Analyst(s): Dr. Bob Sutor
Publication Date: February 25, 2025
Given the potential value of quantum computing and AI technologies, it is reasonable to ask if one of them can benefit from the other. Recent remarks from industry CEOs such as Jensen Huang of NVIDIA show that no one knows when quantum computing will be “useful.” Being useful seems like a fairly minimal requirement, but it is likely five or more years away. This includes “useful for AI.” Therefore, understand that quantum computing for AI is a long-term research program, not an immediate commercial offering.
Key Points:
- AI’s importance and evolving role in society and industry has drawn significant attention to its computational requirements.
- Quantum computing’s promise of solving currently intractable problems has led many to speculate that it will bring AI training and inference breakthroughs.
- Today’s quantum computers can input small amounts of classical data and perform very few operations, making them impractical for useful AI applications.
- Future quantum systems with millions of qubits and error correction may be able to demonstrate AI computation and pattern recognition performance far better than classical approaches.
Overview:
Quantum for AI explores how quantum computing can enhance artificial intelligence, particularly in machine learning. Machine learning relies on large datasets and complex computations to extract patterns, make predictions, and generate insights. Can quantum computing improve AI efficiency while reducing costs and energy consumption?
The first challenge is how to encode large amounts of classical data in quantum form quickly. The short answer is that we can’t. Quantum computers store their data in qubits, and even with the literal exponential growth in the amount of information we can store as we add more qubits, the encoding process is slow and impractical today. When quantum networking from companies like Nu Quantum, Welinq, and IonQ becomes practical, we may be able to use quantum memory and data output from quantum sensors.
The second challenge is an artifact from the quantum mechanical foundation of quantum computing: we cannot copy data. It’s not that we are not smart enough, it’s that it is against the quantum rules of Nature. This severe restriction makes classical neural networks useless for direct translation to quantum implementations. We must substitute quantum algorithms in their place, but these have limitations because of the final challenge.
Today’s quantum computers can execute very few instructions before the information stored in their qubits becomes useless. In the long run, we hope to fix this by implementing fault tolerance and error correction, but these require far more qubits than the systems we have today by two to three orders of magnitude.
Given these three challenges, the practical use of quantum computers for AI is at least five years away. If we turn the words around, however, many vendors use AI machine learning for quantum computing today. It appears that we will use AI to improve quantum computing to improve AI.
The full report goes into these topics in detail. It also lists the companies known to be working on quantum for AI, along with their geographic distribution.
Futurum clients can read more about it in the Futurum Intelligence Platform. Nonclients can learn more here: Futurum Intelligence.
About the Futurum Quantum Computing Practice
The Futurum Quantum Computing Practice provides actionable, objective insights for market leaders and their teams so they can respond to emerging opportunities and innovate. Public access to our coverage can be seen here. Follow news and updates from the Futurum Practice on LinkedIn and X. Visit the Futurum Newsroom for more information and insights.
Author Information
Dr. Bob Sutor is an expert in quantum technologies with 40+ years of experience. He is the accomplished author of the quantum computing book Dancing with Qubits, Second Edition. Bob is dedicated to evolving quantum to help solve society's critical computational problems.
