| Mobile Web

Search results for qubit

AI & Enterprise

Nvidia tests technique to cut physical qubits needed for error correction by 90 percent

Nvidia carried out a 104-qubit quantum computing simulation on Germany’s Jupiter, Europe’s first exascale supercomputer, and tested a technique that could cut the physical-qubit resources required for error correction by 90 percent. The study focused on efficiently preparing magic states through Magic State Cultivation under noise conditions close to real quantum hardware. Nvidia used GH200-based systems and scaled workloads to 8,192 GPUs, with cuQuantum and CUDA-Q software platforms.

Crypto

Claude finds quantum computation method in 53 hours, cuts operations to one-seventeenth

Anthropic\'s AI model Claude autonomously designed eight types of quantum computing procedures and reduced required qubits and operation counts in seven of them compared with existing optimal designs. Some designs improved resource efficiency by more than 10 times. The work focused on point addition used repeatedly in elliptic-curve cryptography, which underpins transaction signatures for Bitcoin and Ethereum. Anthropic said it did not attack cryptocurrencies or estimate the total quantum computer size needed to break them.

Crypto

Google quantum chip Willow cuts computation errors 40-fold, but bitcoin security threat still distant

Google succeeds in cutting errors in “magic state,” a key resource for quantum computing, to about one-fortieth of the previous level. Its Quantum AI team reports achieving 99.99% logical accuracy using its Willow processor, but the performance remains far from levels needed to crack bitcoin or RSA-2048 encryption. Researchers cite a wide gap in required accuracy and note that Willow has 105 physical qubits, far below estimates for such attacks.