At Q2B 2025, Bob Sorensen of Hyperion Research laid out an eight-step plan for HPC centres preparing to adopt quantum computing, recently summarised on the QuEra blog. It is required reading. The framework treats quantum as a procurement decision waiting to happen, but the path from “look inward” to “buy a system” is where most centres stall.

Which six, and why

I will not pretend our software stack is relevant to every step. Engaging leadership (Step 3) and the procurement mechanics of Step 8 sit largely outside what a vendor-neutral orchestration platform can shorten. But six of the eight — looking inward, building internal support, exploring algorithms, assessing value, managing expectations, and vendor selection — are exactly where centres lose months. They lose them not because the science is missing, but because the integration work is bespoke, the vendor landscape is fragmenting, and the budget cycle does not wait for clarity. 

Let’s look at where Qoro can help this and we’ll give you an example for each of Bob’s steps from both a scientist and director’s perspective.

Step 1: Look Inward First

Workload characterisation, before any vendor conversation.

Scientist view: Most computational chemists, materials scientists, and optimisation researchers can name their three biggest pain points in five minutes. What they cannot do is sketch what a quantum version of that workload looks like. Maestro, our intelligent simulation framework, lets them prototype quantum reformulations and integrations on existing CPU and GPU clusters — no QPU, no procurement. They build the baseline themselves with loosely coupled codebases into Maestro through the native SLURM integration (other schedulers are available), and see what the requirements would be from the workloads they intend to run.

HPC director view: A baseline measured on your own hardware, in your own scheduler, is the only baseline that survives a board meeting. Maestro produces it without commitment. The UK National Physical Laboratory (NPL) benchmarked Maestro independently and commented: 

“Qoro and its Maestro framework are well-suited for HPC environments due to their ability to exploit parallelism through multithreading and multiprocessing. Features such as Maestro Auto for batched execution and distributed simulation strategies enable efficient scaling across clusters and reduce overhead compared to single-threaded runs.

Initial observations show good scalability with increased thread counts, and batch mode further improves throughput.

Overall, Qoro and Maestro align with HPC requirements and are good candidates for hybrid quantum-classical workflows, where simulators act as accelerators within larger computational pipelines.” 

– Deep Lall, Research Scientist, National Physical Laboratory

Step 2: Build Internal Support

Sorensen’s advice: give your champions access and let them experiment cheaply.

Scientist view: Solo, our cloud control centre, gets a researcher from pip install divi to a job submitted in under three minutes, with $100 of free credit. No infrastructure setup, no vendor onboarding meeting. Divi, our open-source Python SDK, runs the same algorithm on a laptop, a simulator, or a real QPU.

HPC director view: Solo lets you seed twenty internal champions for less than the cost of one conference ticket. Adoption funnels at the user level, not the procurement level, are how genuine demand surfaces — and how you justify the capital request that follows.

Step 4: Explore Algorithmic Opportunities

Mapping HPC kernels — Fourier transforms, Monte Carlo, linear solvers, electronic structure — to quantum approaches.

Scientist view: qoro-pyscf, our open-source quantum solver stack for PySCF, drops quantum chemistry routines into a workflow your chemists already know. Optimisation researchers use Divi to write a QAOA program once and run it across multiple problem sizes and backends without rewriting a line.

HPC director view: This is where most centres burn six months writing integration code. Our software stack collapses that work from 1 million lines to 20, and from months to hours. Your researchers explore the algorithm space; we absorb the plumbing.

Step 5: Assess Quantum Value Against Your Baseline

Compare performance, power, cost, staffing, and infrastructure impact.

Scientist view: Composer, our circuit packing and scheduling tool, automatically packs multiple programs onto a backend. On a published MaxCut benchmark on IBM Heron, this delivered an 83% QPU cost reduction; observable grouping cut measurement overhead by 96%. Researchers see what their problem actually costs to run, not what a vendor slide promised.

HPC director view: vQPU, our virtual QPU layer, mimics noise and timing of real hardware so you can validate scheduling, utilisation, and TCO assumptions before any capital request lands on a CFO’s desk.

Step 6: Manage Expectations

Sorensen calls the sector “the Wild West” and warns of consolidation.

Scientist view: You should never rewrite an algorithm because your vendor was acquired. With Qoro Dedicato the same code path runs across IBM, IQM, AWS Braket, and any backend we add tomorrow. Researchers commit to the algorithm, not the API.

HPC director view: Vendor consolidation is the single biggest procurement risk in this market. A vendor-neutral orchestration layer is the only durable hedge: build once, run everywhere. We already orchestrate across five quantum computers simultaneously through the IBM start-up programme.

Step 7: Vendor Selection

Demonstrated performance on your workloads. Modality and qubit counts are secondary.

Scientist view: Bring your own problem. We run it on multiple backends through one API and show real numbers — runtime, fidelity, cost — not vendor specifications. CESGA runs Maestro for performance; NPL benchmarked Maestro independently. The evidence is public.

HPC director view: Choose vendors using evidence from your own workloads, on your own timeline. Solo for training, Maestro for evaluation, Dedicato — our on-prem or dedicated-cloud licence — for production. The same software stack across both, so the procurement decision stays reversible.

None of these steps require betting on a single hardware vendor or waiting for a fault-tolerant machine. They require a software stack that already works with what HPC centres have today. That is what we build, to provide you a low-risk way to move forward quickly in your adoption of quantum computing.