SBI Chairman Predicts AI Reshapes Market Infrastructure

State Bank of India chairman Challa Sreenivasulu Setty told an audience at the Clearing Corporation of India Ltd (CCIL) silver jubilee that AI will be a defining force reshaping financial market infrastructure over the next 25 years. He said AI and machine learning can convert market utilities from back-office processors into proactive risk managers capable of anticipating stress, while also automating routine operations to improve accuracy and efficiency.

Setty singled out specific capabilities where AI will be immediately useful, including dynamic margining and real-time risk assessment. He argued that models trained on large datasets of historical transactions, counterparty behaviour, and market conditions can deliver more accurate exposure forecasts and enable continuous surveillance. Operational automation examples he cited include clearing, settlement, and reconciliation workflows, which are ripe for low-latency, rules-plus-ML automation.

• •dynamic margining driven by live market and counterparty signals
• •real-time risk assessment integrating market, liquidity, and counterparty data
• •automation of clearing, settlement, and reconciliation to reduce errors and latency
• •enhanced market surveillance to detect emerging patterns and anomalous flows

This is a high-level endorsement from India's largest bank at a moment when markets are becoming more interconnected and new asset classes, including tokenised securities and cross-border liquidity engines, are gaining traction. Setty framed the shift as moving from scale to "intelligence scale," where the pace and opacity of markets will require models that can detect subtle, systemic signals. He also emphasised that AI adoption will not be purely technical: cyber resilience, operational robustness, and regulatory alignment must be co-designed with AI systems.

Data quality and provenance across clearing, settlement and custody systems; low-latency pipelines for risk scoring; model explainability for audit and regulatory review; model governance and versioning for backtesting and stress testing; privacy and cross-border data transfer constraints; and the operational demands of high-availability, deterministic settlement systems. Architecturally, expect combinations of event-driven streaming, feature stores, online learners to handle concept drift, and simulation-driven validation for tail risk.

Institutional market infrastructure carries systemic risk. Embedding AI into central utilities changes how risk is discovered and managed, and increases the attack surface for cyber and model-risk failures. For ML engineers and quant teams, this means stronger emphasis on reproducible pipelines, adversarial testing, explainable models, and integration with existing regulatory reporting and margin frameworks.

Will CCIL and similar institutions publish pilot frameworks or APIs for real-time risk assessment and dynamic margining? Watch for proofs-of-concept that combine market microstructure simulators with ML-driven scorers, along with regulatory guidance on acceptable model explainability and audit trails. Also monitor coordination between market utilities and regulators on cross-border data sharing and tokenised-asset settlement rules.

Setty's remarks turn a strategic spotlight on production-grade ML in market infrastructure. Adoption will be iterative and safety-first, but the mandate is clear: build intelligence into the plumbing while hardening governance, resilience, and regulatory alignment.