NetApp’s Novus tackles metadata bottlenecks in AI data infrastructure

How Novus Improves Data Readiness for AI Workloads

NetApp Inc. unveiled its Novus platform on October 5, 2026, to address metadata bottlenecks in AI data infrastructure. The solution targets enterprises scaling GPU clusters while seeking to make internal data usable for AI models and agents. By integrating intelligent data infrastructure with production AI workflows, Novus aims to streamline data access and reduce latency in model training and inference processes.

The platform focuses on optimizing metadata management, a critical pain point as AI workloads grow in complexity and scale. NetApp argues that inefficient metadata handling slows down data discovery and preparation, limiting the effectiveness of large GPU clusters. Novus uses AI-driven automation to catalog, tag, and organize enterprise data, making it readily available for training and fine-tuning models without manual intervention.

What Challenges Remain in Scaling AI Data Infrastructure?

Novus introduces a metadata intelligence layer that continuously analyzes data usage patterns across storage systems. This enables predictive prefetching of relevant datasets based on model training schedules and agent workflows. Early adopters report up to 40% reduction in data staging time, according to NetApp’s internal benchmarks. The platform also enforces governance policies automatically, ensuring compliance while maintaining agility in AI development cycles.

Despite advances in metadata optimization, challenges persist in aligning storage performance with the unpredictable demands of generative AI models. NetApp acknowledges that heterogeneous data formats and siloed repositories still complicate unified access. The company plans to expand Novus’ compatibility with multi-cloud environments and open-source AI frameworks in upcoming releases. Success will depend on how well the platform adapts to evolving model architectures and data diversity.

How does Novus differ from traditional data catalog tools? Unlike static catalogs, Novus uses AI to dynamically update metadata based on real-time usage and model needs, reducing manual oversight.

Frequently Asked Questions

Can Novus work with non-NetApp storage systems? Yes, the platform is designed to integrate with third-party storage through standard APIs, though full functionality is optimized on NetApp infrastructure.

What types of AI models benefit most from Novus? Models requiring frequent retraining or large-scale data ingestion, such as large language models and vision transformers, see the greatest gains in data readiness.