US basedUS hosted
Desks
$2 / month

Comment and reply across TML.

Start or manage membership

Most pulsed

Advertisement Sinscents Bad Intentions. Great Candles.

Latest from other desks

tml.the modern leap
Tech / Future Tech

NVIDIA, IBM and Fermilab Build Quantum Codesign Platform

NVIDIA announced a quantum-classical codesign platform with IBM and Fermi National Accelerator Laboratory on November 16, 2026, at the SC26 conference in Chicago. NVIDIA's DGX B200 systems will sit alongside IBM quantum computers, including machines running the Nighthawk processor, and NVIDIA released a new error-correction decoding algorithm, TCBP-JOBS, that it says runs up to 13,000 times faster than existing leading methods — cutting one research computation from 12 weeks to 45 minutes.

Nico ReedNico Reed3 min read
An IBM Quantum System One at Fraunhofer in Germany (archive). IBM's quantum computers, including Nighthawk-powered systems, will be colocated with NVIDIA DGX B200 systems under the codesign platform announced November 16. Photo: IBM Research, CC BY 2.0, via Wikimedia Commons.
Summary

Quantum computers fail constantly, and correcting those failures is the field's bottleneck. NVIDIA's move is to treat that bottleneck as a classical computing problem — and to sell the classical computers that solve it.

Tap a passage. React to the part that matters.

Quantum computing's hardest problem isn't building the quantum computer. It's fixing its mistakes fast enough to matter, and on November 16, 2026, NVIDIA, IBM and the U.S. Department of Energy's Fermi National Accelerator Laboratory announced a joint platform aimed squarely at that problem, alongside a decoding algorithm NVIDIA says is up to 13,000 times faster than the leading existing methods.

The hardware pairing comes first. NVIDIA and IBM will colocate NVIDIA DGX B200 systems with IBM quantum computers, including machines powered by IBM Quantum Nighthawk, the most powerful quantum processor IBM has built. The codesign platform built with Fermilab — the national laboratory in Illinois — will support quantum error-correction work and applications research for future modular quantum systems, the kind assembled from multiple connected processors rather than one monolithic machine.

The headline number belongs to software. NVIDIA's new TCBP-JOBS decoding algorithm, released Monday in the CUDA-Q QEC 0.9.0 library, accelerates the computationally intensive parts of quantum error-correction research by up to 13,000 times while keeping high accuracy, the company says. In practical terms, NVIDIA says a computation that took 12 weeks of research time now takes 45 minutes. Error correction is where quantum designs live or die: quantum bits are fragile, they accumulate errors constantly, and a machine that can't correct them faster than they pile up can't run a useful calculation.

The less flashy announcement may matter as much to the labs that run these machines. NVIDIA is expanding Slurm — the workload scheduler that already runs much of the world's supercomputing — to schedule quantum processors inside supercomputing workflows, and IBM is contributing integration through its Quantum Resource Management Interface. That's the plumbing that lets a researcher submit one job that runs partly on classical machines and partly on a quantum processor, instead of babysitting two separate systems.

The setting was deliberate. The announcements came at SC26 in Chicago, the supercomputing conference where the classical machines quantum computers are supposed to one day outperform are designed, sold and benchmarked. Fermilab's presence ties the platform to the national laboratory system, which brings both research programs and a pipeline of scientific problems — physics simulations among them — that classical machines handle slowly and quantum machines are being built to attack.

Read the structure of the deal and NVIDIA's position in quantum computing gets clearer. NVIDIA doesn't build quantum processors. IBM and Fermilab do the quantum research; NVIDIA supplies the accelerated classical computing, the software platform and now the error-correction tooling that every quantum program needs regardless of whose quantum hardware wins. It's the same seat NVIDIA occupies in artificial intelligence — not the model, but the machinery the models can't run without — extended to the next computing platform before that platform has a winner.

The claims to hold at arm's length are the usual ones. The 13,000-times figure is NVIDIA's own measurement against existing leading algorithms, announced in its own release, and error-correction decoding speed is one component of a working fault-tolerant machine, not the machine itself. No timeline was given for a practical, fault-tolerant quantum computer. What Monday establishes is narrower and real: the three organizations are now building the classical side of quantum computing together, on the record, with the software already in researchers' hands.

The conversation

A little more perspective.

Reading is free. Verified members can comment and reply for $2 a month.

The conversation.

Loading comments…

Keep your curiosity going.

Some links here are affiliate links. If you buy through one, we might make a commission — it never changes what you pay. As an Amazon Associate I earn from qualifying purchases.