TRANSCRIPTION 🔗
WEBVTT
00:00:06.400 --> 00:00:12.800
Hello, everybody, and welcome to an episode of our new podcast from Transactions to Trust, a financial services podcast.
00:00:12.800 --> 00:00:16.640
I'm your host, Maida Zaheed, part of the marketing team here at CGI.
00:00:16.640 --> 00:00:24.960
Today we're diving deeper into quantum computing, what it really is, why it matters, and how it could change the way financial institutions operate.
00:00:24.960 --> 00:00:26.960
We're joined by our expert Curtis Snibo.
00:00:26.960 --> 00:00:31.199
Curtis is a vice president leading our CGI quantum practice.
00:00:31.199 --> 00:00:32.079
Welcome, Curtis.
00:00:32.320 --> 00:00:32.880
Hi everyone.
00:00:32.880 --> 00:00:34.159
My name's Curtis Snibo.
00:00:34.159 --> 00:00:44.799
I'm a vice president of quantum computing at CGI, and I lead our global quantum computing practice where we really focus on where can we apply real quantum solutions to date?
00:00:44.799 --> 00:00:45.840
So what's real?
00:00:45.840 --> 00:00:48.000
What can companies actually leverage?
00:00:48.000 --> 00:00:50.880
And how can we actually implement quantum in the state that it's in today?
00:00:50.880 --> 00:00:53.039
And it gets better and better every day.
00:00:53.039 --> 00:00:57.840
So where can we find those advantages, especially when it comes to finance and banking?
00:00:57.840 --> 00:01:00.399
My background is actually in finance.
00:01:00.399 --> 00:01:04.000
I did a master's of quantitative finance, and then I did a master's of astrophysics.
00:01:04.000 --> 00:01:07.760
So I kind of have the best of both worlds from an academic sense.
00:01:07.760 --> 00:01:13.599
And then like I said, with CGI quantum, we try and spend our time finding ways to be able to apply that to industry.
00:01:13.599 --> 00:01:14.879
So happy to be here.
00:01:15.439 --> 00:01:16.079
Very exciting.
00:01:16.079 --> 00:01:17.040
Let's get into it.
00:01:17.040 --> 00:01:21.040
When people hear quantum computing, it sounds very futuristic, I'm not gonna lie.
00:01:21.040 --> 00:01:28.159
And so can you help us understand what it is, what is it really, and especially why should financial institutions care?
00:01:28.560 --> 00:01:29.439
Yeah, absolutely.
00:01:29.439 --> 00:01:35.680
So quantum computing is really interesting and it's an emerging technology at its core.
00:01:35.680 --> 00:01:47.519
And so, what quantum computing is, to just give a brief introduction into this topic for those that might be new to it, a good way to think about it is a fundamentally new way of processing information.
00:01:47.519 --> 00:01:52.719
It's not AI, it's not an algorithm, it's a fundamentally new way of doing computation.
00:01:52.719 --> 00:01:58.719
So it's built on physics that govern how nature behaves at very small scales, which is quantum physics.
00:01:58.719 --> 00:02:05.760
And so to understand quantum computing, it helps to start with how it relates to the systems we use every day, like classical computers.
00:02:05.760 --> 00:02:07.840
So our classical computers run on bits.
00:02:07.840 --> 00:02:09.520
A bit can be either a zero or a one.
00:02:09.520 --> 00:02:17.120
It's pretty deterministic, and all modern computing is built on manipulating these enormous strings of zeros and ones.
00:02:17.120 --> 00:02:24.960
And so, what quantum computing does is introduce a new unit of a new way to represent that information, to represent those bits.
00:02:24.960 --> 00:02:27.680
And it's called a quantum bit or a qubit.
00:02:27.680 --> 00:02:33.039
And so, unlike a classical bit, these qubits they exist in a state that we call superposition.
00:02:33.039 --> 00:02:41.680
You might have heard about this just from reading, which really means that these qubits can represent a zero and one at the same time until it's measured.
00:02:41.680 --> 00:02:51.120
And so it might sound abstract, but really it just means that in practice, a quantum system can represent many possible combinations of values simultaneously.
00:02:51.120 --> 00:03:03.599
So, really a good way to think about it quantum computing as well is very energy efficient compute for solving very hard problems because it can represent many possible combinations somewhat at you can think about it at the same time.
00:03:03.599 --> 00:03:06.080
And so superposition isn't the only thing.
00:03:06.080 --> 00:03:13.360
There's a bunch of different aspects of quantum physics that we take advantage of, but superposition and entanglement are the two that are the big ones.
00:03:13.360 --> 00:03:22.800
And so entanglement links qubits together in such a way that the state of one qubit is mathematically connected to the state of another, basically whether it's a zero or a one.
00:03:22.800 --> 00:03:26.800
And so that relationship holds no matter how complex a system becomes.
00:03:26.800 --> 00:03:33.039
And it really creates an exponentially expanding uh state base state space as more qubits are added.
00:03:33.039 --> 00:03:39.680
So you can basically explore a solution space much faster than a classical computer using superposition and entanglement.
00:03:39.680 --> 00:03:46.240
But really, that's why quantum computers work well with classical computers and they won't necessarily replace classical computers anytime soon.
00:03:46.240 --> 00:03:56.400
They're not better at everything for most everyday tasks, even things like processing transactions, running databases, sending emails, classical systems are built for that.
00:03:56.400 --> 00:04:02.400
They're very efficient at that, and they probably will remain, remain to be very the most efficient at those tasks.
00:04:02.400 --> 00:04:10.879
But where we look to quantum computing, this area shows promise in solving certain categories of problems that grow exponentially in complexity.
00:04:10.879 --> 00:04:17.600
So, really, quantum computing, I like to say it's not about faster spreadsheets or incremental efficiency gains.
00:04:17.600 -->
Load full transcript
00:04:25.040
It's a completely new way of being able to tackle complex problems that might not be even possible to tackle today with classical computing.
00:04:25.040 --> 00:04:33.040
So hopefully that gives a bit of an intro into quantum computing and why it applies to finance and banking, because we want to tackle these complex problems.
00:04:33.360 --> 00:04:34.079
Yeah, definitely.
00:04:34.079 --> 00:04:38.319
Then there's lots of advantages to be to be taken care of.
00:04:38.319 --> 00:04:41.839
And it seems like we have a big opportunity in front of us.
00:04:41.839 --> 00:04:48.240
And you kind of touched on, you know, why financial services is one of the industries most closely watching.
00:04:48.240 --> 00:04:52.079
But what are some realistic quantum use cases in banking?
00:04:52.079 --> 00:04:54.399
Can we talk near-term, midterm, long-term?
00:04:54.399 --> 00:04:56.720
Like what does the timeline sort of look like?
00:04:57.120 --> 00:04:57.839
Yeah, exactly.
00:04:57.839 --> 00:05:03.680
That's a good place to start with use cases when you look at it from near-term, midterm, long-term.
00:05:03.680 --> 00:05:06.240
Because again, it's an emerging technology.
00:05:06.240 --> 00:05:12.399
It's still growing and it's still becoming more and more mature to be able to put into practice.
00:05:12.399 --> 00:05:17.920
So a lot of people often think about the amount of data that financial institutions have.
00:05:17.920 --> 00:05:25.199
And that is a part of it, why quantum computing is helpful, but it's less about the amount of data and it's more about, again, the complexity of the problems to be solved.
00:05:25.199 --> 00:05:28.399
We want problems with a with tons of solutions.
00:05:28.399 --> 00:05:33.759
They have tons of variables, they have tons of constraints, and the possible solution space becomes huge.
00:05:33.759 --> 00:05:36.160
Optimization problems are a big use case.
00:05:36.160 --> 00:05:43.839
So portfolio optimization, asset allocation, liquidity management, trade routing, overall capital optimization are good use cases.
00:05:43.839 --> 00:05:46.959
There's also outside of optimization, there's risk and simulation.
00:05:46.959 --> 00:05:51.519
So being able to use quantum computers to do better Monte Carlo simulations.
00:05:51.519 --> 00:06:01.199
And actually, that's a really good example of it out of a use case in quantum computing where Monte Carlo simulations, they're widely used to price derivatives as well as measure risks.
00:06:01.199 --> 00:06:09.279
You're running millions of scenarios trying to achieve the most accurate result, basically for a scenario that you can't measure directly.
00:06:09.279 --> 00:06:12.639
So you got to run tons of simulations and you try and average the outcomes.
00:06:12.639 --> 00:06:23.519
You try and see through running millions of simulations where you're most likely to end up in that simulation and hopefully where you'd end up in the real world to be able to kind of predict some of those features.
00:06:23.519 --> 00:06:29.040
And so Monte Carlo simulations, they consume a huge amount of computing time as well as computing cost.
00:06:29.040 --> 00:06:36.639
And so quantum methods have the potential to reduce the number of simulations needed and hopefully still be able to reach the same level of accuracy.
00:06:36.639 --> 00:06:42.240
And not only is it necessarily a benefit of accuracy that you'll get, you might get a benefit of speed.
00:06:42.240 --> 00:06:49.279
You might get a similar result in accuracy, but it used a lot less energy or it got it a lot, got to that solution a lot faster.
00:06:49.279 --> 00:06:59.839
And so when you think about Monte Carlo simulations, for example, for an example, it really would help with accelerating pricing, your capital calculations, your risk measurement processes.
00:06:59.839 --> 00:07:04.240
You could run more detailed models in less time, responding faster to market changes.
00:07:04.240 --> 00:07:08.480
All those types of things really play a large part in quantum computing.
00:07:08.480 --> 00:07:13.439
There's also the cryptographic risk, and that's more of a near-term issue.
00:07:13.439 --> 00:07:16.319
There's a lot of academic research going on into quantum.
00:07:16.319 --> 00:07:22.879
And not all use cases can be put into production today, but they're rapidly approaching that.
00:07:22.879 --> 00:07:26.160
And there are a few that we can put in into production today.
00:07:26.160 --> 00:07:31.199
And a good one, a good example of that that always comes to the top of my mind is portfolio optimization.
00:07:31.199 --> 00:07:35.600
So when we're doing your portfolio optimization today, it can take a long time.
00:07:35.600 --> 00:07:38.399
You might take a long time to have to rebalance those portfolios.
00:07:38.399 --> 00:07:53.839
But with quantum portfolio optimization, we can encode these portfolio decisions, the asset types, the balance of risk and return metrics into these quantum models, and we can tend to see where quantum computers work alongside classical computers.
00:07:53.839 --> 00:08:04.959
So we could implement right now a quantum hybrid model with a quantum classical type of approach, where the quantum component can evaluate a possible portfolio combination.
00:08:04.959 --> 00:08:11.199
It could do that potentially faster, or you could use it to explore a larger portfolio space.
00:08:11.199 --> 00:08:19.920
And so you could use the quantum component to evaluate the possible portfolio combinations while the classical system refines and interprets the results and feeds it to your dashboard.
00:08:19.920 --> 00:08:34.480
So really we're starting to see an intersection where we can use quantum computing as part of a larger process or part of an AI process where we can implement it into an AI agent where the agent knows an optimization needs to be performed, passes that to the quantum computer.
00:08:34.480 --> 00:08:37.440
It then gets
the result from that and continues on in the process.
00:08:37.440 --> 00:08:54.879
So we're probably not, we're not really seeing the replacement of traditional optimization tools right now as it is, but we're starting to see a lot of augmentation to the existing classical tools with quantum optimization, areas where quantum optimization can be applied to be able to derive a benefit.
00:08:54.879 --> 00:08:58.000
Another area in the near term is risk model research.
00:08:58.000 --> 00:09:02.879
So modeling large data sets, complex correlation calculations.
00:09:02.879 --> 00:09:14.000
This is where quantum research has been really focusing on the large mathematical relationships between all the different risk risk variables.
00:09:14.000 --> 00:09:20.720
And so understanding whether quantum tools could enable faster recalculation of exposures, better identification of hidden risk.
00:09:20.720 --> 00:09:24.639
Those are the main areas for risk model research to be able to apply quantum computing.
00:09:24.639 --> 00:09:26.399
Another area is fraud detection.
00:09:26.399 --> 00:09:30.000
So in your transaction networks, they're massive, they're complex.
00:09:30.000 --> 00:09:40.720
We can use quantum computers to be able to recognize some of those complex patterns and relationships that might be difficult for traditional AI or classical systems to detect efficiently.
00:09:40.720 --> 00:09:45.519
I would say in the midterm, we're looking at those faster Monte Carlo type of simulations.
00:09:45.519 --> 00:09:49.519
So running those millions of scenarios a little bit quicker to hopefully get the same results.
00:09:49.519 --> 00:09:52.720
But we're also seeing some advantage in derivative pricing.
00:09:52.720 --> 00:10:04.559
There's been some use cases in the quantum space using quantum to provide better pricing around derivatives and different exotic derivatives that normally would require heavy computational modeling.
00:10:04.559 --> 00:10:11.600
We can use quantum computers to kind of recalibrate these prices more frequently as the market conditions change.
00:10:11.600 --> 00:10:19.840
And then I think there's a lot of use cases in the long term, but the long term really focuses on expanding the use cases to become larger and larger.
00:10:19.840 --> 00:10:31.039
So we start to tackle huge problems as we look to the long term, which might be five to 10 years away, where we're starting to use quantum computers for real-time global portfolio optimizations.
00:10:31.039 --> 00:10:36.480
You no longer have to make so many assumptions about your calculations or about your portfolios.
00:10:36.480 --> 00:10:47.840
You can model those largely, those large global portfolios that include those complex constraints across all your asset classes, all your regions, incorporating constraints around all the regulations.
00:10:47.840 --> 00:10:53.360
And then you could rebalance those as markets change instead of having just a periodic rebalancing.
00:10:53.360 --> 00:11:06.000
And it would really represent a bit of a structural shift in the way we do asset management if we could be able to incorporate all that detail and remove those assumptions that we currently use in classical solutions.
00:11:06.000 --> 00:11:16.559
And really lastly, I would say that the market-wide systemic risk calculations, being able to take in larger amounts of variables and information to be able to apply it to your calculations.
00:11:16.559 --> 00:11:23.600
So I think that's where I would put it for the near use cases, the midterm use cases, and the long-term use cases.
00:11:24.000 --> 00:11:24.559
Very nice.
00:11:24.559 --> 00:11:27.759
And thanks for kind of building that context out for us.
00:11:27.759 --> 00:11:31.600
It sounds like there's a lot happening and there's a lot being done.
00:11:31.600 --> 00:11:35.440
At the same time, I liked when you mentioned the hybrid kind of quantum computing model.
00:11:35.440 --> 00:11:38.639
There's lots happening, you know, where the people can start doing this.
00:11:38.639 --> 00:11:43.840
But one of the things you mentioned was, you know, kind of the encryption risk.
00:11:43.840 --> 00:11:46.960
And is there something banks should be worried about?
00:11:46.960 --> 00:11:50.559
And another thing I wanted to ask was is there a sense of urgency?
00:11:50.559 --> 00:11:52.799
Because things are happening so fast.
00:11:53.360 --> 00:11:54.639
Yeah, absolutely.
00:11:54.639 --> 00:11:57.120
And there is a huge sense of urgency.
00:11:57.120 --> 00:12:02.639
We know that quantum computers can break RSA encryption and elliptic curve cryptography.
00:12:02.639 --> 00:12:04.960
Using Shore's algorithm, we know it can do that.
00:12:04.960 --> 00:12:10.159
And so it's just a matter of time until a quantum computer is available that can do that at scale.
00:12:10.159 --> 00:12:12.000
For now, RSA is safe.
00:12:12.000 --> 00:12:15.519
But the problem is that data can just be collected and stored.
00:12:15.519 --> 00:12:26.799
Storage is cheap, and the bad actors just have to sit on that data until a day arrives, which is called Q Day, when a quantum computer is large enough to be able to run Shore's algorithm to break the current encryption standard.
00:12:26.799 --> 00:12:29.519
So it's called a harvest now decrypt later attack.
00:12:29.519 --> 00:12:39.840
The urgency is there, and it is something that financial institutions should be very knowledgeable about because their data that they store is long-term data.
00:12:39.840 --> 00:12:45.440
Criminals can hang on to that data for a long time and it'll still be valid by the time they're able to access it.
00:12:45.440 --> 00:12:50.559
So there's a huge risk there, and that's the post-quantum cryptography topic.
00:12:50.559 --> 00:12:56.320
Quantum computers aren't the cure to the threat from quantum computers.
00:12:56.320 --> 00:12:59.919
You don't need a quantum computer to implement post-quantum cryptography.
00:12:59.919 --> 00:13:01.759
So I'll leave that with everyone.
00:13:02.159 --> 00:13:02.639
Interesting.
00:13:02.639 --> 00:13:03.360
I like that.
00:13:03.360 --> 00:13:05.519
You know, lots to unpack here for sure.
00:13:05.519 --> 00:13:10.559
But let's move on to there's a lot of hype around quantum, just like there was with AI.
00:13:10.559 --> 00:13:17.759
And I wanted to understand from your perspective, is it realistic or are we still decades away?
00:13:17.759 --> 00:13:20.320
Because you talked about, you know, near-term, short-term, long-term.
00:13:20.320 --> 00:13:21.840
But right now, how do you see it?
00:13:21.840 --> 00:13:22.960
Is it realistic?
00:13:23.279 --> 00:13:23.679
Yep.
00:13:23.840 --> 00:13:26.080
I would 100% say it's realistic.
00:13:26.080 --> 00:13:33.039
So we're approaching an area where we can take advantage of these noisy intermediate scale quantum computers that we have today.
00:13:33.039 --> 00:13:38.879
The goal is to get to a fault tolerant quantum computer where basically we have really good quality qubits.
00:13:38.879 --> 00:13:42.480
We can hold superposition states for a very long time, do long calculations.
00:13:42.480 --> 00:13:45.679
That's where every quantum computing provider is trying to get to.
00:13:45.679 --> 00:13:56.399
And so right now we have noisy intermediate scale quantum computers, which is kind of a middle ground of where we can apply, start to apply these production use cases to solve these business challenges.
00:13:56.399 --> 00:13:57.440
And so it is real.
00:13:57.440 --> 00:14:01.120
We can take advantage of that, especially in the optimization use cases.
00:14:01.120 --> 00:14:04.480
There's very strong use cases in production today.
00:14:04.879 --> 00:14:05.679
Nice, nice.
00:14:05.679 --> 00:14:08.720
Well, that kind of helps us guide where we can go from here.
00:14:08.720 --> 00:14:16.399
But let's say if there's a CIO or CTO or chief risk officer listening, what should they actually be doing about quantum right now?
00:14:16.399 --> 00:14:17.840
Anything actionable?
00:14:18.240 --> 00:14:20.080
Yeah, there's a lot of actionable steps.
00:14:20.080 --> 00:14:24.480
And a lot of it has to do with similar to what you're probably doing with your AI approach.
00:14:24.480 --> 00:14:26.639
You're trying to educate leadership.
00:14:26.639 --> 00:14:29.120
You're trying to start small and identify use cases.
00:14:29.120 --> 00:14:30.960
You're developing the talent around that.
00:14:30.960 --> 00:14:33.120
And we can do the same thing with quantum computing.
00:14:33.120 --> 00:14:43.679
You can leverage a lot of the same frameworks you probably built out in the AI, in your AI practices, where we want to educate leadership and build executive literacy around quantum computing.
00:14:43.679 --> 00:14:48.000
We want to be able to separate that hype from impact that we just talked about, what's real.
00:14:48.000 --> 00:14:54.799
And we want to have the team members within your organization knowledgeable about where quantum can potentially apply and where it could apply in the future.
00:14:54.799 --> 00:14:59.679
And so you would start small, identifying one to two high-value use cases.
00:14:59.679 --> 00:15:05.759
If you're really interested, you can build some partnerships with some universities, some vendors, some CGI.
00:15:05.759 --> 00:15:11.600
And then you want to be able to build out those use cases by developing the talent to be able to do that.
00:15:11.600 --> 00:15:23.919
So upskilling your teams, find those quantum aware and quantum enthusiastic team members within your organization and get them the tools and get them the access needed to be able to start building that talent.
00:15:23.919 --> 00:15:28.960
And then really, it's about it's not about buying hardware.
00:15:28.960 --> 00:15:34.320
It's about preparing your architecture and your teams to be able to utilize the hardware that already exists.
00:15:34.320 --> 00:15:35.919
And we can do that today.
00:15:35.919 --> 00:15:40.879
As a risk officer, you're probably very interested in the cybersecurity risk.
00:15:40.879 --> 00:15:44.960
And so beginning with your PQC or post-quantum cryptography planning as well.
00:15:44.960 --> 00:15:52.399
Really comes down to the education, the starting small, developing talent, and um going from there within your organization.
00:15:52.399 --> 00:15:54.320
That's what I would be doing for quantum today.
00:15:54.320 --> 00:15:57.279
And that's what we're doing to help a lot of our awesome.
00:15:57.360 --> 00:16:06.080
I like how you kind of started with with knowledge and sharing and educating leaders and tying in with how the AI hype cycle panned out as well.
00:16:06.080 --> 00:16:07.919
But thanks very much, Curtis.
00:16:07.919 --> 00:16:08.799
This was excellent.
00:16:08.799 --> 00:16:10.399
I think we covered a lot today.
00:16:10.399 --> 00:16:12.480
Thanks so much for joining us today.
00:16:13.200 --> 00:16:13.840
Yeah, thank you.
00:16:13.840 --> 00:16:14.720
Happy to be here.
00:16:15.679 --> 00:16:16.399
Thanks everybody.
00:16:16.399 --> 00:16:17.759
See you at the next one.