QuTwo: How enterprises can run quantum-aware AI today

QuTwo helps enterprises run quantum-aware AI now by routing workloads to the best available compute — GPUs, quantum‑inspired accelerators (classical hardware that mimics some quantum methods), hybrid stacks, or true quantum processors when they’re ready.

Why this matters for business leaders

Most companies know quantum computing could accelerate specific AI tasks someday, but scalable quantum hardware is not yet a mainstream deliverable. Rewriting large AI systems for an uncertain future is expensive and risky. QuTwo offers a different path: middleware that decides where each piece of work should run, so teams can experiment with quantum-inspired techniques and hybrid computing today while keeping the option to switch to real quantum hardware later.

“We’re building for the quantum world, but QuTwo is an AI company.”

That line captures the product-market fit: QuTwo is less about selling qubits and more about delivering business outcomes through compute orchestration and design partnerships.

What QuTwo builds

QuTwo’s core product is QuTwo OS — a compute orchestration layer that evaluates a task and routes it to the best substrate: classical servers, quantum‑inspired chips, hybrid flows that split work, or true quantum processors once they become competitive. The platform aims to remove the need for major rewrites of models or pipelines when hardware changes.

Peter Sarlin launched QuTwo with funding from his family office, PostScriptum. The company says it already has “large design partnerships which are in the tens of millions.” Active partners include Zalando, working on so‑called “lifestyle agents” to improve personalization and customer experience, and OP Pohjola, which has a joint quantum AI research initiative. The team includes 30+ quantum and AI scientists and industry figures such as Kuan Yen Tan (IQM cofounder), Antti Vasara (SemiQon chair), Kaj‑Mikael Björk (former Silo AI cofounder), and Pekka Lundmark (former Nokia CEO).

Plain-English: quantum-inspired vs hybrid vs true quantum

• Quantum-inspired: Algorithms or hardware that run on classical computers but borrow ideas from quantum physics to improve optimization or sampling. They can offer gains today without qubits.
• Hybrid computing: Workflows that split a task into parts and run different subtasks on different hardware — for example, pre-processing on GPUs and a hard combinatorial core on a quantum-inspired accelerator.
• True quantum computing: Uses qubits and quantum operations. It promises unique advantages for some problems but is currently limited by scale, noise, and error correction challenges.

How QuTwo OS routes compute — a simple workflow

Think of QuTwo OS as a GPS for compute. A typical routing loop looks like this:

1. Detect: The system inspects the task, data size, latency needs, and cost constraints.
2. Decide: It predicts expected runtime, energy use, and solution quality on each available substrate.
3. Dispatch: The workload (or its subtask) is sent to the best-fit hardware.
4. Monitor: Results and telemetry feed back into the decision model to improve future routing.

Observability matters. Enterprise leaders should demand dashboards that show routing decisions, cost comparisons, and a measurable business metric tied to each experiment.

Real-world traction and a micro‑case

Zalando and QuTwo describe their work as building “lifestyle agents” — proactive AI agents that anticipate customer needs and surface curated product suggestions. A practical pilot might look like this:

• Objective: increase recommendation-driven conversion by 5% while reducing recommendation compute cost by 10%.
• Experiment: route heavy combinatorial candidate-selection tasks to a quantum-inspired accelerator during peak traffic, keep ranking models on GPU clusters for latency.
• KPI measurement: conversion uplift, average latency, cost per thousand recommendations, and energy consumed per inference.

For finance firms like OP Pohjola, quantum-aware pilots often target portfolio rebalancing or risk-scenario generation — tasks with combinatorial complexity where quantum-inspired or hybrid approaches could cut run time or energy use.

Business implications, risks, and vendor questions

QuTwo’s approach buys enterprises time: you can experiment with quantum techniques without committing to a hardware-dependent architecture. That’s attractive for CIOs balancing innovation budgets and operational risk. But there are real considerations:

• Pricing and commercial models are still evolving — expect subscription, usage-based routing fees, or project-based design partnerships.
• Vendor lock-in risk if your workflows depend on proprietary routing logic or IP that’s hard to migrate.
• Explainability and compliance questions when parts of decision pipelines run on opaque or experimental hardware.
• Performance claims should be independently benchmarked; ask for reproducible tests and third-party audits.

Key questions to put to QuTwo or any compute-orchestration vendor:

• How do you price QuTwo OS and design partnerships?
  

  Get clarity on subscription vs usage billing and what’s included in design engagements.

• Which metrics determine routing decisions?
  

  Request the decision rules or thresholds (cost, expected time-to-solution, solution quality) and the ability to tune them.

• How do you ensure explainability and compliance?
  

  Ask for logs, audit trails, and model explanations when compute moves between substrates.

• What exit or migration guarantees exist?
  

  Verify portability of models and data if you decide to leave the vendor.

• Can you provide reproducible benchmarks?
  

  Demand reproducible tests on representative datasets and third‑party validation.

How to run a 90‑day quantum-aware pilot (practical roadmap)

1. Week 0–2: Identify and scope
   • Select 1–2 candidate workloads (personalization, supply‑chain optimization, pricing). Pick clear KPIs (conversion, cost, runtime, energy).
   • Estimate budgets and data needs.
2. Week 3–6: Integrate and baseline
   • Instrument current pipeline and collect baseline metrics.
   • Connect QuTwo OS sandbox or on‑prem agent and run initial routing tests on small data slices.
3. Week 7–10: Run experiments
   • Execute A/B tests comparing classical-only vs quantum-inspired/hybrid routing.
   • Track KPIs continuously and validate cost/latency tradeoffs.
4. Week 11–12: Evaluate and decide
   • Review results against predefined business metrics and compliance requirements.
   • Decide whether to scale, refine routing rules, or halt further investment.

Quick checklist for executives

• Identify one measurable use case for a quantum-aware pilot (personalization, optimization, or modeling).
• Allocate a modest pilot budget and clear KPIs.
• Demand transparency: routing rules, reproducible benchmarks, audit logs, and exit terms.
• Prefer vendors that support hybrid computing and quantum‑agnostic APIs to preserve portability.

Glossary

• Quantum-aware AI: AI systems designed to take advantage of quantum-inspired or quantum hardware when appropriate.
• QuTwo OS: Compute orchestration software that routes AI workloads across different compute types.
• Quantum-inspired algorithms: Classical algorithms that use ideas from quantum computing to improve performance on classical hardware.
• Hybrid computing: Mixed workflows that allocate subtasks to different hardware types for efficiency.
• Quantum advantage: A clear, measurable performance or cost benefit delivered by quantum hardware over classical alternatives.

QuTwo is “pushing AI workloads from classical to quantum.”

Final takeaways

QuTwo isn’t a moonshot hardware vendor; it’s middleware paired with design partnerships and a science-led team aiming to make quantum-aware AI practical for enterprises today. For leaders focused on AI agents, AI automation, and AI for business, the prudent move is to pilot a narrow, measurable use case, insist on transparency, and treat early engagements as learning investments rather than full migrations.

Three experiments to start this quarter:

• Personalization pilot using hybrid routing for candidate generation and GPU ranking.
• Supply-chain optimization test using quantum-inspired solvers for routing scenarios.
• Financial scenario-generation pilot comparing runtime and energy costs across substrates.