A separate IBM study showed that 43 of the world’s top 50 banks rely on mainframes as their core computing platform.ii But there is continued investment in cloud. From 2023 to 2025, 87% of companies increased investment in the cloud, according to LSEG research.iii None of these technologies is inherently superior to any other. Some workloads are simply better suited to certain architecture patterns. The devil is in the details.
For transformation, a critical implication of this mosaic of technologies is the accumulation of hundreds of different approaches to data. Each neighborhood produces and egresses its data without normalization across systems or reference data to help make connections. The diversity impedes comprehensive cross-firm views and efficiencies.
The problem with fragmentation goes beyond the idea of “tech debt” caused by maintaining aging systems. Costs absolutely must come down to compete with digital-first disruptors and the changing economics of sell-side services. But competition also calls for the ability to support new products and meet buy-side demand for a broader range of asset classes, including digital assets.
A bigger problem is the inconsistency in the workflows that run the firm. Regulatory and management reporting, intraday funding decisions, compliance monitoring, and client reporting all require consistent, timely views of positions, trades, and reference data. Over time, parallel versions of truth pile higher, increasing cost, slowing change, and making reconciliation harder.
Shared utilities and services
A city with a patchwork of separate utility providers and public transportation systems would be hard to live in or run. Yet sell-side data often comes from many systems and providers that do not easily integrate. Most desks insist their data is unique. Some of these differences are real, such as product terms, calendars, conventions, margining, and settlement mechanics. Each desk is profitable and competitive because of its advantages in how it prices risk, structures products, and executes in its markets.
But these unique traits sit on top of a foundation of commonalities. Across asset classes from equities to FX to repos and structuring, the same backbone repeats, such as instrument definition and reference data, trade capture and enrichment, lifecycle state changes, counterparty and account data, valuations, positions, and downstream reporting obligations.
A transformation leader’s job is to separate “true nuance” from “superficial variation,” then build a unified system that handles common needs once, rather than re-implementing them desk by desk.
At the same time, the answer is not to modernize everything and move everyone to the cloud from the top down. These types of initiatives tend to stumble over internal roadblocks that stop change. Business units that generate billions of dollars in profit may not recognize that anything is broken or needs to change.
The better answer is to create a layer that ingests data from every source system, every external data provider, and every counterparty. That approach buys time for transformation and even extends the return on investment from the underlying technologies. It also lowers the need to make unrealistic demands or overcome steep local resistance and helps keep your citizens confident that they can access data for their own analytics and query.
Zoning and building codes
The goal is to define a small set of firmwide primitives that every neighborhood can plug into for the sake of consistency and operational streamlining. These primitives do not need hundreds of different technologies or data models with their own approaches to what defines a security or a trade, how to represent an execution, or which lifecycle states and events call for consistency so that downstream functions can understand them.
