Empowering Intelligence: democratizing data access and driving smarter decisions at Nasdaq
Summary
NASDAQ is adapting its infrastructure and software offerings to meet the demands of high-performance compute and agentic capabilities. The presentation outlines two pillars: the physical infrastructure (from microseconds to megawatts) and the financial technology software layer. Key challenges include maintaining resilience and trust in a highly regulated, low-latency environment. The solution involves expanding data center footprints, implementing advanced cooling (liquid cooling), and building an 'intelligence layer' to provide a common, trusted interface for developing complex, agentic workflows.
Key takeaways
-
The Convergence of Compute and Agents
7:21
The industry is experiencing a convergence of high-performance compute and agentic capabilities, which is fundamentally changing trading strategies and requiring massive increases in compute power, power, and space.
-
Physical Constraints and Latency
9:02
Trading is a physical, API-driven world where latency matters. The source of truth for market data can vary based on physical location, necessitating robust systems to manage multiple, potentially conflicting, data instances.
-
Infrastructure Scaling for AI
12:42
To support advanced deep research and modeling (which can increase strategies from 50 to 300-500 annually), NASDAQ is expanding its data center footprint and introducing liquid cooling to handle high-density compute requirements.
-
The NASDAQ Intelligence Layer
17:08
The software side is addressed by the 'NASDAQ intelligence layer,' an infrastructure and technical stack overlaying existing products (like Verafin, Axiom, and capital markets tech). This layer provides a common interface for building agentic workflows.
Technical details
-
System Resilience and Trust
120s
In critical financial infrastructure, failure is not an option. Trust must be earned through years of production-grade success and zero downtime. This requirement dictates the need for highly robust and rigorous system design.
-
Low Latency and Physical Location
542s
The physical nature of trading means latency is critical. The source of truth for prices can be geographically dispersed, requiring systems to account for different, time-sensitive data points across multiple venues.
-
Agentic Workflow Development
1338s
Developing agentic workflows requires more than just functionality; it demands a rock-solid semantic context layer and comprehensive documentation to satisfy regulatory requirements (e.g., explaining why an agent made a specific decision).
-
Infrastructure Components
762s
Key infrastructure components include expanding data center footprints, implementing liquid cooling for high-density compute, and building an overlay layer (the intelligence layer) to standardize data access and interaction.
Mentioned resources
- Weights & Biases
- Coreweave
Channel & topics
Watch on YouTube · Back to latest
This independent, AI-assisted summary is provided for commentary and informational purposes. It may contain errors or omit important context. Please watch the original video for the creator's complete presentation. Video, thumbnail, and related copyrights belong to their respective owners.