The hardest part of building 6G was never the network. It was the phone. The radio frequencies that 6G needs are so high that putting a working transmitter and receiver into a battery-powered device in your hand has looked, until now, like a problem without a good answer. This week, a team from the Institute of Science Tokyo is presenting a chip that answers it, and they did it using a manufacturing process old enough to be in high school.
The work is being presented at the 2026 Symposium on VLSI Technology and Circuits in Honolulu, the chip industry’s main venue for new silicon. The chip is a transceiver, the part of a phone that sends and receives radio signals, built for the D-band. That is the stretch of spectrum around 140 gigahertz, roughly 50 times higher in frequency than the bands most phones use today.
Why go that high? Because spectrum is like land. The low frequencies are downtown: convenient, but crowded with 4G, 5G, WiFi and everything else. The D-band is open country. There is enough room there to move data at speeds fibre can manage today. The Tokyo chip demonstrated 144 billion bits per second at close range, about 64 billion at three metres, and a usable link out to 50 metres.
It reaches those speeds with two tricks worth understanding.
The first is dual polarization. Radio waves wiggle in a direction, up and down or side to side, the way you can shake a rope vertically or horizontally. Waves polarized at right angles to each other do not interfere, so one antenna setup can carry two separate data streams at once over the same frequencies. It is two lanes painted onto one road. The chip packs four vertically polarized and four horizontally polarized transceiver elements onto a die measuring three by four millimetres, smaller than a pinky fingernail. Two of these chips work together in a module with the antennas built into the package.
The second trick is the manufacturing choice, and it is the genuinely surprising part. The chip is made in 65 nanometre CMOS, the standard silicon process that powered gadgets around 2006. Today’s flagship phone processors use processes ten generations newer. Conventional wisdom said frequencies this high demanded exotic, expensive semiconductors such as indium phosphide, or at least cutting-edge silicon. Building a working D-band radio in a cheap, mature, widely available process challenges that assumption directly. Old silicon means low cost, many factories able to make it, and no competition with AI chips for scarce leading-edge capacity.
That is what this actually changes. The industry has quietly assumed that the first 6G radios at these frequencies would live in base stations and fixed links, where size, power and cost matter less, with handsets following years later once someone solved the miniaturization problem. This chip is built explicitly for user equipment, the industry term for the device in your hand. If the phone side of a D-band link is feasible in commodity silicon, the timeline for 6G services that actually reach consumers gets more believable, and so do nearer-term uses: wireless virtual reality headsets without cables, instant file transfers between devices, and short-range links that replace docking connectors entirely.
The honest caveats matter. The headline 144 gigabits per second was measured at 0.3 metres, a distance better suited to a kiosk download than a phone call. D-band signals fade quickly in air and barely penetrate walls or bodies, so these links will serve dense hotspots and room-scale connections, not wide-area coverage. Power consumption, beam steering as a user walks around, and the cost of the antenna packaging all remain open engineering problems. And 6G standards are not finished, so the exact frequencies could still shift.
Even so, the direction is hard to miss. The pattern repeats across wireless history: a frequency band looks impractical, someone builds it in cheap silicon, and a decade later it is ordinary. It happened with WiFi at 5 gigahertz and with 5G at millimetre wave. This week in Honolulu, it started happening to the D-band.
Sources
- Conference paper (full citation): Yudai Yamazaki, Ryuji Kinugawa, Takumi Kojima, Takaya Uchino, Anyi Tian, Chenxin Liu, Sunghwan Park, Kotaro Ito, Sena Kato, Chun Wang, Minzhe Tang, Takashi Tomura, Yuncheng Zhang, Hiroyuki Sakai, Kazuaki Kunihiro, and Kenichi Okada, “A 144Gbps D-Band Dual-Polarized MIMO High-Density Phased-Array Transceiver in 65nm CMOS for 6G UE,” IEEE Symposium on VLSI Technology and Circuits, Honolulu, HI, June 2026. Okada Lab publication page: https://www.ssc.p.isct.ac.jp/en/publication/ arxiv
- VLSI 2026 official technical tipsheet (the public pre-conference source all the chip specs in the article are drawn from): https://www.vlsisymposium.org/wp-content/uploads/2026/04/2026-VLSI-Technical-Tipsheet-REVISED-FINAL-4.25.26-1-1.pdf
- Okada Laboratory acceptance announcement, May 12, 2026, confirming the paper among six accepted at VLSI 2026: https://www.ssc.p.isct.ac.jp/en/news/2026-05-12-04-25-00/
- Conference details (dates, venue): 2026 IEEE/JSAP Symposium on VLSI Technology & Circuits: https://www.vlsisymposium.org/
- The team’s direct predecessor work: a phased-array transceiver in 65nm CMOS from the same group (with Panasonic Industry, Panasonic System Network R&D Lab, and Shinko Electric Industries), an eight-element module reaching 56 Gbps at 0.3 m and 154 GHz, published in Nature Electronics (2025): https://www.nature.com/articles/s41928-025-01426-x arxiv
- Institute of Science Tokyo institutional release on that 150 GHz module (VLSI 2025): https://educ.titech.ac.jp/ee/eng/news/2025_07/067967.html
Ray Jackson holds a BSc in Electrical Engineering from the University of Manitoba and a PhD in Physics from Carleton University. His reporting interests include Current and Future Technologies, Engineering and Artificial Intelligence.