There’s no doubt about it. Funding AI can be difficult, even for BFSI firms, due to the high levels of cost and complexity associated with such initiatives. In an effort to shine light on that challenge we sat down with ADA Advisory Board member, Adri Purkayastha, BNP Paribas Group’s Head of AI and Digital Risk Analytics, to share his perspective on the topic.
Keep in mind that Adri will also be presenting on “Advancing Cloud-Based Data-Driven Decision Making” at our upcoming AI BFSI virtual event taking place February 1-2, 2022. REGISTER NOW to reserve your spot.
Seth Adler, Editor-In-Chief, ADA:
On cost structure, there are those that have moved from CAPEX to OPEX. There are those who have not but this is a trend we’re seeing. So how does one make that migration so that one can actually take advantage of cloud-based architecture in a true way to not only scale the project, but to scale the organization?
Adri:
That's a very good question, and it's actually philosophical, in a way, in my opinion. So when you think about CAPEX, if you think about, let's say, private equity industry. CAPEX is essentially something below the line item and essentially that can be amortized, or in some way, the cost can be gotten rid of as a CAPEX investment.
But when you think about OPEX, which is mostly cloud services, which is consumed, the key challenge is, you are trying to optimize for low OPEX. And if your OPEX is low, how do you achieve that? It's through good cloud engineering. And you have the best approach to cloud cost management, to services selection, and overall governance around cloud services. If you don't have, then you have very large ongoing costs, and that can lead to failure.
So if you're moving from CAPEX to OPEX, it's not just an opportunity to re-engineer your cost structure, but also to re-engineer your cloud infrastructure or actually engineer it properly, because cloud native using the correct services, and hence, knowing what to use, removes the dichotomy.
Whether OPEX is better than CAPEX? It is, in some cases, more so when you have optimized it properly.
Seth Adler:
100%. But what if I'm getting pushback? Let's say the finance department is saying. "Listen, Seth, we're not doing that. Sorry. Sounds great, totally hear you, love the idea, but no." From the CFO's perspective, I'm obviously not making the argument well enough. You've just made one pitch it. So how would you come back to that?
Adri:
Let's imagine you give your kids a distinct amount of money they can spend on a toy or product they want. They’ll then take that money and do something out of it which is valuable is to them. And then to you, you see joy as the allocator of capital. And if, in the project, in the case of, let's say, OPEX, if you allocate certain amount of capital, so you have a threshold above which there is essentially a trigger, and there has to be an exception. And by doing so, you force an upper floor, but within that leeway, the project team or the product team may achieve success.
And then, obviously, from a net present value standpoint, that success is too big of a success to that floor that you had set up. So setting a floor is good, but you have to be appreciative that the flow that you've set up should be enough for your team to achieve something reasonable.
Or the other way to do it is, you allocate a lot of the capital with CAPEX and think that there will be something that you'll get after a few years, but you may not get it, and then you have a net loss. So I think with cloud services, you can use best-of-grade tools which would be very cost-prohibitive to build, probably impossible for a small organization. And then you pay for a service. If the service is successful, then you keep on paying for it. So you capture some of the value, create some of the value capture.
Seth Adler:
Let me make sure I understood that. So on the one hand, if we kind of attached the ceiling to the floor, we are limiting the potential height of the ceiling. The floor is dragging down our ceiling, right?
Adri:
Exactly
Seth Adler:
That's on the one hand. On the other hand, if we just give it one bracket, we might lose out completely. If we stay in CAPEX for three years, we might find ourselves out of business.
Adri:
Yeah, because you would probably use something which is cost-prohibitive, and hence, you wouldn't use it. And hence, you would be probably going to a race and turning out with, let's say, a horse car, where somebody has brought a Lamborghini.