It is no longer simply about how many transactions a core banking platform can process. Banks increasingly need to support transactions, real-time enquiries, payments, APIs, modular services and AI-powered interactions — often simultaneously.
The 2026 Temenos Highwater Benchmark on Microsoft Azure explored what this means at scale.
The benchmark tested Lending, Deposits Modularity, Temenos Payments Hub, Holdings and Party microservices and, for the first time within the benchmark transaction mix, Copilot for Core.
Testing was conducted against large-scale datasets, including 50 million customers and up to 125 million accounts. Overall throughput reached 16,385 transactions per second, including 12,241 TPS across Holdings enquiries, 2,109 TPS for Deposits, 1,539 TPS for Lending and 252 TPS for Copilot for Core. Temenos Payments Hub also processed 244 payment orders per second.
But the real story is not simply the headline number.
Three takeaways stand out.
1. Scale Is Now a Multi-Workload Challenge
The 2026 benchmark used a workload mix of 76.2% enquiries and 23.8% transactions, reflecting the increasingly diverse demands placed on modern banking platforms.
Customers check balances and transaction histories, digital channels continuously access APIs, employees retrieve information and payment services operate in real time — all while core transactions continue to be processed.
In the benchmark, balance enquiries reached 4,845 TPS with an average response time of 5 milliseconds, while transaction-list enquiries achieved 2,988 TPS at 6 milliseconds.
For banks, this changes the scalability question.
It is no longer simply:
“How fast is the core?”
It is increasingly:
“Can the entire banking architecture continue to perform predictably as the workload mix evolves?”
This means looking beyond transaction volumes to consider throughput, latency and resource utilisation across the wider banking platform.
2. Modularity Is Proving Its Value at Scale
A second important takeaway is the role of modular architecture.
The benchmark combined multiple services and deployment models across Microsoft Azure, including Azure Kubernetes Service, Azure Container Apps, Azure SQL Database and Azure Database for PostgreSQL. It also incorporated Arm-based Azure infrastructure, including Microsoft Cobalt technology, as part of the platform environment.
Deposits Modularity achieved 2,109 TPS, with the benchmark demonstrating near-linear scalability across vertical and horizontal scaling scenarios while meeting defined response-time requirements.
There were also notable year-on-year improvements within the microservices architecture.
Moving the Customer Enquiry API from Transact to the Party microservice resulted in an 82% improvement in response time, while database resource utilisation for Holdings was reduced by 57% during the mixed workload run.
For financial institutions, the significance goes beyond performance.
Modularity allows individual banking capabilities to evolve and scale according to their own workload requirements. This can support a more progressive approach to modernisation, enabling banks to evolve specific capabilities while remaining part of a coherent banking platform.
3. AI Is Becoming Part of the Banking Workload
Perhaps the most significant development in this year’s benchmark was the inclusion of Copilot for Core within the transaction mix for the first time.
Copilot for Core is an AI-powered assistant embedded in Temenos Core, designed to help bank users retrieve information through natural-language interactions and accelerate everyday operations.
During the benchmark, Copilot-related workloads achieved 252 TPS alongside traditional banking services.
This signals an important shift.
Much of the industry’s AI conversation has focused on use cases and productivity. But as AI becomes embedded into everyday banking operations, it also becomes an infrastructure and performance consideration.
AI services interact with data, consume compute and create demand alongside core banking, payments and other services.
For banks, this means AI increasingly needs to be considered not simply as an additional feature, but as part of the operational workload of the banking platform.
Building for What’s Next
The 2026 Highwater Benchmark points towards banking architectures that are becoming more modular, more distributed and more intelligent.
The broader takeaway for technology leaders is therefore more important than any individual TPS figure:
Design for the workload mix you expect tomorrow, not simply the transactions you process today.
That means preparing for transactions, enquiries, payments, modular services and AI-powered interactions to coexist and scale together.
Microsoft Azure provides the cloud foundation across infrastructure, containers, databases and testing, while Temenos continues to optimise its banking capabilities for increasingly diverse workloads.
As banking evolves, cloud-native architecture, modularity and AI will increasingly become interconnected parts of the same transformation journey.
Together, Temenos and Microsoft are helping build and test the foundations for the next generation of banking at scale.
Explore the 2026 Highwater Benchmark
Discover the architecture, workload scenarios and detailed performance results behind the 2026 Temenos Highwater Benchmark.
