Artificial intelligence is no longer a distant prospect for banks, brokers and asset managers. It already sits behind fraud detection systems and trading engines, and its reach is spreading into customer service, analytics and back-office processing. Yet the adoption of AI in finance is uneven, and many institutions are discovering that enthusiasm is not the same as readiness. The gap between firms that prepare properly and those that improvise is likely to widen quickly.
A World Finance analysis by Valerie Walsh, senior vice president of portfolio management at data centre operator Digital Realty, sets out five ways the finance industry can prepare for AI. Its starting point is the size of the prize. The piece cites a McKinsey estimate that AI could add around $13trn to global economic activity by 2030, largely by making businesses faster and cheaper to run.
The readiness gap in AI in finance
Despite that potential, research by Digital Realty found that more than a third of IT decision makers at the UK’s largest financial services companies said their organisations were not ready to implement AI. Comparable figures elsewhere in Europe were 27% in Ireland, 23% in the Netherlands and 18% in Germany. These are not small firms struggling for resources; they are leading institutions in some of the world’s most sophisticated financial centres.
The readiness problem matters because the applications are moving well beyond the established uses. Financial institutions are increasingly applying machine learning to customer communication, predictive analytics, trade processing and automated investment tools. Each of these demands clean data, robust systems and staff who trust the technology. Without those foundations, projects stall or produce results that cannot be relied upon.
Steps one and two: target the right tasks and train people
The first recommendation is to identify where AI would genuinely add value. That means reviewing existing workflows and asking which tasks could be automated so that employees can focus on higher-value work. The most useful insight often comes from staff themselves, who know which jobs are repetitive and time-consuming.
Finance has already seen clear gains in a few areas: tailored customer service, more accurate risk models and the handling of routine transactions. But the article stresses that this is not a one-off exercise. Processes and technology both evolve, so firms should keep reassessing where automation makes sense rather than locking in a single plan.
The second step is education. Workers are understandably anxious that automation could cost them their jobs, and resistance can quietly undermine any rollout. Involving staff early in planning, and explaining that AI is intended to take on mundane work such as around-the-clock security monitoring, helps build acceptance.
The goal, according to the piece, is a culture in which people and AI systems work together rather than in competition. This is more than a public relations exercise. Regulators in many jurisdictions expect firms to understand and oversee the models they deploy, which requires staff who can question outputs and spot errors.
Step three: upgrade the infrastructure underneath
AI places heavy new demands on data processing, storage and connectivity. Many financial institutions run on legacy systems that are expensive to upgrade, and training or running advanced models can quickly outstrip the capacity of an older in-house data centre.
Walsh argues that outsourcing to specialist providers, which focus on interconnection and continually redesign their facilities, can give firms purpose-built environments without the burden of constant upgrade costs. The benefits she lists include lower operating costs, easier scaling, stronger security, a centralised platform for integration and better overall functionality. That supports large-scale data analysis, customer support and real-time views of transactions. Trading and stockbroking firms in particular stand to gain from faster client service and smoother end-to-end data flows.
Readers should note that the author’s employer operates data centres, so the emphasis on outsourcing reflects a commercial perspective. Even so, the underlying point is widely accepted: AI is only as effective as the computing power and data pipelines that sit beneath it, and many firms have underinvested in both.
Steps four and five: set a strategy and learn from others
The fourth recommendation is to establish a clear AI strategy from the outset, including a deployment and rollout plan that takes account of key stakeholders such as customers. A well-designed approach can deliver services that are more tailored, more accurate and cheaper, alongside stronger tools for preventing and detecting fraud and money laundering.
The fifth is to study what others have done, both competitors and companies outside finance. That includes paying attention to mistakes. The article points to earlier AI systems that produced biased consumer targeting, a reminder that poorly governed models can damage customers and reputations alike. For financial firms, where fairness in lending and pricing is closely regulated, the stakes of such errors are especially high. Borrowing best practice from others’ experience can help a firm move faster while avoiding the most costly pitfalls.
What this means for the industry
Taken together, the five steps describe a shift from experimentation to disciplined execution. AI in finance is becoming less about isolated pilots and more about rebuilding processes, systems and skills around a new set of capabilities. Firms that treat it as an IT project risk underestimating the organisational change involved.
There is also a competitive dimension. Customers increasingly expect instant, personalised service, and compliance teams face growing volumes of data to monitor. Institutions that get AI in finance right can respond faster to clients, spot suspicious activity earlier and price risk more precisely. Those that fall behind may find themselves carrying higher costs while offering a slower, less relevant service. Because the benefits compound over time, early and well-governed adoption is likely to matter more than simply spending the most money.
Conclusion
The opportunity is significant, but so is the risk of being left behind. Institutions that map their processes carefully, invest in their people and infrastructure, plan deliberately and learn from others’ missteps will be best placed to capture the efficiency gains on offer. Those that wait for the technology to settle may find that competitors have already redrawn the market around them.
