What is Moody’s Analytics approach to quantum computing?
At Moody’s Analytics, we’re focused on innovation to better serve our customers. We’re excited to announce a quantum leap forward with our exploration of quantum computing, which promises extensive improvements for computationally expensive problems. To that end, we’ve created the Moody's Analytics Quantum Computing Team. As managing director, I lead the team in the U.S. and Europe.
In terms of quantum computing strategy, we are taking a three-fold approach: 1. Create internal knowledge. We see quantum computing as something that will evolve and grow over time. So, we are creating resources, knowledge, best practices, and strategic partnerships, because in five years every unit will have their own quantum professionals working on their own domains. At that point, the plan is to train and create a better foundation for others to build on top of, similar to what happens with different aspects of computing today.
2. Generate thought leadership. Whenever any of our clients and stakeholders think about evolution, new technology, or innovation in the area of finance or risk or data, we want to be top of mind for them in this area. This is a great opportunity to work together on developing joint use cases and case studies to enhance our solutions portfolio.
3. Improve and accelerate product speed-to-market. We are working with two specific cases that are optimizing portfolios, leveraging the data, analytics and unique insights at Moody’s, along with economic capital calculations. These are two of the more complex problems we solve today and appear in several of our products. This is an area where we could potentially realize quadratic or exponential enhancements to accelerate how we create solutions.
The more powerful computing by quantum offers the potential to revolutionize our ability to dimensionalize risk and provide even better solutions for customers.
Where are we in the evolution of quantum computing?
The quantum industry pace is accelerating. Some people say it’s three years away, others argue it is five or seven.. Every week there is a major milestone in the field (either from basic research, engineering or company launches) so it appears that within a few years our industry will be fully disrupted. It is time to put on the wetsuit and grab the surfboard, to borrow a surfing reference. The wave is coming.
This is where our three-fold approach comes to play: building use cases we can research and develop. That takes time.
The big change will come overnight when we achieve this quantum advantage. It will only take one company to release a few thousand qubit device and those are on roadmaps already. Today, we can run a portfolio optimization with just a handful of assets. And we have the theory, the software, the toolset to run it for longer and bigger problems. We simply need the hardware. But the industry today is in the engineering problem phase, meaning we know what we have to do. We just have to "build it.”
What’s in it for our customers?
It is our responsibility to ensure we always offer the best solutions based on the best and latest technology. Today we talk about quantum, but in 10 years we'll be talking about some other new technology as well. That’s because technology evolves amazingly fast. Making the right decisions about how to use technology helps lead to competitive advantage. Conversely, companies that don’t do this can easily find themselves at a major disadvantage. What if the very quantum startups working today with some of our clients find a way to solve in seconds what takes hours or days for us?
There is also the “get into production” angle. Make this real. Moody's Analytics has some of the best datasets and mathematical models, along with some of the best quants and analysts in the world. When we go out and speak with quantum engineers or researchers in the industry, we realize they lack one important thing: domain knowledge. If you have a bad model or algorithm and make it quantum, you're going to have a very fast, bad algorithm. What we're trying to do is take our best-of-breed mathematical, quantitative, analytical models and make them much faster.
What is Moody’s Analytics competitive advantage when it comes to quantum?
First, we need to understand what quantum computers are for or what we can use them for, because they are not useful for everything. Rest assured you will continue to use Excel and send emails from your laptop or from your phone using our faithful chips and processors and silicon-based devices.
At Moody's, we solve very specific problems for banks and insurance companies.. These problems are related to credit risk, regulatory compliance, capital and capital requirements or portfolio optimizations and fraud analysis, among others. Those computationally expensive problems are ones we can solve much better with quantum computers -- either faster, or with more accuracy. Now, here’s a critical question: if there is a technology that allows to solve the same problems we solved today in a fraction of the time it takes and with a fraction of the resources, shouldn't we be exploring that technology to the fullest of its possibilities? The answer is of course, and that is exactly what we are doing.
Does Moody’s need quantum computing to be on the leading edge? Are we building or buying a quantum computer?
Moody's is a leader in providing a holistic view of multi-dimensional risk and solutions that help customers make better, faster decisions. We have adopted cloud computing, integrated AI and machine learning and many other key technologies to better serve our clients. Quantum computers are just one additional step, and a really big one -- a specific, groundbreaking technology that can allow us to create much better products for our customers.
But we most likely won't be putting a quantum computer in the office. The difference between the 1960s and today is that we already have the Internet, cloud computing, and a plethora of programing languages and all the data science and AI and machine learning tools that we can leverage. The vast majority of quantum hardware manufacturers are putting their computers in the cloud. So we don't have to do the heavy investment and capitalization on building or creating our own devices. In addition, there are a variety of quantum devices.
In addition, there isn’t one single “type” of quantum computer. Some may remember the VHS and Betamax battle from the 1970s and 1980s (if you were one of those who bought a Betamax…well, I’m so sorry). Something like that could happen in quantum as well. Therefore, the smart choice is to partner with these quantum hardware providers with cloud access and then invest in the algorithms and in the integrations. That’s where our leading data, analytics, insights and deep domain expertise comes into play.
Moody’s is working with some of the best quantum hardware providers in the world to provide computational speed-ups when they come and exploring different technologies at the same time.
What are the limitations of quantum computing?
There are many limitations related to quantum computing. One of the biggest ones is how abstract the concept is, so when people hear the two words “quantum computing” or “quantum mechanics,” everyone starts thinking it is science fiction. It is not an easy topic to cover. It requires education about what can and cannot be done. The second limitation that we have is that it is still a nascent technology. We are still in the early stages in exploring, researching, and evaluating, to marry this technology with our expertise in multi-dimensional risk solutions. Now is the perfect time to start exploring it.
Will quantum computing integrate with other evolving technologies?
Absolutely. The one that that pairs perfectly is machine learning. Those two fields match and leverage each other really well. First, because they are fundamentally based in statistics, stochastic models, etc. Second, because they use many of the same tools. It is no secret that data science is one of the best ways to get into quantum. A lot of non-physicists that are in this are “converted” data scientist and machine learning practitioners who are leveraging their skills. And then machine learning is a very heavy computational technology. You must train and retrain your models. That takes a lot of CPU or GPU time and a lot of back and forth. There is a lot of research (Quantum Machine Learning) that suggests quantum computers will be able to enhance and leverage what we do with machine learning and vice versa.
A lot of the things we do is hybridize the work between the classical and the quantum worlds. By classical I mean everything from GPUs and TPUs, all the way through to the everyday things you do on your laptop, and then using quantum to improve those most computationally expensive elements. Imagine a quantum neural network that simplifies the training and uses far less training data or does not require principal component analysis.
Apparently, you can crack an internet password or cryptowallet with a quantum computer. Is that correct?
One of the things that that we know about Quantum is that we will be able to solve one of the most computational problems that we have today: cracking passwords. That is a problem that we made exceedingly difficult to solve on purpose. We humans built an algorithm (RSA) that made almost impossible for your computer to be able to crack my credit card or my crypto wallet or any transaction that I do with my bank or line or anything like that. But quantum computers have a way to crack that code. It would take billions of years to crack your crypto wallet, and only a few seconds for a 1M logical qubit device.
Because of that, cybersecurity is on top of the list of use cases today. Not just our list but also for most governments in the world. The geopolitical implications of a device that could eavesdrop on any conversation, transaction or online device are massive. The good news is that we don’t think we will have a device capable of running that algorithm for at least 10 more years. But we know that will be possible and we are working on mitigating that risk as well.