IBM is laying the groundwork for the convergence of AI and quantum computing in financial services by optimizing hybrid algorithms.  

The tech giant says that quantum computing can benefit financial institutions by addressing errors in AI models, Manuel Proissl, global industry applications lead for financial services at IBM Research, told FinAi News.  

“All of the quantitative risk models, for example, have a very strict framework around it,” he said. 

“You have to thoroughly understand where the sources of the errors are coming from and how you can mitigate them. And this brings us to quantum because the idea is not to replicate these models in parts where we are already good at. It is attacking the errors.” 

Read part 1 of quantum-AI series. 

Despite the capabilities of “all of the supercomputers on the planet,” errors are possible in every AI model, and hybrid quantum-classical computing methods are key to maximizing AI, Proissl said.  

With this approach, “a bulk of a problem is addressed by classical quantitative finance method, which is usually statistical tools and so on,” he said. “Then there is an AI component that maybe extracts some form of a signal or recognizes some patterns.” 

Lastly, “we are ingesting quantum to run a subroutine,” outsourcing a complex task to a quantum processing unit (QPU), which then feeds results back into an AI model to improve overall performance, he said. 

In essence, tech leaders at FIs should be thinking about synchronizing CPUs, GPU and QPUs to streamline distinct parts of a workflow while measuring “the value of reducing errors,” Proissl said.  

Banking use cases  

Proissl said when optimizing hybrid algorithms, strong use cases for quantum-AI convergence in financial services include:

• Quantitative investing;
• Trading;
• Portfolio and asset management;
• Risk modeling;
• Fraud prevention; and
• Derivative pricing.  

Lloyds, for example, found that quantum computing complements AI and machine learning models when fighting financial crime, according to a spring news release by the $1.3 trillion bank, which conducted a nine-month experiment with IBM.  

Lloyds concluded that quantum is beneficial when deployed in areas where AI and ML are strained, including graph-based anomaly detection. 

In another study, HSBC used IBM’s quantum solutions to deliver a 34% improvement in predicting the profitability of algorithmic bond trading, according to a September 2025 release by the $3.3 trillion bank.  

Quantum AI also can help banks build “robust portfolios” that withstand market swings and “geopolitical shocks,” Proissl said.  

Infrastructure overhaul? 

For banks that have a strong AI infrastructure, the need for enhancements to support quantum largely depends on the desired latency and ownership of the technology, Proissl said.  

Using IBM’s cloud quantum computer, for example, requires no infrastructure change for most applications, he said.  

However, use cases that involve continuous evaluation of real-time metrics might require infrastructure upgrades to minimize latency between classical and quantum computing, he said. 

There might come a time when more FIs want to own quantum hardware, but for now, cloud-based solutions are generally more feasible, Proissl said.  

IBM’s Cost Breakdown to Access Quantum Network

Developing qubits — the basic units of information in quantum computing — typically costs banks $10,000 to $50,000 per superconducting qubit, according to a 2025 study by Rob Perryman, chief technology officer at financial services firm Sun Life and former tech leader at HSBC.  

In the short term, FIs should allocate at least 5% of their AI budgets to quantum computing, Katanya Kuntz, founder and chief executive of Qubo Consulting, previously told FinAi News. 

Register here for the FinAi Lending Summit, set for Oct. 7-8 in Las Vegas.