IonQ's secp256k1 estimate spends 13 physical qubits on every logical one

Divide 19,397 physical qubits by 1,457 logical ones and you get 13.3 physical qubits per logical qubit ^[16]. That ratio is where the news is. A 13-to-1 overhead is the signature of a high-rate code doing the heavy lifting, which is consistent with the optimised Walking Cat architecture the estimate targets: trapped-ion hardware, quantum low-density parity-check codes ^[6]. IonQ says every operation was mapped down to the error-correction primitives the hardware actually uses, which is what makes a success probability computable rather than assumed ^[8]. Good engineering, and the harder claim of the two. But the write-up on dev.to reports counts and runtime without the physical error rate, the code distance, or the physical gate time ^[19]. Those are the numbers that decide whether 13.3 to 1 hits a usable logical error rate. The headline 20,000 is 19,397 rounded up ^[2][5].

The runtime has the same shape. Twenty-six days is 2,246,400 seconds, and 39 million logical Toffoli gates in that window is about 17 per second, or 57.6 ms each ^[17]. Every error-correction cycle underneath one logical Toffoli has to fit inside that 57.6 ms. With no physical clock rate in the write-up, there is no way to tell whether that budget is comfortable or whether it is the constraint that set the 26 days ^[19].

The date attached to all this comes from a product roadmap ^[9] that targets systems of this scale by 2028; the resource estimate itself sets no date. Anyone anchoring a migration deadline to it is anchoring it to one company's manufacturing schedule, and that company also builds post-quantum cryptography and quantum key distribution hardware ^[20]. The pull-in is still worth taking seriously: the write-up says earlier expectations sat in the 2030s ^[11] and attributes the change to compounding progress across algorithms, compilers and hardware design ^[12]. If that is the mechanism, the crossing date is not any single vendor's to set, which cuts both ways on the 2028 number.

For planning, the estimate answers one question, which is what one key on one architecture costs. It does not say what the second key costs, and secp256k1 is the 256-bit curve behind Bitcoin and other chains ^[4], so whether 26 days per key is alarming depends on how many keys an attacker needs and how long yours stay worth forging. The write-up notes that no digital assets were harmed in the research ^[13], a sentence you only write when readers might reasonably wonder. It also reports that the study stresses rigorous stress-testing of the algorithms meant to replace current standards ^[14]. For anyone actually running a migration, that line carries more work than the qubit count does.