Photonic ICs), EICs (Electronic ICs), and advanced packages. In parallel, advanced simulation methodologies such as PSSR (Power Supply Sensitive Response) must be introduced to address the intricate electrical-optical design constraints.
This is why we paid close attention to TSMC’s educational showcase on 3D Optical Interconnects at this year’s IEEE VLSI Symposium. From fundamental building blocks to interconnect schemes, packaging flows, and system-level simulation, TSMC demonstrated its leadership in next-generation integration.
However, the most critical stage for unveiling the future of high-density optical-electrical system integration is just ahead: the IEEE IEDM (International Electron Devices Meeting) at the end of this year. As the most authoritative global conference on semiconductor device technology, IEDM is more than a venue for paper presentations—it is where roadmaps are defined, ecosystems align, and futures are envisioned.
For every engineer and researcher invested in silicon photonics, heterogeneous integration, and AI-driven silicon systems, IEEE IEDM is the one conference you absolutely cannot miss in 2025. The future is approaching fast—breakthrough begins at IEDM.
Below is a summary of key takeaways from the 2025 IEEE ECTC (Electronic Components and Technology Conference):
DALLAS, TX (JUNE 25, 2025) – The 75th annual 2025 IEEE Electronic Components and Technology Conference (ECTC), held at the Gaylord Texan Resort & Convention Center here May 27-30, had record attendance, a record number of paper submissions/presentations, record international and student participation, and a record number of exhibitors in a sold-out exhibition hall:
2,518 attendees, the highest in the conference’s 75-year history and a significant increase over the 2,008 who attended last year, which itself was a record.
The number of abstracts submitted was the highest ever (775), as were the 390 technical papers presented in 36 oral and 5 interactive presentation sessions, including one dedicated to students.
See https://www.ectc.net/press/index.cfm for highlights of some of the noteworthy technical papers which were presented.
Several paper presentations attracted more than 600 attendees, as sessions on topics of intense industry interest – such as hybrid bonding– were standing-room only.
16 professional development courses were attended by 596 participants.
There were speakers from more than 20 countries globally.
There was a record level of industry support, with 51 corporate sponsors and 138 booths in the exhibit hall.
Among the highlights were 11 Special Sessions. In these, panels of industry experts discussed the present status and future roadmaps of technologies essential for artificial intelligence (AI), high-performance computing (HPC) and other fast-growing, evolving applications.
“Advanced chip packaging technologies are essential for the development of the electronics industry, and the ECTC conference has long been the world’s leading forum for advancements in microelectronics packaging and component science and technology,” said Przemyslaw Gromala, ECTC 2025 Program Chair and Chief Expert/R&D Project Leader at Robert Bosch GmbH. “ECTC serves as a collaborative global platform for exploring cutting-edge advancements in microelectronic packaging, fostering innovation and addressing key industry challenges. This year’s Special Sessions offered a rich selection of compelling topics and expert panelists.”
Here are highlights from three of the ECTC 2025 Special Sessions:
Advanced Materials for Enabling Co-Packaged Optics Integration – This Special Session was co-Chaired by Karan Bhangaonkar (Google) and Vidya Jayaram (Chipletz). Panelists were Mark Gerber (ASE), Z. Rena Huang (Rennselaer Polytechnic Inst.); Padraic Morrissey (Tyndall National Inst.), Kumar Abhishek Singh (Intel) and Christopher Striemer (AIM Photonics).
As modern computing strives for higher performance, co-packaged optics or CPO (i.e., the integration of optics and electronics on a substrate) is emerging as a solution to meet computing/communication demands for high bandwidth at low power. The innovations, challenges and future needs to realize CPO technology were discussed in this Special Session.
Gerber from ASE gave an overview of the system requirements that are driving CPO material considerations, noting that heat can have significant effects on photonic integrated circuits. He also identified other issues that increase thermal sensitivities in CPO architectures, such as flux outgassing, and described why the order in which assembly steps occur also has an impact.
Huang from RPI said that while much work is taking place to understand and address CPO packaging considerations, much more progress is needed, especially for AI chiplets for large-language models (LLMs), where the key concerns are speed, power and efficiency. She discussed the possibility to build optical networks on optical interposers/wafers/panels, noting that large chiplets could be optically connected with optical interposers, using optical waveguides to reduce loss.
Morrissey from Tyndall said that for optimum CPO performance, the optics must be moved closer to the edge of compute, and glass has great potential for use as a substrate because it’s optically transparent, has good RF behavior, and lends itself to fabricating high-quality vias. Glass also can work beyond wafers and with really large panels, he said, and can lead to pluggable connectivity. However, he noted that heat is an issue with glass.
Singh from Intel said CPO isn’t just desirable, it’s absolutely necessary to scale-up advanced packaging. He noted that CPO allows both edge and vertical interconnects, meaning there are many areas where materials come into play. He said that while CPO architectures face unique challenges – such as foreign particles blocking the light path – they also face many of the same issues as non-CPO architectures, such as misalignment and cracks. But these challenges bring opportunities to advance the state-of-the-art.
Striemer from Aim Photonics described why optically active materials will drive CPO, and also outlined the need for passive material innovations such as 3D printing and novel designs like suspended structures, although right now it’s unclear which ones will offer the most benefit.
Hybrid Bonding (HB): to B, or not to B? Needs and challenges for the next decade – This Special Session was co-Chaired by Benson Chan (Binghamton Univ.), Masha Gorchichko (Applied Materials) and Dishit Parekh (AMD). The panelists were Su Jin Ahn (Samsung), Anne Jourdain (imec), Chet Lenox (KLA), Laura Mirkarimi (Adeia), Masao Tomikawa (Toray Ind.) and Brett Wilkerson (AMD).
Hybrid bonding is the key technology for high-density 3D integration and advanced packaging, and in recent years, significant advancements were made in pitch scaling, die-to-die bonding, alternative materials, and low-temperature processes. But many engineering and technological challenges remain, such as defectivity, metrology, design challenges, and cost. This panel summarized recent advancements in HB, identified the most pressing issues limiting its adoption for mainstream electronics, and outlined its likely development over the next decade.
Gorchichko from Applied noted the different types of HB (wafer-wafer, die-wafer, die-die) and said that while HB debuted about 10 years ago in an image sensor from Sony, today the focus is on integrating DRAM memory. She said that advanced metrology is key, because we are now talking about molecular bonds, and therefore an understanding of all the relevant chemical and mechanical requirements is needed. Moreover, to get higher yields and more throughput, not just better but also faster metrology is needed. She noted that thermal concerns are an issue with increasing power density.
Su Jin Ahn from Samsung outlined major HB technology issues and challenges. One is that the many process steps required leave particles on the bonding surface, leading to potential failure. Another is that for AI, the wider, thinner dies used degrade bonding and lead to quality issues. She said what’s needed are advanced metrology/inspection tools and methods, along with design/process co-optimization for bonding. She said the main driver going forward is the need to combine chip and packaging technologies.
Jourdain from imec emphasized the need for fast, reliable metrology solutions, and discussed the pros and cons of copper interconnect and barrier metals.
Lenox from KLA noted the many needs and challenges that come with HB – interposers, warpage control, dielectric interface profiles, clean singulation, bonding alignment – and said that while HB is important, there are other less complex and costly technologies that might preclude the need for HB, such as bridges. He also said that HB creates a need for advanced metrology and inspection capabilities all the way from the front end to packaging.
Mirkarimi from Adeia focused on three areas: metrology improvements for improved throughput and yield; the evolution of 2.5D/3.5D HB packaging technology; and thermal solutions for high power-density chipsets. With regard to metrology, she noted that HB architectural complexity demands a reliable “health of the line” metrology protocol for all process steps. Also, better ways to understand nanoscale topography and to detect surface defects are needed, as are ways to rework HB to reduce costs, such as bond energy engineering. Regarding HB packaging trends and challenges, she said that ultra-high bandwidth, inter-die communication at a 1µm die-to-wafer pitch will require system simplification, such as bonding dies directly to the substrate or using bridge dies to replace an interposer. For thermal management, she described a potential cooling solution that makes use of an integrated manifold and a cold plate bonded to an IC, among other features, and which can be custom-designed to manage specific heat maps.
Tomikawa from Toray said that as the need to bond chips to interposers increases, the need for HB processes that make use of polyimide (PI) resin will become more apparent. That’s because PI enables low-temperature, low-pressure HB processes which minimize warpage and device damage. Many challenges remain, though – precise copper protrusion control and low-temperature copper diffusion bonding are key factors for success.
Wilkerson from AMD spoke from a product perspective. He pointed out that HB requires complex processing. much time and many expensive fab processes, and that it can take several weeks to accomplish, which impacts a company’s time-to-market capabilities. He said thermal resistance is a critical issue, and that there’s a need for standards for the use of HB for memory integration with silicon.
Thermal Management Solutions for Next-Generation Backside Power Delivery – This Special Session was co-Chaired by Dwayne R. Shirley (Marvell) and Tiwei Wei (Purdue University/UCLA). The panelists were Muhannad Bakir (Georgia Tech), Dureseti Chidambarrao (IBM) and Herman Oprins (imec).
The increasing power density and thermal challenges in advanced semiconductor packaging have led to the development of backside power delivery (BPD) technology, where the power delivery network is relocated from the frontside to the backside of a silicon wafer. While it enhances power efficiency, performance, and design flexibility, BPD also introduces thermal management challenges and the need for innovative cooling solutions. Participants in this Special Session discussed the latest advancements and challenges in thermal management for next-generation BPD.
Bakir from Georgia Tech said that for 3D architectures, interlayer cooling is needed, but asked, how do we enable high-density interconnect within such a structure? He said the solution is to fabricate through-silicon via (TSV) structures with integrated cooling, electrical conductivity, and power, using vertical vias. He addressed the many issues that such a design brings – aspect ratios, TSV heights, etc.
Chidambarrao from IBM said that because there are lots of subtleties with BPD architectures, reducing the problem to important basics and solving it with enough accuracy is key. He noted that chip complexity significantly impacts thermal conductivity, and that level-to-level differences can be quite large. He said that an understanding of the entire chip overlay is needed, because there are tricky hot spots, and he described IBM’s strategy, which is to use machine learning plus FEA modeling to calculate the average properties of multiple levels together. He noted that BPD issues only become worse with 3D architectures.
Oprins from imec said it is absolutely critical to identify the basic problems with each particular BPD architecture, because there are so many different flavors of BPD.
Wei from Purdue/UCLA said that bond interfaces drive thermal impacts, and she gave an overview of research into different ways to deal with this issue, encompassing factors such as die thickness, CMOS-compatible structures like air gaps or glass bridges, trenches, copper/diamond microbump bonding, two-layer microchannel structures (i.e. manifolds), and others.
Follow ECTC on social media
LinkedIn: https://www.linkedin.com/groups/1916290/
About ECTC, EPS & IEEE
The ECTC technical program addresses new developments, trends and applications for a broad range of topics, including components, materials, assembly, reliability, modeling, interconnect design and technology, device and system packaging, heterogeneous integration, wafer level packaging, photonics and optoelectronics, IoT, 5G, quantum computing and systems, 2.5D and 3D integration technology, and other emerging technologies relevant to electronics packaging.
The IEEE’s Electronics Packaging Society (EPS) sponsors ECTC. EPS is the leading international forum for scientists and engineers engaged in research, design and development of revolutionary advances in microsystems packaging and manufacturing. It encompasses all aspects of packaging and integration of electrical, electronic, optoelectronic, biological, micromechanical and sensing components.
IEEE is the world’s largest technical professional organization and a public charity dedicated to advancing technology for the benefit of humanity. Through its highly cited publications, conferences, technology standards, and professional and educational activities, IEEE is the trusted voice on a wide variety of areas ranging from aerospace systems, computers, and telecommunications, to biomedical engineering, electric power, and consumer electronics.
Companies & Industrial Organizations
Robert Bosch GmbH (Bosch) – Przemyslaw Gromala’s affiliation
Google – Karan Bhangaonkar chaired CPO session
Chipletz – Vidya Jayaram co-chaired CPO session
ASE – Mark Gerber presented in CPO session
Rensselaer Polytechnic Institute – Z. Rena Huang participated
Tyndall National Institute – Padraic Morrissey participated
Intel – Kumar Abhishek Singh participated
AIM Photonics – Christopher Striemer participated
Applied Materials – Masha Gorchichko co-chaired HB session
AMD – Dishit Parekh co-chaired HB session; Brett Wilkerson participated; Sam Naffziger gave keynote
Samsung Electronics – Su Jin Ahn participated
imec – Anne Jourdain, Xiao Sun, Herman Oprins involved
KLA Corporation – Chet Lenox participated
Adeia – Laura Mirkarimi participated
Toray Industries, Inc. – Masao Tomikawa participated
Marvell Semiconductor, Inc. – Dwayne Shirley co-chaired BPD session
Georgia Tech – Muhannad Bakir represented
IBM – Dureseti Chidambarrao participated
Academic & Research Institutions
Rensselaer Polytechnic Institute (RPI)
Tyndall National Institute
Purdue University/UCLA (Tiwei Wei)
Georgia Tech
IBM Research
Note: Some panelists also represented university affiliations (e.g., RPI, Purdue/UCLA, Georgia Tech).