Designed for developers
Write your workloads in Python and run them without having to manage runtime environments, computing resources, or infrastructure.
The service handles workflow execution, artifacts, logs, and task lifecycle management.
Unified Computing Service
A unified cloud environment for your hybrid workloads (CPU, GPU, QPU).
Unified across architectures
Build workflows that combine multiple computing architectures within a single execution. Prepare your data on a CPU, accelerate specific tasks on a GPU, run a circuit on a QPU, and continue processing without switching environments.
Your workflows become composable computational graphs rather than a collection of separate applications for each infrastructure.
Open by design
The service is built on Qall, an open-source framework designed to remain extensible. Use the scientific and quantum SDKs you already know and run your workloads across different backends and execution environments.
The stack remains open: developers can contribute, integrate new environments, and extend execution capabilities.
Key features
Managed execution
Deploy your workflows without managing servers, containers, or runtime environments. The service handles task launch, execution, and lifecycle management. Define your requirements, and the service finds and allocates the corresponding resources.
Seamless QPU access
Run your quantum and hybrid workloads on quantum processors available through the Scaleway cloud. The service handles backend access and job execution, while allowing you to continue using your preferred quantum SDKs (Qiskit, CUDA-Q, PennyLane, etc.).
Composable workflows
Build workflows made up of multiple tasks and combine different types of computation within the same execution graph. You can also reuse existing workflows as building blocks for new computations.
Artifact storage
Data produced by your tasks is stored and automatically shared between the different steps of a workflow. Artifacts also allow you to retain execution results and reuse steps that do not need to be recomputed.
Observability
Track your workflows and tasks from launch to completion. Access logs, execution status, duration, and results to understand what is happening at every step.
Packaging and registry
Your code and dependencies are packaged as reusable workloads. The Registry stores, versions, and distributes your workflows so they can be deployed and reused without rebuilding your environment each time.
Unified Computing Service is compatible with our CPU, GPU, and our European quantum computing offerings.
Typical use cases
Run hybrid algorithms
Combine classical and quantum computing in a single workflow to solve problems such as VQE, QAOA, or HHL. Run each step on the most appropriate resource -CPU, GPU, or QPU- and automatically move data between the different stages of the computation.
Build composable workflows
Break your computations down into reusable tasks and compose them into complex workflows. Call an existing workflow as a computational building block and progressively build applications from components you have already developed.
Version and reproduce your computations
Keep the code, dependencies, and results associated with each step of your workflows. Reproduce an experiment, compare different versions, or recompute only the steps that have changed, while maintaining a complete trace of your computations.
At the service's core: Qall
Qall: Composable execution for hybrid computing
Qall is an open-source and infrastructure-agnostic SDK for heterogeneous computing and computational workflows. It provides the execution and Registry layer at the heart of the service, allowing you to define, compose, deploy, and execute workloads across different computing resources.
Designed for heterogeneous computing, Qall allows classical and quantum resources to be combined while remaining independent of providers and execution environments. Its open architecture allows anyone to contribute to the stack or provide their own execution environment.
Qall does not aim to replace existing scientific or quantum SDKs. It sits above them, providing an execution and orchestration layer that allows them to be used within broader workflows. Qiskit, PennyLane, PyTorch, CUDA-Q, and other tools therefore remain at the core of scientific and algorithmic development.
Frequently asked questions
Who is Unified Computing Service for?
Unified Computing Service is a developer-first service designed for developers and technical teams building computational workflows.
Its goal is to let developers focus on their code and workloads without having to manage the infrastructure required to run them across different computing architectures.
What is the difference between Qall and Scaleway Unified Computing Service?
Qall is the open-source framework used to define and execute workflows. Unified Computing Service is Scaleway's managed service providing the infrastructure needed to run them.
In short: Qall is the open software layer; Unified Computing Service is the managed service operating it.
Is Qall a new quantum SDK?
No. Qall does not replace existing quantum or scientific SDKs.
Qiskit, PennyLane, CUDA-Q, and other tools remain the tools used to develop algorithms and work with quantum circuits or scientific models. Qall sits above these tools, allowing them to be integrated into broader workflows combining different computing resources.
Qall is therefore less concerned with how a quantum algorithm is designed and more concerned with how the overall computation is executed.
For the initial launch, the developer interface of Unified Computing Service will be exclusively Python-based.
Is Unified Computing Service a quantum service?
Not exclusively. The service allows different computing architectures -CPU, GPU, and QPU- to be combined within the same workflow. Quantum computing is an important use case, but it does not define the product.
Why do I need Qall if I can already access a QPU through Qiskit or PennyLane?
Accessing a QPU is only one part of a quantum computing workflow. In many cases, a computation also includes preprocessing on CPU, optimization or emulation on GPU, followed by post-processing on CPU or GPU.
Unified Computing Service brings these different stages together into a single workflow, without requiring you to manage the infrastructure associated with each of them separately.
Quantum SDKs remain the tools you use to develop your algorithms; Unified Computing Service handles their execution as part of a broader computing environment.
Can I combine CPU, GPU, and QPU in the same workflow?
Yes. This is one of the main goals of Unified Computing Service.
Each workflow step can declare the resources it requires. The workflow can then chain tasks running on different computing architectures and move data between them.
Do I need to rewrite my code to use Unified Computing Service?
The goal is to minimize the changes required to your existing code.
You continue to use Python and the scientific or quantum tools of your choice. Qall adds a layer that allows your code to be turned into tasks and workflows that can run across different resources.
Does using Qall lock me into Scaleway?
No. Qall is designed as an open, infrastructure-agnostic execution layer.
The framework is open source, and its architecture allows different execution environments and resource providers to be integrated. You can therefore use Qall independently of Scaleway's managed service.
Unified Computing Service, on the other hand, provides a managed and integrated experience on Scaleway infrastructure.
When will Unified Computing Service be available?
Unified Computing Service is not yet available. The service is planned for launch in Q1 2027.
Does Qall do the same thing as CUDA-Q?
CUDA-Q is a programming platform designed around quantum applications.
Qall operates at a higher level: the workflow and execution level. It allows multiple tasks, potentially developed with different frameworks and running on different computing architectures, to be composed into a single workflow.
CUDA-Q can therefore be used within a Qall workflow. Qall does not aim to replace CUDA-Q, Qiskit, PennyLane, or other specialized frameworks. Instead, it provides an execution layer above them.
CUDA-Q makes a quantum program hybrid; Qall makes a computing workflow heterogeneous.