The Open Acceleration Stack: Making the Impossible Possible in Quantum-Classical Supercomputing 

Today, we are introducing Quantum Machines’ Open Acceleration Stack, an architecture designed to tightly integrate quantum processors with classical accelerators such as GPUs, CPUs, FPGAs, and specialized ASICs.

We created it to address a challenge that is becoming impossible to ignore as the quantum computing field moves toward logical qubits: the growing need for powerful classical computation alongside quantum systems. Until now, that challenge has made many goals in quantum computing very difficult to achieve. With the Open Acceleration Stack, it is now possible to couple any accelerator to any QPU at both the hardware and programming levels through the real-time orchestration capabilities of the Pulse Processing Unit or PPU. 

Why is this important?  

Because quantum bits remain fragile, complex, and still far from the performance we ultimately want them to achieve. Heavy classical compute and AI are critical, needed alongside qubits at the algorithmic level to correct them, learn from and improve them, and calibrate them. The entire process must run in tight feedback loops, coordinated in real time by the PPU. 

To me, this moment feels very similar to the early days of high-performance computing. Back then, progress did not come only from faster processors. It came from entirely new system architectures that connected processors, memory, and accelerators into a single coordinated machine. I believe quantum computing is now reaching a similar inflection point, with the PPU acting as the coordinating engine of quantum-classical systems. 

It is no longer just about the qubits 

For many years, the main challenge in the quantum computing field was building and stabilizing physical qubits. That work continues, of course, but the problem is evolving. As quantum processors grow, running them successfully requires increasingly complex workflows: continuous quantum error correction, AI-driven calibration, and coordination across multiple layers of the quantum software stack. All of this relies heavily on classical computation that must operate fast enough to keep up with fragile quantum states. The PPU serves as the real-time execution layer that keeps these feedback loops operational. 

A quantum computer is not merely a quantum chip. It is a system in which quantum hardware and classical processors work together in tight feedback loops. There is no quantum computing without classical processing, and the PPU synchronizes these operations in real time while connecting specialized classical compute architectures directly to the QPU. 

Quantum error correction (QEC) is a good example. QEC is a continuous cycle: measure the qubits, decode the results, and apply corrections before errors accumulate. As we scale in qubit numbers, decoding becomes increasingly complex. If it is too slow, decoherence and compounded errors make the entire process useless. Fast classical feedback coordinated by the PPU is essential. While the value of using AI in this context is important, the Open Acceleration Stack is making it possible to implement it in real time. 

I see a similar challenge in calibration and optimization. Researchers increasingly apply AI and reinforcement learning to tune quantum systems and explore the extremely complex parameter space of quantum devices. That would be impossible to do by brute force. The Open Acceleration Stack allows any accelerator to operate within tight feedback loops between the accelerator, the PPU within the OPX, and the QPUs. When these loops are too slow, the process fails: the data becomes outdated, optimization loses relevance, or the workflow simply takes too long. The PPU, together with OPNIC, the interface that connects classical accelerators to our real-time quantum control system, ensure these loops run fast enough to turn theoretical possibilities into practical reality. 

It’s also crucial to recognise the importance of an open architecture. Classical computing should not sit outside the quantum system; instead, classical accelerators should operate alongside real-time quantum operations, with the PPU orchestrating these interactions at the control layer. Whether a workload relies on machine learning frameworks, FPGA-based decoders, GPUs, or CPUs, the system provides a unified pathway to connect these resources directly to the quantum control layer. 

Flexibility was extremely important to us when designing this architecture. Quantum computing is evolving quickly, and researchers around the world rely on very different software stacks and hardware environments. Some workloads benefit from GPUs, others from FPGAs, and many teams have already invested heavily in specific classical computing infrastructures. The Open Acceleration Stack allows them to integrate CPUs, GPUs, FPGAs, and ASICs without redesigning their entire infrastructure, while continuing to use the software and workflows they already know. 

At the center of this architecture is OPNIC, a critical link between the PPU and external compute resources that enables extremely low-latency data exchange between classical processors and ongoing quantum operations. Separate and heterogeneous compute systems become a single coordinated environment through QM’s Orchestration Platform. This means that algorithms running on GPUs or other accelerators can interact directly with a quantum processor while it is running, through the real-time control path enabled by the PPU. Developers can work with familiar tools, from Python frameworks to CUDA and C++, and integrate them into hybrid quantum-classical workflows. 

The importance of the ecosystem

I can’t thank our collaborators enough for working with us to bring this vision to life. Over the past years, we have worked closely with companies including NVIDIA, AMD, and Riverlane, each contributing important expertise to the emerging quantum ecosystem. Our earlier work with NVIDIA, which led to DGX Quantum and paved the way to NVQLink, demonstrated the power of tightly coupling GPUs, CPUs, and QPUs through a PPU-centered control architecture. With the Open Acceleration Stack, we are expanding that concept beyond a single configuration. Instead of a fixed system, we are enabling an open ecosystem where different accelerators, software stacks, and quantum platforms can work together. 

To me, one of the most exciting aspects of this work is bringing quantum computing closer to the broader computing ecosystem. By connecting quantum processors with familiar accelerated computing platforms through the PPU-enabled architecture, researchers, HPC centers, and eventually data centers will be able to interact with quantum systems using tools they already understand. I believe the next major advances in quantum computing will not come only from better qubits, but from systems that allow researchers to tackle problems that were previously out of reach, turning what once seemed impossible into something achievable. 

The Open Acceleration Stack is a big step toward that future.