Fanatics Taps OpenAI to Power Visual Search for 30K MLB Products

Fanatics, the licensed sports merchandise leader, has rolled out OpenAI-powered visual search across its Shop the Look experience. The feature lets fans upload a photo or select an item to instantly surface matching jerseys, caps, and accessories from a catalog of roughly 30,000 Major League Baseball SKUs.

Why it matters for GPU / AI infrastructure

Running multimodal models at this scale — low-latency inference over tens of thousands of product images — demands sustained GPU capacity and optimized serving stacks. Fanatics' 10% engagement gain validates the ROI of dedicated AI compute for retail visual search, a workload pattern we see growing across e-commerce platforms in the Middle East.

The deployment also highlights the shift from generic foundation models to fine-tuned, domain-specific pipelines. Fanatics likely combines OpenAI's vision encoder with proprietary product embeddings, requiring continuous retraining loops that benefit from elastic GPU clusters rather than fixed on-premise capacity.

For infrastructure teams, the takeaway is clear: visual commerce workloads are becoming a first-class citizen in AI capacity planning. Expect more retailers to follow this pattern, driving demand for high-throughput, low-latency inference endpoints backed by modern GPU clouds.