Nvidia’s new quantum layer cut Fermilab’s design work from five months to three weeks - The Next Web

CUDA-Q Logical is open source and already in use at Fermilab, Sandia, IonQ and five other organisations. Nvidia has added an orchestration layer called CUDA-Q Logical to its open-source quantum platform, aimed at the part of fault-tolerant quantum computing that nobody enjoys.Designing an application for a machine that does not exist yet means guessing at algorithms, error-correction codes, and hardware together, and changing any one of them rewrites the resource requirements for the other two.Fermilab put a number on what that costs. Anna Grassellino, its chief technology officer, said a team explored combinations of those resources in three weeks where the same work would normally have taken about five months.Seven organisations are named alongside it: Sandia National Laboratories, Infleqtion, IonQ, Iceberg Quantum, IQM Quantum Computers, QCDesign and Quantum Motion.The most consequential claim comes from Iceberg Quantum working with Diraq, which puts 1,000 logical qubits within reach of 150,000 physical ones.Nvidia describes that as roughly ten times fewer than previous estimates, without saying whose estimates. Overhead is the central problem in fault-tolerant computing, since a logical qubit is assembled from many error-prone physical ones, and any credible reduction moves the date at which useful machines become buildable.None of these logical qubits exist. These are resource estimates produced in simulation.Sandia has also built a benchmark, QUOPS, intended to measure progress toward utility-scale quantum computing across platforms. A cross-vendor yardstick is genuinely useful in a field where every company reports a different number.Worth noting all the same: the benchmark arrives inside the announcement of a commercial platform sold to the companies being measured.The same release names Phasecraft, the Bristol company whose 10x claim for its THRIFT algorithm reached Nature Communications last year, as a user of Nvidia’s cuQuantum toolkit.Its separate announcement of a molecular database, published the same day, turns out to be one proof point inside a larger platform launch rather than a standalone result.Timothy Costa, an Nvidia vice-president, framed the release around an era of logical qubits and researchers needing an open, customisable platform.The word doing the work there is open. CUDA-Q is open source, which lowers the barrier to adoption and makes it correspondingly harder to leave once an organisation’s codesign work is expressed in it.The pattern holds across Nvidia’s quantum activity. The company does not build quantum computers and shows no sign of wanting to. It builds the classical layer everything else has to pass through, then invests around the edges.In May, its venture arm backed Alice & Bob, the Paris company whose cat-qubit approach aims at fault tolerance with thousands of physical qubits rather than millions. The strategy is deliberately hardware-agnostic, because whichever qubit wins, the orchestration still runs on GPUs.The market reads it the same way. In April an Nvidia quantum announcement moved the whole sector, sending Xanadu’s stock up nearly fivefold in six sessions and making its founder a billionaire, on no investment from Nvidia at all.Quantum computing is meanwhile arriving in ordinary infrastructure. Oracle is installing a Quantinuum machine inside its data centre rather than brokering remote access, and Britain’s second commercial machine, a 32-qubit Rigetti, came online in 2024. The hardware is real, small, and error-prone, which is precisely why the codesign problem matters.Every figure here comes from Nvidia and the partners it names, and none is attached to a peer-reviewed result in the announcement.The Fermilab comparison is an internal estimate of work not done the slow way. The 150,000-qubit figure has no published baseline to measure the tenfold improvement against. Both may well hold up, and neither has been shown to yet. I am the Editor in Chief for TNW, covering technology not as a parade of launches and valuations, but as a system of influence, persuasion, (show all) I am the Editor in Chief for TNW, covering technology not as a parade of launches and valuations, but as a system of influence, persuasion, and change. I write about startups, venture capital, digital policy, and Europe ecosystem, with an eye on the larger story beneath them: who gets to build the future, who profits from it, and how Europe is learning to speak in a louder voice of its own. Before moving into senior editorial leadership, I've built my career for over +10 years across journalism, storytelling, content strategy, SEO, and digital publishing, with experience in SaaS, hospitality, art, and culture. Get the most important tech news in your inbox each week.The heart of techA Tekpon CompanyCopyright © 2006—2026, Cogneve, INC. Made with <3 in Amsterdam.