Google offers $150,000 in a Kaggle contest to shrink quantum circuits with AI

[Andrew4]

Google Quantum AI wants help making quantum algorithms cheaper, and it is paying for it. The Quantum Insider's Matt Swayne reports that Google has opened a Kaggle competition with a $150,000 prize pool. Entrants use AI to find circuits that need fewer gates and fewer qubits

There are three tasks. One is a quantum fast inverse square root aimed at plasma simulations for inertial confinement fusion. Another is a Reed-Solomon decoder for a technique called Decoded Quantum Interferometry, and the third is a cheaper simulation of the FeMoCo cluster at the heart of the nitrogenase enzyme

Entries are checked with kickmix, an open source tool that is part of Google's Qualtran library. It fuzz tests circuits at around 60 billion gates per second on a single CPU core, so mistakes are caught quickly. Google says its own experiments with Gemini have already beaten the published circuits on all three tasks

In a neat touch, Google has published a hash of its own circuits and will reveal them after the contest closes, so nobody can accuse it of moving the goalposts. First prize is $50,000, second is $25,000 and third is $10,000. There are also $10,000 prizes for a milestone on 20 November, for the longest time at the top of the leaderboard to 4 January 2027, and for the best entry on each task

Opening this up to the Kaggle crowd is a smart move, because circuit optimisation is a puzzle that suits clever outsiders. It also fits the wider trend of AI tools cutting resource estimates, from Claude designing circuits to Xanadu's QROM savings. FeMoCo is the classic chemistry target, so any improvement there will get attention

Would you enter a contest like this? And can AI keep shrinking the hardware needed for useful quantum algorithms?

Google Launches $150,000 AI Competition to Reduce Quantum Computing Demandsthequantuminsider.com

[Starforge Daniel]

I am not sure Kaggle regulars have the background for this. Quantum circuit design has a steep learning curve

Google might mostly attract existing researchers wearing a Kaggle hat. That would still be a good outcome