On a recent earnings call, Mastercard CEO Michael Miebach talked about the need to “enhance and expand our value-added services, such as in data analytics, fraud, and cybersecurity, particularly as we further embed AI into our products and services.”
One of the people leading that charge for Mastercard is Global Head of Research and Development Rahul Deshpande. The Purchase, N.Y.-based payments giant chalked up $27 billion in revenue in 2024, but it signaled this month that it expects slower revenue growth between 2025 and 2027.
We spoke with Deshpande to explore how Mastercard secures transactions, leverages AI for fraud prevention, and identifies trends shaping the future of commerce.
This interview was part of our 2024 research report, Creating New Value in Large Organizations: What It Takes.
• • •
Give us insight into your role at Mastercard, and an overview of the company’s priorities?
Over the last few decades, we have evolved into a technology company with some value-added services that are outside of our card portfolios, in networks, in fintech, in cybersecurity, in data, and more. In a nutshell, we play a multi-faceted role as network operators to ensure that whenever and wherever a purchase is made, that transaction is safe and secure for the consumers, merchants and the financial institutions. …Our priority is to help facilitate the transfer of the value between the people who want to exchange it. I lead the global function of research and development at Mastercard for an umbrella unit called Foundry. Collectively, we are responsible for Mastercard’s innovation agenda, which means trying to figure out and anticipate the changes in the world of commerce.
We report in to the Chief Innovation Officer for Mastercard, Ken Moore.
How do you, with your software and MBA background, discuss AI in a nuanced way, especially given its broad use today?
AI has been around for a while; it’s more than 70 years old. It has been there since the advent of computational technology… At the end of the day, it’s about data analysis and probability. What’s different [with AI] right now is the advent of the computational power, as well as the data that’s around us. So being able to use generative AI, for example, to look at the unstructured data [we have]. What I’m excited about is there’s a confluence of these advancements in AI, as well as computational power and data technology, that are coming together right now. The applications [are increasing] in different areas, including finance, retail, healthcare, education, and payments.
How do you think about leveraging AI to stay ahead of those who may use it for increasingly sophisticated fraud?
You’re right — with the sophisticated technologies come the sophisticated threats. This technology can be used for both good and bad. One of the key stats we have is that the global cost and lost revenue from cyber attacks are estimated to reach about $10.5 trillion by 2025 — that’s coming from Cybersecurity Ventures. And, we know from IBM, who measured in 2023 that the average cost of a data breach is just under $4 million. These are frightening numbers. If we don’t make security the foundation of every product and service, trust — and therefore innovation and opportunities, especially in commerce — would be lost. For us, that means that we ensure that security, integrity, and trust in the 143 billion transactions processed globally last year is accounted for.
How do we address some of these challenges? In 2018, we invested in excess of $7 billion into cybersecurity capabilities, we have contributed to launching more than 20 cybersecurity-based startups, and we measure the cybersecurity posture of every financial institution and 14 million merchants every 10 days. Now, we have GenAI, so we’re integrating it into this product called Decision Pro, which looks at every transaction with AI models. We think we’re seeing the boosts of product fraud detection rates by an average of 20 percent, or as high as 300 percent in certain instances.
Let’s touch on some other trends that you and your colleagues there are thinking about in 2024.
We have the Mastercard Signals report [which covers] the latest trends that are transforming our industry in the world of commerce. Recent issues [explored] the future of payments, commerce, generative AI, and the latest emerging trends. We looked at about nine different technology trends, starting with computing and spatial interfaces.
Spatial interfaces — Apple popularized this term — the idea [has been] there in popular culture for a while. You may have seen this in movies like Minority Report. Unlike normal computing, where you’re using the mouse and keyboard and screen, spatial computing tracks your hand gestures and eye movements to engage with virtual objects.
We think that spatial computing technology is going to mature over the next several years, [and] integrate into our daily lives. It is going to transform education, healthcare, manufacturing, entertainment, and, of course, shopping.
Shoppers return 5 to 10 percent of what they purchase in the store, but about 15 to 40 percent of what they buy online, and it costs $30-$33 to process a return on a $50 item.
Think about this world where you put on glasses, select a shoe from a virtual rack, and try it with the built-in LiDAR and image processing. That itself is justification for retailers to adopt this technology. Shoppers return 5 to 10 percent of what they purchase in the store, but about 15 to 40 percent of what they buy online, and it costs $30-$33 to process a return on a $50 item. Then, 70 percent of returns are because of the size or style issues, so this experience is going to help out.
Spatial interfaces are just one of the trends we looked at. We [also] looked at network technologies like 5G, 6G, and ultra-wideband, [and] data services [like] tokenization. Tokenization replaces the insensitive information of a trade card number with a unique random code, which we call a token, so it safeguards you from things like data breach, as the random numbers mean nothing to the criminals. But tokenization can also represent any digital assets, like stocks or intellectual property or other sources of value on a blockchain or a network to make it more transferable, tradable and secure. You can also program those tokens; you can create smart contracts.
And then finally, we’re looking at the data integrity. Because of the deep fakes, we need to start to look at the data provenance, for example. There are technologies that are allowing us to look at the data provenance from the get go — like homomorphic encryption, or polymorphic encryption that helps you compute on encrypted data.
How do you approach the challenge of securing cryptography in a future where quantum computing could be used by adversarial entities?
Quantum computing is really interesting. It tries to solve what we call the “NP-hard” problems. These are the problems that are really hard to solve using conventional computing — like optimizing loyalty rewards for the right consumers. These are really hard to solve in a linear computing way. This is where quantum computing has some promise. On the flip side, because the same technology is available to the criminals, [quantum] can also be used to solve for cryptography. So we recently made our encryption safer… quantum-proof, basically. That’s one of the use cases that we first looked at.
Quantum is probably about three to four years away to enter the mainstream. But as the Foundry, we’re looking at it as an emerging technology that could transform [many things related to our business].
Are there ways that you try to plug into basic research and academic work that’s happening at universities here in the US or other countries?
None of these things can be done by yourself, right? Most of the things are happening in universities. They’re also happening with a lot of the startups that we need to partner and engage with. We have a very robust startup engagement program called StartPath. That’s a principal way that we engage with the companies who are working in these advanced fields. We also have partnerships with a lot of universities around the world [that are] specialized in cybersecurity or consumer research, mixed reality, quantum computing, or AI. We leverage all of that. We also have an in-house team of people that are looking at research papers… and then partnering with universities, as well as partners like startups and [other] companies, to devise new products for Mastercard.
I would look more at the startup ecosystem right now, versus looking at the [R&D labs of large] companies as my inspiration.
Are there role models for really productive R&D in today’s world, or R&D taking a really big swing at hard problems?
I would look more at the startup ecosystem right now, versus looking at the [R&D labs of large] companies as my inspiration. Our goal is to look at the new technologies, changes in the consumer behavior, and then apply them to products and services. And, who does that?… These new startups, the new companies coming up, finding new models or new products that use the new technologies.
Conventionally, it’s more about the product development process more than anything else. A startup will go through framing up an idea, building a prototype, going to customers and asking, “Is this the right thing for you?” Then building the first market test, making sure that it actually works, and then scaling for commercialization.
You go through these phases: framing, concept, prototype, market test, and commercialization. That’s the regimental process, a process that I think is very beneficial for R&D. You go through the same process in R&D, from building up a business case, building a prototype, going to the product owners, or customers, [and asking], “Is this a good idea?” Then going through the first market test and commercialization.
Doing that in a small, compressed way optimizes for the right ideas and quickly filters out the bad ideas, or [those] that are less likely to be desirable. I look at the startup space more than anything else. Obviously, companies do this really well [and] that’s one inspiration. But more than that, in today’s world, R&D is more applied; we’re just looking at new product development in the end.
