Harnessing AI to optimize applications and services is crucial but building the infrastructure is equally important
Tue 11 Mar 2025 // 09:16 UTC
Sponsored Feature As an industry, financial services is accustomed to big numbers.
It has been estimated that the FS sector – which includes banking, insurance, investment, lending and other businesses related to wealth allocation – would reach a combined value of $33.54 trillion by 2024, growing at a CAGR of 7.7 percent (on 2023). That accounts for about 31 percent of the total world economy according to some calculations.
The future's looking just as rosy. Analyst Growth Market Reports expects global FS market worth to surpass $58.69 trillion by 2031, growing at an even healthier CAGR of 9.7 percent.
Building the IT strategy to enable such growth gives rise to another statistic almost as mind boggling - the totals for the vast amount of data these growth trajectories will generate.
IDC has predicted that overall volumes will grow at their own CAGR of 21.2 percent to 221,000 exabytes by 2026 for example. Much of that information will have to be stored, interrogated and analyzed, rather than just archived away, which means making sure it is easily available via low-latency connections. Latency is just one critical element that demands attention – others include data sovereignty, data privacy and cybersecurity.
Data growth forecasts specifically related to the FS industries may vary on detail, but concur on two things: data assets are amassing at an accelerating rate, while the information they hold is becoming increasingly integral to establishing and maintaining a competitive edge.
So too is the use of innovative AI-powered applications and services to retain and attract new customers in a digital age. Mindful FS IT planners must therefore standardize their strategies data-driven initiatives while simultaneously building-out their range of use-cases designed for AI systems.
IT leaders in FS must act fast if they're to retain a competitive stance, urges Colin McLean, Chief Revenue Officer at Digital Realty. "These days FS is shifting bigtime to AI," he says, "and as they are discovering, AI calls for a fundamentally different approach to IT provisioning than the conventional pre-AI technologies on which they have built their success thus far."
According to a survey of FS firms by the Bank of England the highest perceived AI use-cases are in data and analytical insights, and combating fraud and money laundering. The areas with the largest expected increase in benefits over the next three years are operational efficiency, heightened productivity, and reducing their cost base.
AI-assisted crime counteraction is already benefiting each of these areas. Workers handling manual fraud reviews can now be helped by Large Language Model-based assistants running Retrieval-Augmented Generation (RAG) on the backend to tap into information from policy documents that expedites decision-making over whether cases are fraudulent – speeding-up the process and cutting losses.
AI-ready datacenters
While the FS sector is alert to the game-changing gains its investments in AI could win, it has been slower in realizing how these opportunities come with a substantial datacenter overhead in terms of provisioning and supporting the IT necessary to run AI workloads.
That trend is confirmed by Digital Realty's 2024 'Global Data Insights Survey' of IT decision-makers, in which respondents cited data storage constraints as a foremost pain point. Sixty-four percent of those polled by the company say that a lack of data storage provision required to hold the massive datasets that AI requires is their 'biggest hurdle' to adopting an AI strategy.
"Few would query that AI utilization is driving another new wave of innovation across verticals, and arguably FS currently has a more compelling want for AI than other sectors," points out McLean. "For instance, it has to take a very encompassing view as it addresses the challenges that regulatory compliance has put on its doorstep. FS firms' response has to be holistic, not monolithic. Speed of adoption is essential.
While FS companies need to deploy AI quickly, they face a double predicament in that while the availability of AI-ready datacenter facilities will become outpaced by demand, new datacenters are also notoriously slow to come online.
"Getting planning permission to build-out existing datacenters can drag on," explains Maclean. "The booming technology industry in certain regions of the US, for instance, has resulted in extra demand for energy, and project wait times for new utility power can be five years in some locations."
A study by McKinsey & Company suggests that global demand for datacenter capacity could rise at an annual rate of between 19 percent and 22 percent between 2023 and 2030. To avoid a demand/supply deficit, at least twice the datacenter capacity built since 2000 would have to be constructed in less than 25 percent of the time, McKinsey suggests.
Ready for AI-ready
Demand for AI-ready datacenter capacity is the main driver of this potential deficit. McKinsey's analysis suggests that this will rise at an average rate of around 33 percent per-year through to 2030. This means that by 2030 around 70 percent of total demand for datacenter capacity will be for facilities equipped to host AI workloads.
This is the inception point at which Digital Realty is globally resourced to step up to help customers deploy fast. The company says it anticipated this new wave of demand for public and private AI infrastructure, and has been preparing for it for quite some time. That says Maclean, gives it an unmatched advantage.
"The company has long seen datacenters as foundational to full-term digital transformation and AI adoption across the enterprise," he says. "We're already able to support high-density workloads on CPUs or GPUs – 50 percent of Digital Realty datacenters worldwide are now fully 'AI-ready'."
Integral to Digital Realty's view of the new AI-driven requirement is its concept for the management of DataGravity. Data Gravity refers to the idea that data has mass, and as its size and importance grow, it becomes increasingly difficult to move or replicate.
"Data Gravity is the attractive force caused by enterprise data creation and exchange, drawing applications, servers and other data," McLean explains. "As data creation and exchange grows, it accelerates exponentially due to this attractive force, creating a cycle of more data creation."
Where it becomes rooted, Data Gravity can cause challenges that can impede the efficiency of data exchange, security, customer experiences, and innovation on a global scale, he adds. For these reasons IT investments should account for Data Gravity data's exponential growth.
"We see how Data Gravity results in data increasingly being distributed across multiple locations, including on-premises sites and in the cloud," reports McLean. "The distributed footprint makes it difficult to manage, govern and secure data effectively, raising likely compliance challenges and business risks."
Closer is better
PlatformDIGITAL is Digital Realty's global datacenter solution that enables clients to host their critical infrastructure and interconnect to digital ecosystems across the world. It's designed to provide customers with a Pervasive Datacenter Architecture (PDx) solution methodology for managing Data Gravity and a host of other challenges.
Perhaps one of the most effective ways enterprises can address Data Gravity and performance issues through PlatformDIGITAL is by 'localizing' their IT infrastructure, so that AI strategies are tied to the location of IT itself.
"Localization helps distribute data and compute resources closer to end users and reduces latency," says McLean. "With data as the fuel that powers AI models and AI-enabled insights, it's critical that the right data be readily available in a steady state for AI processing closest to where it resides."
FS businesses may also need strategies to manage where their data is being collected and stored, and how that data aligns with privacy regulations and compliances globally. This is why data localization is becoming more of a priority for IT leaders in FS, according to McLean.
"Distributed data strategies accelerate AI deployment efficiencies," he says. "Not only do IT locations need to have the right hardware to support emerging technology initiatives like AI, IT leaders need to create plans that correspond with regional regulations."
Scaling and distributing HPC risk calculation
A European financial services customer that Digital Realty and partner Lenovo recently developed solutions for provides an example of how AI-ready infrastructure helped it analyze vast amounts of data to gain insights into its own customers' behavior, while boosting operational efficiency and managing financial risk.
"The associated IT workloads were centralized around a calculation grid housed in a datacenter in Europe," explains Maclean. "As application performance and latency issues worsened, the customer saw the pressing need to scale and distribute its HPC risk calculation capabilities."
Building its own data facility to meet the necessary requirements would have taken the company years. So to bring the required HPC power within reach of its European HQ and close to other HPC clusters within the surrounding metropolitan areas, the customer chose to pursue a hybrid IT strategy, managing its own datacenter for existing workloads, while leveraging colocation to facilitate the required expansion to support risk calculation workloads.
"Digital Realty and Lenovo working together enabled the FS provider to scale and distribute HPC capabilities crucial to their risk management strategies," says McLean. "By leveraging PlatformDIGITAL to extend datacenter capacity with colocation, the company saved time, and mitigated the CAPEX that would have been necessary to build its own facility. Plus, it deployed around six times faster due to Digital Realty's ability to retrofit existing datacenters."
Sponsored by Digital Realty.