Thought Leadership
Gartner sees no evidence of a genuine quantum advantage for production AI applications and advises enterprises to keep quantum computing and AI budgets strictly separate.
Anyone expecting quantum computers to replace conventional GPUs and TPUs for training or running AI models anytime soon is likely to be disappointed. Market research firm Gartner predicts that, at least through 2028, no production AI application will run on quantum hardware at meaningful scale. According to the latest forecast, conventional accelerated AI systems will continue to outperform quantum approaches across all relevant production benchmarks.
Chirag Dekate, Vice President and Analyst at Gartner, puts it bluntly:
“True quantum computing is not ready for any production AI application and is unlikely to be for the rest of this decade.”
Chirag Dekate, Vice President and Analyst at Gartner
Dekate also notes that there is not a single peer-reviewed result demonstrating a genuine quantum advantage for a production AI workload. When vendors promote “quantum AI,” the technology in practice almost always involves hybrid or quantum-inspired approaches rather than genuine, enterprise-scale quantum computing.
Four Hurdles Stand Between Quantum Computing and Production
According to Gartner, quantum hardware would have to advance in four areas simultaneously before it could deliver measurable performance or cost benefits for AI workloads: the hardware itself, error correction, middleware and the underlying algorithms. If any one of these components is missing, fault-tolerant quantum computing will remain a research topic rather than a production technology, the analysts say.
Three Terms That Are Often Confused
In its analysis, Gartner distinguishes between three categories that are frequently conflated in public discussions:
Classical AI: Deep learning models, transformers and reinforcement learning that run entirely on CPUs, GPUs or TPUs and already deliver measurable business value today.
Quantum-inspired AI: Classical algorithms that borrow concepts from quantum mechanics, such as annealing, tensor networks or amplitude encoding, but run entirely on conventional hardware. According to Gartner, these approaches already provide benefits in optimization, simulation and sampling without requiring quantum hardware.
Hybrid quantum-classical approaches: Experimental workflows in which small quantum circuits work alongside classical AI or HPC systems. According to Gartner, these approaches are currently intended for research and vendor-supported pilot projects rather than production use.
Gartner Warns Against Diverting AI Budgets
Analysts are seeing growing marketing pressure around the alleged convergence of quantum computing and AI. Executives concerned about missing the “quantum trend” risk diverting AI budgets into research and development efforts that may not generate returns before 2030. Gartner therefore forecasts that fault-tolerant quantum computing for AI purposes will remain in the research stage until at least 2030, simply because the number of available logical qubits is not sufficient to run economically viable end-to-end AI algorithms.
Separate Budgets, Separate Timelines
Gartner’s advice to CIOs is clear: Quantum research budgets and budgets for production AI infrastructure should remain strictly separate. The two areas have fundamentally different timelines, cost structures and governance requirements. Generative AI can deliver measurable value, for example through faster processing, greater accuracy or higher levels of automation, within 12 to 18 months. Quantum AI, by contrast, has yet to generate measurable value in a single production application, and that is not expected to change anytime soon.
Dekate sums it up this way: “Mixing the two budgets weakens accountability for both and allows pure quantum optionality to crowd out critical, productive AI capabilities.” He adds: “CIOs consistently rank generative AI, agentic AI, cybersecurity and cloud among their top spending categories. Quantum computing does not appear on the lists of top investment priorities.”
According to Dekate, organizations that do invest in quantum projects are doing so with significantly smaller budgets than those allocated to generative AI and generally without demanding a short-term return on investment.
Gartner’s Practical Recommendations
For organizations that still want to explore quantum offerings, Gartner has three concrete recommendations:
Prioritize quantum-inspired approaches over genuine quantum hardware: Classical, quantum-inspired algorithms can be integrated directly into existing GPU infrastructure. In areas such as optimization, generative AI, linear algebra, graph analytics and reinforcement learning, these approaches can offer benefits similar to those associated with genuine quantum computing, without its cost, complexity and technological immaturity.
Define clear pilot exit criteria: Every quantum pilot should begin with clearly defined success metrics, a classical baseline benchmark and firm stop conditions. This helps prevent open-ended experiments from consuming budgets indefinitely without delivering measurable value. Companies should consistently reject pilots that amount to little more than vendor demonstrations.
Track the right indicators of progress: The key metric is not the number of physical qubits but the availability of logical qubits with practical error rates. Progress in error correction, AI-assisted calibration and quantum control is more relevant to enterprise value than headlines about increasingly large quantum processors.
Conclusion
Gartner’s message to IT decision-makers is unequivocal: Companies looking for production-ready AI benefits today will find them on classical, GPU-accelerated infrastructure, not on quantum hardware. Organizations that choose to explore quantum computing should treat it as a separate, long-term research field rather than as an extension of their current AI strategy.
(lb)
