TL;DR
IMEC’s 300mm RF silicon interposer platform hits record-low signal loss at 325GHz, a key step toward scalable, affordable 6G chip manufacturing.
IMEC, the Belgian semiconductor research institute that partners with more than 600 chip industry players, has expanded its 300mm RF silicon interposer platform with three new manufacturing capabilities that bring 6G chip production closer to commercial viability. The platform achieves record-low signal loss at frequencies up to 325GHz, covering the millimetre-wave and sub-terahertz bands that 6G networks will require. The work was presented at the IEEE International Microwave Symposium in Boston this month.
The core problem IMEC is solving is one of materials and economics. 6G radios will need to operate at frequencies far above what conventional silicon can handle, requiring compound semiconductors such as indium phosphide, gallium arsenide, and silicon germanium. These materials deliver superior RF performance but are manufactured on small, expensive wafers that do not scale the way standard 300mm silicon production lines do.
IMEC’s approach uses a silicon interposer as a carrier substrate, allowing engineers to integrate small chiplets made from compound semiconductors onto a standard 300mm silicon wafer. The interposer handles the digital interconnects and passive components while the III-V chiplets handle the RF signal processing. The result is a mix-and-match platform where different materials can be combined without requiring each one to scale independently.
The three new capabilities announced in June address specific manufacturing bottlenecks. High-density embedded capacitors, known as MIMCAPs, allow passive components to be offloaded from the expensive III-V chiplets onto the cheaper silicon interposer, reducing chiplet size and cost. A scalable modelling framework for passive components gives designers the simulation tools needed to predict performance before fabrication. Laser-assisted bonding enables precise placement of III-V chiplets onto the silicon carrier.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
“With this work, we demonstrate a uniquely integrated platform that brings together performance, scalability, and manufacturability,” said Xiao Sun, principal member of technical staff at IMEC. “Our next priority is to further advance the platform’s technology readiness, and to enable support for low-volume manufacturing, helping our partners more easily develop and scale next-generation RF systems.”
The timing matters because Nvidia has made telecom one of its next major growth targets. The company invested $1 billion in Nokia last October for a 2.9 per cent stake, and at Mobile World Congress in March it assembled a coalition including Ericsson, Deutsche Telekom, T-Mobile, SK Telecom, and SoftBank to build 6G networks on what it calls AI-native platforms. Nvidia has been expanding its AI partnerships across hardware categories, from robotics and data centres to automotive, and telecom infrastructure represents the next logical extension of that strategy.
Jensen Huang has argued that every radio access network in a 6G world will essentially behave as an AI computer, blurring the line between communications hardware and AI inference. If that vision holds, the bottleneck shifts from software to silicon, specifically to how cheaply and reliably the underlying RF chips can be manufactured at scale. That is precisely where IMEC’s platform fits.
IMEC occupies an unusual position in the semiconductor ecosystem. It is a non-profit headquartered in Leuven, Belgium, with more than 5,000 researchers from 96 countries. Its business model is to develop pre-competitive chip technology in partnership with industry, then hand off the results for commercialisation. TSMC, which is expanding its European chip manufacturing footprint, is among IMEC’s longstanding partners, along with Samsung, Intel, and most of the world’s major foundries.
The significance of the 325GHz insertion loss benchmark is that it covers not just the frequencies 6G is expected to use initially but also the sub-terahertz range that researchers are exploring for ultra-high-bandwidth short-range links. Achieving low signal loss at these frequencies on a standard silicon manufacturing platform, rather than on exotic substrates, is what makes the work relevant to deployment economics rather than purely to laboratory performance.
None of this means 6G hardware is ready for production. IMEC’s own roadmap acknowledges that the platform needs further development to reach low-volume manufacturing readiness, let alone the high-volume production that global telecom deployment would require. But the gap between research breakthrough and commercial chip typically runs five to seven years in the semiconductor industry, and 6G networks are not expected to begin standardisation until 2028 at the earliest. IMEC’s platform is on a timeline that could intersect with the moment the telecom industry needs it.
