Cloud computing is part of a three-pronged strategy for Citi’s consumer business, along with mobile applications and opening APIs to fintech partners. Cloud is the enabler that consumers don’t see – but it is also the part that the bank wants to keep internal.
Sanjeev Mehra, Singapore-based managing director and head of global product development for Citi’s global consumer technology, told DigFin that the bank is experimenting with cloud computing within the Monetary Authority of Singapore’s regulatory sandbox. But it continues to do so without actually using external cloud services.
Cloud computing is the leveraging of internet-connected computers to collectively provide data storage, management and processing. Microsoft and Amazon are among those technology companies providing cloud computing as a service to financial institutions and other enterprises.
Mehra says the cloud is an enabler, in that it allows the bank to move applications to lower-cost infrastructure as the number of apps explodes. It is also a cost arbitrage, as cloud services can be cheaper than relying on on-premise servers. Outsourcing to vendor cloud networks is expensive. “But at our scale, the private cloud is powerful,” Mehra noted.
There are also regulatory issues. Financial regulators are not yet comfortable with banks putting client data on public cloud networks.
Mehra added that the friendliest regulator in Asia Pacific when it comes to letting banks use cloud computing is Singapore’s. Among the sandbox trials at the Monetary Authority of Singapore is experimenting with cloud computing. Such experiments could remain ongoing for an indefinite period of time, with no visibility as to when regulators will be fully confident around data-sharing issues. That’s one reason why Citi is relying on its worldwide network of private servers, which is big enough to provide cloud-computing support, Mehra says.
The cost of staying private
Matt Quinn, chief technology officer at Palo Alto-based Tibco, a systems integration vendor, says more financial institutions are moving to cloud computing – with many now adopting hybrid solutions that mix private networks (or on-premise, as he terms it) with public services. He expects it could be a while before banks or other enterprises fully embrace the cloud. “After forty years of proclaiming the death of mainframe computers, clients still run on them,” he said. “Cloud is not seen as secure, and there are regulatory issues.”
Nonetheless, he expects many financial institutions will use it because they will find it increasingly expensive to maintain the internal level of high processing required for periods of acute computing needs, or ‘surges’.
For example, European regulators are raising requirements among banks for Monte Carlo-type risk simulations, demanding bigger calculations going back longer in time, and incorporating more financial instruments. This is forcing banks to do short-term computations more frequently.
“Surging is expensive,” Quinn told DigFin. A cloud computation by a large corporation can cost tens of thousands of dollars, and surges may be required several times in one month. “But that’s cheaper than the millions of dollars of capital outlay up front to enable peak-level processing all the time,” he said. (Tibco’s services include connecting clients such as banks with cloud service providers such as Microsoft or Amazon.)
Mehra says the economics of cloud computing are not obvious. Banks look to cloud solutions to reduce the cost of their infrastructure, but that benefit is eroded by the cost of operations and governance. The real value in cloud computing is using short surges to support product development and marketing.
For example, a new product campaign may drive 20 times more people to visit the bank’s website or app, and during these times the bank needs to rely on a distributed network of computing power to avoid a crash. Such events are increasing in frequency now that Citi has opened APIs (application programming interfaces) to third-party fintech developers, and is embracing agile development models.
“Cloud is an evolution,” Mehra said, calling it the latest iteration of processing that began with mainframe computers, moved to midrange computers and then to desktop personal computers.
But the bank is careful not to confuse computing with a business capability. “Where’s the ‘so-what’ for the customer?” he notes.