Major Banks Detail Cloud Modernization Patterns

The LSEG's migration of critical infrastructure leverages AWS Outposts to deliver resilient and scalable low-latency services. This hybrid approach allows LSEG to maintain high performance for trading and risk management applications while utilizing cloud for enhanced operational resilience. The firm's Risk Intelligence division is also incorporating Amazon Bedrock to provide faster and more accurate risk analysis. DBS Bank's in-house built Quant Pricing Engine (QPE) on AWS showcases a different modernization pattern, achieving a 100x improvement in pricing query response time. By using Amazon ElastiCache for Redis and leveraging GPU instances for computationally intensive jobs, DBS can process hundreds of millions of pricing calculations daily, reducing some tasks from minutes to under half a second. This architecture supports the bank's interest rate, equities, FX, and XVA trading businesses. For ultra-low latency, kernel bypass techniques remain a key differentiator in on-premise performance. Technologies like DPDK and RDMA allow applications to communicate directly with network hardware, avoiding the Linux kernel's networking stack to reduce delays, a critical factor for HFT. This approach is often paired with specialized hardware like high-performance NICs to achieve sub-microsecond latencies. Major competitors are pursuing diverse multi-cloud and hybrid strategies. JPMorgan Chase is investing $17 billion in technology, with a plan to migrate 70% of its 6,000+ applications to a mix of public and private clouds to avoid lock-in. Their strategy explicitly keeps the lowest latency trading workloads as close to the data as possible, while customer-facing applications move more aggressively to the public cloud. Goldman Sachs is taking a unique approach by building its own "Financial Cloud for Data" on AWS, effectively turning its internal engineering investments into a revenue stream. This platform provides clients with programmatic access to Goldman's banking, trading, and risk management capabilities as a service, built on modular AWS components and open-source platforms like Legend. The use of FPGAs for hardware acceleration is a growing trend in both cloud and on-premise financial systems. Major cloud providers now offer FPGA instances, allowing firms to accelerate specific workloads like real-time data processing and machine learning without large capital expenditures on specialized hardware. This provides a pathway to offload complex calculations while keeping core trading logic in low-latency environments.