STAC Report
Posted: October 28, 2025
STAC-AI™ LANG6 on HPE ProLiant DL380a and NVIDIA H200 GPUs
STAC research sets the standard for evaluating technology performance in capital markets—designed by and for the industry’s most demanding technologists.
Developed with input from a global community
Our work helps financial institutions and technology vendors assess and compare solutions with clarity and confidence. Every STAC Benchmark™ is grounded in real-world workloads and developed with input from a global community of over 500 member firms.
This community isn’t passive. It's a driving force—shaping benchmark priorities, sharing technical expertise, and collaborating to solve the industry's toughest performance challenges. Together, we ensure that STAC research stays relevant, rigorous, and trusted.
STAC Benchmarks cover the full spectrum of high-performance workloads in capital markets, reflecting how today’s firms structure their systems, teams, and technology strategies. Domains are grouped by their focus - "Fast Workload" (relatively simple computation, with speed of communication being paramount), "Big Workload" (where the size of the dataset that must be moved or analyzed, and AI (compute-intensive training and inference driven by machine learning models, often using unstructured data)
Where speed is everything, and latency is measured in microseconds.
These domains focus on ultra-low-latency infrastructure, streaming data, and real-time responsiveness.
Key areas include:
Stac-m1
Low-latency direct-feed or "ticker plant" solutions
Stac-m2
Messaging middleware under market data distribution workloads
Stac-T1
Tick-to-trade latency measurement
Stac-n1/T0
Network I/O of network stacks in a market data (N1) or tick-to-trade (T0) environment
Stac-n2
Network I/O between data centers and cloud instances
Stac-Ts
Time synchronisation and traceability
Where scale and compute power are the challenges.
These benchmarks test the performance of large-scale analytics, AI workloads, and strategy development across complex infrastructures.
Key areas include:
Stac-a3
Strategy backtesting
Stac-a2
Risk computations
Stac-m3
Enterprise Tick analytics
AI & ML
Where intelligence and data combine to deliver actionable insight. These domains focus on evaluating and understanding the performance, efficiency, scalability of training and inference machine learning and generative AI models and supporting infrastructure
Key areas include:
Stac-ml
Machine learning (ML) workloads in finance
stac-ai
Artificial intelligence (AI) technologies in finance including LLMs and retrieval-augmented AI (RAG) solutions
SiGs
Special Interest Groups (SIGs) are collaborative communities of experts, practitioners, and stakeholders who come together to explore, define, and advance key areas of technology and finance. Each SIG focuses on a specific domain.
Key areas include:
FPGA
Cloud
Data centre
STAC Benchmarks cover the full spectrum of high-performance workloads in capital markets, reflecting how today’s firms structure their systems, teams, and technology strategies.
Smart order routing and execution platforms
FPGA and NIC-level performance
Persistent, high-resilience internal messaging
Ethernet MAC/PHY functions and network multiplexing