Authored By
Ted O’Connor
Ted is a Senior Vice President focused on Business Development at Arcesium. In this role, Ted works with leading financial institutions in the capital markets to optimize data, technology, and operational needs.
Share This post
Investment banks and broker-dealers are navigating digital transformation, AI adoption, and regulatory shifts while managing legacy systems. Data mesh architecture offers a scalable solution to integrate disparate systems, ensuring compliance and unlocking ROI. This article explores how sell-side institutions can optimize data infrastructure to drive innovation, efficiency, and long-term success.
If you are a chief information, chief technology officer, or chief data officer at an investment, custody, or wholesale bank, prime broker, or a broker-dealer, you have likely been in the digital transformation business for many years now, to some degree. Sleepless nights may be a reality, given the extraordinary complexity and roadblocks that come with modernizing technology infrastructure. Sell-side institutions are wrestling with stubborn legacy systems, cultural resistance, taxing vendor selection processes, and a torrent of high stakes decisions. It’s 2025 and by now we all know financial services must unlock time and money-saving efficiencies through digital automation to stay in stride with the Joneses. Just when banks were warming up to SaaS-based point solutions to digitize various middle- and back-office functions, a pandemic accelerated everything. Then generative AI arrived. At the epicenter of all of this is the proverbial new oil, data, which is, in reality, the new oxygen.
In this article, we will offer some practical guidance for banks and other sell-side institutions in building a sturdy data foundation that will facilitate all digital transformation initiatives. Data mesh architecture is the right paradigm for institutions that want to give ownership to departmental teams and need to transform colossal volumes of raw data into an actionable asset that can drive higher-level decision making and ultimately ROI.
RELATED READING: 5 Ways to Assess the Sophistication of Your Technology
While institutions grapple with the mammoth task of digitization and AI adoption, they must continue in an economic environment rife with potholes and uncertainties, exacerbated by geopolitical volatility and potential trade wars. Banks are having an evolutionary moment. Private credit has taken business away from lending opportunities while blockchain and neo-banks are forcing adaptation. Inflation remains a going concern, and they do business in a higher-for-longer rates environment.
However, capital markets transactions have reached all-time highs. Volumes at stock, futures, and derivatives exchanges all posted record transactions in the fourth quarter of 2024. Cash on balance sheets are at all times highs. As the yield curve un-inverts, mergers and acquisitions activity is on the rise, with deal values up in 2024 by 13% in the US. A new administration portends a relaxing of regulatory scrutiny. Investment banking revenue soared in Q4 2024 compared to a year ago, with the largest banks all reporting increases of 25% or more.
We are observing banks changing the way they do business. As is the case with numerous sectors, banks are trying to figure out how to do business in a world that is exploding with data. CIOs know that they need to upgrade their operating models and optimize the data flowing through their pipelines. They want to turn the potential energy of oceans of unstructured data into the kinetic energy of standardized, structured, actionable data. But everybody also knows there is a right way to implement technologies and a wrong way, which leads to demoralizing speedbumps, false starts, or creaky change management. Only 48% of digital initiatives enterprise-wide meet or exceed their business outcome targets, according to Gartner’s annual global survey.
The rise of gen AI and, for that matter, cloud digitization, is far from empty hype. The race is on to adopt gen AI tools in the most effective manner. Deloitte wrote that, “A once-in-a-generation opportunity to differentiate may emerge in 2025.” However, even if your institution cannot leverage technology to differentiate, it better not risk falling behind and becoming irrelevant. Success in a rigorously competitive environment now hinges on the efficacy of digital transformation initiatives.
DOWNLOAD OUR IMPLEMENTATION GUIDE: Building a Modern Data Platform
The traditional banking system has been built on a lot of disparate systems that were either amalgamated through acquisitions, individual business units, or built internally. The notion of data standardization was... well, only a notion. We have witnessed mainframe systems from the 1980s still functioning in institutions to this day. NTT’s 2024 global study revealed that 63% of banks still operate on mainframe systems, but 91% report their AI and cloud initiatives had been endorsed by their boards. The blitz of large language models (LLM) and other gen AI tools has stoked a fire in banking leaders to push modernization programs.
For those venerable banks that are mired in 20- or 30-year-old systems, it is understandably easy to ignore or postpone such an intimidating challenge and just stay the course. Non-standardized data, multiple disconnected server-based or on-premises systems, and millions of lines of archaic code like COBOL are just some of the factors that make a migration to today’s cloud infrastructure an arduous undertaking. But nobody wants to be branded as having technical debt, one of the least desirable terms in business today. Successful transformation requires a CEO who embraces change and empowers a CTO or similar leader to execute a transformation mandate. See our earlier article for advice on modernizing your current operating model and infrastructure.
Upgrading operational technologies is a significant investment, often seen as a 5- to 10-year project. Banks may spend billions of dollars on technology. Any CIO or CTO that has developed and implemented digital transformation initiatives is well versed in the deluge of decisions that they must make and the subsequent pressure to deliver ROI. What's the new cloud platform’s shelf life? What is the new best-in-class solution? How do I sunset one and ramp up another? For example, a bank may have well over 100 systems that support their capital markets activities alone. If its leaders elect to go back and retrofit at the individual operating technology level, they will face a project costing tens of millions of dollars which takes years. And they still wouldn’t have solved their data problem.
CIO Dive correctly wrote that, “Digital transformation in banking boils down to a data problem.” A global investment bank may have more than 10,000 databases. One database for trade execution; a CRM database for client relationships, databases for compliance, reconciliations, tracking hedge fund clients, margin loans, securities lending transactions, and on and on. Some banks have absorbed regulatory fines due to inaccurate data, indicating a failure to manage and govern data effectively. July 2024, U.S. banking regulators tacked on a $135.6 million fine on top of Citigroup’s earlier $400 million fine for failing to address longstanding deficiencies in risk management and data governance. See our earlier article “Escape the Shackles of Your Legacy Data Platform” to find out the signs your system needs an upgrade and how to proceed.
Deloitte found in 2024 that 40% of US business and technology leaders are investing in the foundations for a robust data estate — data architecture, data management, and data insights. The same study said a ‘strong data foundation’ was the #1 critical success factor overall. Implementing a data mesh as a transition layer is a crucial strategy to manage interoperability, governance, and gradual migration by acting as a bridge between the centralized data of legacy systems and the decentralized, data-sharing principles of collaborative businesses.
This approach involves extracting and normalizing data from existing systems and feeding it to critical upstream use cases, risk systems, funding systems, central treasury, compliance, market oversight, regulatory reporting, and, yes, AI. The data mesh provides data standardization and connectivity to the next system that it feeds into. Therefore, it de-risks the replacement of underlying systems, allowing for an easier, faster plug-and-play approach to upgrades. This can also buy time by extending the life of existing technologies (if good data can be extracted from them).
For example, take a global bank with a trillion-dollar balance sheet operating in 80 different markets that has backburnered digital transformation — not an uncommon scenario. If the institution now wants to catch up and digitize and has been built up by acquisition, then it would have hundreds of support systems performing the same operations across different jurisdictions around the world. Integrating new tech with these fragmented systems can cause painful interoperability, compliance, and data harmonization headaches, among other problems. A data mesh transition layer offers the best ROI and minimizes risk when transforming systems.
A major bank may allocate millions a year to maintain its enterprise data lake solution that a big software company sold them. But such a solution was likely not purpose built for an investment bank’s business units. Banks need solutions that are purpose-built, tailored for their specific business units rather than generic enterprise-wide systems. A transitional data mesh architecture will smooth and simplify this complex transition, enabling more seamless adoption of new cloud-based tools for compliance reporting, treasury, accounting, etc. to work alongside centralized data lakes. A data mesh can serve as both a temporary and a permanent solution enabling banks to integrate data from various sources, acting as a middle layer between source systems and downstream applications.
Data mesh architecture is the key to driving a data-driven business culture, driving data ownership from front- and middle- to the back-office, achieving a balance in data sourcing, management, and analytics across all functions. A thoughtful data mesh implementation maintains the data integrity within that technology and ensures the tools can function at scale. In this manner, a bank’s data infrastructure will be positioned in a state of optimal readiness to transform, no matter their unique automation needs, growth trajectory, or their preferred pace of change. The data mesh allows banks to harness and optimize the increased amount of data flowing through their pipelines to support sophisticated asset classes, portfolios, and strategies.
The ROI of technology adoption must be high, it must allow the institution to adapt to change, and it must preserve its financial stability. Today’s shifting regulatory and macroeconomic environment demands that sell-side institutions settle tech debts so they can scale operations and access data in real-time to mitigate risk and fulfill compliance requirements and fiduciary duties. With a data mesh in place, banks can replace underlying systems without disrupting upstream systems. Better and faster data is the north star for financial institutions seeking to stay competitive in digitally transformed markets, and data mesh is the right way forward.
1. Data is the foundation – Digital transformation success hinges on a strong data infrastructure that integrates legacy systems and new technologies.
2. Technical debt is costly – Banks must address outdated systems to stay competitive, as non-standardized data can hinder AI and cloud adoption.
3. Data mesh as a transition layer – Implementing a data mesh helps manage interoperability, governance, and gradual system migration.
4. AI & cloud adoption is inevitable – Generative AI and cloud technologies are critical to efficiency, compliance, and innovation.
5. ROI requires scalability – Digital transformation should enhance real-time data access, risk mitigation, and operational agility for long-term growth.
