Cloud computing is nothing new for the financial sector. Over the last half-decade, the largest banks in North America have been steadily migrating certain parts of their systems to operate in the public cloud through platforms like Microsoft’s Azure, looking to take advantage of hyperscale computing power.
Late last year, Citibank’s global head of bank research Ronit Ghose discussed the idea of large banks moving AI and other platforms to the cloud, saying “the banking industry must react and evolve to not get wiped out by an extinction event such as digital disruption.” He likened banks to black and white televisions—operating on a dated, yet resilient infrastructure that doesn’t look flashy but works well and results in fewer breakdowns and outages than the modern competitors. In the case of banks, these new competitors would be fintechs.
Enterprise companies are typically looking to cloud solutions for a few key reasons, the most important of which is pure computing power. Cloud platforms are able to handle much more data at once when compared to traditional physical data centers, with the added advantage of being able to fluctuate and handle massive usage spikes as well. The question then revolves around when banks should finally pull the trigger and move to the cloud and what kinds of platforms or services deserve to be migrated first.
TD Bank makes a move
A recently announced partnership between TD Bank and Microsoft might help to answer some of those questions. TD will use Microsoft’s Azure as their cloud foundation to build tools meant for accessing data and AI resources. The two have worked together in a cloud capacity before, as five years ago TD’s employees began using Office365, which serves as a very introductory move to public cloud infrastructure. Since then, TD has also started using Intune, a Microsoft cloud-based mobile device and operating system management platform.
The obvious first advantage cloud computing provides is speed. Before the adoption of any cloud services, financial institutions traditionally ran on mainframe services—large bulky computer networks localized within a few data centers owned and operated by the bank itself. A switch to cloud computing offers flexibility these mainframes could not possibly achieve.
“Microsoft’s cloud allows us to be foundational in giving us a platform to be more agile and innovative,” says Jeff Henderson, the CIO of TD Bank. “It gives us the tools to continue to build strong capabilities for the benefit of our customers, and not just through digital channels but every channel.”
Some of cloud computing’s other advantages from a bank’s perspective include low-cost data storage and disaster recovery, as well as a deep investment in security. Microsoft has over 3,500 security professionals dedicated to ensuring the Azure platform remains secure, a number most enterprises simply cannot match, even with vast resources and large employee numbers.
Taking these into account, TD’s shift to cloud revolves around one proposition: Providing legendary customer service. This is actually the driving force behind most cloud migrations, as enterprise companies need better ways to handle their increasingly massive data stores while finding better ways to personalize services and build new features.
“Cloud gives us the tools to build strong capabilities not just through digital channels but every channel,” says Henderson.
“The challenge becomes that the definition of customer service is changing more rapidly than it ever has before. We’re living in a world where for us to keep pace, we need to be more agile, and more agile with respect to how we innovate in that customer-centric space.”- TD CIO Jeff Henderson.
A “good chunk” of what TD will first leverage with this partnership, according to Henderson, will be capabilities aimed at improving the employee experience. With close to 90,000 employees, moving certain services to the cloud will streamline a lot of work. That is what already happened with Office365 and Intune, as TD can now use apps like Word and Excel in the cloud.
For a highly-regulated industry such as banking, these services are a mere introduction to the cloud. The real advantage comes when integral platforms are moved—but these shifts also introduce the biggest risks.
Does core banking belong on the cloud
There is a long list of reasons banks may be skeptical of a vast cloud migration, including concerns with privacy and the sharing of personal information with third-party companies, data sovereignty and the laws surrounding how data is managed in one country versus another, and different ways the data itself will be accessed by the hosts and possible partner vendors. Banking platforms with heavy use as well as anything with customer data and insights are usually the last to be migrated.
“It’s more than just infrastructure as a service, and platform as a service, and the finished software as a service. When you take all those together, it’s the ability to look at new ways to approach their business model,” says Chris Barry, a VP of the enterprise and partner group at Microsoft.
“We’re well on the journey of enterprise companies broadly understanding the capability and capacity of public cloud infrastructure. With advances in security and regulatory compliance, they’re realizing hyperscale compute isn’t a flash in the pan, it does create this platform on which to deeply consider those new business opportunities.”
For TD, one of the main new business opportunities cloud computing offers is an upgraded version of TD Securities, an investment arm that offers capital market services. Henderson describes how hosting this platform on Azure is a perfect use-case for the bank as they receive the benefits of a standardized stack and strong computer capabilities. The first apps banks like JPMorgan ran in the cloud were similar to this launch, such as apps in wholesale trading and risk modeling.
By running risk and pricing calculations through Azure, banks can take advantage of the extreme variability required for securities transactions. This is one of the biggest advantages of cloud computing from a bank’s perspective, as the publicly hosted infrastructure allows for large spikes of computing power to be handled quite easily—something traditional mainframe data centers cannot easily accommodate.
“It’s an effective use case for the cloud from an efficiency perspective,” says Henderson. “Rather than hosting the amount of hardware it would take in our own data centers, we have access to more pure infrastructure horsepower to speed up those kinds of calculations or to perform calculations we wouldn’t otherwise be able to do. Some of those capabilities are now already running, and they are working well.”
Henderson is quick to point out that the reasoning behind moving certain platforms to the cloud and leaving others on a mainframe is all about placing an emphasis on what the bank’s customers interact with on a daily basis. These “systems of engagement” are the ones like TD Securities, as well as the mobile and web-based platforms customers use to simply log in and check balances or pay bills.
“We’re living in a world where for us to keep pace, we need to be more agile, and more agile with respect to how we innovate in that customer-centric space.” – Jeff Henderson
That also means that TD’s core banking platform will remain on a mainframe for the foreseeable future. This back-end system performs essential actions such as service loans, deposits, and the opening of new accounts. A very small percentage of banks have moved these mission-critical platforms to the cloud. Most of these platforms still reside in owned data centers or are hosted through carefully-guarded private clouds. A 2018 study from Adobe found that just seven per cent of financial institutions have implemented a cloud-based tech stack.
“When you look at the problems we’re trying to solve, core banking apps are not typically the ones we’re looking to get the most agility out of, at least in the short term,” says Henderson. “Most of the systems that run on our mainframe are those backend core banking platforms. They do what they do very well on the mainframe. They’re not going to run on the mainframe forever, but they’re not my top priority right now. We’re trying to drive agility in front end systems.”
Speed to market is a top priority for banks like TD. Mobile and web apps are improving exponentially and updates must be consistently rolled out. This kind of agility allows banks to rapidly partner with fintechs and outside organizations, or bring in new talent and develop solutions at a pace that was once inconceivable. TD is doing just that by acquiring companies such as Layer 6, an AI startup that drives deep personalization for customers. By moving certain services to Azure, TD can roll out Layer 6-enabled features at a much quicker pace and let the talented AI startup fully utilize the seemingly-endless amounts of data points the bank has access to.
“It’s about speed and cost,” explains Henderson. “TD Securities lives in a world where the faster you can do calculations, and the more accurate they are, it provides more business benefits. Advanced analytics and AI will be running models on Azure Cloud. Having the volume of data we have and the access to computing capabilities on Azure, we’ll be able to get the leverage of our AI talent internally and produce what we think will be a leading personalized approach to all of our channels.”
Down the line, moving core infrastructure to the cloud will also allow enterprises to better prepare for and adopt nascent technologies such as blockchain integration. TD competitor RBC is already experimenting with ideas that leverage blockchain technology to manage credit scores. These kinds of advances will only be able to thrive in cloud environments—ethereum co-creator Joseph Lubin said that bolstering the power of blockchain networks with cloud computing will “be of great benefit to the scalable adoption of emerging decentralized systems around the globe.”
Due to the nature of the sector, banks are a bit slow to the cloud world. There are a few industries that exist already almost entirely on the cloud. A number of Microsoft customers run their entire SAP instances inside Azure according to Barry, and he also points to retail, energy, and insurance as a few industries that have been able to successfully migrate critical platforms. For example, Johnson & Johnson has shifted over 85 per cent of their applications to the cloud already
From the perspective of the cloud vendor, Barry often finds that the customers who contact Microsoft for cloud hosting often have an intense desire to realize their own data estate, which is the massive bulk of raw data an enterprise creates. This data is typically siloed and in disparate functions which often leads to seams in customer experience. A cloud platform can rectify these seams by applying logic and reason to rationalize that deep data trove. From there, the next step is to recognize opportunities and go to market in new ways.
The only real inhibitors to moving each platform to the cloud, according to Barry, is existing investments in data centers. Massive enterprises are have amortized a legacy investment into their owned physical data centers, even while they recognize that a move to the cloud is a sound investment. You may not want to move every service until the current solution is truly end-of-life. And highly-regulated verticals like banking need to realize evolutions in the space and take it slowly, according to law, “But that work will never be done,” says Barry. “It will always continue.”
How to prepare for a shift to cloud
One of the more interesting aspects of moving massive enterprise platforms to the cloud is how those companies handle the transition. Obviously, it’s not as simple as flipping a switch and moving everything in one fell swoop. A lot has to be done in order to fully prepare large organizations, and there is never really one inflection point companies identify that shifts their mentalities away from mainframes towards the cloud.
Henderson says that four years ago, TD was not ready to consume the public cloud. To get to that point, they had to first build an internal private cloud. The bank has been running that private cloud for the last three years, and have up to 200 apps running on it at any given time.
This solved a lot of challenges for TD, the first of which was acquiring the right talent to successfully manage cloud infrastructure. The group that worked on the internal cloud designed principles, the first of which was to not sacrifice the bank’s risk appetite. After building apps on the private cloud and building a capable team, Henderson says the bank eventually “got to a point where we understand all the risks, can manage them, and rolled out early uses cases,” and that was the “genesis of the partnership.”
Finding the right talent is a big deal for banks adopting the cloud. Earlier this year, JPMorgan announced they were building a cloud engineering hub in Seattle, minutes from major cloud players Amazon and Microsoft. Finding experienced cloud infrastructure talent is vital. Although JPMorgan is looking to build their own systems and host data in dozens of centers, the need for talent to prepare for that kind of shift is felt across the industry. That might lead to the big banks following suit and looking to open up shop in Seattle, or at least recruit heavily from the area.
“The important part is the ability to understand the scale that cloud providers operate at,” JPMorgan’s head of cloud, platform and app security Todd Hrycenko told Business Insider. “It’s hard to acquire that skill set independently.”
Microsoft also works to enable cloud preparation for their clients, even when it comes to understanding costs. Of course, Barry points out Microsoft is not in the business of budgeting and running balance sheets for their clients, but he also notes a few interesting points around how their clients come to cloud solutions from a purely monetary perspective.
“Depending on the industry and customer, they have to broadly reassess how they fund projects and move from a world where they are paying multiple millions to build data centers, and shift that to an opex environment for subscription cloud services,” says Barry.
“If a client has a fiduciary liability, there’s a degree of care and mindfulness that goes into those shifts. We work with customers to establish a cloud governance approach, so someone like TD can develop what we might call an enterprise control plain to help them think through and realize which platforms do make sense for the cloud across a number of criteria.”
Another major factor banks must consider is whether or not to take a single or multi-cloud approach. Enterprise companies can opt to host their data with a single provider, or spread it across the top three players (Google, Microsoft, or Amazon) or even opt for other alternatives such as NTT and ServiceNow. A majority of banks (up to 60 per cent) remain unsure as to which option to choose. Some companies, such as JPMorgan, are tapping startups like Snowflake to manage their multi-cloud service, allowing them to seamlessly transfer data between each of the three big players.
“One of the things that has held banks up from cloud adoption for the last several years—they have been trying to figure out how they could do it in an abstract way and how they could try all the cloud providers so that they don’t look like they picked a favorite,” Anthony Skipper, the founder and CTO of Galactic Fog, a firm that specializes in multi-cloud services, told Business Insider. “If you pick a favorite and you’re wrong, you’re fired.”
Regardless of how quickly or slowly major banks embrace the cloud—and which clouds they choose—it will happen soon. Henderson knows this but is still realistic about how TD will adopt new practices and shift where their core platforms are hosted.
“It’s about being pragmatic as we go, and it will be even more controlled as we move from here,” he says. “We have a saying we use for our strategy here. Crawl, walk, run. Take it carefully and understand what we’re doing. Our approach is the right approach right now.”