Classiq launches a Fault Tolerance Engine to plan quantum programs for error-corrected hardware

Started by LivMorgan_Face, Oct 01, 2026, 07:11 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Topic: Classiq launches a Fault Tolerance Engine to plan quantum programs for error-corrected hardware   Views(Read 50 times)

LivMorgan_Face

Quantum software company Classiq has announced what it calls a Fault Tolerance Engine. The idea is to take a quantum program written at a high level and work out exactly what it would need to run on future error corrected hardware. That sounds dry, but it addresses one of the biggest practical questions in the field. How big a machine do you need to run something useful?

The engine converts programs from Classiq's modelling environment into concrete fault tolerant execution plans. It decides how logical qubits are laid out, how interactions are routed and how operations are scheduled. Then it calculates the physical cost, including the number of physical qubits required, error correction cycles, code distance, runtime, accumulated error rates and the number of T-gates and magic states needed. In other words, it turns a neat diagram into a detailed construction plan. Engineers will appreciate that

Classiq says the result is an architecture aware plan rather than a rough estimate based on formulas. That matters because the gap between a neat algorithm on paper and what has to run on a machine is enormous once error correction is added. Chief executive Nir Minerbi said fault tolerance changes the question quantum software has to answer, because you need to understand what a program becomes when every logical operation has to be protected. That is a refreshingly honest framing for a software company

No hardware partners, prices or benchmark numbers were included in the announcement, so it is hard to judge how accurate the estimates will be in practice. But tools like this could help companies decide which problems are worth pursuing and roughly when the hardware will be ready for them. It fits with the theme of the Nature Electronics comment we discussed, about moving from engineered advantage to practical utility

Is this the kind of tool that will bring practical quantum computing closer, or mainly useful for planning and investor presentations? And does anyone here work with quantum software at all? I would be curious to hear from anyone who has tried Classiq's platform


BretHart_Core

Useful for investors as well as engineers, to be fair. Being able to show exactly when an application becomes possible on a hardware roadmap makes for a much stronger pitch. Boards like a timeline with numbers on it

Diane

Resource estimation is underrated. Most people quote qubit counts without understanding that one logical qubit might need hundreds or thousands of physical ones. A tool that shows the real cost of an algorithm would bring some realism to the hype. That alone makes it worthwhile

Tel75

I would love to see this run on the elliptic curve and Shor style algorithms people worry about for cryptography. A clear resource estimate would show how far off Q-Day is

That would cut through a lot of speculation. It would also show which encryption needs replacing first
Coffee first. Questions later.

ReacherBadger

No benchmarks and no hardware partners makes it hard to evaluate. Software announcements without numbers are a bit of a leap of faith

Hopefully a technical paper follows. Classiq has published solid work before, so I will give them the benefit of the doubt
Blue is the colour.

Highland Fatima

Classiq has always focused on making quantum programming higher level, so this is a logical next step. Writing for fault tolerant machines by hand would be a nightmare. Abstraction is the only way to make it manageable
Measure twice, post once

Eastern Nathan

Magic states are the hidden bottleneck in most fault tolerant designs. They are expensive to produce and many useful algorithms need huge numbers of them. Knowing exactly how many a program needs helps decide whether it is practical at all. Some famous algorithms look far less attractive once you count them