Peer insights from Ecosystm x Digital Realty executive roundtable

Across Asia Pacific, financial services firms are accelerating artificial intelligence (AI) adoption—across fraud, risk, customer service, and operations. Building systems that regulators, boards, and customers can trust, is far harder than getting a model to work. Data sovereignty, privacy, anti-money laundering (AML), and know-your-customer (KYC) obligations, layered over fragmented legacy platforms and complex cross-border operations, slow everything down. The promise of AI is still there, but these constraints explain why progress is often uneven, costly, and difficult to scale.

In December, Ecosystm, in partnership with Digital Realty, convened a closed-door roundtable with senior technology, infrastructure, data, and risk leaders from across the financial services sector. The discussion centered on a shared challenge: how to optimize data exchange and infrastructure to support AI initiatives while meeting the requirements of regulators such as the Monetary Authority of Singapore (MAS) and other regional authorities.

What emerged were practical, experience-driven insights from leaders managing AI at scale in heavily regulated settings.

1. Data exchange has become the primary constraint to AI in financial services

AI in financial services depends on fast access to large, high-quality datasets–transactional, customer, risk, and market data–often scattered across legacy systems, private environments, and multiple clouds.

Roundtable participants agreed that data exchange, rather than model capability, is now the biggest bottleneck. Regulatory requirements around data residency, confidentiality, and cross-border data movement – reinforced by MAS guidelines on technology risk management–make large-scale data migration both risky and expensive.

As data volumes grow and use cases become more real-time, institutions are increasingly shifting from a “move the data” mindset to a “bring compute to the data” approach. Reducing unnecessary data movement and improving connectivity between environments helps performance, lowers risk, and keeps organizations aligned with regulatory expectations.

2. Hybrid infrastructure is the regulatory reality for AI

For most financial services organizations, a fully public-cloud AI strategy is neither realistic nor desirable. Cloud platforms are essential for innovation, experimentation, and scalability, but many production AI workloads face limits around cost predictability, latency, and regulatory oversight.

Roundtable leaders consistently pointed to hybrid infrastructure as the preferred approach:

Public cloud for development, testing, and non-sensitive workloads
Private or colocation environments for regulated data, latency-sensitive inference, and steady-state AI operations

This approach supports MAS expectations around resilience, risk management, and data sovereignty, while allowing institutions to continue innovating without putting critical systems or data at unnecessary risk.

3. Governance must be built into AI infrastructure

In financial services, AI governance is tightly linked to infrastructure design. Participants observed that trying to add explainability, audit trails, and controls after models are deployed often causes delays, rework, and regulatory friction.

Leading organizations are now embedding governance into their data exchange and infrastructure layers, focusing on:

Clear data lineage across hybrid environments
Explainable and auditable AI workflows
Consistent access controls aligned with internal risk policies
Operational resilience in line with MAS Technology Risk Management (TRM) guidelines

When governance is integrated from the start, AI teams can move faster and with greater confidence, reducing friction between innovation, risk, and compliance.

4. Standardization enables scalable and defensible AI

Another clear theme was the need for standardization in data definitions, metrics, and governance frameworks. In large financial institutions, inconsistent standards can erode confidence in AI outputs – especially for risk, credit, and fraud use cases.

Without common data models and KPIs, AI systems are harder to validate, explain, and justify to regulators. Standardization supports model and dataset reuse, ensures consistent oversight, and strengthens an organization's ability to demonstrate control and accountability as AI adoption grows.

As institutions move toward more automated, real-time decision-making, these standards are essential for maintaining trust and regulatory alignment.

5. Colocation is emerging as a strategic enabler for regulated AI

AI workloads place new demands on infrastructure – dense compute, advanced cooling, reliable power, and low-latency connectivity. Many traditional enterprise data centers aren’t built to handle these requirements at scale.

Roundtable participants highlighted colocation as a practical approach for AI in regulated financial environments. By colocating AI-ready infrastructure near sensitive data, institutions can:

Cut latency and boost performance for time-critical workloads
Streamline data exchange across hybrid architectures
Retain control over data residency and access
Strengthen resilience and business continuity

Colocation lets financial services organizations modernize step by step, supporting both legacy systems and next-generation AI platforms without adding regulatory risk.

Moving from AI experiments to AI platforms in financial services

The discussion highlighted a broader change in financial services. Success with AI now depends on building platforms that integrate efficient data exchange, resilient infrastructure, and built-in governance – rather than running isolated pilots.

Institutions that invest in these foundations are better able to:

Scale AI with confidence and oversight
Manage cost and complexity across hybrid environments
Comply with MAS and regional regulations
Produce reliable, explainable insights quickly

Continue the conversation with an expert briefing

If your organization is exploring how to optimize data exchange and infrastructure to support AI success in finance, a peer-informed discussion can help clarify your next steps.

Digital Realty’s Expert Briefings connect financial services leaders with specialists experienced in designing secure, compliant, and resilient AI environments across Asia Pacific.

Optimizing Data Exchange and Infrastructure for AI Success in Financial Services