Quantinuum称Helios错误率降低200倍
Quantinuum首席执行官Rajeeb Hazra表示,公司的100量子比特Helios系统达到每百万次操作出现五个错误的逻辑保真度,并创下公司所称的纠错纪录。他将这一结果与两年前公司力求实现“1,000次操作出现一个错误”的表现作对比,称改善幅度约为200倍。Hazra还表示,AI可以编程量子计算机,以获取用于模拟、推理和预测的数据。据Hazra介绍,已有两家云服务商与Quantinuum达成合作;此次由Quantinuum赞助的采访中讨论了相关系统。
Quantum computing is shedding its reputation as speculative as gains in quantum error correction turn fragile machines into reliable ones. For the supercomputing community, which now measures data centers in gigawatts of power, the more pressing question is how quantum fits beside the processors already on the floor.
For years, the field faced a trade-off between speed and reliability, because correcting errors inside a quantum computer slowed it down. That trade-off is fading fast, according to Rajeeb Hazra (pictured), chief executive officer of Quantinuum Inc., which listed on the Nasdaq in June.
“We are now at a stage in 2026 where we have the ability to have tremendous amounts of computation at a very high reliability rate,” Hazra said. “Just two weeks ago, we announced we were doing up to a million operations with five errors. To put it in contrast, just two years ago, we were trying to do 1,000 operations with one error. That’s how fast it is.”
Hazra spoke with theCUBE Research’s John Furrier and Gemma Allen at SC26, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed advances in quantum error correction and how AI and quantum computing work in tandem. (* Disclosure below.)