ETH Zurich neutral atom qubit breakthrough could cut Shor's algorithm requirement to 10,000 qubits - the real question

[JohnyBlue]

ETH Zurich published in Nature this week on a new swap gate for neutral atom qubits using geometric phase rather than the standard laser pulse approach. The key claim from postdoc Yann Hendrick Kiefer is that this could enable breaking RSA with Shor's algorithm using around 10,000 qubits rather than the millions previously estimated.

Neutral atom qubits trapped in optical lattices have been a strong dark horse against superconducting qubits, and stability during operation has always been the weak link. The geometric phase approach sidesteps a lot of the laser timing sensitivity that was driving error rates.

If the 10,000 qubit number for Shor holds up under independent review, the harvest now decrypt later threat just got a lot less theoretical. Combined with the Oratomic and Google papers from March, the cryptanalysis side of quantum is moving faster than the post quantum migration side

Stable qubit breakthrough pushes quantum supercomputers within reachinterestingengineering.com

[HeartbreakKidCurtis18]

Ten thousand qubits is still a long way off but it used to be twenty million, the goalposts keep moving

[DiogoCardoso]

Geometric phase gates are mathematically beautiful, glad to see them getting practical implementations

[MrRicardo]

If this holds up under peer review the entire PQC migration urgency conversation needs to be reframed

[Rogue Di]

Per Lukin at Harvard fault tolerant quantum is at least five years sooner than people thought, this slots right in

[QuantumDay]

Neutral atoms versus superconducting is going to be a real horse race over the next few years

[Shane]

Honestly Kiefer's quote about ten thousand qubits feels overconfident to me, the resource estimate work is still preliminary

[Amber Tiger]

Fair but even an order of magnitude improvement on the previous resource estimates is significant

[RomanReigns02]

Aaronson talked about Bitcoin being vulnerable to 25-30k qubits a while back, this is in the same ballpark

[DiamondDallas86]

Spot on. Thanks for that

[StringTheory97]

Nature paper or it didn't happen. Oh wait, it is in Nature. Carry on

[Slay]

The actual gate fidelity numbers in the supplementary are what to look at, the marketing headline is fine but the engineering is in the appendix