AI Surveillance Platform Scales Data Storage at a Fraction of the Cost

UBX runs the world’s largest boxing franchise, with over a hundred locations across multiple countries. To fuel and manage that growth, the company built Performance Hub, an AI-powered surveillance platform designed to bring intelligence to the gym floor. Using computer vision and machine learning, it monitors activity and delivers real-time visibility that helps owners run smarter, safer operations.

By transforming ordinary CCTV footage into structured data, Performance Hub gives every camera a second job: analysis. It recognizes members, flags irregular activity, and surfaces patterns that help owners see what’s working (and what isn’t) across every location. For gym operators, that means fewer blind spots and faster decision-making as they grow their businesses.

That insight comes at a price: a massive volume of data that needs to be stored, managed, and secured. Each recognition event and frame processed expands the storage footprint, driving up costs and straining performance. What began as a breakthrough in AI operations was now testing the bounds of their cloud environment.

When growth outpaces the infrastructure

Performance Hub’s AI infrastructure was designed to keep gym operations running effortlessly without constant human oversight. Each site uses compact edge devices that process facial recognition and tag video data locally before syncing it to the cloud. Members can walk into any UBX gym and be recognized instantly, updating attendance records across the network in seconds.

But as more sites came online, data demands quickly exceeded what the existing infrastructure could handle. Each location was continuously feeding short clips and AI-generated metadata to the cloud, pushing bandwidth and storage to their limits. What ran smoothly for gym owners on the front end was, behind the scenes, quietly consuming resources at unsustainable rates.

“We’re managing about 7,000 IoT devices between TVs, cameras, and other equipment,” said David Nedved, CTO at UBX. “Footage, especially computer vision footage, is massive; it takes up lots of storage. So how do you get that cost down?”

The cost curve was steep. Every new gym location and camera added another layer of data and another jump in cloud spend. Even with careful monitoring, cloud fees climbed faster than expected. What started as a scalable model was now straining under its own success.

The price tag wasn’t the only challenge. As data demands grew, users began seeing lag when retrieving archived footage. Billing was also becoming more complex and resource-intensive, as the team tracked thousands of small file operations to calculate usage accurately.

To address these issues head-on, the team identified three non-negotiable priorities: predictable pricing, stable performance, and seamless integration. They didn’t want to redesign their platform; they wanted storage that could keep up with an AI engine that never stops learning.

A smarter way to store

At the heart of Performance Hub is an edge-to-cloud AI system built for speed, security, and reliability. Each gym location runs on compact Asus mini PCs powered by ARM processors and NVIDIA GPUs, handling facial recognition and metadata tagging locally before syncing with the cloud. This approach keeps latency low and privacy high, sending only the essential, AI-enriched metadata and time-lapsed video clips to centralized storage.

To manage that ever-growing data stream, the team replaced its tiered cloud model with Wasabi hot cloud storage, built for consistent performance and transparent pricing. They also adopted Wasabi Reserved Capacity Storage (RCS) to lock in predictable pricing by pre-purchasing storage in set time increments, giving them long-term cost control from day one.

Integration was simple and straightforward. Performance Hub’s SaaS platform relies on Wasabi’s S3-compatible API to create and manage storage buckets automatically by region. Each new gym location can be configured in under a minute, with data stored in-region to meet local compliance and residency requirements.

“Wasabi eliminated our direct-from-device upload timeouts and small file issues, which are critical for managing AI metadata and media at scale,” Nedved explained. “Wasabi [also] helps us meet compliance for local data storage, whilst still maintaining complete cloud-native programmatic API access into our platform,” he added.

With this foundation in place, Performance Hub eliminated the bottlenecks that once slowed its AI engine and gained the flexibility to expand capacity without re-engineering its software or business model.

Proof in performance

Partnering with Wasabi turned Performance Hub’s biggest constraint into a driver of growth. Storage costs dropped by 4-6 times, removing the budget pressures that once limited performance and expansion. With egress and API fees eliminated, budgeting for storage is now predictable and future planning requires less guesswork.

“The performance from Wasabi when looking at our terabytes of data is significantly faster than our previous provider,” said Nedved. “As a result, billing is substantially cheaper than before, without any changes to the software.”

Performance Hub billing calculations now run about twenty times more efficiently, and new gym locations spin up in under a minute through Wasabi’s API, with no manual setup required. These results stack up to tangible improvements for gym owners: archived clips load faster, playback is smoother, and compliance checks happen automatically in the background.

“Downloading and viewing content from Wasabi is a faster and better experience for our users,” Nedved added. “That was an unexpected win for us.”

Looking ahead

What started as a move away from rising cloud bills has turned into the backbone of how Performance Hub manages its data. With predictable pricing and faster access, the team can focus on building the product instead of managing the storage behind it.

“The biggest thing for us is cost,” Nedved said. “As Wasabi continues to produce other products at better price points, that’s always an interest at scale.”

Affordable, high-performance storage gives the team room to think bigger toward new brands and use cases for their AI. The next evolution of Performance Hub isn’t about storage; it’s about everything storage can make possible.