Gartner Predicts Enterprise AI Workloads at Scale Will Not Run on Quantum Hardware Through 2028

• 05 Aug 2026
• TM Team
• 0 Comments

No enterprise AI workload at scale will run on quantum hardware by 2028 and classical accelerated AI will dominate every production benchmark according to Gartner, Inc., a business and technology insights company.

“When vendors claim to deliver ‘quantum AI,’ they usually refer to hybrid or quantum-inspired techniques, not quantum-native AI running at enterprise scale,” said Chirag Dekate, VP Analyst at Gartner. “True quantum computing is not ready for any production AI workload and will most likely not be for the rest of this decade. Furthermore, no peer-reviewed result demonstrates quantum advantage on production AI workload.”

“Quantum AI” refers to AI or machine learning (ML) techniques that require execution on quantum hardware to achieve a claimed performance, cost or capability advantage over classical computing.

“Achieving fault-tolerant quantum computing at the scale needed to deliver measurable AI performance or cost benefits requires advances in four areas: hardware, error correction, middleware and algorithms,” said Dekate

Quantum AI is different from three commonly conflated categories:

· Classical AI: AI models, such as deep learning, transformers and reinforcement learning, that run entirely on CPUs, GPUs or TPUs and deliver measurable enterprise ROI today.

· Quantum-inspired AI: Classical algorithms that borrow ideas from quantum mechanics, such as annealing, tensor networks and amplitude encoding, but run on conventional hardware. These techniques already deliver value in optimization, simulation and sampling workloads and do not depend on quantum hardware.

· Hybrid quantum-classical methods: Experimental workflows where small quantum circuits are orchestrated alongside classical AI or high-performance computing (HPC) systems. These approaches are used for research and vendor-assisted pilots, not for production AI.

When vendors claim to deliver “quantum AI,” they usually refer to hybrid or quantum-inspired techniques, not quantum-native AI running at enterprise scale.

Vendor framing around quantum-AI convergence is accelerating and boards pressured to “not miss quantum” will be tempted to divert AI budget toward R&D that cannot return value before 2030.

Because of this, Gartner predicts that fault-tolerant quantum computing will remain in the R&D phase for AI, with insufficient logical qubit scale to support economically viable, end-to-end AI algorithms through 2030.

Separate the Quantum Budget From AI Budget

Quantum R&D and AI production infrastructure have incompatible time horizons, unit economics and governance needs. GenAI returns measurable value within 12 to 18 months on turnaround, accuracy and automation metrics. Meanwhile, quantum AI has never returned measurable value on any production workload, and is unlikely to do so for the foreseeable future.

“Co-mingling the two budgets distorts accountability for both and lets quantum optionality crowd out production AI capability,” said Dekate. “CIOs constantly place GenAI, agentic AI, cybersecurity and cloud in the top spend categories. Quantum does not appear on top-priority investment lists. CIOs who are funding quantum are doing so at materially lower rates than GenAI, and are not demanding near-term return on investment.”

To get the most out of quantum offerings, enterprises should:

· Deploy quantum-inspired classical methods inside the existing AI stack: Organizations should prioritize quantum-inspired algorithms running on today's GPU infrastructure rather than pursuing quantum hardware. Across areas such as optimization, genAI, linear algebra, graph analytics, and reinforcement learning, classical approaches deliver similar benefits without the cost, complexity, or immaturity of quantum systems.

· Define pilot kill criteria and refuse vendor theater: Every quantum pilot should begin with predefined success metrics, a clear classical benchmark, and explicit conditions for termination. This prevents open-ended experimentation from consuming budget without delivering measurable value.

· Track quantum progress that matters: The most important measure of progress is not physical qubit count but the availability of logical qubits operating at useful error rates. Advances in error correction, AI-assisted calibration, and quantum control are more relevant to enterprise value than headline announcements about larger quantum processors.