# CUDA-Q and NVQLink by NVIDIA

Canonical page: https://technooptimist.io/quantum/cuda-q-nvqlink
Frontier: [Quantum](https://technooptimist.io/quantum.md) · Field: Fridges, control & fabs · Company: [NVIDIA](https://technooptimist.io/companies/nvidia.md)
Stage: Deploy (4 of 5: Research → Proto → Pilot → Deploy → Scaling)
Updated: last update 2026-09-15, facts checked 2026-10-11
Goal: Make GPUs the classical brain of every quantum computer: decoding errors, calibrating qubits and running the classical half of hybrid programs fast enough that large error-corrected machines become workable.
Status: NVQLink adopted by 9 US labs and 15 centres abroad; CUDA-Q Logical (Sep 2026) targets fault tolerance.

## Milestones

| Status | Milestone | Date | Slip |
| --- | --- | --- | --- |
| Done | NVQLink introduced with 17 QPU and 5 controller builders | 28 Oct 2025 |   |
| Done | Real-time qLDPC decoding on Quantinuum Helios over NVQLink | 17 Nov 2025 |   |
| Done | Supercomputing centres in Asia and Europe adopt NVQLink | 17 Nov 2025 |   |
| Done | CUDA-Q Logical: compiler for error-corrected machines | 15 Sep 2026 |   |

## Most important updates

- **15 Sep 2026**: NVIDIA releases CUDA-Q Logical to compile and plan error-corrected quantum computers. At IEEE Quantum Week in Toronto, NVIDIA added CUDA-Q Logical to its open quantum platform. ([our story](https://technooptimist.io/stories/nvidia-releases-cuda-q-logical-to-compile-and-plan-error-corrected-043320ff.md) · [original source](https://quantumcomputingreport.com/nvidia-unveils-cuda-q-logical-to-accelerate-fault-tolerant-system-orchestration-across-hardware-modalities/))
- **17 Nov 2025**: Supercomputing centres in Asia and Europe adopt NVQLink; Quantinuum Helios decodes errors in real time over it. At SC25, NVIDIA said more than a dozen supercomputing centres in Asia, Europe and the Middle East are adopting NVQLink to attach quantum processors to its Grace Blackwell systems. ([our story](https://technooptimist.io/stories/supercomputing-centres-in-asia-and-europe-adopt-nvqlink-quantinuum-4747dea3.md) · [original source](https://investor.nvidia.com/news/press-release-details/2025/Worlds-Leading-Scientific-Supercomputing-Centers-Adopt-NVIDIA-NVQLink-to-Integrate-Grace-Blackwell-Platform-With-Quantum-Processors/))
- **28 Oct 2025**: NVIDIA introduces NVQLink to connect quantum processors to GPU supercomputers, with 17 QPU builders and 9 US labs. At GTC Washington, NVIDIA introduced NVQLink, an open architecture that links quantum processors to GPU supercomputers. ([our story](https://technooptimist.io/stories/nvidia-introduces-nvqlink-to-connect-quantum-processors-to-gpu-28986ee9.md) · [original source](https://investor.nvidia.com/news/press-release-details/2025/NVIDIA-Introduces-NVQLink--Connecting-Quantum-and-GPU-Computing-for-17-Quantum-Builders-and-Nine-Scientific-Labs/default.aspx))

## Current obstacles

- **Far from useful scale**: Using CUDA-Q Logical, Sandia and Quantinuum put Helios-class hardware about five orders of magnitude short of utility targets.
- **Tighter time budgets ahead**: Helios's 2 ms budget is generous; faster qubit types must be decoded within microseconds, so links and decoders must keep speeding up.

## Physics limits

- **Error correction multiplies qubits**: Each reliable logical qubit needs many physical ones: one 2026 design needs 150,000 physical qubits for 1,000 logical, and every extra qubit adds classical decoding work.
- **The feedback loop has a speed floor**: Qubits lose their state in microseconds to seconds depending on type; signals, electronics and decoding all take time, so correction must fit inside that window.

## How it works

### Software: One program, three kinds of chip

CUDA-Q extends CUDA so a single Python or C++ program sends quantum steps to the qubits and classical steps to GPUs and CPUs.

### Link: Microseconds to the qubits

NVQLink wires a quantum computer's control electronics straight to GPUs: under 4 µs and 400 Gb/s, fast enough for feedback mid-computation.

### Decoding: GPUs catch errors live

Error correction yields streams of checks; GPUs must find which qubits failed in time. Quantinuum's Helios did it in 67 µs against a 2 ms budget.

### Logical layer: Compiling for fault tolerance

CUDA-Q Logical (2026) compiles programs for error-corrected qubits down to pulse schedules and estimates the full hardware they need.

## Spec sheet

| Spec | CUDA-Q and NVQLink | Provenance |
| --- | --- | --- |
| NVQLink GPU-QPU latency | < 4 µs | reported |
| NVQLink GPU-QPU throughput | 400 Gb/s | reported |
| NVQLink AI compute | 40 PFLOPS (FP4) | reported |
| Public QPUs CUDA-Q supports | 75% | reported |
| Helios decoder reaction time | 67 µs (2 ms required) | reported |
| Launch partners | 17 QPU + 5 controller builders | reported |
| CUDA-Q languages | Python, C++ | reported |

Provenance: reported = stated by the company; estimated = our estimate; sample = a sample figure.

## About NVIDIA

NVIDIA designs the GPUs, networking and software behind most AI training and inference. Its CUDA software made GPUs the default for AI. In October 2025 it became the first company worth $5 trillion.

- Vera Rubin in production in 2026; cloud partners get it from the second half of the year.
- Licensed Groq's inference technology and hired its leaders for about $20B (Dec 2025).
- Plans to invest up to $100B in OpenAI, tied to at least 10 GW of NVIDIA systems (Sep 2025).

- Founded: 1993
- Headquarters: US
- Status: public
- Listing: NVDA (NASDAQ)
- Valuation: $5.4T (market cap, 2026-09)
- Staff: ~36,000
- Website: https://www.nvidia.com
- People: Jensen Huang (Co-founder, President and CEO); Colette Kress (CFO); Bill Dally (Chief Scientist)
- Partners, customers, investors: TSMC (supplier); SK hynix (supplier); OpenAI (customer); Intel (partner); Groq (partner); CoreWeave (customer)

Company page: https://technooptimist.io/companies/nvidia.md

## Related programs

More from NVIDIA:

- [Vera Rubin](https://technooptimist.io/computing/vera-rubin.md) (Deploy)
- [Blackwell](https://technooptimist.io/computing/blackwell.md) (Scaling)
- [Isaac GR00T](https://technooptimist.io/robotics/gr00t.md) (Pilot)
- [Nemotron 3](https://technooptimist.io/ai/nemotron-3.md) (Deploy)
- [Cosmos 3](https://technooptimist.io/ai/cosmos-3.md) (Deploy)
- [Groq 3 LPX](https://technooptimist.io/computing/groq-3-lpx.md) (Proto)
- [Rubin CPX](https://technooptimist.io/computing/rubin-cpx.md) (Proto)
- [Spectrum-X and Quantum-X Photonics](https://technooptimist.io/computing/spectrum-x-quantum-x-photonics.md) (Deploy)
- [Omniverse (Factory Digital Twins)](https://technooptimist.io/manufacturing/omniverse-factory.md) (Deploy)
- [NVIDIA DRIVE](https://technooptimist.io/vehicles/nvidia-drive.md) (Deploy)
- [Earth-2](https://technooptimist.io/ai/earth-2.md) (Deploy)

Everything in Quantum: https://technooptimist.io/quantum.md

---

The Techno Optimist tracks every frontier of technology, program by program: https://technooptimist.io/ (a Markdown version of any page: add .md to its URL; site map for agents: https://technooptimist.io/llms.txt).
Data API (JSON, realtime stream): https://technooptimist.io/developers. People read 5 pages free, then Reader $10/month; these Markdown pages are open to agents.