The National Hockey League is charting a path to modernize its video infrastructure with the VAST AI Operating System, moving processing to the edge and redefining how content is created, segmented, and delivered. For the first time, NHL teams will receive video directly from the League, reducing third-party dependencies and establishing a scalable foundation for future AI-driven workflows.
The Problem: Fast Games, Slow Workflows
Traditionally, video capture at NHL arenas has meant recording one large file per camera, per game. This presents multiple challenges:
- No direct delivery: Teams never had received video directly from the League, forcing them to depend on third parties for access.
- No segmentation: Teams and coaching staff had to work with full-game downloads, which are slow and inefficient
- Third-party reliance: Many teams paid outside vendors to get segmented video, adding cost and complexity.
By shifting to League-managed, segmented video distribution, the NHL is ensuring that teams and coaches get the content they need faster and with greater accuracy, while maintaining full control over how video is captured and shared.
The VAST-Powered Workflow
Fortunately, the NHL had already modernized their in-arena capabilities by deploying the VAST AI OS at each and every North American location. They’ve already been enjoying the benefits of a more seamless content acquisition and distribution experience, whereby all game footage is collected at the edge and automatically brought back to their NYC HQ, leveraging the VAST DataSpace to efficiently keep everything in sync and enable faster collaboration.
This new challenge, segmenting game footage at the arena, would be difficult to accomplish on traditional data storage systems. Thankfully, the VAST AI OS provides a way to not only store and distribute content, but also to automate sophisticated processing pipelines, effectively bringing data-aware compute and orchestration directly to the edge.
Here’s a breakdown of how it all works:
1. In-Arena Capture
Each game generates one large video file per camera
Files are streamed directly into the VAST DataStore
2. Pipeline Initiation
A VAST DataEngine function consumes an API Feed from the NHL that notifies applications of major events, including period start/end
This function filters for specific start/end events and publishes to a topic via the VAST Event Broker
A DataEngine trigger detects events in the topic and fires off the pipeline as shown in the diagram below
3. Data Processing and Indexing
The in/out breakpoints for a period are fed to a ‘clipping’ function, which leverages the open source ffmpeg utility to create a segmented file
Each period specific clip lands in a DataStore bucket within the arena, triggering a bucket notification that goes into an Event Broker topic
When the clip is pushed into the bucket, metadata and tags are set on the object with game specific metadata
These tags and metadata are automatically indexed within the VAST Catalog for easy search/query
Segmented clips are organized in the VAST DataSpace global namespace
4. Distribution
The “last mile” is to get these segmented clips into AWS S3 so that the NHL can perform final processing and distribution
The Bucket notification from the previous step acts as a trigger for a function that:
i. Pushes the clip to NHL’s AWS S3 Bucket
ii. Sets appropriate metadata and tags on the clip/object
iii. Updates an in-arena VAST DataBase table with clip information and final AWS S3 location. This serves both as a way to enable search, but also to power dashboards the NHL can use to monitor progress of the pipeline.
Because the arena systems are all linked using the DataSpace, all content is automatically accessible both at the edge, as well as in NHL HQ
Clips in AWS S3 are surfaced through internal applications and/or other distribution platforms for teams as a one-stop access point
This is the first time the NHL has run video processing workflows in-venue, marking a shift from centralized cloud-based workflows to real-time edge orchestration.
Why Edge Processing Matters
By shifting compute into arenas, the NHL aims to reduce both cost and latency:
- No third-party payments: Segmentation is performed in-house
- Faster turnaround: Teams receive usable video within minutes instead of hours
- Efficient distribution: Content flows directly into league-controlled applications
More importantly, the architecture creates a platform for more advanced data-driven workflows in the future.
What’s Next: Intelligent Video Pipelines
The clipping workflow will be just the start. With the VAST AI OS in place, the NHL can extend into more advanced scenarios:
- Ultra-High Resolution: Scaling beyond 4K to 10K cameras without raw data ever leaving the arena
- Automated Subclipping: Using NHL EDGE positional data (NHL Puck and Player Tracking) telemetry to dynamically generate subclips (e.g., zoomed goal shots, specific player angles
- Customized Feeds: Delivering tailored video perspectives depending on the audience (wide tactical views for coaches, broadcast-ready highlights for fans)
These additional workflows will be powered by the same foundation: VAST’s AI OS, which enables edge computing, data federation, and AI-driven indexing, labeling, and curation at scale.
Current Status and Rollout
We’ve successfully validated this end-to-end workflow in NHL’s test environments. Over the course of the next several months, it will be rolled out to all 32 NHL arenas in North America, with a target ‘go live’ date by February 2026.
The NHL sees this as foundational to its long-term content strategy. For VAST, it’s a clear demonstration of how our AI Operating System supports real-world, data-intensive workflows at the edge.
NHL + VAST at All-In Montreal
The NHL and VAST co-presented this innovation at ALL IN 2025 in Montreal last month, highlighting how technical architecture and operational priorities converge to deliver business impact for professional sports. ALL IN coincided nicely with the official release of version 5.4 of the VAST AI OS, and these fundamental capabilities are now available for all VAST customers to enjoy and build upon.
