Will agentic AI end offshore securities services hubs?

Agentic artificial intelligence is forcing the securities services industry, particularly global custodians, to rethink the role of large offshore operations hubs.

For more than two decades, securities services providers have relied on lower-cost locations such as India, the Philippines, Poland and Portugal to provide support around the clock, absorb rising transaction volumes, and manage activities ranging from custody and fund administration to middle- and back-office processing.

Agentic AI challenges that model because it can plan, make decisions and complete multi-step tasks with less human intervention. That significantly weakens the case for a traditional follow-the-sun model as the default choice, even though many securities services activities still depend on judgement, market knowledge, and client handling.

The central question is no longer whether agentic AI works, but how firms can redesign operating models to replace fixed labour costs with flexible technology, without compromising regulatory expectations or financial system stability.

Major custodians are already applying AI, automation, and related technologies in live operations. Leading global financial services group BNY offers a clear example, with reported use of AI, automation, and distributed ledger-based platforms to reduce reconciliation breaks and improve post-trade efficiency.

State Street Corporation’s investment in proprietary data platforms and AI-enhanced infrastructure shows how custodians are using data at scale to generate client insights and support more efficient investment operations.

BNP Paribas Securities Services’ work on corporate actions and settlement processing also illustrates how AI-enabled systems can help manage fragmented market rules and exception-heavy workflows.

These examples are not isolated pilots. They are early signs of a broader shift in how securities services operations will be organised and resourced. The pace of change will depend on regulation, legacy system complexity, client confidence, and whether providers can place strong governance around increasingly automated workflows.

This shift goes beyond securities services. JPMorgan Chase & Co has rolled out a proprietary generative AI platform called LLM Suite across the investment bank, and is expanding AI-agent capabilities to support employee workflows and automate complex multi-step tasks.

And Singapore’s DBS Bank says specialised AI agents now support more than 70 tasks in corporate credit assessment.

In short, agentic AI can convert large volumes of rules-based office work into automated computing tasks. Over time, it may also support more complex judgement-based work.

Cost shifts

This raises an obvious question: do securities services providers still need large offshore workforces?

The answer is not straightforward. AI may reduce routine processing headcount, but the savings will not automatically translate into higher profits or lower client fees. Costs will shift towards new areas such as cloud computing and AI usage, cybersecurity and resilience testing, specialist technology talent, governance oversight, and integration with legacy platforms and market infrastructure.

In some areas, the existing offshore model may still be cheaper and more predictable. Early agentic workflows in reconciliation have shown reduction of manual exception handling by up to 80%. However, these workflows require far more application programming interface calls and computing power than simple chatbots, making computing capacity a major recurring cost.

The labour bill will also change shape. Entry-level processing roles are likely to shrink, while demand will grow for people who can build, manage and supervise AI workflows and translate complex market rules into machine-readable instructions.

This creates a J-curve effect. Costs will initially spike as legacy and AI systems run concurrently, only falling as the technology scales. However, this decline is not guaranteed if governance, integration, and vendor fees remain unchecked.

Location strategy

Securities services providers will also need to rethink where work is done. Labour costs still matter, but location choices will increasingly depend on data rules, computing infrastructure, energy reliability, and geopolitical risk.

Data localisation laws and sovereign AI requirements make it harder to move sensitive work into one single low-cost hub. Providers may instead need a network of smaller, compliant hubs in specific markets. These rules could also slow adoption by making a common AI operating model harder to design.

Asia is already seeing this shift. India hosts many of the main offshore hubs of securities services providers. Many Indian global capability centres are repositioning themselves beyond basic shared services to become AI-focused innovation hubs, supported by evolving domestic data protection frameworks.

The Monetary Authority of Singapore is actively refining its model framework for AI governance for financial services, giving firms a structured environment to experiment with autonomous agents while maintaining systemic safety.

AI also changes which locations are attractive. Agentic AI depends on reliable power, strong digital infrastructure, and large data centres, favouring places with deep technology infrastructure rather than only cheaper labour.

Geopolitical risk will further shape location choices. Providers may spread AI operations across nearshore or geopolitically aligned jurisdictions with similar regulations and reliable energy supplies, creating networks of specialised AI hubs rather than several large processing factories.

Risk management

Risk management will need to change as well. The biggest operational risk is a chain reaction of errors: if an AI agent misunderstands a local market rule, it could process incorrect settlements at speed before a person can intervene.

So providers will need clear human approval points for high value, unusual, or irreversible transactions. Full automation is not suitable where mistakes cannot be tolerated. In many cases, AI will support human teams rather than replace them.

Three risks stand out.

First, vendor lock-in could become expensive and operationally risky if a provider builds too much of its AI system around one vendor’s tools, data structures and application programming interfaces.

Second, providers should avoid becoming overly dependent on closed systems or a single technology stack. Where appropriate, open-source models may help preserve flexibility.

Lastly, if entry-level manual roles disappear too quickly, firms may lose an important training ground for deep market knowledge.

Market impact

The market impact extends beyond the margins and profitability of providers.

Cross-border settlement due to time zone mismatches between New York, London, and Hong Kong or Singapore have historically created manual reconciliation delays. Agentic AI can reduce these gaps by checking trade details and funding needs in real time. This could support faster settlement cycles, such as T+1, and eventually T+0, while improving liquidity and lowering counterparty risk.

Regulators, however, are likely to remain cautious about autonomous systems in critical market infrastructure until the industry as a whole can demonstrate strong explainability, audit trails, and accountability.

Real-time data processing can also give institutional investors a clearer view of portfolios and operations. By reducing dependence on manual capacity, securities services providers can handle sudden spikes in volume without adding the same level of operational risk.

Success will depend on governance that satisfies regulators while enabling useful automation and technology design without excessive dependence on single vendors.

Firms also have to redesign talent plans that protect institutional knowledge, and make clear decisions about where AI is genuinely better than people, and where the expertise of human oversight remains safer, cheaper, or more accountable.

Implementation risks remain significant. Running old and new systems together may cost more and last longer than expected, especially if legacy platforms are difficult to connect. Divergent national rules could also force providers to build separate AI systems for different jurisdictions.

The ultimate challenge is cultural. Firms must embrace AI governance as a core operating discipline, not as a compliance requirement added at the end.

Offshore operations hubs are unlikely to disappear, but their role will change at different speeds across markets and functions. Agentic AI will reduce some routine work and make securities services operations more technology driven, while human judgement, regulatory caution, cost uncertainty, and local market knowledge will keep subject-matter expertise relevant.

The likely result is a more balanced model: smaller, smarter, and more specialised hubs working alongside AI systems. Firms that successfully manage this balance will define the next era of global market infrastructure.

*This article was published by Asia Asset Management on September 29, 2026 titled “Agentic AI unlikely to end offshore securities services hubs”.