Fujitsu targets 1024 qubit quantum computer for 2026
Aggressive roadmap, big chill
Fujitsu has set a “very aggressive target to deliver a 1024 Qubit computer next year, as it unveiled a 256 qubit superconducting device in Japan today.
The 256 qubit platform was developed in collaboration with Japanese research institute RIKEN. The organizations said it overcomes “key technical challenges, including appropriate cooling within the dilution refrigerator which is achieved through the incorporation of high-density implementation and cutting-edge thermal design”
The core design arranges 4-qubit unit cells in a 3D configuration. The organizations have previously built a 64 qubit device. The new device uses the same unit cell design, “effectively demonstrating the scalability of this architectural approach.”
It also means the 256 qubit machine can operate within the same cooling unit as its predecessor. The refrigerator operates at temperatures as low as approximately 20 millikelvin, or minus 273.1 degrees centigrade.
Techniques for improving qubit variability include developing a technology to “individually fine tune the resistance of Josephson Junctions using laser irradiation”. These are key components in superconducting systems. The partners also improved coefficient of variation in device resistance from 4.1 to 0.6 percent, while processing time has been reduced to a third of that for the 64 bit system.
A successful computing platform is about more than hardware, and Fujitsu is working on “seamless collaboration between quantum and classical computers, enabling the efficient execution of hybrid quantum-classical algorithms.”
Head of Quantum Laboratory Shintaro Sato added that one key focus is allowing multiple users to use the same quantum chip, while it was also developing technology to allow “high speed hybrid jobs” combining quantum and classical computing.
Fujitsu CTO Vivek Mahajan said the organizations plan to have a 1000 qubit device up and running in a new building on the Fujitsu Technology Park next year and is working on getting past this level.
He described this as a “very aggressive target” spanning not just the chip but the control electronics and the “very big scale” refrigerator technology.
The rapid scaling up of qubits sounds impressive. But it’s also worth remembering that one key quantum breakthrough – cracking asymmetric encryption – would require something in the region of 10,000 qubits. Or millions depending on who you listen to. And such systems are years away. Again, depending who you listen to.
John Burgoyne, Head of Product and Marketing at Oxford Instruments NanoScience, said Fujitsu’s announcement “Shows that superconducting qubits are a scalable architecture, while highlighting the challenges of managing heat loads and the cold “real estate” not just for the quantum processor itself, but particularly for the control wiring.”
Scalable cryogenic platforms would be necessary as heat loads and payload density outstrip conventional refrigerators, he said. “We see similar customers pursuing high-count superconducting qubit goals, for which we have just announced the first installations of our ProteoxQX platform with its unique scalability and interconnectivity."