The question of “Why invest in quantum computing now?” is one every CIO should be asking.
When compared to the evolutionary path of traditional computing, Sergio Gago, managing director for artificial intelligence (AI), machine learning and quantum at Moody’s, says: “We used to joke in the industry that quantum computing is roughly in the late 60s or early 70s, of classical computing, when people were saying, ‘How do you actually code this? What is the stack?’. The opportunities for the market is the one very important thing to have in mind when you’re deciding to invest in quantum computing technology.”
However, he points out that there is no quantum advantage to business processes today: “No matter what some quantum hardware companies are saying, today we have field programmable gate arrays [devices] and big hardware clusters. These are technologies that we know very well. You can do everything but with a lot of computational limitations.”
Research conducted by Corinium Global Intelligence in partnership with Moody’s Analytics found that 87% of the financial sector data scientists polled currently do not have a budget for quantum initiatives.
According to Gago, many people try to find a return on investment (ROI) for establishing a team of quantum computing specialists: “The answer is you don’t. You cannot work from an ROI perspective on this.”
Instead, he says tech leaders should try to work out the value the quantum team will bring to the business and use this to justify the investment to the company. As an example of a pitch to the business looking at using quantum computing to speed up a particular computational problem, Gago says: “We believe that for this specific problem, we can run it exponentially faster, which means that we have a competitive advantage that’s exclusive to us or we can run it for many more cases instead of a handful of portfolios, or we can run it in real time, which results in a lot of value for our customers.”
In chemistry and in areas such as finance, Gago believes the ability to solve complex problems will quickly hit a ceiling, which will mean that at some point in the future, classical computing will lack the processing power needed to tackle such problems in a timely manner.
“There are certain problems where in the financial industry, for example, everyone is completely limited by what we can do. So, we all approximate the answers in credit risk calculations,” he adds.
Projecting the opportunities
Moody’s has an army of quants specialised on types of Monte Carlo simulations, which they hybridise with machine learning and artificial intelligence (AI). This team, he says, pushes the boundaries in classical computing. However, the maths show that a quantum computer would have the processing power to tackle these hard problems.
The team at Moody’s is now working on using classical computers that can simulate quantum computer applications. These simulations test what Gago describes as “tiny problems”.
“With that information, we can actually extrapolate and say, well, once we have enough error corrected qubits that are high quality and have high clock speeds, and are connected in a lattice, then we will be able to run these algorithms and have an exponential speed up to specific problems,” says Gago.
When asked about the progress being made in the industry to tackle the computational problem areas faced by the financial sector, Gago says Moody’s speaks to many quantum companies, all with their own intellectual property and their own idea and great research.
However, he says: “What most of them lack is understanding of the industry domain. So, you actually see companies that say they can speed up an algorithm, but when you know our industry, you realise that is not really a problem.”
Running simulations on classical computers and extrapolating the processing power required may, he says, show that 700 logical qubits will be needed to tackle a specific computational problem. There is an industry roadmap, which tech leaders can use to estimate roughly how many years they will need to wait before a production-ready 700 logical qubit quantum computer is available.
“That’s a good enough plan, but this horizon into the future is based on today’s information. But what’s happening in this field is that every week there is some new breakthrough,” says Gago.
For example, he points to Quantinuum’s recent news that its researchers had achieved three entangled logical qubits. As to the significance of this breakthrough, Gago adds: “I think now we’re now getting hints of the beginning of the fault-tolerant era of quantum computing.”
From a purely practical perspective, Gago says the breakthrough is analogous to the quantum supremacy experiments from IBM or Google: “Those were done on problems that are completely disconnected from an actual applicable problem, but does this make those breakthroughs irrelevant or useless? Absolutely not.
“If you asked scientists 10 years ago about when would we have logical qubits, many would say never and the others would probably would say by the end of the next century. But now we know that it is possible. This is kind of the beginning of additional research and scalability that the industry can start working on.”
Justifying quantum investment
Some IT leaders will inevitably find it hard to justify investment given the level of uncertainty in quantum computing. But Gago believes this should not stop them from planning, saying that this is how those businesses that are leading the way in AI have managed to use the technology to achieve a competitive advantage.
“Some companies, including ourselves, have embraced generative AI [GenAI] and we know how to use it. But many other companies have no idea where to start,” he adds.
With AI, businesses needed to prepare for an AI strategy before the technology reached enterprise maturity. For Gago, this means putting in place data governance and data pipelines. Moody’s, he says, has been investing in AI for 10 years, with data in the right place and a data warehouse orchestration layer. This is something he believes is lacking in many companies. Once these things are in place, implementing GenAI, according to Gago, can be achieved relatively quickly.
“I think now we’re now getting hints of the beginning of the fault-tolerant era of quantum computing”
Sergio Gago, Moody’s
The research from Corinium Global Intelligence reported that 82% of financial institute data scientists believe quantum computing immaturity is a barrier to the development of the technology in their organisations.
For Gago, this was also the case in the early days of enterprise AI. “If you ask yourself what is going to be quantum’s ChatGPT moment, it’s certainly not going to be next year,” he says.
Nevertheless, that should not stop companies from preparing. After all, Gago says, Google researchers published their paper, Transformer: A novel neural network architecture for language understanding, in 2017 – research which has led to the emergence of generative AI. But it has taken four to five years for the technology to become mainstream.
The important thing is that rather than try to calculate an ROI, Gago urges IT leaders to determine the computational problems that simply cannot be solved with classical computer architectures. Understanding the benefits to the organisation of solving these problems can help frame the funding debate.
And even though it may be seen as a long-term commitment to something that could be many years away, from the conversation with Gago, having the right expertise and technology in place will inevitably lead to a strategic advantage as quantum computing matures.