6
Unanswered tech questions: cloud vs. on-premises and build vs. buy

From technological infrastructure to sourcing choices, banks must tailor their strategies to fit their size, resource, and AI ambitions.

Unanswered tech questions:
cloud vs. on-premises and build vs. buy

As banks use AI more widely and technology matures, the required computing power grows drastically. This raises the question of whether to use capacity in the cloud or on-premises.

Some prefer the scalability and flexibility offered by cloud. “We need the scalability cloud provides to run LLMs,” says Scotiabank’s Matthew Parker-Jones. “You simply can’t get this from traditional data center technology or even private cloud technology. You need to partner with the hyperscalers.”

We need the scalability cloud provides to run LLMs. You simply can’t get this from traditional data center technology or even private cloud technology. You need to partner with the hyperscalers.

Matthew Parker-Jones, Head of Product for Global Transaction Banking, Scotiabank

Others have adopted a hybrid approach, blending cloud with the security and control benefits provided by on-premises. “Our hybrid approach mixes on-prem with public and private cloud,” explains RBC Commercial Banking’s Vivekanandan. “We’ve built our own GPU infrastructure that allows us to build, deploy and maintain AI-powered banking applications. This provides us with enhanced security, privacy and sovereignty. That’s vital because trust is central to our relationship with customers.”

In reality, the significant upfront cost of building large-scale computing infrastructure means it’s only viable for larger banks. Banks need an approach that reflects their resources, AI ambitions and existing cloud strategy.

Our hybrid approach mixes on-prem with public and private cloud. We’ve built our own GPU infrastructure that allows us to build, deploy and maintain AI-powered banking applications. This provides us with enhanced security, privacy and sovereignty. That’s vital because trust is central to our relationship with customers.

Niranjan Vivekanandan, EVP and Chief Operating Officer, Commercial Banking, RBC

Decision-making rests on careful assessment

They will have to start by conducting a thorough assessment of the costs and benefits of each approach. Advances in cloud offerings mean that many of the perceived risks are manageable: the cybersecurity credentials of cloud providers are likely superior to those of any bank, and cloud infrastructure can be ringfenced to minimize the risk of data leaks.

But other risks are emerging. The perceived cost advantages of cloud are increasingly being challenged as providers charge more for access to the powerful infrastructure that’s necessary to train and run LLMs. Some banks are also wary of having a large proportion of AI applications powered by cloud services provided by a small number of vendors — or even just one.

Another question is whether to buy or build. The best approach depends on the maturity of third-party solutions, the expertise necessary to develop capabilities internally, and whether the capability in question is critical to the bank’s competitive advantage.

“The question of buy or build depends on the use case and the maturity of vendor solutions,” says BDO Unibank’s Saw. “We’ll buy tools that, for example, enable conversational AI, because we’d never be able to build that and the solutions are mature. But we’ll build and implement tools that house sensitive data because we don’t want that stored elsewhere.”

Banks often underestimate the time and effort they will need to customize and orchestrate vendor tools. This is especially true for complex processes such as credit, where there are multiple handoffs between distinct processes and tools.

The question of buy or build depends on the use case and the maturity of vendor solutions. We’ll buy tools that, for example, enable conversational AI, because we’d never be able to build that and the solutions are mature. But we’ll build and implement tools that house sensitive data because we don’t want that stored elsewhere.

Cindy Saw, Head of IT Group, Data and Analytics, BDO Unibank

Key actions

Objectively assess the risks and benefits of cloud and on-premises approaches and challenge any assumptions that might be out of date.
Explore whether the current balance of cloud and on-premises is suited to the rapidly growing requirements of GenAI.
Refresh the balance of build vs. buy as use cases evolve and vendor solutions mature.